commit 4b4463ea331336285321b4352ad305fd306f06da
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Oct 28 13:41:17 2016 -0400

    1.7 release.

commit 24a516dff281d3262fe57a57e5e3510d655e2e2c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Oct 28 00:32:46 2016 -0400

    Add FALLOC_FL_UNSHARE_RANGE flag to fallocate

commit c3f0b53764599645b033902a8164a9463f57ef04
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 27 14:39:55 2016 -0400

    consolidate the seed initialization into init_seed
    
    Reducing the number of places we check user_set_seed.
    
    Bonus side-effect of not printing about getrandom() when we set a seed.

commit 698453cc56072fa852a03099a8d0867143279fa2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 27 14:31:37 2016 -0400

    kill_pid debugging
    
    like the pid_alive debugging, this may be useful to prevent some
    future stupidity.

commit bc239d4a072a92c50a4a407784d99520ae6c3c19
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 27 14:30:55 2016 -0400

    update find.sh to a more interesting example.
    
    Also use the helper scripts.

commit 584d09a8d1be8c54d88fac3a11eabb95d761cf95
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 27 14:29:05 2016 -0400

    Sanity checks in pid_alive
    
    debug code that I've been carrying forever.
    These should never fire any more, but they might be useful to have
    in case I screw something up in the future.

commit af4611a1d454032729a93113da95e46532504b2d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 27 14:27:45 2016 -0400

    for now disable periodic dirtying of mappings.
    
    Another case where after some syscalls (mremap for eg) we segv.
    Need to investigate further post 1.7

commit ad6f3585e16158256712515c82f84bc5698278b9
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 27 14:27:22 2016 -0400

    pick read/write dirtying based on PROT settings of mapping

commit c5d4fbb48c3e996769894fdf9c6566195b6cdde1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 27 14:26:17 2016 -0400

    disable the mremap page-dirtying for now
    
    This segfaults, and I'm not sure why yet. Disable for 1.7

commit 08ddcfa937e0232a6c93594da9aa564f2d900189
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 27 14:25:04 2016 -0400

    let children reseed (from rand()) when we pass in -s

commit 3ca8b3752cb2f329df384755a3be7596110e0489
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 27 14:19:12 2016 -0400

    if we passed in a seed, don't generate new seeds from urandom/gtod

commit 3aba9133af20b707d44a17ace91a857c22c2ea9d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Oct 25 19:50:12 2016 -0400

    bpf: Fill out .k with a bpf map fd occasionally.

commit 3494a7959d4a590650a3fd2839d89f64cf5ebe3a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 20 14:34:06 2016 -0400

    update the map->prot when marking a page as writable

commit d7ca00d437e03b45bd38b06ecf284ed9e672580a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 20 13:48:08 2016 -0400

    fix off-by-one bugs in single page fault routines.
    
    page offsets are 0-4095, so we don't need the extra subtraction.

commit 0b6e4f0515bc7af2904808c1bd9ceca309ba090c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Oct 18 19:17:03 2016 -0400

    allow '0' as a file mode

commit c78a28d4daa46267c5766875b0b8ab83b5915ae5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Oct 18 19:16:06 2016 -0400

    remove the 'do syscall in a subchild' code for now.
    
    This isn't doing anything useful other than adding overhead right now.
    To be revisited post 1.7

commit 9cb8dbf6fc3fb5e6660d2cc15ec61d21e99f07fd
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Oct 17 15:06:47 2016 -0400

    mark pages as PROT_READ before we fault them with a read

commit 3a0e33d1db3214503316840ecfb90075d60ab3be
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Oct 14 15:53:25 2016 -0400

    silence cppcheck

commit a79863662025676a0d49b361db24371d295abe04
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Oct 14 15:35:50 2016 -0400

    change the random munging a little
    
    - increase the likelyhood that we don't do any limiting
    - tweak some chances that we leave the rand num unmunged

commit 35018d4bbe5179f1aef4e00823f0474538770d0a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Oct 14 15:32:55 2016 -0400

    generate random numbers in the -1..-10 range

commit aa1798cf35dbfa9c4a11c05028ac4e89d2bac6eb
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Oct 14 15:08:24 2016 -0400

    add the 'special' ring-id's for keyctl

commit 5cd7824c547adfd33fa4ce8d82778e4daaed937e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Oct 14 14:14:19 2016 -0400

    fix segfault in setsockopt() when called with --disable-fds=sockets

commit 73da6ab7f0efbfc32b25e13da3772559f69aa567
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Oct 14 14:13:36 2016 -0400

    Fix up segfaults when encountering null entries in syscall tables.
    
    Triggered occasionally with -r

commit 55fd93c44ccc24b9d21ea37c73d72c12e98e6f01
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Oct 14 14:12:43 2016 -0400

    don't use rewind() on /proc/pid/stat
    
    This seems to not do the right thing always.
    The fseek is more explicit.

commit 40c86c5a30b36dd95307f5a6033c101252f28495
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Oct 14 14:09:35 2016 -0400

    multi-page mprotect fixes for fault-write
    
    Multiple problems here:
    - In some cases we would pass down a multi-page mapping after mprotect'ing
    the first page as writable, and then try writing to a different page within
    that mapping, and segfault.
    - In some cases we would mprotect write a whole multi-page mapping, even when
    we were only going to write to a single page.

commit 49767cb18e47ca4dbd127786b0870389a406f227
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Oct 10 16:06:21 2016 -0400

    add pkey alloc/free syscall stubs

commit 178058972b816132789fc1a945a22ff3020581cb
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Oct 10 15:56:11 2016 -0400

    add pkey_mprotect stub

commit 055d51b38afe5c2e63b40987bac5a6164c0ec2e2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Oct 7 15:57:54 2016 -0400

    open the testfiles multiple times with multiple modes

commit 8d530c3bac53de5a06fddd5c99056364f1cae33f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Oct 7 14:44:55 2016 -0400

    fill out the .k with rand32, so we get fd's etc in there.
    
    This needs to be smarter, but this is better than rand() for now

commit 60e27523a341749e8f429c2168a6129b37afea56
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Oct 7 14:01:13 2016 -0400

    add some eBPF instructions/addressing modes

commit ce7fdca7e4f84538cfc42ff59df65a2f4d4571b5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 6 15:40:19 2016 -0400

    scope reduction.

commit 0d44070108f82d59d27db95c8357eb2885fec3be
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 6 15:39:28 2016 -0400

    convert bpf fd generation to be table driven

commit 5aa3e6f96c15415cc863c3f3aa4cb42eec69f93e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 6 08:46:55 2016 -0400

    fix uninitialized var when run with --disable-fds=sockets

commit 6512bb9044bbd82922fc4e74e1f122d68fbca0e8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Oct 5 14:09:56 2016 -0400

    correct the params for BPF_MAP_TYPE_STACK_TRACE creation

commit 93d419beb2f2fb1e71ddb34ba4e47e6d74b94510
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Oct 5 14:08:40 2016 -0400

    add the newer bpf map types

commit c3da5c4224b0646aac2ea5cca2ebec1e169f46d1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Oct 5 13:47:06 2016 -0400

    create a bpf fd of each type.

commit 05948f04430564d5e66aa93244543585c86049e4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Oct 5 11:35:59 2016 -0400

    move the reading of MemFree out to its own function for other uses.

commit 1a44abb7710eb93987f165699f959827b0704245
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Oct 5 11:26:32 2016 -0400

    fix up some segfaults when --disable-fds=sockets is passed.

commit 76aadaba3e48f73adecff456e595582fdc472916
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Sep 23 16:25:49 2016 -0400

    annotate fadvise flags

commit 7887d9e30c8fc9d3c58ae62c5d17ae8c7dc29c21
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Sep 13 11:18:12 2016 -0400

    make enable_random_syscalls static

commit e966131caf8c287f11f367426c6ca43ba9d358bd
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Sep 12 15:48:04 2016 -0400

    reduce the amount of time we wait for the grandchild a little

commit 9e44545de53ef8ca77368c0edf32e245e896a7a8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Sep 12 15:47:15 2016 -0400

    skip the sleep if the grandchild has already exited.

commit a37e38b4ab6a7b3260fa165a1edb2dcc022a6e48
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Sep 12 12:41:44 2016 -0400

    add missing locking around state modification

commit 04ffc99dc5b050744b5cac628a41590e8039e2c8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Sep 12 12:41:08 2016 -0400

    count syscall attempts before, not after the syscall.

commit 3bd0e91a586822e05fe8763ec262a2c56f0434e3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Sep 12 12:40:21 2016 -0400

    set GOING_AWAY state correctly.
    
    The recent change to widen the BEFORE state window meant we scribbled
    over the GOING_AWAY state.

commit 3c4daa609aa4c22f8f7b9ad4707c9847c8b11c20
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Sep 12 12:39:21 2016 -0400

    unify the stats update in handle_syscall_ret()

commit f7974b6849d3819247bddf699c475d37d50ce610
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Sep 12 12:37:37 2016 -0400

    small pause before repeated kills to grandchild

commit 90e237c8be2806ee69dc17def523e26b0789efc0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Sep 12 12:36:27 2016 -0400

    let the grandchild live for a second before we kill it.
    
    When under load, we got into a situation where the fork/exec never
    actually completed before we started sending kills.

commit 5bc525ca3d2b8f117e3d8303e948bfa48e8b4270
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Sep 9 21:20:01 2016 -0400

    change check for inactive children
    
    Jan Stancek reported that he saw processes hanging without getting killed,
    and this is the most probably reason.  We were considering children
    as inactive even if they were doing the syscall, so they were never getting
    sent a SIGKILL.
    
    Change the state comparison, and also widen the window that we spend
    in 'BEFORE' up until right before the syscall.

commit ecea2b06d5f348068cb74ffef072813e13a38856
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Sep 1 17:39:36 2016 -0400

    record/output the successes/failures per syscall

commit d18ba5b45727f67780c0f449b1a7308614f9c83b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Sep 1 17:20:21 2016 -0400

    add option to dump errno distribution on exit

commit 87427506ffa17ead38bd548a781c6842148d4555
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Sep 1 00:37:47 2016 -0400

    fix generation of localhost addresses.
    
    The main thing wrong here was that it was in host order, not network
    order, so we were trying to send to 1.0.0.127. Oops.
    
    Make things more readable by just referring to it in ascii.

commit 37eaa4d616294a7028ebf4aba1fc3bffcfc927bc
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 30 21:19:43 2016 -0400

    Only do the max_children limiting once.
    
    This was in a loop, which sent it to 0. Derp.

commit c073f7380db6e1a9e3d8b0477d91bb8ee6b94b18
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Aug 26 16:49:10 2016 -0400

    remove ->socket
    
    Use the triplets arrays instead.
    Mucho code deletion.

commit 6f4f50259f2ef233ac1b56e3261b27770fa487d1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Aug 26 16:31:27 2016 -0400

    add PF_X25 valid triplet

commit f8187f53c59056b962856858f12913c269f1277a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Aug 26 16:02:04 2016 -0400

    add valid protos for PF_KCM

commit be0547db1a41e029d958a5f31668ca21e0fbab32
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Aug 26 15:55:03 2016 -0400

    dummy entry for (dead) econet

commit bcc2f0234270bf658dce689ec2fcdafc0145a132
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Aug 26 15:53:11 2016 -0400

    add PF_DECnet socket triplets

commit e2fdeec6d640a030bd08a0c47d34bf1a9c95ab96
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Aug 26 15:50:05 2016 -0400

    add PF_CAIF valid triplets
    
    based on guesswork

commit a52f72f41e2e596c497b5e5b09a2dfc72b476d36
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Aug 26 14:39:13 2016 -0400

    use correct loop boundary for privileged sockets

commit 829c4bfe158657a5c30556a8b995b13bf5627021
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Aug 26 14:37:40 2016 -0400

    add support for privileged sockets to the socket generator.
    
    Not super happy with this yet, but this is better than nothing for now.

commit e24b8840802d59548037894bd93a44f21cf9e504
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Aug 26 14:23:55 2016 -0400

    Replace ->generate with lists of valid socket triplets
    
    These can be shared between the 'generate fds' code as well as indirectly
    from ->socket calls (tbd)
    
    Currently lacks support for 'needs root' sockets.

commit 19fdbda05370999da03ec3891698cc522a0ff99e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Aug 26 10:26:55 2016 -0400

    overwrite the correct pointer with the sockaddr

commit 17e27d3ce608c62001efe3d8479abf271510adce
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Aug 26 10:16:46 2016 -0400

    pass the UNSPEC sockaddr back up instead of just leaking it

commit c20bc68e57400602d37b2a9c2b18031710012b94
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Aug 26 00:22:08 2016 -0400

    scope reduction

commit c378e63a11141ccb3cc906fbad996aef8b2f1102
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 25 17:07:21 2016 -0400

    ->generate for PF_NETROM

commit ece557c3181d076ece8bada0b9f8ab45c31dbbef
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 25 17:03:17 2016 -0400

    ->generate for PF_TIPC

commit 228f17481ad574748e391380bb339beb04e10acd
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 25 16:24:43 2016 -0400

    ->generate for PF_PHONET

commit e12514ba3ff869bda7972c5d1dacbf8928c11f8a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 25 16:20:42 2016 -0400

    ->generate for PF_LLC

commit b0673ce61ae555083750beb96a4db6bfba68ef1c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 25 16:17:14 2016 -0400

    ->generate for PF_RDS

commit b62cfa7ba89e543eb3e7a32e8c50073b0166abbb
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 25 16:16:25 2016 -0400

    ->generate for PF_PACKET

commit d5d1ada16b6d87e01dc0244dcf585b6e01b620c6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 25 16:09:00 2016 -0400

    ->generate for PF_ALG

commit 768f817c2d991ec01f14e97bc3acd3464e478e4d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 25 16:06:58 2016 -0400

    ->generate for pppox

commit 1543000320d06bd7928718cfad5f81a0f5eb1ae0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 25 15:56:58 2016 -0400

    ->generate for rxrpc

commit a5506ef075bec5562b2b820257b691a857eeab2d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 25 15:44:41 2016 -0400

    rename the IP specific protocols

commit ec1cd9d47bebad49b6e449618d9db60a514bef4a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 25 13:45:21 2016 -0400

    Sometimes just do generic socket options instead of the per-proto ones.

commit b1558c017ddc215a46ab3396b806bf9f63873fba
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 25 13:42:19 2016 -0400

    move the generic setsockopt options into setsockopt.c
    
    It's not a protocol, so the filename was a misnomer.

commit 18eb29935f5faa9ef9b28c4d336b6a5e59b1b312
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 25 13:29:56 2016 -0400

    sometimes make generate_sockaddr create a PF_UNSPEC sockaddr

commit 0d7e5bca772cf6bb3eb40e584a9729cd5e67b618
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 25 13:29:07 2016 -0400

    point alg->salg_name to a hash name if we choose hashes.

commit 6a36879c7b8bae11e83afec5e665e968acc615b4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 25 13:26:17 2016 -0400

    sendmsg sanitise improvements
    
    - Create more realistic iovec's (limited to at most 3 entries)
    - Sometimes no iovec
    - Set msg_control/msg_controllen to zero sometimes.
    
    Still not great, but at least seems to succeed slightly more often.

commit 526521e28bbcb6613423aaf6780c93e2b03090b5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 25 13:24:58 2016 -0400

    cap the number of ARG_LIST options to at most 3

commit b0109d15b452a2a22a43034c8400b6b2cf76cefd
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 25 13:24:03 2016 -0400

    when we play with 1GB mappings, don't create so many children

commit dadd15e33834af3598aabc15d180075f05b9d9ba
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 25 11:51:14 2016 -0400

    remove extraneous format string

commit d868baf5417b65398533126e5fd263bb7df2ea22
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Aug 22 16:17:59 2016 -0400

    misc coverity warnings

commit feed6eeb0e7e7cb0e300e32d85b79cec73658c2a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 18 16:52:56 2016 -0400

    ->generate for nfc

commit 57071b9d752de3f40f21777df832e41c29a6a493
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 18 16:48:33 2016 -0400

    ->generate for CAN

commit 2cde944fe1fa808c268b245e6355eaac83b2fe7a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 18 16:42:11 2016 -0400

    ->generate for unix sockets

commit 0f6caf1fd82eba51dc45d84479fd7ebc4166712a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 18 16:39:39 2016 -0400

    ->generate for bluetooth

commit 81bb32f6630f2f028e2f1a1260b12adcf6d90892
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 18 16:28:44 2016 -0400

    ->generate for IPX

commit 650f4c18aae810b4dfee4d67990354510ec02159
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 18 16:25:54 2016 -0400

    when comparing PF names, take the length into account.

commit 743ee32a3b99ea7e1c11f4377e6c13300aa4a4cb
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 18 16:23:49 2016 -0400

    ->generate for atm

commit 1a0b370aaf7e946d3741c5425a3e31bea1defa4a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 18 16:07:57 2016 -0400

    ->generate for ax25

commit 7f4fa16a7e9561a5b627ff711050a3299495aadb
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 18 16:01:25 2016 -0400

    eat whitespace

commit c60bb9e976561970afc7397308567ce7e9ce825c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 18 15:29:37 2016 -0400

    print number of unknown protocol/types

commit 14ef7ec8ed8175b5adb759980ea432cceaa288d7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 18 15:02:44 2016 -0400

    fix -E to work when no -P is passed

commit 98e987a34898e135d410a5ba924a40ef25a989c6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 18 14:54:52 2016 -0400

    ->generate for netlink

commit 83f14a2906f9f83c2ef6c1f1f1389286684bab9b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 18 14:44:13 2016 -0400

    ->generate for IRDA

commit b9e3a9d5f74562c53c658b226673f086f8fe9736
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 18 11:00:34 2016 -0400

    make sendmsg pass in more sensible iovecs
    
    also the msg_flags arg is apparently ignored, so leave it 0 most the time

commit 475eaf918a00ee30cd37c6668c7a8af4f9d3263d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 18 10:42:39 2016 -0400

    scope reduction

commit 8a3c9e18d25ff3cbfebad6a7d8ad68435aa12bc4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 18 10:41:22 2016 -0400

    fix uninitialized use of .type in socketcall

commit d4f6f98967ab8d9cb38d10b139d82902d2181ff6
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Wed Aug 17 13:14:20 2016 -0400

    perf_event_open fix unexpected char warning
    
    The perf_event_open code parses some files under /sys
    There are some hex values there, traditionally these were all lowercase
    but on RaspberryPi2 machines for whatever reason they use uppercase.
    
    This led to warnings like this:
    
        Unexpected char A
        Unexpected char D
        Unexpected char D
        Unexpected char B
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 73b51799f7b0248056c106082477eb4bf2ca78a2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 21:42:34 2016 -0400

    pass the protocol family down to generate_sockaddr from sendmsg()

commit 6aeaa6ec1f3e1f5bc5e8ffcdb55af6dc736722b3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 21:18:49 2016 -0400

    remove duplicate get_argval function

commit 5dc766502c816c0ff3bcf083df16e046d0c8327d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 21:03:56 2016 -0400

    fill out flowinfo & scope id for ipv6 sockaddr

commit 82574abd60740c1a70b139068a50cfdffaff445c
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Wed Aug 17 20:23:49 2016 -0400

    perf_event_open: update to features in 4.8-rc2
    
    This patch adds in the new ABI features found in 4.7 and 4.8-rc
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 6ef294a4c039761f02a50a0e536935954c1336a4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 20:10:11 2016 -0400

    generalise the 'generate socket' routines.

commit cd5a452f05300bd0c23b8b9e3f8229f4eee7f3a9
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 20:01:14 2016 -0400

    ->generate for appletalk

commit b40e58ff795c0105d7df31d357b714fcd925423f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 19:45:36 2016 -0400

    add a ->generate for ipv6

commit 3d9287d454e83f3c26777f4bbac900c66bd86787
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 17:58:42 2016 -0400

    handle ipv6 sockets in analyzer

commit d5f29cd01dce0e88328aee13f33f970e023d6a51
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 17:47:06 2016 -0400

    flesh out the ->generate function for IP packets

commit 5600b7a25abaa5d197590f272f18084b6e0381c3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 17:01:58 2016 -0400

    simplify rand_proto_type() using RAND_ARRAY
    
    The special casing for packet sockets isn't worth it.

commit bf49450565f071d7bf69675af8a6095da0644540
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 16:58:37 2016 -0400

    Fix up -P to generate at least 50 sockets.
    
    Killing off NR_SOCKET_FDS had this unfortunate side-effect that -P only
    created a half-dozen or so sockets.
    
    For now we'll reintroduce this nasty hard-coded limit.

commit 232ad5ab19af593dff4dee4376076e3b45293b10
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 16:57:21 2016 -0400

    If we get a zero byte socket cache, invalidate it.

commit 403c1261f605815974719096211689cf372cfe1b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 16:00:03 2016 -0400

    if the child pidstat file didn't open, don't try to close it.
    
    This can happen if we run out of file handles

commit d4ce5f34406cd51cb6faf73f503bd88328d8b944
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 15:29:06 2016 -0400

    remove debug code that crept in

commit 1f8a18693be0c9c2c88cfce7567b12da03992284
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 15:21:05 2016 -0400

    revamp the socket generation
    
    Instead of calling down to ->socket N times until we have enough sockets,
    call ->generate which will create N sockets.
    
    The advantage of doing it this way is that we get far less duplicate sockets.

commit f06319cce463fb87b025913f5880f51dbe061c3f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 12:03:08 2016 -0400

    move initial reseed out of fork_children into spawn_child
    
    Making spawn_child the only place we do reseeding.

commit 77e56117c79f3eeae880c5439601e65e2597d36f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 12:00:13 2016 -0400

    move periodic_work() after the seeding interval

commit 11a17e301f5f30ab577aef7bbd2111b67d6d6952
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 11:59:28 2016 -0400

    Prefer using getrandom() over urandom if available.

commit 9d1f58d78882a9f06404f742beb89c63506c704d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 11:47:55 2016 -0400

    socket(INET[6], IPPROTO_IP) can be multiple types.

commit c798e7099053412f060427e3507be34c584450bb
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 11:41:46 2016 -0400

    add possible SOCK types for IP sockets

commit 577815638f1b52f64eee55498e017875a8ffdf7c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 11:38:18 2016 -0400

    only do raw sockets if we're running as root

commit 9d6190f4406924acfeb5977a0aa0b1b70e69e36e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 11:26:52 2016 -0400

    make PF matcher work on length of string passed in.
    
    We were matching 'INET6' as 'INET' because it appears earlier
    in the list.

commit 9e981daa797e5187e0172a37180586ec82d3b67d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sat Aug 13 00:26:54 2016 -0400

    remove masking of random seeds.
    
    Reducing entropy in the seed isn't a good idea, we end up re-using
    a lot of the lower numbered seeds.

commit f07a51551698d878797ac2887ed1b21fc7ff575b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 10:40:42 2016 -0400

    tweak localhost bias to be 90%

commit d5d1075236557414df597210fa7b426d80e3acc0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 17 10:38:51 2016 -0400

    remove all the server_addr stuff
    
    It wasn't particularly useful in this form. Eventually I'll do something
    better here, but for now we'll rely on localhost

commit 62ec5b0504e7804c49cd32f5b1c2ea94e447a35b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 16 14:15:10 2016 -0400

    move all the results related code out to separate file

commit 88a87b1034dd705f6aded972e1e4473988277e69
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 16 14:01:19 2016 -0400

    store results for ARG_LEN

commit e3dba430e08679e22cec38afcaf114336a9b66f5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 16 13:50:28 2016 -0400

    rewrite ->store routines to take a pointer to the result struct to update
    
    This avoids having to duplicate the argnum switch in every ->store

commit 4a36d22d3290f75b3a61c924ccf6dacb9e79e3e8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 16 13:23:07 2016 -0400

    remove implied local variable name

commit f4e320ebda77939d44ae6f5dfd6cff6728372ce1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 16 13:12:23 2016 -0400

    store results of which file descriptors a given syscall succeeded with.
    
    This will later be used to avoid using fd's that block, or fail in some
    other way.

commit e324ce2d66fc8d811738742fbb3a79e51c234d76
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 16 12:34:11 2016 -0400

    remove out of date comment about -n
    
    This arg doesn't exist any more.  This condition may be unnecessary now,
    but it's safe enough to leave it in case I change the path building code
    later.

commit 5ebc41c1bc4e16eba2c71c6169e10df1b7a11972
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Aug 12 17:31:23 2016 -0400

    Fix a missing reseed point.
    
    We were only reseeding from fork_children(), so when we called
    spawn_child directly, we got a child with a seed we'd already used.

commit 7ec1889ede862687dd53469eb98ceea8b03c0217
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Aug 12 15:10:14 2016 -0400

    dump the stack of any 'stuck' children.

commit ecf6dfd83d4c886d78d4605163cb8c3f1728db62
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Aug 12 15:05:01 2016 -0400

    if we haven't done a syscall yet, treat child as "making progress".
    
    Chances are that we haven't been scheduled because some other
    children are hogging the cpu.

commit 7475c7ad0ff107f7112d4b7d660ca26dcb61ad71
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Aug 12 09:36:03 2016 -0400

    shut up coverity.

commit 3ac0b187703d7a1d12d38b33efefebc74f03da40
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 11 18:13:23 2016 -0400

    check for NULL protocol ptrs

commit de4de2d55f055df7b854f00cc9cf33beee148a89
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 11 17:41:11 2016 -0400

    give equal chance to mmap exclusive flags with/without extra flags

commit 3ae85e02f6776e0f02df544c48d6820ed8c4a945
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 11 17:38:20 2016 -0400

    rewrite PACKET_RX_RING support
    
    For now we just support v3 of the packet format.
    This at least gets us able to trigger the tcpdump bug that was
    reported in 4.8-rc

commit d1390caa6146910c95786fc78aa38ea0dc141b09
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Aug 11 17:37:28 2016 -0400

    Add a hook for protocols to call post socket()
    
    Sometimes we might want to always call certain setsockopt's on
    a created socket.

commit ac65d70b42944ff572aea2c5fcb508620405a62b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 10 18:20:13 2016 -0400

    fix segfault when specifying multiple -c's
    
    We now need to care about NULL's in the syscall table.

commit 9a1e242e77915510081d3d9b765af00411ec3fbd
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 10 18:03:41 2016 -0400

    mmap sanitize can use simplified 'size' setting

commit 7f06c6edf50bb772f8d0e9729946d3f2e0a46059
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 10 18:03:06 2016 -0400

    make IPPROTO_IP sockets be type DGRAM

commit a01dc8b28fc836c5628cf967f52bfa3ad1068037
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 10 18:01:19 2016 -0400

    remove the 4MB mapping
    
    This is pretty pointless, because no-one cares about !PAE 32bit x86

commit c8a03b87dcbe5c12119a01d2ee742c779e6ffa91
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Aug 10 17:52:33 2016 -0400

    secret option to skip the delay when run as root

commit fbe7d7c38197132980cdf3c8e3528bd0cef25f2a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 9 15:25:36 2016 -0400

    Add initial support for kcov functionality.
    
    This doesn't do anything yet, and is more or less just a carbon-copy
    of the code in the kernels Documentation directory.  Landing this now
    for further experimentation later.

commit bc97bc35f26f1ba2f263a2628866155915709569
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 9 13:51:29 2016 -0400

    remove the pause in the parent before we wait for the child.
    
    It isn't necessary due to the other syscalls

commit 44db6756c3658c5eb6fa8b28eb785ea61dffb813
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 9 11:47:57 2016 -0400

    dump errno if we fail to fork a child

commit ee17c3c6028bd9d33a17e83237289840e29b92e6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 9 11:47:22 2016 -0400

    forcing an unlock is a bad idea, just wait a little longer.

commit fb3e7bc04ef995c4cf091219c938946290c71126
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 9 11:46:41 2016 -0400

    exit child on SIGHUP

commit d19438e250e56530d8d0576c1906393dbd878fde
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 9 11:45:53 2016 -0400

    factor out the "do syscall in child" code

commit 05bc6bbf779525574870267b99d798c16c758e2d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 9 11:28:02 2016 -0400

    Fix dumb typo in bpf configure check

commit 08c9d6d3faac0ab0f7b20dcc02113bcf096ac065
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Aug 8 13:51:12 2016 -0400

    fd provider for bpf maps
    
    so far, just supports array types

commit 177317f8e827151d68addb9cc0337ef0d173018f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Aug 5 14:41:17 2016 -0400

    Check for existence of linux/bpf.h
    
    This is a little more involved, due to us now needing to handle
    holes in the syscall table when something isn't defined.
    
    Based on an initial patch from Vinson Lee <vlee@freedesktop.org>

commit 2fbc368ffb65717016d69d0cc135ccb12e614ac3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 2 09:55:33 2016 -0400

    force unlock 'stuck' locks.
    
    don't wait for it to unlock, because the unlocker may be dead.

commit b1bb1028f8faf7f7f612f8137bc0bbbddf54ece8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 2 00:10:55 2016 -0400

    add support for TCP_REPAIR_WINDOW

commit 616f9953a7daa2f7c209af7934ffafde4acd8bc8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Aug 1 23:50:12 2016 -0400

    Check for 0 pid when encountering locked lock
    
    Because the pid zeroing happens before we set the lock back to UNLOCKED,
    there's a tiny window where we end up with a locked lock with no owner.
    Just recheck, and we'll get it right next time around.

commit b241797b5151887d0683e81492c4bd60579dd5cc
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jul 29 11:54:57 2016 -0400

    pass the signal from the sighandler using siglongjmp instead of global var

commit 1b498f616f945babc7c449b22e442c327043ff92
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 28 17:31:00 2016 -0400

    Add extra local var to make handle_childsig more readable

commit 40ab1d3197c8afec43d7171f198e363c23688e79
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 28 17:29:00 2016 -0400

    The 1GB mappings cause oom's too much, disable them most the time

commit a9eadb7c983cd62bb7d75a2024b01a8925ed1a80
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 28 17:28:20 2016 -0400

    make the backtrace code work with the main process too

commit 2339c4910682711f972dbd5c3cd154ce91b6df79
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 28 17:26:03 2016 -0400

    silence annoying cppcheck warning

commit fc67eb0840e5ffae71a6253c4502b17e26f2fdf9
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 28 17:07:28 2016 -0400

    grand-child forking code needs to handle errors

commit 66fde990bc151b0cc7522c49e025714898cd1bff
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 28 15:30:13 2016 -0400

    replace kill() invocations with kill_pid

commit 70127bfd7e44c8fa8167ba6e844b9eefc488849c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 28 13:51:13 2016 -0400

    replace open-coded pid_alive

commit 6e71177ae2a6978d69a5e087963fe03fa809f6c6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 28 13:13:09 2016 -0400

    pid_alive is a boolean thing, so make it return bool

commit e56924c4ba7288fd1b1ff5400d672334391961c0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 28 13:02:40 2016 -0400

    check child is alive before we bomb it with SIGKILLs
    
    also tweak delays

commit 1263564e0ef261cfcb7cdeed4c867562cdd465ed
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 28 12:23:09 2016 -0400

    fix misplaced bracket in dump_childnos

commit ee76fc3aaf78518a0e06b3e88bedd12c0e87a262
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jul 26 11:38:36 2016 -0400

    sanitise add_key type argument

commit 157637dae5cbe4413020861ff04ccd8b4f0fa366
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jul 26 11:14:16 2016 -0400

    remove all the 'unknown errno' checking.
    
    It's a huge task to get even a partial coverage here, and it's not
    particularly useful, and slows things down unnecessarily.

commit e405aa7ee24923c2fa492dd723c090bf3ebcdede
Author: Fabien Siron <fabien.siron@epita.fr>
Date:   Tue Jul 5 10:02:46 2016 +0000

    net/proto-unix.c: set protocol to 0 in unix_rand_socket()
    
    The protocol of an AF_UNIX socket should be 0.
    
    Signed-off-by: Fabien Siron <fabien.siron@epita.fr>

commit 84dc4dd4f582e8951c85368643ab439af9eb8b18
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jun 3 16:01:51 2016 -0400

    Before falling back to gtod, try getrandom() if available.

commit d0dd97bb2f05662a07d1565b53f5fb6fccb3dae4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jun 3 15:48:05 2016 -0400

    if we use the fallback seed, don't use rand() yet.

commit b314824e300a4fb09cee9b4fad8a7f8b2801bd38
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 26 17:36:41 2016 -0400

    fix wrong variable name in childops wait

commit f34ebf964687b334ba483b9004364b78a846df67
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 26 16:25:04 2016 -0400

    bpf: only do bpf_gen_filter for BPF_PROG_TYPE_SOCKET_FILTER

commit c79e0864840c51fff1e4eb87bd97384a59dc57bd
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 26 15:59:39 2016 -0400

    make the BPF_PROG_LOAD case look more sane
    
    This passes in a sensible looking bpf_attr.
    bpf_gen_filter is the wrong thing to use here, some refactoring needed.

commit b7d645c4c985d9b272a6e785380a34da5d84dc57
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 26 14:33:09 2016 -0400

    introduce object types for bpf maps & progs

commit c328cdcda9d47844fec025c13b3dbadccb22cd8b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 26 14:20:58 2016 -0400

    flesh out the bpf sanitiser a little.
    
    Lots to do here, including storing the fds in the object caches,
    and then later, using the map fds.

commit b659d6e2f2258b840f5aee1f5d1a30359ac7b683
Author: Vinson Lee <vlee@freedesktop.org>
Date:   Thu May 26 00:05:24 2016 +0000

    fds/drm.c: Add forward declaration of struct.
    
    Fix build error.
    
      CC    fds/drm.o
    fds/drm.c: In function ‘open_drm_fds’:
    fds/drm.c:120:3: error: ‘drm_fd_provider’ undeclared (first use in this function)
       drm_fd_provider.enabled = FALSE;
       ^
    
    Signed-off-by: Vinson Lee <vlee@freedesktop.org>

commit 06259c05da8e09594e196dbd57077130ce155b12
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 26 12:33:51 2016 -0400

    check for potential failure to acquire testfile fd.
    
    also, offsets are off_t's

commit 9a0a5ac6d12711ab8988700d85b05d4f82fb08e7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 26 12:31:58 2016 -0400

    fix memory leak on child fork failure.

commit b17a6f9163674cbe213440ddca93659c456e8ec1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed May 25 11:42:22 2016 -0400

    compile fix.
    
    forgot to git add yesterday, oops.

commit cbd7270994210421beaca9decb61102147a6fc33
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue May 24 15:49:52 2016 -0400

    for each random syscall, do one in an extra child too.
    
    idea here is two racing syscalls with the same inputs

commit 6532e89292164ec09962ab84dd2ebd4f506f0e20
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue May 24 15:49:11 2016 -0400

    Add a child-op to truncate one of the testfiles to a random size

commit 9a71ff0a25e91fada5e3cd9ead28d1b2282c575e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue May 24 13:46:45 2016 -0400

    Reactivate the use of 1GB mmaps if enough memory available.
    
    With the recent fixes in this area, this should be safe again
    
    also, fix another instance of 'wrong offset for 1gb mapping'

commit 93f1689da9083ed902b469153941525d79d4137a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue May 24 13:11:05 2016 -0400

    check the actual num/values, not the pointers

commit 74303be33819a1b3ef53887b5af0fd83183f8157
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue May 24 13:08:59 2016 -0400

    fix type in copy_file_range syscall definition.

commit 83af013b74b2239413bd012a39669a126912d024
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 20 13:49:02 2016 -0400

    bump the child progress indicator in set_dontkillme
    
    This is mostly for when we are unsetting it, as we could have spent
    some time dumping stack traces etc.

commit 02c98869fd47196655eb1447be25432057e2012d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 20 13:47:54 2016 -0400

    change set_dontkillme to take a child struct instead of a pid

commit d0affde28acac94aa4195aba3ce5db26dbcfd3cd
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 20 13:00:41 2016 -0400

    fix up more clang detritus

commit 37aba5d803d6da20bfdd9a4004dd8e3ce2bdc798
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 20 12:49:31 2016 -0400

    shut up clang

commit 251739ce8c7c13968027d9bbeb4d05bc5e8b9c0d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 20 12:43:02 2016 -0400

    merge enable_fds_param/disable_fds_param into toggle_fds_param

commit d52a876a795edcd9a9c2231cfea45baced78cce4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 20 12:23:58 2016 -0400

    document how register_fd_provider gets run

commit bf2cea4e0b0acbf85bb006aee46b31eec7cb98e5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 20 12:21:09 2016 -0400

    include file cleanup
    
    after switching the fd providers to use constructors, clean up the
    leftover mess.

commit ad64c0acfc3884b8aac0cbe85d8278aaf4fc1257
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 20 12:15:43 2016 -0400

    Convert fd provider initialization to use constructors

commit 69dbf0489d0442e332c7b1566d7d3a9d8eea3277
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 20 11:26:40 2016 -0400

    if random_ipv4_address picked localhost last time, let it happen again.
    
    The ip_lifetime check happened too late, so we only had a 50/50 chance
    of getting localhost more than once.

commit 6f7e3f35a02ca482bcb246cdc6356a2ce9df246c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 19 14:07:59 2016 -0400

    add flags from 4.7 for preadv2/pwritev2

commit 93be60b2e449b913923ea57bec449bcde04f0e20
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 19 12:08:15 2016 -0400

    make setup_fd_providers silent, so it doesn't spew during -h

commit 271b3d469805621b088858262f533f995cc934d3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon May 16 12:58:58 2016 -0400

    split num_fd_providers_enabled into two vars.
    
    this was being used to mean "num of providers to enable"
    as well as "num of providers we _did_ enable" depending on
    whether --enable-fds= was passed.
    
    Splitting it into two vars makes the accounting easier

commit cad191604278df8e3f148638e6196f69c6694031
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 13 16:11:11 2016 -0400

    when disabling 1GB pages, just do nothing
    
    don't bother adding a second page_size mapping.

commit 928414a2734ac8bf8b5e1362ee57de6835355514
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 13 15:47:00 2016 -0400

    fix offset for 1GB pages.
    
    When I removed the 10MB mapping, I forgot to update this, despite
    the big fat comment.

commit 1856c2c04688e2cb8ac7438ee8aa0e0019aff632
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 13 15:44:28 2016 -0400

    print the error if the initial mmaps fail

commit 44bd034522a63e68cd721b82baa103510192bc0f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 12 14:58:24 2016 -0400

    only do the 'stuck syscall' info on children running random syscalls

commit 825b51cfe8fcbc7461c9a327411475ae481654c8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue May 10 13:42:39 2016 -0400

    don't call fcntl on -1 fd's

commit ac2997db26011a45287d9cf69b2f4455c2b4a1a5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 26 13:48:19 2016 -0400

    Introduce 'root child ops'.
    
    Make 'drop privs' be the first possible operation a root process can do.
    After calling this, a child will then fall back to calling only
    !root operations.

commit 8364483cc4df0ad88872ff6aaaa7c93d14a846e2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 26 11:35:11 2016 -0400

    rewrite set_new_op to take ptrs to alternative childops.

commit 8fb7bdfe2c95cb9d06ecc4f8a8605757aa6ce79a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 26 11:31:33 2016 -0400

    factor out the 'set new op' code to a separate function

commit 01f451207652e5335d170177adf0bb18335a0797
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 26 11:12:38 2016 -0400

    set a flag in the child struct when we've dropped privs

commit adc72d6121288be4f9a1d7d8c3a1571acaf7f99f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 26 11:09:41 2016 -0400

    convert uid.c to use output() instead of outputerr()

commit acc095b6d7adbfc7b41ffbce1d3f3925d62da5d6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 26 11:07:35 2016 -0400

    reformat childop structs

commit 42d4847a0f82ce84795ab391e6920e60a1ac22d1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Apr 25 09:12:47 2016 -0400

    fix leaks in thrash-pid-files

commit 3510cc8cb0f97c52f12623d461bf53d64a4bf3e1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sun Apr 24 17:53:30 2016 -0400

    add a 'read /proc/<pid>/status' childop
    
    simple beginnings. Eventually, I'll add support for hammering on all
    the /proc/<pid>/ files.

commit 642c47958aa91ccc80703882016a4eb163b952a2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sun Apr 24 17:51:49 2016 -0400

    bump the 'pick new op' counter even higher.

commit cf387d29bb9e622b776512080b65791be0c54abf
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sun Apr 24 17:50:13 2016 -0400

    push the dropprivs code down into the child processes.
    
    * main process always retains root privs.
    * child processes will later get the option to retain root privs
      and run root-only ops.

commit b35eaf67b913eecd4ecd875d91e0624342cd0118
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sun Apr 24 10:09:36 2016 -0400

    define maximum size of ip_ssoptrs array

commit a8e000922be680a97e05546fd55259cc0fc6449c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sat Apr 23 23:39:05 2016 -0400

    be careful not to read past the end of the ip_ssoptrs array
    
    If we create a socket with a protocol higher than that of
    IPPROTO_MAX, then we can screw up here, because we sometimes
    can call the ip setsockopt() on a non-ip socket.
    (see syscalls/setsockopt.c:do_random_sso)

commit 3d9ab352bf088ed88317ca92ecf0d7e19b48234d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sat Apr 23 23:12:24 2016 -0400

    remove the 10MB initial mapping.
    
    It doesn't really serve any purpose the other mappings don't already cover

commit 2264cf7ac42a8aa6993f4d69687baeb06669aca8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sat Apr 23 23:10:18 2016 -0400

    store the type of childop we're doing in the child struct

commit 34adf2c50845a761517827e9ad11cfbac70677da
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sat Apr 23 22:51:31 2016 -0400

    remove the shm_rw debug code.
    
    It was racy, and has served its purpose.

commit d0089f1a0aec919ee87f9e799a6c3711a81386f6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sat Apr 23 22:30:38 2016 -0400

    fix the 'pick new childop' code.
    
    multiple problems here.
    - Returning from signals would clobber the locals.
    - I forgot to decrement the counter
    - And forgot to reset it when we select an op.
    
    I don't know what was going on when I wrote this code, but it was
    pretty far from my greatest moment.

commit a8550b020b2f9f6cb2002b4b1b92157c480c281e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Apr 22 01:12:31 2016 -0400

    change stats to count operations, not just syscalls

commit 7a31dd9ba8e4e0a0a55f9e3f49613602c55f772d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 20 02:03:02 2016 -0400

    Add new CLONE_NEWCGROUP flag

commit efe6f46925900c9d352d0460d3429948a45d2835
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 19 19:48:16 2016 -0400

    reset the child timestamp when we clean it out

commit 0d0c68ab29bbfbe234c09d6011cf7ecdb75e087e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 19 18:50:10 2016 -0400

    reset the op ptr when we reach the loop counter.

commit ba1eee11129abf47143f4ccf675a8f170b3f8d43
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sun Apr 17 16:54:37 2016 -0400

    use fstat instead of lstat

commit 0f3dd2f7d850392a67555e7f3229da7d0e2d6e6b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sat Apr 16 17:02:08 2016 -0400

    open/stat instead of stat/open

commit c1c503900e9995e9f6c75dec61194667fa2f253a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sat Apr 16 16:58:02 2016 -0400

    fix open() return code check

commit 6d751b3049a370bcd731e257f2718910861ef8ad
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Apr 15 16:28:02 2016 -0400

    move op_nr from syscall record to childdata
    
    This is a progress indicator, and is child-type agnostic, so it
    doesn't belong in the syscall struct

commit cc6570c65903d5b0d8e4e27a50127c624a7c1f71
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Apr 15 15:59:27 2016 -0400

    children/ -> childops/

commit c9cca6d4947ac81ad644ea32e6ec6a1c59a0809e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Apr 15 15:57:05 2016 -0400

    update the prototypes for the read-all-files childop

commit daa4bf17a61044b77e73bfe963505d505b6ff16b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Apr 15 15:45:24 2016 -0400

    repeat the previous childop 1000 times
    
    instead of potentially changing every iteration.

commit 5e4228da4b482be8e3043480b68b2763d952704a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Apr 15 15:35:40 2016 -0400

    fix badly named child_random_syscalls()
    
    It only does a single syscall, and is a childop, not a child

commit 0635357a713eef4577c78da8a9ee3fdf8d6a96bb
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Apr 15 15:26:10 2016 -0400

    pass the child ptr down to the child op.

commit 3d49efc613b6d2902ac7a47e8840851aaf9a5125
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Apr 15 15:02:14 2016 -0400

    change reap_child to take child struct
    
    iterating over the pid list when we know the
    child struct is a waste

commit 91c1e632776c8a2ce6f7677ee54eec2d636d48cf
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Apr 15 14:30:14 2016 -0400

    move the "child progressing" timestamp from the syscall rec to the child.
    
    This will be more useful when multiple child types are implemented.

commit bcfc86c513c72888cbb88a9eb6417675c71d421c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Apr 15 14:29:49 2016 -0400

    decrement enabled fd provider count if one fails

commit 72396aa7888459987772fd0b1788c2b421b03739
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Apr 15 14:29:16 2016 -0400

    add missing break's

commit 4cf405d07a4733ac9a377478619483d692c7758b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Apr 15 14:03:15 2016 -0400

    perf: reset the failure counts if we succeed at all.

commit 8799ba00836caff9d2fa6babec714ae2bf2562cf
Author: iljavs <ivansprundel@ioactive.com>
Date:   Wed Apr 6 12:59:23 2016 -0700

    Update perf.c
    
    win10 LXcore always seems to return EINVAL for this, throwing trinity in an endless loop.

commit f2ecc0e9a94871ebd8f27daf390516d03a22fa1d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Apr 8 14:39:39 2016 -0400

    rename ->send to ->gen_packet, and redefine args.
    
    This function will be used by sendmsg etc too, and we need to get
    back a ptr to a generated buffer, along with its size.

commit b94edace8fcb9f7d6039a3b5f9c0b68b588a098c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Apr 8 12:52:19 2016 -0400

    Only pick a single netlink group
    
    From the man page:
    "Since Linux 2.6.13, messages can't be broadcast to multiple groups."

commit ff5a5b10a1fd14ed4a65125b5f825b6c6632e305
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Apr 8 12:06:49 2016 -0400

    when per proto ->socket succeeds, indicate success so we don't fall back

commit 40d38f86cbfee1b1531b0be1369aee41cb569e67
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Apr 8 12:05:40 2016 -0400

    add missing NETLINK_ROUTE decoding

commit 8b1122f982a61dde4183b15228974ad74f092251
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Apr 8 11:56:07 2016 -0400

    decode netlink protocols

commit 2f5fa17ef3e858e6c0f5d530b24613371032d3b4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Apr 7 10:42:37 2016 -0400

    if a proto's ->send is missing, fallback to random
    
    we returned too early, and ended up free'ing bad pointers.

commit f1f16f1c10373f32f31471b4e7dd6290ed1ea138
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 6 15:32:53 2016 -0400

    add MSG_BATCH flag

commit 048f6cd36f6ee576e05f1e435a3095472f40f538
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 6 15:29:22 2016 -0400

    add SO_CNX_ADVICE setsockopt

commit 3f338c2d18d411352f3c20d5794c3541bfd8ffb1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 6 15:24:29 2016 -0400

    add pwritev2/preadv2

commit 729c2cb220831022fd9a6b1ef76eda743917c247
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Mar 31 15:30:44 2016 -0400

    add a ->send method to struct netproto

commit 21b08e7a999804fc0e7ad3f11beddcfd449c086c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 30 09:35:36 2016 -0400

    add l2tp to analyze-sockets

commit 2630751fb74b00c074788ebe8e21b85fa73d5e6e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 30 09:31:11 2016 -0400

    remove some noise from potential failures from unshare()

commit 6c791b39ec794c53e9294ec38c941fc82610da65
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Mar 29 15:07:36 2016 -0400

    EBADF is a possible return from execve

commit 8543692ebbfc13986541789fbb2926da2637532b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Mar 29 15:07:03 2016 -0400

    by default, start nr_cpus * 4 child processes.

commit 9ab78be27faba9053e20877d3c4fb30646a22f17
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Mar 29 15:06:19 2016 -0400

    make get_new_random_fd bail early if no fd provider has finished yet.
    
    another rare deadlock.

commit 4e633ab7d46bb360e9061651204a0216002e055a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Mar 29 11:31:48 2016 -0400

    use SOCK_RAW for IP sockets half the time.

commit c3c20277f2294ee46f996379d6dff6c0b9be51a4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Mar 22 12:46:17 2016 -0700

    add bare-bones PF_KCM support

commit 4387f9f447486519d93f95586a549ce0f7a0514b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Mar 15 16:01:18 2016 -0700

    remove unnecessary include

commit 49072463130406c5eacbdfb98bea7aa2d173e710
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Mar 15 16:00:40 2016 -0700

    add ext[234] ioctls

commit b24bc9b12be906962e354e5c7612dea680efeefe
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 14 20:13:08 2016 -0400

    check for existence of btrfs ioctl headers

commit 6a5b4a8649b8fb17bc538795e699b29402d61cdf
Author: Vinson Lee <vlee@freedesktop.org>
Date:   Mon Mar 14 16:54:15 2016 -0700

    net/proto-ipv4.c: Include limits.h as a workaround for older kernels.
    
    For example, CentOS 6 kernel-headers-2.6.32-573.18.1.el6.x86_64
    linux/netfilter_ipv4.h does not include limits.h and results in this
    build error.
    
      CC    net/proto-ipv4.o
    In file included from /usr/include/linux/netfilter_ipv4/ip_tables.h:20,
                     from net/proto-ipv4.c:12:
    /usr/include/linux/netfilter_ipv4.h:53: error: ‘INT_MIN’ undeclared here (not in a function)
    /usr/include/linux/netfilter_ipv4.h:64: error: ‘INT_MAX’ undeclared here (not in a function)
    
    This was fixed in upstream 3.2 598aaff2ee05c91728e5845956dd9754ed04315c.
    
    commit 598aaff2ee05c91728e5845956dd9754ed04315c
    Author: Ben Hutchings <ben@decadent.org.uk>
    Date:   Wed Aug 24 18:45:36 2011 +0000
    
        headers, netfilter: Add missing #include <limits.h> for userland
    
        Various headers use INT_MIN and INT_MAX, which are defined for
        userland in <limits.h>.
    
        Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
        Acked-by: Patrick McHardy <kaber@trash.net>
        Signed-off-by: David S. Miller <davem@davemloft.net>
    
    Signed-off-by: Vinson Lee <vlee@freedesktop.org>

commit 22622044f323bbc91609f0e52d91d3688cc0e0e6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 14 17:00:14 2016 -0700

    add VFS generic ioctls

commit 8b0b1220ce4cc5f5be8a7ff3177eb75bc6b13a2e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 14 16:34:01 2016 -0700

    allow for multiple ioctl definitions to succeed at fd_test
    
    This will be useful for eg for doing both VFS ioctls and BTRFS ioctls
    on the same fd.

commit f037dbd0515c33c504ffd6ef7e364cd4a6df1f07
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 14 15:49:34 2016 -0700

    mangle the random page we pass as an ioctl arg

commit c7343b0aadb91af2b0477744ad03a4fa087eb941
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 14 15:38:56 2016 -0700

    add btrfs ioctls

commit 14e75a6417383cd347efcd83bda01c7c5ec58c1d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 14 15:10:24 2016 -0700

    ioctl: make random arguments a bit more.. random.

commit 3a6ff062a8ff39d6b1f378468bea5b364513c8bb
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 14 14:53:26 2016 -0700

    fill in some socket ioctl gaps

commit 02b3bf7df0d9247d51b3725b12e282ded1b6848e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 14 14:38:50 2016 -0700

    ioctl: move the rdev test later
    
    sockets weren't getting matched to the socket ioctls because they
    don't have a backing rdev

commit bbff560bff9b66f603dc3c79da20086ab4e64da5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 14 14:29:18 2016 -0700

    ioctl: allow 'pick random group' as a fallback.

commit a7e8ecf7147fffa9bca8bb24e9917fb1117f31a2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 14 14:26:53 2016 -0700

    simplify ioctl generic sanitise

commit bd9b58a142035d25e789ef1d33a6bc56eb8880fe
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 14 16:09:48 2016 -0400

    sometimes unshare the userns on child startup

commit 2b1f865a892602fa3ac6ac445136d3db4752e962
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 14 16:07:57 2016 -0400

    let get_writable_address return some sysv shm objects

commit f8f487314f289543b4588b4c71a3cf333557c43a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 14 16:01:48 2016 -0400

    make mmap use a random fd for non-anon mmaps

commit 6f886cf69b5eed5dfeb6122f38f3a9649fb3e9e8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 14 15:59:52 2016 -0400

    when reaping dead children clean up zombies immediately.
    
    Without this, we don't get the shm->running_childs check
    working if all processes are in Z state

commit 0c2136c42a69b0d87b6509abbcef77a4ac791a64
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 14 15:59:19 2016 -0400

    teach perf_event_open how to pull from the perf fd cache

commit 9f40cf9f2e1826e07e3a8d98ceff21383a7e4ac6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 14 15:57:02 2016 -0400

    fix accounting of enabled fd providers.
    
    There was a possibility of an infinite loop here, because we were
    comparing against the number of known providers, instead of number
    of enabled providers.  Rewrite how we count enabled, and use that instead.

commit 24b0c8dcde6d0db77d61dd2e57ea226add68c4d8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 14 12:55:09 2016 -0700

    don't modify argv directly.
    
    operate on a copy instead when we munge it.

commit 1f5b0ccaf0ad5905c52adae0ccbf93f0d7711916
Author: Vinson Lee <vlee@freedesktop.org>
Date:   Fri Mar 11 16:30:03 2016 -0800

    compat.h: Add SO_IP_SET for older distros.
    
      CC    net/proto-ipv4.o
    net/proto-ipv4.c:223:12: error: ‘SO_IP_SET’ undeclared here (not in a function)
    
    Signed-off-by: Vinson Lee <vlee@freedesktop.org>

commit 81a029cafa71630f765adb9f34dc78c257cfc9d4
Author: Adrien Schildknecht <adrien+dev@schischi.me>
Date:   Thu Mar 10 15:31:49 2016 -0800

    net: include <linux/if.h> instead of <net/if.h>
    
    Commit 4ada80dd202675 includes net/if.h and linux/netfilter_ipv4/ip_tables.h
    net/* and linux/* should not be mixed.
    Debian Jessie x86_64:
      CC    net/proto-ipv4.o
      In file included from net/proto-ipv4.c:11:0:
      /usr/include/linux/if.h:71:2: error: redeclaration of enumerator 'IFF_UP'
        IFF_UP    = 1<<0,  /* sysfs */
      /usr/include/net/if.h:44:5: note: previous definition of 'IFF_UP' was here
        IFF_UP = 0x1,  /* Interface is up.  */
      [...]
    
    Signed-off-by: Adrien Schildknecht <adrien+dev@schischi.me>

commit 560b25d32f92cc34f305762da71afbdc0f3187dc
Author: Adrien Schildknecht <adrien+dev@schischi.me>
Date:   Thu Mar 10 14:56:55 2016 -0800

    compat.h: add SHM_NORESERVE flag
    
    Debian jessie MIPS build fix.
    sysv-shm.c:22:3: error: 'SHM_NORESERVE' undeclared (first use in this function)
       SHM_NORESERVE,
    
    https://sourceware.org/bugzilla/show_bug.cgi?id=17930
    
    Signed-off-by: Adrien Schildknecht <adrien+dev@schischi.me>

commit 8e0584879dd2f9bc41bad5cf19f722fdfa0b674e
Author: Adrien Schildknecht <adrien+dev@schischi.me>
Date:   Thu Mar 10 14:56:23 2016 -0800

    move x86 specific syscalls to x86 dir
    
    Signed-off-by: Adrien Schildknecht <adrien+dev@schischi.me>

commit 353ee4d4f9dbc615e1ad8ea65a1a481c8df301f9
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 9 16:08:42 2016 -0800

    move child debugf into child code

commit ff58c4a70f262ce0fa4a5929551fdec02247ce73
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 9 16:08:06 2016 -0800

    move child cleanup to child.c too

commit 2ff781851d7ee1b1bfdf7a426c45c9e59d594f3a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 9 16:04:35 2016 -0800

    make init_child use the child struct passed in instead of searching for it

commit ad0f981cb4ff0e2319f279bc6f689f8381c2cafe
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 9 16:03:37 2016 -0800

    keep init_child local to child.c

commit 115993aaa73566d9fb945494be599694e5bec356
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 9 14:40:11 2016 -0800

    add ipvs setsockopts

commit e2fc611d5f81878f713300ab7f1c7a847f87b449
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 9 14:37:35 2016 -0800

    add SO_IP_SET setsockopt
    
    ugh, this demuxes the command from the data passed in instead of it
    being multiple socket options. Something for later maybe

commit f39f129410207d838c66e1e3e35d049a8b2d4d4c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 9 14:34:40 2016 -0800

    add ip6tables setsockopts

commit 9985bd61e9aeb54316526f9f70e7e4139a7a73ed
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 9 14:31:11 2016 -0800

    add the arptables setsockopts

commit 82cbfb238aaf34bfd3abf1cbf839e4dbd9484809
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 9 14:30:00 2016 -0800

    add the ebtables setsockopts

commit 4ada80dd2026756643b29c8d346b039ea9cd7f41
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 9 14:27:45 2016 -0800

    add the iptables setsockopts

commit 5d44f2c7a3b77ebb2fe38119dcd6d450c41ff3dd
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 9 10:59:41 2016 -0800

    dump unannotated 32bit syscall table entries too
    
    add code to the 64bit path to only print entries not also in the 32bit table.

commit f311015bdddac1a60065256af8161a960c1ce771
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 9 10:54:18 2016 -0800

    fucking Makefiles, how do they work

commit a3aa4bd9bdd08c4bfe969a2fb4050473933fe150
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 9 10:40:29 2016 -0800

    add cppcheck target

commit dd03ae2b876d35234c53a7ef610bbbf03c77daae
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 9 10:40:14 2016 -0800

    fix type-spew from cppcheck

commit 8a0612d823dd65055714941f6be040be4ae3d35a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 7 16:27:48 2016 -0800

    add some crappy code to find unannotated syscall args

commit 800bd3466aa9e6caa7caa58843fd690ebce3b8b7
Author: Michel Normand <normand@linux.vnet.ibm.com>
Date:   Tue Mar 1 17:53:03 2016 +0100

    syscalls/mlockall.c only define MCL_* if not already done
    
    required to avoid build errors on PowerPC.

commit d2fb10d1763cba2df6b9d4409cc09a07c17476c0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Feb 26 12:50:24 2016 -0500

    sanitize the cmd & buf args for shmctl

commit aa30ce391add0306c867bb9cfafd997bb8b53fdb
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Feb 25 10:43:08 2016 -0500

    give up trying to open perf events if they keep failing with EPERM

commit 16397b63080fe8491868da91fcfab8fe525d33ff
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Feb 25 10:31:01 2016 -0500

    randomize fd provider initialization order

commit f41d662108863ff10cb10d0793b5f2a96b1bdce8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Feb 25 10:04:53 2016 -0500

    don't let get_random_fd run on uninitialized fd providers
    
    This triggered a segv on init when pppox needed an fd, and
    tried to get a pipe (which gets initialized _after_ socket init)

commit b91a4ef77da5a5b804e3f478879bb214883e5ab2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Feb 24 18:10:14 2016 -0500

    rename.

commit e7275c6504851ed714c0990990abdf99afec6468
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Feb 23 12:46:36 2016 -0500

    remove lock stealing leftovers

commit 0064f7558ea7041b1e1f01792d4bb4affbd755f3
Author: Artem Savkov <asavkov@redhat.com>
Date:   Wed Feb 17 17:20:39 2016 +0100

    Fix logging logic in output().
    
    Commit "preparation for different forms of logging." contained a typo that
    broke logging to mainlogfile.
    
    Signed-off-by: Artem Savkov <asavkov@redhat.com>

commit f4c49fb29de6932b2dcb019ee2ba40e867e486f7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Feb 17 10:59:40 2016 -0500

    remove void * arithmetic in dirty_last_page

commit b8bc3a5fc4e8913630aee76601661e3a000a4e9f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Feb 17 10:57:37 2016 -0500

    scope redux

commit dcfce4823a414318b63758cad7c9bbbb04df9cfe
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Feb 17 10:57:11 2016 -0500

    fix a rand64 case
    
    undefined behaviour shifting > width of type

commit 3ebd636144735adde1f3263c768978e2330cc64e
Author: Vinson Lee <vlee@freedesktop.org>
Date:   Mon Feb 8 21:24:41 2016 +0100

    Makefile: Use -std=gnu99 if -std=gnu11 is not supported.
    
    -std=gnu11 is not available with gcc < 4.7.
    
    Signed-off-by: Vinson Lee <vlee@freedesktop.org>

commit 31cef803a3a79d3bb26a1a19b805da12d7e457f1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Feb 4 11:13:20 2016 -0500

    resync crypto algs

commit 0207de22f08d6314be07722176e626355e14d727
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Feb 2 15:30:45 2016 -0500

    fix up more out of bounds writes in the page-dirtying routines

commit 0cd4b515c4aa95320031a262c8f10df76e9291b3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Feb 2 15:13:13 2016 -0500

    double check that we always have enough children running

commit 0b974c7d8335e7b9e94379941224ab8490d16491
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Feb 2 13:31:08 2016 -0500

    Add some SYSV shm objects.
    
    These aren't used anywhere yet

commit 0583dbc0a112c924ea19dd289f6971d5364183fb
Author: Jiri Slaby <jslaby@suse.cz>
Date:   Tue Feb 2 13:25:21 2016 +0100

    net/protocols.c: fix build on old machines
    
    * include compat.h for PF_* constants
    * use PF_ALG only if USE_IF_ALG is defined
    
    Signed-off-by: Jiri Slaby <jslaby@suse.cz>

commit a53144d86f4b862d907b45f9cdf3fad421c2fdf4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Feb 2 11:27:18 2016 -0500

    use TRINITY_PF_MAX instead of PF_MAX in array definitions.
    
    Older distributions may have a lower PF_MAX than we need.
    Pointed out by Jiri Slaby <jslaby@suse.cz>

commit 10031645562ca23e6fd7492100eb1b831449dda9
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Feb 2 11:03:20 2016 -0500

    fix off-by-one in socket validation

commit 137a2d3387303c938408f63eeff00b9de1a579f4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sun Jan 31 20:18:22 2016 -0500

    reduce scoping

commit 73031324ef69081b8be0d55dc51de57f57aa7134
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sun Jan 31 20:15:46 2016 -0500

    create mmap's of the testfile fd's too

commit eb6b6e561dfb7dc8be8caa3812e2e90e0562aacc
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jan 29 16:20:14 2016 -0500

    check the limit on the domain in the socket cache

commit 790f32889e53e3261f5600af4991ed9d029c5a6b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 28 21:19:44 2016 -0500

    go back to a state where get_map can't ever fail.
    
    This is too much of a pain in the ass to run around everywhere
    and handle fallbacks

commit 6a47b7f316491769dd6bd49728e79fd8e8b9a4ff
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 28 20:55:18 2016 -0500

    fix potential infinite loop in get_writable_address

commit a781db526a5f57eed12048a0b54978f7dc6714ae
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 28 20:53:03 2016 -0500

    mincore fails giant malloc's all the time, cap the max size at 1GB

commit 3113b1d263d976d4434862ae8930c2b0f8475a69
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 28 20:52:36 2016 -0500

    dirty_random_mapping needs to handle NULL get_map() results

commit acae380ccbffabd338a76b7571fcce82c50aebed
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 28 20:07:50 2016 -0500

    make mmap'd files use their own object cache

commit 528344b52ef3d245ed57bd42f51180f61dd22227
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 28 20:06:26 2016 -0500

    make get_writable_address retry on failure.
    
    Again, an ugly hack. We could theoretically get stuck here
    if we've run out of writable mappings.

commit b6fcd4ab3c060b469fa8e4bf266622603fc23149
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 28 20:05:22 2016 -0500

    Move the file fd provider up so it runs first.
    
    This is kind of a hack, but makes sure we have some mmap's
    on the lists before the perf fd provider tries to grab some.

commit 7276c6ff77cda70f369e0b66c36097fc489ac2aa
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 28 20:05:00 2016 -0500

    rename OBJ_MMAP -> OBJ_MMAP_ANON

commit d59e1fe1449189a347b979b5a3447d0b218fa13e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 28 19:51:01 2016 -0500

    allow get_map to return NULL
    
    if get_random_object fails, we shouldn't return 0x10

commit c17fbae6d68848007d69568ed2219ec08b86b589
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 28 18:57:17 2016 -0500

    change mmap_fd to take a global/child flag

commit a3ab7df996f95efa3cf32390c2c18c05ccc959f7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 28 18:46:42 2016 -0500

    move mmap_fd to maps.c

commit cf3c9324393b0f9d5bcf4688660e6619c202dbe3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 28 18:43:36 2016 -0500

    rename the TRINITY_MAP_ definitions

commit 85f7deac96cb48c109679872c157fe55447401de
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 28 18:35:28 2016 -0500

    remove debugging code

commit dd1d471e1f3775eccbd2c73b6e17039f377c95c7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 28 18:10:18 2016 -0500

    when creating file fd's, also created some mmap'd ptrs.

commit 6402c11220248e88a2cd255acb20f8ededb98275
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jan 27 20:37:40 2016 -0500

    add new EPOLLEXCLUSIVE flag to epoll_ctl

commit 3c4642ddb043416410215ea894c8e553be9b60ad
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 18:46:10 2016 -0500

    remove cut-n-paste mess from man page.

commit bcfce2fa39dc89ece4ba7732299738d020c8dec3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 18:44:35 2016 -0500

    constify sock_option members

commit 83f3345ccfd24177b57d13876ad9efdbc2099d23
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 18:42:20 2016 -0500

    move the INET related SOL_ options to proto-ipv4

commit 113cb1cb71e595ee8e26ca08fb424ae0f654ffd1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 18:41:25 2016 -0500

    move SOL_NFC local to where it's used

commit 9462dc1d579aa502450cfc24ee5a78522ec49223
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 18:40:57 2016 -0500

    move SOL_ALG local to where it's used

commit 787c960f4fc0231814f290f4a1b116d658d7a608
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 18:40:21 2016 -0500

    move SOL_CAIF local to where it's used

commit b0f40cc30a13c77fb3e35b3329b2cb7f4bc57549
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 18:39:56 2016 -0500

    move SOL_IUCV local to where it's used

commit d79c05a1258ad52c7fc52e34f1226833818da425
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 18:39:26 2016 -0500

    move SOL_RDS local to where it's used

commit 645aa378ce0247df8390115043ebf23e70d33f93
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 18:38:49 2016 -0500

    move SOL_PNPIPE local to where it's used

commit 728abecee0c8da8d5e09fcd0718efa0801c95e80
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 18:37:45 2016 -0500

    move SOL_BLUETOOTH local to where it's used

commit 58c08b0e07116a4183552267fe1e3cdd6369dced
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 18:37:04 2016 -0500

    move SOL_PPPOL2TP local to where it's used

commit ce520196aed53b0e01de9e0de3bc4659c79f0bfa
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 18:36:27 2016 -0500

    move SOL_RXRPC local to where it's used

commit c97a304c679ad1f9cf345fa1cddbee66d46a2897
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 18:35:42 2016 -0500

    move SOL_NETLINK define to where it's used

commit b1985051f32586c53635b5b29e2afdf410c3de3e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 18:35:00 2016 -0500

    keep the LLC/NETBEUI defines local to where they're used

commit f83191770ce06750c1e752ef447aec10acc1e741
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 18:28:55 2016 -0500

    move compatiblity defines out of net.h into compat.h

commit c495e71f2c485755900d471d6cea7329bc77d07c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 18:26:18 2016 -0500

    constify struct netproto

commit 35611a165510e97f78cb19a14b9aa62c034db077
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 17:08:55 2016 -0500

    remove no longer necessary includes

commit 3e03295c7ba304325864b11b0ec02890788058c6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:57:31 2016 -0500

    header cleanup

commit e207f8a7eaa8487cbebf38267a671237f2613b70
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:55:24 2016 -0500

    convert PF_NFC to use gen_sockaddr and remove now unneeded code.

commit 10f69f76860ccfad99569073fdc9eeb0cb00fde5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:52:45 2016 -0500

    convert PF_ALG to use gen_sockaddr

commit 96f55c44f5657f9a5d126e3673896d3843edaa97
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:51:30 2016 -0500

    convert PF_CAIF to use gen_sockaddr

commit 79e7fef94ea1f553ecc8b71c13605cdd5710e08f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:50:42 2016 -0500

    convert PF_PHONET to use gen_sockaddr

commit 77ada88a7a1c0014b6051ab8c53740e4da32b238
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:50:01 2016 -0500

    convert PF_TIPC to use gen_sockaddr

commit 0c18a057be993f0c1cb5df79ed788403f9098609
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:49:20 2016 -0500

    convert PF_CAN to use gen_sockaddr

commit 854ad493a33d192c8a3002f6cec556ff77083c51
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:45:46 2016 -0500

    convert PF_PPPOX to use gen_sockaddr

commit c4aff6f453796498d2309d75e740c14ac8989fca
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:45:00 2016 -0500

    convert PF_IRDA to use gen_sockaddr

commit f3a7941f44ee93cd1eadb8755302b877360484e2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:44:19 2016 -0500

    convert PF_RDS to use gen_sockaddr

commit e44b9257467e4de3fc31b5be155480437257ac9c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:43:32 2016 -0500

    convert PF_ECONET to use struct netproto

commit 9703c3b534c00f05f52a0c62e53e26ad65f171dc
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:41:20 2016 -0500

    convert PF_PACKET to use gen_sockaddr

commit 541d68fa6bf004553386cc6e623ea93555178915
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:40:21 2016 -0500

    convert PF_NETLINK to use gen_sockaddr

commit a378fad7a65f465d67c5e140919d397e53ac2688
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:38:59 2016 -0500

    convert PF_NETBEUI and PF_LLC to use gen_sockaddr
    
    given they both share the llc routine, just fold proto_netbeui.c
    into proto-llc.c

commit d457c5d7a3e72ffb4b13fcc63836e514a0546b2d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:30:28 2016 -0500

    convert PF_DECNET to use gen_sockaddr

commit a4b987ef2ca12d47c5ed13b7930fd08928c268ed
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:29:30 2016 -0500

    convert PF_ROSE to use gen_sockaddr

commit 1895a2c61cace77d654af1cc455b48db260307e6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:28:49 2016 -0500

    convert PF_INET6 to use gen_sockaddr

commit 13bec99131dc25aa33abd6c95fad7a5dc4195f42
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:27:49 2016 -0500

    convert PF_X25 to use gen_sockaddr

commit 2e50eb842b525599880fc937c23481694f6b3978
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:26:56 2016 -0500

    convert PF_ATM[SVC|PVC] to use gen_sockaddr

commit 996ab382600a39a9b3aaca35092f8f2b76f8ac9c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:20:47 2016 -0500

    convert PF_ATALK to use gen_sockaddr

commit d788d4058e5969b481c8eed81cff8da79abf5046
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:19:50 2016 -0500

    convert PF_IPX to use gen_sockaddr

commit cc27d57644e0bb77ce131596ab19ec7a0794043c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:19:05 2016 -0500

    convert PF_AX25 to use gen_sockaddr

commit e899e8e5bc3d9dc5c125795fcf28448731aa54c4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:18:04 2016 -0500

    convert PF_INET to use gen_sockaddr

commit 23e984ac8b090a73534458bd4849fac0dbe02c1e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:16:47 2016 -0500

    convert PF_UNIX to use gen_sockaddr

commit cc72d0c1ed26ef1e4673c3722b65ae6f5455e2d2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 25 16:15:08 2016 -0500

    introduce a gen_sockaddr funcptr to netproto struct

commit 1de59ebe3d216d2d140f6c8dab1bfcf17b3267f3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jan 22 15:43:35 2016 -0500

    make perf fd generation ctrl-c'able

commit 4a93bf0d0c4f564d48770d63434515670893d15d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jan 22 15:37:03 2016 -0500

    consolidate two "print a number" routines.

commit 77ba85be8091b7992b40ed78eddeddff428ccfb0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jan 22 15:36:54 2016 -0500

    change -W to -Wextra

commit 9df7d5a53e7acd69b363ebd8d61861877ee6055a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jan 22 15:21:07 2016 -0500

    Move IP related stuff out of setsockopt.c into proto-ipv4.c
    
    This requires us to pass the triplet down, so that we can figure
    out which IP protocol we're talking, so this cset impacts all protocols.

commit 77a5c19bd21a6007978905e888e8a10be66aca46
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jan 22 14:34:17 2016 -0500

    pass -std=gnu11
    
    Some compilers gripe about the C11 features we were using like
    loop declarations.

commit 71bc8a11c9651c628531c16193f0523368e8c8d5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 17:27:30 2016 -0500

    check that we have a valid socket pointer before jumping to it.

commit 28c139cf1d87de0641ef699e11a93ab8e42c19d5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 17:09:57 2016 -0500

    kill off sso_funcptr
    
    This isn't needed now that we use ->setsockopt through the netproto

commit 96b0b286b38fbb7ba3286aca59575f03dbe39f4b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 16:26:58 2016 -0500

    convert PF_NFC to use struct netproto for setsockopt

commit e3b0902437588949a40fafe075087fa3eb05aee6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 16:25:25 2016 -0500

    convert PF_ALG to use struct netproto for setsockopt

commit d5358fc7d5ee340dd99301056d0fadde3d04abb0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 16:23:04 2016 -0500

    convert PF_CAIF to use struct netproto for setsockopt

commit 5e835913549e5f9efab3d4b90bbc779e3173f6df
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 16:21:33 2016 -0500

    convert PF_PHONET to use struct netproto for setsockopt

commit 05ce6aa75b83224dc5f884b475033e1efa6e30ca
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 16:19:45 2016 -0500

    convert PF_RXRPC to use struct netproto for setsockopt

commit b3ad01d5834b5f2e99eacf2c1628397581e5d69d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 16:17:50 2016 -0500

    convert PF_IUCV to use struct netproto for setsockopt

commit 4be32eab1b2f742c4b2bd55a7111b1ff43056694
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 16:15:55 2016 -0500

    convert PF_BLUETOOTH to use struct netproto for setsockopt

commit 31ca3650cbf320d221902b4c2379183704dad516
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 16:14:18 2016 -0500

    convert PF_TIPC to use struct netproto for setsockopt

commit f9587dac4279100f238c1817563f2ddc50c150a3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 16:13:11 2016 -0500

    convert PF_LLC to use struct netproto for setsockopt

commit 139a81781f33d73a15257072b8782db9b57fa836
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 16:06:53 2016 -0500

    convert PPPOX to use struct netproto

commit 27d0ec7651c04fb83663686cd04785cc1821f605
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 16:03:57 2016 -0500

    convert PF_IRDA to use struct netproto for setsockopt

commit a8da18886466d567ec2bce9aadec32356456c002
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 16:02:38 2016 -0500

    convert PF_RDS to use struct netproto for setsockopt

commit edbed290df479a89d9289ee649ce16a60cc01187
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 16:01:23 2016 -0500

    convert PF_PACKET to use struct netproto for setsockopt

commit 33432afcbceb6a4c5c6a1a151b3fd38aaccf6043
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:59:44 2016 -0500

    convert netlink setsockopt to use struct netproto

commit 50971bc57cd58b105f87e8dbd891ed4bae02ca10
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:58:38 2016 -0500

    convert PF_NETBEUI to use netproto struct

commit b5c1a946d8a6ee7988833662b83ad53c034d42a8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:55:28 2016 -0500

    convert DECNET to use struct netproto for setsockopt

commit 1b9df73b4baf3becf39e99b63bd3e5936038541f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:54:18 2016 -0500

    wire up the ATM proto ptrs to the proto struct

commit 21b503d7f9e86cc75071c0e0b5037b3df1ba703f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:52:02 2016 -0500

    convert ROSE to use netproto for setsockopt

commit a5c1e4122671d04cdba989bb6571d3c4fc04b6c6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:49:25 2016 -0500

    convert INET6 setsockopt to use struct netproto

commit b2cf13a4da947255f4e5796d1d10d08551354f4d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:48:00 2016 -0500

    convert x25 to use struct netproto for setsockopt

commit 0557a9c7822193724db0bbac76c09d69dffa44ad
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:46:55 2016 -0500

    convert ATM to use struct netproto

commit aba6736bb6c96db2d0b9f9304071bb316a93fd13
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:44:33 2016 -0500

    convert netrom to use struct netproto

commit 461a127412250ad0317c8b610c33239b3269771c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:40:31 2016 -0500

    convert ATALK to use netproto setsockopt

commit fc5474b43c350db6de75cfc5e81319c7a6057a17
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:39:09 2016 -0500

    convert IPX setsockopt to use netproto

commit 3b4472030baeb428fcbb54c242ae765f033160a9
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:37:44 2016 -0500

    as we did with ->socket, start converting setsockopt

commit d6874bc68d3255afaeaa7aacc474917f2e55e12f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:22:55 2016 -0500

    remove socket_ptr struct from syscalls/socket.c
    
    Now that we use the netproto structs exclusively, this is not needed.

commit 95c605bd38322335c985fb8fb9a360af2e22722f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:20:15 2016 -0500

    convert PF_NFC to use struct netproto

commit 3234b892e1d9b28a820dae6c0b419c5bfb367e38
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:18:53 2016 -0500

    convert PF_CAIF to use struct netproto

commit e512b99bc8762d0d6d03bdb3796973fb32999199
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:17:34 2016 -0500

    convert PF_PHONET to use struct netproto

commit 2e013525d08ddc96f9dd28af01668af651ab1d6d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:16:04 2016 -0500

    convert PF_BLUETOOTH to use struct netproto

commit 77d82e48d65db69092ba066c55dfc08de073bb74
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:14:47 2016 -0500

    convert PF_TIPC to use struct netproto

commit fda2a99ec1de7e98d48000c27436eaa209ca911d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:13:48 2016 -0500

    convert PF_CAN to use struct netproto

commit d10785f34cc53566724fc506d3b13ce14ea73f66
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:12:34 2016 -0500

    convert PF_LLC to use struct netproto

commit 236721333c77686cf956f1dc8c6bfed4f4c9e02a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:11:22 2016 -0500

    convert PF_IRDA to use struct netproto

commit 0524bbf9a634473ebdb7168e5bf2518c7f543a91
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:10:07 2016 -0500

    add ifdef around ipv6 parts of protocols.c

commit cb82a44c9348a5016c627818907a259d30f19b97
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:09:44 2016 -0500

    convert PF_RDS to use struct netproto

commit 15036f32b8e0152830195f722a7d7acc747a737e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:07:35 2016 -0500

    convert PF_PACKET to use struct netproto

commit 1dd439f9a90e717027d33b28907dd012f0e18e40
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:06:29 2016 -0500

    convert PF_DECnet to struct netproto

commit e56eb9ef3ffc1e1aefe50b1ec737b49eb92c9707
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:05:05 2016 -0500

    convert PF_INET6 to struct netproto

commit b2982041a98dcf277a95437f8706379ce360197b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:03:18 2016 -0500

    convert PF_X25 to struct netproto

commit 546a40fd032fff0b6ca2b8e87fe75a8b6cbcf091
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 15:01:47 2016 -0500

    convert PF_APPLETALK to use struct netproto

commit 3cf0e88a289e9757d953362f66414c70e726d026
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 14:58:43 2016 -0500

    fill in missing protocol ptrs from previous commits

commit f7709c6cc20475002ecd735c007d3549fb1ed1fc
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 14:56:29 2016 -0500

    convert PF_IPX to use netproto struct

commit c891246a9746026a36113b30f7bd05758ddd9dca
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 14:55:24 2016 -0500

    convert AF_AX25 to use netproto struct

commit d21b555f92ebfc2478123230c130a4192b8168c6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 14:53:19 2016 -0500

    convert AF_INET to use netproto struct

commit 8974b95c5109ed15c1d413c09ca16c02d8d28d4e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 14:50:11 2016 -0500

    convert AF_UNIX to use netproto struct

commit c61702e46a522d51b0df27ebf144ceeedc71ad82
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 14:46:38 2016 -0500

    start abstracting away per-protocol stuff
    
    socket.c shouldn't need to know directly about the individual protocols.
    Start by moving socket creation. Initially, just netlink.
    Eventually, things like setsockopt, and more will also move to the
    netproto struct.

commit 321e1793c75b540df8a8aba386d4162b01087a69
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 13:34:14 2016 -0500

    dump the fanotify flags too

commit 0d613839196636908df1089fb7bd910db32d75ee
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 13:11:19 2016 -0500

    fix broken cut-n-paste and add in missing bits for fanotify to work

commit 16334580741f97f8dc46aff8975b5cae2e1bcca3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 13:02:56 2016 -0500

    fd provider for fanotify

commit 1dd8464920b3852a6d90d98038b1865d336bd211
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 12:12:17 2016 -0500

    sanitise the 2nd arg of fanotify_init

commit 654c3e5f2c29cd0bbc69f30faa08c9b2b31a2618
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 11:42:09 2016 -0500

    add a userfaultfd provider

commit f14cd0bf71770f0e438431b37eb7a229972ca155
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 21 11:28:51 2016 -0500

    remove bogus comment. this is actually still true.

commit eb3563b0a214648601475b0d9103b3987a95234f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jan 20 22:11:55 2016 -0500

    add recent socket setsockopts

commit 2dc90ccfe3f7d2d6897f8dc4c36d6472b4b0882b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jan 20 22:07:33 2016 -0500

    add MADV_FREE flag

commit 9199173ac99b3b9c293ad6ee193f207a3eeddb47
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jan 20 19:22:30 2016 -0500

    the 'stop' argument to handle_childsig is a boolean

commit 6f51b21f17351969a1545dba3022f42788e3bb69
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jan 20 19:19:49 2016 -0500

    remove childpid parameter from handle_childsig

commit e25e11f227b2b9b13dccb082056bda2fae172e86
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jan 20 19:13:39 2016 -0500

    pass the childno down to handle_childsig
    
    rather than recompute it.

commit e9c4c0fc48c7d1835fccf75d10213417799d19b6
Author: Eric B Munson <emunson@akamai.com>
Date:   Wed Jan 20 14:51:12 2016 -0500

    Add MCL_ONFAULT to mlockall calls
    
    The new mlock2 system call was added, but it looks like the MCL_ONFAULT
    flag was missed.  This is now a valid flag to pass to mlockall(), but it
    must be used in conjunction with either MCL_CURRENT or MCL_FUTURE (or
    both).  Also rework the way that the flags value for mlockall() is
    caclulated.  It is valid to pass any combination of the MCL_* flags
    except for MCL_ONFAULT alone.
    
    Signed-off-by: Eric B Munson <emunson@akamai.com>
    Cc: Dave Jones <davej@codemonkey.org.uk>

commit d7907a0fc3e92f5e73cbdd6133d790920a35d8e4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 19 14:02:39 2016 -0500

    bias all three sizes in struct size generation equally
    
    Before, we favored putting them in the evaluation order.

commit 751cb910ed90cbe2b1d55fb7a7f6990575463109
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 19 13:54:10 2016 -0500

    allow rand16/rand32/rand64 to return RAND_BYTE

commit 28e2045d335cce4ee8eeb9f6865b06852f783e1b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 19 13:47:57 2016 -0500

    let rand64 call rand16
    
    and also call rand32 less often

commit 0ac9ed7a46a71111bef39c703232f395f5418ad2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 19 13:45:48 2016 -0500

    make rand32 sometimes return rand16 ranges

commit f52472c13e81d2d5c574b9b0440e28a444e1516c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 19 13:44:10 2016 -0500

    masking rand16 with 0xffff is the same as doing nothing
    
    change it to mask with 0xfff instead

commit 239063f979f08a3e2a172c2be781a07408be3998
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 19 12:55:46 2016 -0500

    fix rand16 to not shift out of range.
    
    It also didn't match the original intent of this case

commit 1940c04f896801a18fb5fcdbd1fec9f25012c61a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 19 12:42:51 2016 -0500

    fix stupid off-by-one

commit 50e6817b833944997d07997fc4163384f342a7a5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 19 12:35:47 2016 -0500

    case 7 can be up to 20 chars

commit fa58171f9d62cc56cc78b161721d61a95b1ca1b8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 19 12:35:07 2016 -0500

    use same null-terminator as elsewhere

commit eaa777f7030dfcdbe34d781100470d437b4117c6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 19 12:30:28 2016 -0500

    The higher page-scribblers only work with an even number of bytes

commit d79d06d2733c097acf77533d0bf7d24296a324de
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 19 11:54:44 2016 -0500

    more random 'random struct' generation.

commit 08b603f4fc8c3bfd8e530f45321780762f3ec8b5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 18 21:16:59 2016 -0500

    make struct fabricator work with less than a page

commit bc279bb3e97c36bad6d852731f16567d66bcd443
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 18 10:34:06 2016 -0500

    check buffer length before trying to do some of the longer generators

commit 0347350352967f862e602912d6c98b9d792db496
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sun Jan 17 22:06:14 2016 -0500

    merge generate_rand_bytes and generate_random_page
    
    More work to do here, probably broken in some cases, but it's a start

commit c919a75d6b6cd5699b4cf9853d21de8c02663cb0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sun Jan 17 21:44:05 2016 -0500

    consolidate "scribble stuff into a page" routines

commit 8cdbc6c686d1839924a02d4061b1e2da5afb476a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jan 15 15:48:57 2016 -0500

    fork new children as needed.
    
    This way we keep closer to max child count at all times.

commit 22c1567e3f7ddfe5d1abd9e9f028129240c65ae4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jan 15 13:14:33 2016 -0500

    spawn_child return a bool, not -1

commit 9b66709816d44224707c361fa7d87a9af62e8b27
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jan 15 13:11:03 2016 -0500

    update the pids array, not the local copy.

commit 8d195e3b7a8f8da02f027b2dadcde9b7541b9c91
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jan 15 12:47:53 2016 -0500

    use sigtimedwait with a 1s timeout instead of sleep(1)

commit 360686ac3d6f8973109c3ce11bd3a59daf95c4c7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 14 13:35:41 2016 -0500

    pid caching for the child code too

commit b9eda9d7e705e164e75fcefd779ff58c8c86a632
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 14 13:32:57 2016 -0500

    cache pid locally in various places instead of re-reading from pid array

commit b5174a416b6e64e62383b4ee5805108e25a78049
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 14 13:27:26 2016 -0500

    remove clearing of pid entry from clear_childdata
    
    This doesn't actually need moving, it's done by reap_child now

commit 61483a5ccfadecf9c727705388bc7611c069d81b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 14 11:57:27 2016 -0500

    move the pid back out of the child structure
    
    By reintroducing a global pidslot array we can do better locking
    when searching for EMPTY_PIDSLOT.  This isn't so much of an issue
    right now, but will be soon with some upcoming changes.

commit 22b1e304c0d8e4a374002be29b67e0df4ce8daf2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jan 13 12:18:41 2016 -0500

    add copy_file_range

commit a02f722dd1d3dba1506c34f8f857eecf4dbe0016
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 12 21:04:10 2016 -0500

    reset stall count before each recount.
    
    When this became a global I forgot to add something to clear it each time
    we do a recount.

commit 0fb3aa5a5ac78fa894092e8dc24926c323c50bf2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 12 16:30:14 2016 -0500

    use a better exit code when we fail sanity

commit d880a059ddf046eabcdff0fee791b6168ff9f1af
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 12 16:24:16 2016 -0500

    remove the lock-stealing mechanism
    
    This was a bad idea, and it's now unnecessary (hopefully)

commit fa8118572cdbc83fc2a154d6da4837a78b65396c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 12 16:20:31 2016 -0500

    remove the child sanity check for the tvsec progression
    
    Let's just do this in the main process.
    (It also fires too many false positives in the child)

commit 437400ba75c3e24888b9831d90f3f2a04af08966
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 12 16:17:55 2016 -0500

    count SIGXCPUs and exit after we see 100 of them.

commit 005f917e0a033e8c419d1dff0f46430a7d5825d9
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 12 16:10:39 2016 -0500

    move responsibility of the clean out of the childdata from the child to main

commit c6119c1e6057bb67fc0032df068a46a502f6cf72
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 11 22:47:34 2016 -0500

    rewind the getline file pointer each time we read it.

commit e15055aaaec0594c4051c0eb99d6cd03503cd978
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 11 16:04:49 2016 -0500

    initial seeding cleanup
    
    - remove workaround for no-longer-around init process
    - main seeding now happens in init_shm

commit 6e94a0ae7369f566431aebd2dd352bba2a0c4f83
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 11 15:22:13 2016 -0500

    free the line after we parse it from /proc/pid/stat

commit 5932fddbc3ef0d8d55d89fa425d5fd15ebabbbfb
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 11 15:21:40 2016 -0500

    cache the /proc/pid/stat fd for the lifetime of a child

commit b9dcd3ab94fa2616cff6312e98e6bc5a4fbcdc2a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 11 14:53:31 2016 -0500

    allow socket usage when using victim files

commit 0f30f4323703295b6586d3d9fdb9534a2a8c511e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 11 14:51:31 2016 -0500

    remove old FIXMEs

commit 2c3cdc0d60ab2593f4ce0876656b1a5e770f8ae0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 11 14:45:20 2016 -0500

    remove unnecessary print now that we only use one process

commit 5f22207c889140d9c0b15bae1ba50409ef849bc5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 11 14:43:56 2016 -0500

    remove duplicate set_seed(0) call
    
    left-overs from separate init/main process

commit a4680bfd693a62a2b73e35ebba97e1b2710699e7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 11 14:39:50 2016 -0500

    if we detect a stuck syscall blocking on an fd, don't reuse it.

commit ec6181ed36bc6f5e4f0d36162e625aa4496059be
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 11 14:36:22 2016 -0500

    free pathnames post syscall

commit edd684339ead00bd262f9564fd7cdb5d5337143f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 11 12:38:50 2016 -0500

    check_child_progressing actually checks all children
    
    rename it.

commit d84490719df6dcafc3d6425bbac55e8c2b08e9a6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 11 12:38:31 2016 -0500

    move statistics printing out of check_child_progressing

commit 01cf58f0254a63b2ebdbc1a9698d587f127bdabf
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 11 12:10:33 2016 -0500

    remove unnecessary clearing of mainpid

commit ba3605750d0b8da56aab1d35dcd3c4776cf9ccfc
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 11 12:09:38 2016 -0500

    kill off initpid
    
    This is unnecessary now that init and main are the same process.

commit 8e0a313823cd01eb8acda8a32ca17e94f26c9bef
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 11 12:07:52 2016 -0500

    mainpid doesn't need to live in the shm any more
    
    Everything is now a descendant of the main process.

commit 8f0922f2da0c54b457d7cbab283b22513e62de1e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 11 11:52:43 2016 -0500

    Remove last_reaped
    
    There shouldn't be any way we can concurrently reap the same pid now.

commit ea345d8222946a8608ebd8320d4f7b0c769b0218
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 11 11:51:40 2016 -0500

    kill off the reaper lock
    
    This used to provide mutual exclusion between the watchdog process
    and the main process. Now that there is no separate watchdog process,
    this lock is redundant.

commit 8a6cb732eb8a869af9a3bb30964d2c515dd7769d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 11 11:24:21 2016 -0500

    split out a function to create a single child

commit 0c57388006ecbca98de5ddc6841a06a596bd3603
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jan 8 14:04:06 2016 -0500

    move fdstr clearing to later

commit 351650bb8411945fe0015f5b779c442bf8151e54
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jan 8 14:00:15 2016 -0500

    scope reduction

commit 9e504fe107ad068ec3fda3495ed2cc76101d3eee
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jan 8 10:41:27 2016 -0500

    don't bother SIGKILL'ing processes stuck in D state

commit 42ccfbcca651c9f4a9dcb3a3d0f89fe3e52539a3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jan 8 10:30:42 2016 -0500

    When reaping a child, release its syscall record lock.
    
    Not strictly necessary, as it would get cleaned up by check_all_locks()
    from the the main loop, but this cuts down 2-3 lines of log spew
    every time a child segv's while holding a lock.

commit a0db69f65c558d2684a0c2729b3b2535e37e7d84
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jan 8 10:27:38 2016 -0500

    make stuck_syscall_info non-blocking.
    
    A side-effect of removing the watchdog process is that now main_loop
    and all the functions it calls can't ever busy wait, or we get child
    processes stuck in zombies. This is catastrophic if they were holding
    a lock and segfaulted.

commit 7aa8cfc9524f61cfebd2846de6775eeac6210e75
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jan 8 10:26:05 2016 -0500

    introduce a trylock function

commit 884eafbcbe8e116327f84283ab0e9cbf9057050b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jan 8 00:20:41 2016 -0500

    syscall32 needs the shm writable.
    
    also, narrow the period that the shm is RO around 64-bit syscalls.

commit 1b1d784774369a43edcec2f8b94d776db529450f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 7 16:51:02 2016 -0500

    remove unnecessary break's
    
    With these in place, the *p=0 line is never reached.

commit 442fa4e1f9091c18c80d656a0dcf92eb5a46ccc3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 7 10:39:24 2016 -0500

    Drop the futex debugfs frobbing for now.
    
    It causes a ton of spew to stdout, and can't actually work
    unless you're running as root, without --dropprivs (which is a
    terrible idea).
    
    I'll add a root process later, and a means for children to communicate
    requests to it to do things like this on its behalf.

commit 44bac396a50b81319d41db0356b16b0900292d47
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 7 10:19:13 2016 -0500

    Remove the watchdog process.
    
    Have the work it used to do done by the main process instead.
    There's still room for improvement here, but this is a big step
    forward in terms of simplification, and allows a whole bunch of
    code & locking to just go away.
    
    There's also no longer a good reason to fork main_loop() as a separate
    process, so now we're back to a single process (for now).
    
    A big change here also is the 'reaping' of processes. This is now
    only ever done by the main process. The child no longer cleans up
    after itself (ugh, kids eh?) before it exits.  There's no watchdog
    process, so we can probably remove the reaper_lock now as a follow-up.

commit 3247f5c5b185d6a202e80144ddd1facba4ce958e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 7 10:10:38 2016 -0500

    dump pipe object info on startup

commit f21238a05cbf7ccb2464cbef0f41363e9fa51f7c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jan 6 20:09:24 2016 -0500

    Change main process so that it no longer blocks waiting for signals

commit 96c4eb08d6d60a2651fe9276a27592b3bb25cc9a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 7 10:22:36 2016 -0500

    accumulate the check_lock results instead of over-writing each time.
    
    Typo introduced in previous commit.

commit 962f70d6fde58f202fc85339d18acb4563f88cee
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jan 6 21:44:11 2016 -0500

    remove leftover debugging code.

commit 14655b6524145b70c0a0649b762eda43dd18b7c2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jan 6 17:22:43 2016 -0500

    add a dumb syslog() wrapper

commit 69b42fa0cd99d776a995449da8cfbb8c1a2d7569
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jan 6 17:18:55 2016 -0500

    remove unnecessary include

commit 8879ece3b24c38bea78813d48a91f733dbfc4039
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jan 6 17:10:08 2016 -0500

    move the resetting of a pid to EMPTY_PIDSLOT later.

commit 3c8e5e5ca3ca893bdd54bb947b4f3560ef53b98c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 5 22:32:46 2016 -0500

    start numbering testfiles from 1, not 0

commit 1620b181c8f73d019628b388295e8c274b1dd2d0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 5 18:17:05 2016 -0500

    two-step locking.
    
    There was a tiny race where a lock would show as LOCKED, but hadn't
    had its owner set yet. This would manifest by the watchdog finding
    locks owned by pid 0.

commit abcf1017d4ebf898ee44df3b8508afccba51a149
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 5 14:08:22 2016 -0500

    change the lock-checking to return a bool

commit fee8277677dcfb6d48cc56dae1748c54f50a2e90
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 5 01:13:19 2016 -0500

    signedness fix

commit 166a433d6816dff42a06ce5b3ad92706bc7a4d72
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 5 01:12:33 2016 -0500

    signedness fix

commit 3eeba24232bdabcc7a4871686ae134217c16788d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jan 5 01:12:03 2016 -0500

    rewrite the random ascii number generator to be simpler

commit cca66d7ede02156f85a8536fbf5a1657b834a582
Author: Jiri Slaby <jslaby@suse.cz>
Date:   Mon Jan 4 12:21:31 2016 +0100

    Makefile: fix build with older glibc
    
    man 2 clock_gettime says:
    Link with -lrt (only for glibc versions before 2.17).
    
    If we do not do that, we get:
    post-mortem.c:61: undefined reference to `clock_gettime'
    syscall.o: In function `__do_syscall':
    syscall.c:121: undefined reference to `clock_gettime'
    watchdog.o: In function `is_child_making_progress':
    watchdog.c:317: undefined reference to `clock_gettime'
    
    So introduce (standard) LDLIBS into Makefile and add -lrt there.
    
    Signed-off-by: Jiri Slaby <jslaby@suse.cz>

commit 4d6144c53987566819959ad5fdb5be7b569e0a54
Author: Jiri Slaby <jslaby@suse.cz>
Date:   Mon Jan 4 12:21:30 2016 +0100

    compat.h: add more macros
    
    Some more RTNLGRP macros are not defined on SLE11 yet.
    
    [v2]: NVME_IOCTL_RESET is not defined on openSUSE 42.1. Include
    compat.h in ioctls/nvme.c.
    
    So define them in compat.h.
    
    Signed-off-by: Jiri Slaby <jslaby@suse.cz>

commit 6050c1cc3e09d6d99f0f61183fb8da9e1bc20086
Author: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Date:   Wed Dec 30 02:36:12 2015 +0300

    convert syscall time tracking to CLOCK_MONOTONIC
    
    Commit e8646038d918 ("add some extra sanity checks inside the child
    process") made trinity near unusable for me: the rec->tv sanity check
    fails rather quickly after first syscall capable of changing wall time.
    
    The patch reworks trinity to use clock_gettime(CLOCK_MONOTONIC) instead
    of gettimeofday(). It makes trinity usable again.
    
    Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

commit 40c64904cc3e24f41adb7ca0ad948bcfb9c0ee9e
Author: Davidlohr Bueso <dave@stgolabs.net>
Date:   Mon Dec 28 08:17:06 2015 -0800

    futex: fix uaddr passing when dealing with lock
    
    We are working on the address of the futex word in our
    local lock structure. As such, we are currently passing
    it a completely bogus params for uaddr and uaddr2, which
    can be seen with the amount of unsuccessful futex calls.
    
    Also get rid of printing the addresses in the mmap'ed
    playground when creating our futexes, just show how many
    are created, no one really cares about these specifics,
    which can be seen/studied in the logs anyway.
    
    Signed-off-by: Davidlohr Bueso <dbueso@suse.de>

commit d987efb5bc9ba23e4f8117d00431be00803030a2
Author: Davidlohr Bueso <dave@stgolabs.net>
Date:   Mon Dec 28 08:17:05 2015 -0800

    futex: fix private object segfault
    
    When doing a get_random_object() lookup, we can
    dereference a nil pointer in the local list. As
    such, always use the global variant to avoid
    crashing the child.
    
    Signed-off-by: Davidlohr Bueso <dbueso@suse.de>

commit 1ab8aa36dae3ff8af304292b312938fddf718602
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Dec 23 17:16:25 2015 -0500

    check for empty pidslots in stall_genocide
    
    This was a pain in the ass to track down. We were sigkill'ing
    a pid of -1, which took down everything.
    Amazingly, this has been there for six months, and only recently
    became a problem.  This is because children previously weren't
    running long enough, or getting stuck enough for this situation
    to arise that often.  With the recent segfault fixes, this
    function gets hit a lot more.

commit b4f40a2d9b83b258abceed9078e7f16f8d6dd8e7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Dec 23 13:56:32 2015 -0500

    don't try to destroy empty lists

commit 37171f1c9139739dc683c78c7caa6535210d3f6e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Dec 23 13:36:20 2015 -0500

    fix divide by zero in get_random_object

commit a9cdb6593eea88f54bc3f099144ffefbe6603100
Author: Davidlohr Bueso <dave@stgolabs.net>
Date:   Tue Dec 22 18:16:07 2015 -0800

    RTNLGRP_NSID opensuse build fix.
    
    Signed-off-by: Davidlohr Bueso <dbueso@suse.de>

commit 7372566441a515ebee66c5fa66089e679abfeb47
Author: Davidlohr Bueso <dave@stgolabs.net>
Date:   Tue Dec 22 17:56:11 2015 -0800

    futex: fix post call make-it-fail
    
    ... the purpose is to clear the option.
    
    Signed-off-by: Davidlohr Bueso <dbueso@suse.de>

commit 42de68f83cc32a80e6af4f00764da40620d57942
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 21:24:49 2015 -0500

    improve on the nvme.h test
    
    Some older versions of this header lack the ioctl definitions.

commit 708719ec8de89221bb71b201aaa26812f37497d8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 20:44:39 2015 -0500

    old habits die hard

commit f7346ecac3c8db036bdcd77eec89c372c2930b65
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 20:11:17 2015 -0500

    RTNLGRP_MPLS_ROUTE opensuse build fix.
    
    pointed out by Davidlohr Bueso

commit 0774c700ed4da79e876b78788c2d21ee7f1b72c5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 19:43:49 2015 -0500

    argh, forgot to add include/futex.h

commit 662b80805ecf2cea1586068722cdc74b5bfd33d8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 19:43:15 2015 -0500

    more scope reduction

commit dc26f8e31e59ff5dd6c7da7d5f16e17d6cbb2d34
Author: Davidlohr Bueso <dave@stgolabs.net>
Date:   Tue Dec 22 13:14:25 2015 -0800

    futex: convert to object cache
    
    This allows better handling of the uaddr parameters we pass
    to futex(2). As such, teach the object structure about the
    internal lock primitive. We have two alternatives now, which
    replaces the current use lock or whatever trinity gives us
    as addresses. Now we continue using the lock half the time,
    but make use of the mmap playground already available otherwise.
    
    Signed-off-by: Davidlohr Bueso <dbueso@suse.de>

commit dcdcb333038e274400ad464178dda8c3ddbbdeb8
Author: Davidlohr Bueso <dave@stgolabs.net>
Date:   Tue Dec 22 13:14:24 2015 -0800

    futex: trigger error injections
    
    Make use of futex support for artificially triggering errors
    within the various futex paths faults and deadlock scenarios,
    via the CONFIG_FAULT_INJECTION kernel framework along with the
    CONFIG_FAIL_FUTEX option. Trinity can make use of this feature
    by randomly toggling the process' make-it-fail file as well as
    selecting appropriate fault injection debugfs options.
    
    Signed-off-by: Davidlohr Bueso <dbueso@suse.de>

commit 2b2e87e92a86d0720f572faae744ba345bc0496f
Author: Davidlohr Bueso <dave@stgolabs.net>
Date:   Tue Dec 22 13:14:23 2015 -0800

    futex: FUTEX_WAIT can also use FUTEX_CLOCK_REALTIME
    
    FUTEX_WAIT is treated as a simple version for
    FUTEX_WAIT_BITSET internally (with a bitmask of
    FUTEX_BITSET_MATCH_ANY). As such there is no reason
    for regular wait calls not to make use of it. Update
    trinity to reflect the new kernel logic.
    
    Signed-off-by: Davidlohr Bueso <dbueso@suse.de>

commit 25bae91882def7ea536e6b9ac33bcc834766f6a5
Author: Davidlohr Bueso <dave@stgolabs.net>
Date:   Tue Dec 22 13:14:22 2015 -0800

    objects: rename no_objects
    
    ... and trivially simplify it while at it. no_objects()
    is slightly confusing regarding the name and I believe
    changing it to objects_empty() serves better.
    
    Signed-off-by: Davidlohr Bueso <dbueso@suse.de>

commit 8e3e294003524516f86dc7dbb8d39225283e9515
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 14:54:00 2015 -0500

    scope reduction

commit ada55975ec2542cbb40395e64adda6cf257a0410
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 14:50:34 2015 -0500

    move all object creation/destruction to main out of init.

commit 8a31ab2b74613c825d50d6da9e6c66e9ab98705b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 14:47:52 2015 -0500

    move the socket shutdown code into destructors

commit 82eb06fbfdb2fc9bb16e92fe533cfa99735296c0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 14:41:00 2015 -0500

    make the mmap dumper generic, and move to objects.c
    
    This is useful for debugging. I'll clean up the TODO's later.

commit 6a5822c54a754b7a477fd5fecbc309d300edb38c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 13:30:05 2015 -0500

    wire up the pruner to potentially trigger when we add an object.

commit 039838dcc2652f1dd752cf774e1e9ed69f39b9d1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 13:22:23 2015 -0500

    use get_objhead instead of open-coding

commit 1fb5e8bc1cbbbaaa10e246ee41e71689854e2816
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 13:19:28 2015 -0500

    child list creation can inherit destroy/max from global list

commit d9af62f2e26d62a9dcd2b83bc415c705f05ab754
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 13:15:54 2015 -0500

    move initialization of child lists into init_child

commit 2fe3df81b716b90bddbed468fef6185aed5280e7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 13:14:00 2015 -0500

    only prune lists that have reached their max length

commit d490294726da1796fddb06be416cc64bbc380452
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 13:10:44 2015 -0500

    genericise the destruction of global objects

commit 45ab7def1ba15cd4b6bd28c8440e00b47282edc6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 13:04:20 2015 -0500

    simplify prune_objects, have it iterate over all lists.

commit 8ac58621178dc21fb8f5864eeda1a58b53ebe7d9
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 12:58:52 2015 -0500

    max_entries = 0 means "don't prune"

commit 65e18b2be8369646dc01f98e373b79213cb2fe16
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 12:48:17 2015 -0500

    add a simple garbage collector for old fd's etc.

commit e11d73682e11c36746b79e8be49f4bd56b623bf6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 12:05:06 2015 -0500

    make destroy_initial_mappings more generic

commit a8aa7dcfd456b3521e8d2c675056102302e846c9
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 12:00:34 2015 -0500

    share the mmap destructor across local and global mappings

commit eb1ccbbaa6ae167f025e11268d6837f5e9dd75f5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 11:57:25 2015 -0500

    mmap destructor needs to munmap the mmap

commit 2661275b48b52643d63cf92bf5f583acfead115f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 11:56:46 2015 -0500

    use the mmap object destructor instead of open-coding it

commit 0b9dc1fff7d509e07c6c89251d8fb83b1863636c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 11:51:00 2015 -0500

    remove print spew from alloc_zero_mapping

commit e748e92bbc09c3cec39f0542540f1a8cb016c789
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 11:47:31 2015 -0500

    use alloc_object for all allocation of objects.

commit 158f0829f7197dea001183c735e771c10a65e76c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 11:43:30 2015 -0500

    add destructor for timerfd's

commit a022ae05fc4e05e2dce92bd37f6f672ca749f8bf
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 11:42:24 2015 -0500

    add destructor for testfiles

commit 53713e12aa9f2e997ce3be788003b91d82a43dbc
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 11:40:25 2015 -0500

    add destructor for sockets

commit 0f65e00638717bcfb6b1f7ff77076f4d8e17c19c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 11:36:47 2015 -0500

    add destructor for pipe fd's

commit 5b7754032a22d1c1521e49683150ae7a24e51a5a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 11:35:16 2015 -0500

    add destructor for perf fd's

commit 8cc9d3e68786374bbd87b462b8ed6e4ed9a8d29a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 11:34:05 2015 -0500

    add destructor for memfd's

commit 20da7fa85fb5476ea88dee90601b8a9422a2669f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 11:32:15 2015 -0500

    add destructor for inotify fd's

commit 04caa1c02d9d6bfa9fbb1765b558773b84854030
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 11:30:37 2015 -0500

    add destructor for filefd's

commit f71a1254e9e21a62448d2bbe6622a861c2867cd3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 11:27:31 2015 -0500

    add a destructor for drmfd's

commit e6ef3e03e14f18499ceb6d1641b9e639b405f256
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 11:25:49 2015 -0500

    add a destructor for eventfd's

commit 0e0cdce6106bd9105679f2ddfd6fb677eb96c98e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 11:22:39 2015 -0500

    add a destructor for epoll objects

commit d074e91c4cae4863084f8ed0f922aa6075490eb2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 11:22:25 2015 -0500

    export get_objhead

commit 12d85444984a9aa85d7f0e6a3f02c79f2e091f3f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 11:05:22 2015 -0500

    the busy-wait on watchdog startup is actually waiting for the shm

commit bff28f841dcc492e717e557f42c3cba1255ab351
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 22 00:09:17 2015 -0500

    split up watchdog() a little to move out the init code

commit eb75dedc7f9952da5ccc66947a63ebba60c7cbe7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Dec 21 19:53:19 2015 -0500

    store inotify fd's created by children in child local list

commit 0734374379d0113ae5492a577623405e38c941e5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Dec 21 19:49:42 2015 -0500

    mark return type for inotify init variants

commit ce8c2337e5be92f326dd1a3c4e7bb870dec6a61b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Dec 21 19:48:59 2015 -0500

    merge both inotify_init variants into same file

commit 16e0722ed99e8ee93d48a2ad0b68a5fb39f66927
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Dec 21 19:47:45 2015 -0500

    store eventfd's created by children in child local list

commit 14bfbbe9e4133e13e32dd142fb90bc9915270a8b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Dec 21 16:24:46 2015 -0500

    store epollfd's created by children in child local list

commit 24de470e4629b2547ea23319bbd5a592a6c4cda7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Dec 21 16:20:57 2015 -0500

    store fd from epoll creation in the right object cache

commit 84f895774ff39c268af70cc80df5a7d55584691b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Dec 21 16:06:54 2015 -0500

    check for presence of linux/nvme before using it

commit e59291aed58602c402226549f17ced530bffb13f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Dec 21 15:52:47 2015 -0500

    assume untainted if we fail to open the tainted sysctl

commit 4fa673ed193b11819f275ad15c9e51b0a09b92f6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Dec 21 15:20:38 2015 -0500

    set the child number earlier.

commit ad0643f59c5e01f676d1d056dfb189f72fa86cd9
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Dec 21 15:05:14 2015 -0500

    remove the 'previous' syscall logging
    
    It's never been particularly helpful, and it's use will be obviated
    by the upcoming logging rewrite.

commit fa60add5242757d0449bbf9f55a7bb5366db94cc
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Dec 21 13:13:59 2015 -0500

    creation of the child logfiles must happen in main pid.
    
    otherwise the watchdog can't fileno() them.

commit 397961a569231c73259131572cba531a179033b0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Dec 18 20:48:14 2015 -0500

    improve msgrcv sanitise
    
    mark as NEED_ALARM, give it a non-null address, and set a smaller len.

commit 1cb54a7856dbfcb977350c38cf0e95c9942897e0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Dec 18 20:46:50 2015 -0500

    ignore rt_sigsuspend. It's boring and just causes the child to hang

commit e5a5e576d8a8d083fc28e524efb36239e8f149d2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Dec 21 12:59:28 2015 -0500

    remove noisy output from sigreturn

commit 4a302e24b9b334c9e9a8d94b39e78a15c4b8528d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Dec 18 19:03:26 2015 -0500

    fix up straggler rand() -> rnd() conversions

commit 3f5eb372334a8c968554f8ee227aeee52f935e8a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Dec 18 18:59:33 2015 -0500

    get rid of the RAND_32 / RAND_64 macros

commit c0c5d9edf3595bcba9950c123548cf189cb51330
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Dec 18 18:35:44 2015 -0500

    I missed include/ in the rand -> rnd conversion

commit 663e440324176f7da04df9746220e037f4ba2f0a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Dec 18 18:35:09 2015 -0500

    add missing includes of random.h
    
    These all use RAND_ARRAY, so are pulling this in indirectly.

commit 9952a51ded92d6e14972e1d2604f02a35dc7b894
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Dec 18 18:29:44 2015 -0500

    simplify futex random array usage

commit 6ad0eb79ea25863d8c503b077fa2716657dbfa57
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Dec 18 17:53:52 2015 -0500

    main process debug message cleanup

commit 94fd4ea839e32832137ed5e218d062051208d3e0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Dec 18 17:41:15 2015 -0500

    Disable use of 1GB pages again
    
    Need to rethink this, as this causes a number of problems.

commit bf347f38a2599b88ee1887126f1a5ba964a9b7b1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Dec 18 13:16:50 2015 -0500

    convert all rand() calls to rnd() wrapper
    
    This does nothing right now.  It'll make life easier later when
    switching to thread-safe variants of rand()
    
    bonus side-effect: it shuts coverity up forever about rand() calls.

commit f9811ed592d04c35c12a6b03cf05c31005f603c0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Dec 18 12:09:56 2015 -0500

    now get_address can pick random mapping sizes

commit 46387deeb82a71a522b249a81cabc3e646b11b2e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Dec 18 12:08:57 2015 -0500

    move mapping size calculation out of syscall into startup code

commit d184e9ac9d3599e740c92bdb4c262ce3959a9b92
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Dec 17 17:46:23 2015 -0500

    further simplification.

commit e142f7ad926a41f015e1250601361a0dc28222e0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Dec 17 17:38:18 2015 -0500

    simplify get_address to the more useful cases

commit 174bfd24b3fd08701d62a8cd62039a6db872cb06
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Dec 17 17:14:11 2015 -0500

    scope reduction

commit 1069bf0182dc36be03dabc96d5011872a2fcf80a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Dec 16 14:13:06 2015 -0500

    cache the child struct ptr in the child main loop

commit 4f37acd2ec57860a3c7f6887336ebc2e47f0fd4d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 15 18:42:31 2015 -0500

    add a function to speed up bluetooth socket generation.

commit 848052ec421270753af95f091f9d1c82c64156da
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 15 18:41:15 2015 -0500

    factor out common code to obtain the right objhead ptr

commit 96f7375f1de9386dee01378b68a8087119e94dbf
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 15 18:37:01 2015 -0500

    convert this_child to be a function instead of a global var.

commit 41d6098aaffbe511c133fa0704706bf38af40f8c
Author: Johannes Thumshirn <jthumshirn@suse.de>
Date:   Tue Dec 15 15:19:04 2015 +0100

    nvme ioctl support
    
    Add support for NVMe ioctls
    
    Signed-off-by: Johannes Thumshirn <jthumshirn@suse.de>

commit 1ccd4a9ae934aa1985c93069ac0e6faef53db35e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Dec 14 16:16:19 2015 -0500

    when generating ipv[46] addresses, use localhost half the time.

commit 1f0c06495b9c19797a189a693c34adb57430cbb8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Dec 14 14:46:54 2015 -0500

    move pathnames code out of fds as it doesn't provide fds.

commit f5ef75e73d3b5a9549da411dc8efa9d0d672daa3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Dec 14 14:45:38 2015 -0500

    move generate_pathname into pathnames.c

commit c8113d10695983957a1738dbb2b9029404709a3f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Dec 14 14:44:13 2015 -0500

    include file cleanup

commit 437be860a28e37837598f87e025cdc3468948946
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Dec 14 14:38:24 2015 -0500

    separate the 'generate filelist' from 'open fds' code.

commit e3845568c4887393c0144f822b239311d4bc7392
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Dec 14 14:27:02 2015 -0500

    init the list head before adding it to the list.

commit 41a79274e79e0f5524e765a750fd3f1c255b845e
Author: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Date:   Mon Dec 14 14:51:54 2015 +0000

    mips: fix prctl's
    
    It looks like a typo in the syscalls/prctl.c file. It's lacking the PR_
    preffix, and due to that is causing build failures like this one:
    
      CC    syscalls/prctl.o
    syscalls/prctl.c:37:2: error: 'GET_FP_MODE' undeclared here (not in a
    function)
      GET_FP_MODE, SET_FP_MODE,
      ^
    syscalls/prctl.c:37:15: error: 'SET_FP_MODE' undeclared here (not
    in a function)
      GET_FP_MODE, SET_FP_MODE,
                   ^
    
    Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>

commit c8ea635daff0137d680918e811bd522454af7c58
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Dec 11 20:52:21 2015 -0500

    wrong indentation hid a bug that would cause early exit.

commit f780b9d336731ac01040b7f2143323691deb60c2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Dec 11 15:44:14 2015 -0500

    factor out the code that determines which syscall to run

commit 976589bcd4cb9a90d2cb8b1055025de30db81a9b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Dec 11 15:35:07 2015 -0500

    factor out the check sanity code

commit e4c8f48612ddbe6bfc81e0bdbf351cafee899c76
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Dec 11 15:19:57 2015 -0500

    clean up the post-syscall sanity checks a little

commit c449703bf6ae6c8e794a4421b5203b073f1b0269
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Dec 11 15:11:30 2015 -0500

    indentation fix

commit fdbff31e49ede02f1aeb8254117f4f47c45e1b99
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Fri Dec 11 00:03:43 2015 -0500

    update perf_event_open support to 4.4-rc4
    
    Update the perf_event_open support to match the current
    4.4-rc4 kernel.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit f5d9678e8467617b2926529bf391ac52a182464e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Dec 11 10:18:26 2015 -0500

    use rand32 instead of rand() for alg feat/mask

commit 3bffb03819734f8550ed9a064ab7a40980038e7d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Dec 10 23:44:35 2015 -0500

    fill out the nl_groups field correctly for netlink sockaddrs

commit 48b747dd8ee9c5d3681a76829bedcf65dc6611ec
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Dec 10 23:08:36 2015 -0500

    more alg_gen_sockaddr improving.
    
    duplication, but when the big struct gets split up, this will
    make more sense

commit 9671fe1600b99781e9eb760c97e787f050176030
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Dec 10 22:58:07 2015 -0500

    reduce duplication in the check_if_fd function

commit 9432aa2e2d063f73207890366f1d5f36d213c7f7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Dec 10 22:54:15 2015 -0500

    first pass at getting all the algos in for PF_ALG
    
    This is a mess, and needs cleaning up, but.. later.

commit b8ff7235ca316f47b8a3a55fc99effcda4f4f643
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Dec 10 22:53:41 2015 -0500

    remove all the child related stuff from sigalarm handler

commit 8811d7f3710e307a8b056d7ff66f93ab70bdb0a6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Dec 10 22:52:22 2015 -0500

    check_if_fd needs a return true if we have logging disabled too

commit 8fc6da24eb91b0ba1414f4b1c293a290a1648c79
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Dec 10 22:46:36 2015 -0500

    fix prot tests

commit a71415428d06855416718e6adc6ba24bfe303dbb
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Dec 10 22:46:16 2015 -0500

    handle_sigreturn needs to write to the shm.

commit 9c01feaaec97ffb7fe6d1b05701b29296edc3beb
Author: Davidlohr Bueso <dave@stgolabs.net>
Date:   Sat Dec 5 15:23:07 2015 -0800

    futex: pimp futex fuzzing
    
    Make futex fuzzing more ad-hoc and spice things up to improve
    testing coverage.  Roughly half the calls will use the generic
    arguments, with the occasional exception of using CLOCK_REALTIME,
    when applicable to the correct cmd (avoid -ENOSYS). The other half
    can perform anyone of the following operations, each with a ~25%
    chance, acting on at most two uaddresses from a small pool selected
    randomly.
     - Perform a pseudo trylock/unlock. Contended scenarios
        involve immediately blocking or doing a pi lock.
    
     - Wait/requeue, which pairs with one of the below wakes.
     - WAKE_OP, or,
     - a regular wake or directly requeueing.
    
    Additionally, update the TODO file as futexes are not (de)initialized,
    and we can just use the handful provided locality by futex.c. These are
    an array of locks which are used for the corresponding uaddresses requred
    by the system call.
    
    Signed-off-by: Davidlohr Bueso <dbueso@suse.de>

commit 71ebf7842d7adc28ac1d37f39adb06dd2e9390f5
Author: Davidlohr Bueso <dave@stgolabs.net>
Date:   Sat Dec 5 15:23:06 2015 -0800

    futex: add _PRIVATE ops
    
    Extend futex_ops array to include coverage for private futexes
    (which do not share the addr space, and therefore faster wrt
    futexes). These are the same ops currently used, except with
    a xor'd FUTEX_PRIVATE_FLAG.
    
    Signed-off-by: Davidlohr Bueso <dbueso@suse.de>

commit 160868e87b002064ab86b016c410aaf890f08445
Author: Vinson Lee <vlee@freedesktop.org>
Date:   Fri Dec 4 02:18:52 2015 -0800

    Remove unused variable in fds/drm.c.
    
    Fix build error with GCC 5.2.1.
    
      CC    fds/drm.o
    fds/drm.c: In function ‘open_drm_fds’:
    fds/drm.c:70:15: error: unused variable ‘i’ [-Werror=unused-variable]
      unsigned int i;
                   ^
    
    Signed-off-by: Vinson Lee <vlee@freedesktop.org>

commit 92c59c46cf94c8d69eccbd5505884f37a5ff9488
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Dec 3 14:27:22 2015 -0500

    Revert "compile with -Og when in development."
    
    This reverts commit 46cfc86b6b62a5f3a69f4693d2f61eda7bf7f6df.
    
    At least the gcc5 in debian produces horribly broken code with -Og
    Nice idea, but not ready for use.

commit 46cfc86b6b62a5f3a69f4693d2f61eda7bf7f6df
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Dec 3 12:11:17 2015 -0500

    compile with -Og when in development.

commit de1925683405d9add8e14d65aab2e0bd0155ba34
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Dec 2 16:24:52 2015 -0500

    post-syscall, ARG_SOCKETINFO becomes an fd, don't dereference it.

commit d8b6ad78c9f6dee4d806c4414129f0e7348b5343
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 1 13:04:53 2015 -0500

    watchdog needs to check SOCKETINFO fd's too.

commit 632cd349b645bf62bb4ee9c00701c03df6a1b06d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Dec 1 12:49:53 2015 -0500

    scope reduction

commit 7491348ced78154be7584252e185242c3192eeb5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Nov 30 16:10:41 2015 -0500

    move some unnecessary things out of the --dry-run path

commit d1a21d70e8efcaff0742c2e879b0bca0760ccf66
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Nov 30 15:26:16 2015 -0500

    rename generic_fd_from_socketinfo -> fd_from_socketinfo

commit 29fb3354a339cda39537844b931a4dd56811e668
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Nov 30 14:52:21 2015 -0500

    introduce shorthand 'no objects' helper

commit 61565ad10c907a109f8a53a7e49936a2407f65fa
Author: Vinson Lee <vlee@freedesktop.org>
Date:   Sun Nov 29 23:13:04 2015 -0800

    compat.h: Add NETLINK_SOCK_DIAG.
    
    Fix CentOS 6 build error.
    
      CC    net/proto-netlink.o
    net/proto-netlink.c: In function ‘netlink_rand_socket’:
    net/proto-netlink.c:49: error: ‘NETLINK_SOCK_DIAG’ undeclared (first use in this function)
    
    Signed-off-by: Vinson Lee <vlee@freedesktop.org>

commit 76d6ddf590cd3fe7da7ac1628228136fa945dfaa
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Nov 30 13:47:13 2015 -0500

    destroy the network sockets in the same process that created them.

commit af781c9a06f89b77285753a62c77bae7d146c52e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 16:20:21 2015 -0500

    check for epoll init failure

commit 42053b9d43668d5007c4612f87986b85289fa549
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 16:19:55 2015 -0500

    initialize the listhead on new objects

commit 20a4f8674e665a21c9f24db9eaffd0f65b281146
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 15:38:29 2015 -0500

    convert socketfd code to object cache

commit e601a292ef83f1623e908a3d3e885b7634d9839d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 15:00:45 2015 -0500

    convert inotifyfd to use the object cache

commit f1b7d5f8dd59d19c62df45a8d64b03bfef218c65
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 14:45:50 2015 -0500

    convert drmfd to use the object cache

commit 1289f23587a8b6aa19c97d6e4e6e3bc96cf8f966
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 14:33:24 2015 -0500

    convert memfd code over to use object cache

commit e39d753c54bd7c31e281794bd9238b4ef73b23ce
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 14:27:40 2015 -0500

    convert testfilefd over to use the object cache

commit 59b65a072b8a5a27a010f12a202de75bf0a3d151
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 14:23:58 2015 -0500

    convert the timerfd code to use the object cache

commit d790247209c8562eb356a5a0ac4596dc2bc82297
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 14:17:30 2015 -0500

    int is big enough for eventfd flags

commit dbb435e2397d2e1f4da9b4412aced10592c68aba
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 14:16:20 2015 -0500

    convert eventfd code to use the object cache

commit f9dfaaed5e2c38c9d24e7dec1c8c49f3cc1cca66
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 14:08:46 2015 -0500

    convert epoll fd code over to use the object cache

commit 885c3819c10f73cf1197c38c392fb5612be5083b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 14:06:28 2015 -0500

    convert the perf fd code over to use the object cache

commit 448bbe9afdc4825537410764e4391739f58458e8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 14:01:11 2015 -0500

    add an alloc_object helper

commit 8d482379810a25bb5c6eddcf47c71e15a8b185d3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 13:58:50 2015 -0500

    convert the file fd code over to use the object cache

commit 79f79013b97bf3c83212c4d547859b251c3c72ca
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 13:49:56 2015 -0500

    convert the pipe fd code over to use the object cache

commit 519d94021322535184812de3c5f4d97650eac74b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 13:35:06 2015 -0500

    move the object initialization up a level, out of the mm code.

commit 1049a753d77d619458ba0c842952c6017961c15e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 13:26:15 2015 -0500

    use a macro to count errno list members

commit 04d8fc2a7195783a8d4922689c3f2d4bbf53e0a1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 13:23:47 2015 -0500

    change syscall ->errnos.values to a pointer
    
    allowing deduping, as we did for ->arglist.values

commit 5ba1366ae69331dbe419d0326634fa8d19e53229
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 25 12:12:05 2015 -0500

    remove find_child_logfile_handle
    
    Simplify find_logfile_handle greatly.

commit f05d1e05cf94a6fa7075517d706492bf4e7a4b0a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Nov 24 16:10:36 2015 -0500

    use a macro to count arglist members
    
    bonus bug fix: the count for fcntl flags was wrong.

commit 339ba423cf6eaf077e6488f438656ea51e681702
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Nov 24 14:55:24 2015 -0500

    change syscall ->arglist.values to a pointer
    
    Thus allowing for some deduping where some syscalls share common flags
    (open, recv, send, etc..)

commit 5449eaa2daeb28e92a0a0eb7b41cf94b5d42c9f8
Author: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Date:   Wed Nov 18 15:58:04 2015 +0200

    Free correct argument in post_ppoll()
    
    sanitise_ppoll() allocates first and third arguments and we need to free
    them in post_ppoll(), not first and fourth.
    
    Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

commit ef72bed6ec9e54a23f5bba6d9186337a53b9ebf7
Author: Kirill A. Shutemov <kirill@shutemov.name>
Date:   Wed Nov 18 14:49:30 2015 +0200

    Fix "O_CREAT/O_TMPFILE without mode" error in open_file()
    
     > *** invalid open call: O_CREAT or O_TMPFILE without mode ***:
     > ./trinity terminated
    
    There's one more source of such crashes: open_file() in fds/files.c

commit 715074714443946109710f32ed1fbc5325ba2d82
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Nov 10 12:14:51 2015 -0500

    NETLINK_SOCK_DIAG uses the type field for its own purposes.

commit 510d0acfef9b593e8cfdb612fc9789b577f2d583
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Nov 10 00:08:08 2015 -0500

    update s390 syscalls

commit 234fa7023ca9227c976ea14f0fa13ed16f926712
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Nov 10 00:05:23 2015 -0500

    add kcmp to ia64

commit 98db42321b97052cc0b0732b52cf630a63f192f9
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Nov 10 00:04:20 2015 -0500

    update arm syscalls

commit 4581dbb61a3fc25b6d263bd88a8546db8ddc8256
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Nov 9 21:56:12 2015 -0500

    move get_random_fd proto to fd.h

commit 763f4afb19019228aa8e86949cb860ec0bb75831
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Nov 9 21:55:55 2015 -0500

    bump gcc version check

commit 83c4bebbc5b2e0748f36cfa349e4ceaf9612a921
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Nov 9 21:55:31 2015 -0500

    and another gcc false positive

commit a5492669da5e8af4c0c4be05424163979a2f1ee1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Nov 9 21:54:25 2015 -0500

    silence another annoying gcc false positive

commit c45c1c8b0bbf1f05e5c4897f62776627e9f7ed38
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Nov 9 21:51:02 2015 -0500

    splice can work on fd's other than pipes now

commit d0974556841a520bec96d3c51e4077e0dd6e110e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Nov 9 21:50:54 2015 -0500

    for syscalls with >1 ARG_FD, don't always reuse the 1st ARG_FD

commit 9b91acc2c529dda96856a905b2ae0aa8c27ece82
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Nov 9 16:13:23 2015 -0500

    add sys_mlock2

commit f355557f22f42292418167af65edf9d88e4d35b7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Nov 9 16:09:53 2015 -0500

    add sys_membarrier

commit b26a1b17955f2c1b09b520e295b285a0deb31441
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Nov 9 16:01:51 2015 -0500

    ia64: Add userfaultfd

commit 4b1df1bd5b346b2e09305e06a02805bc5f2cb969
Author: Diogo Leal <me@diogoleal.com>
Date:   Thu Nov 5 14:58:17 2015 -0500

    Add first pass at man page based on README

commit 917b0ce8d1e30e2b8d6f7dd2c16c98e034705a7b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 4 19:26:20 2015 -0500

    Wire up phonet_rand_socket

commit ec5a3764c516e7e34ff8d7bf1c2c2b2d4866d4bb
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Nov 4 19:20:43 2015 -0500

    move version back to devel

commit a0fa61798f5266cfbcb0750307d8596fb7ecb742
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Nov 3 12:22:04 2015 -0500

    remove debug print

commit 4295cf7ba277850458f65e147c0e84e5a55ade0e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 29 10:47:39 2015 -0400

    use container_of to map from map back to object

commit 9ac02b8df59d8fd10116b0b53c754583b3011ce4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 29 10:43:00 2015 -0400

    add destructor for maps

commit adb585a561ae9e3fcf9a5d6c4d953a1cb864fdce
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 29 10:42:50 2015 -0400

    fix up proto for destructors

commit 86f659b5613e99ad287011997e7d99dd946ecf33
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 29 10:27:54 2015 -0400

    do first alloc at first add_object instead of in init

commit fc2b0a7208a39dd743a6dd7aef7720073eb72674
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 29 10:26:59 2015 -0400

    make getmap get the map from the obj

commit 55e8ddaa58843458368285c55d141853ddd47b6e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 29 10:26:42 2015 -0400

    destroy the obj, not the map

commit 9aecb641e8c73ffd07275d8a09398dd51dfa6920
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 29 08:17:43 2015 -0400

    get map struct from the object

commit 07d9e458b560f1c6ed3a32da3c90cdb0df96c8e1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 29 08:16:57 2015 -0400

    remove linked list from mapping struct

commit 0a50b350b6748746abd0c4eb0d754c66849385d1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 29 07:28:57 2015 -0400

    put the listhead directly in the object

commit e84b5eb791f251fdd73afb558bf4b6c07c358641
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 29 07:28:07 2015 -0400

    use the map struct in the object when dumping the list.

commit 84968d3ee4877f8bff67f8981ad6d07665e4c729
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Oct 28 01:55:06 2015 -0400

    convert global mmaps list to use the global object cache

commit e43dab6700f9410508f58298e12f98d43552c60e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Oct 28 00:55:12 2015 -0400

    move mapping struct into a union in the object structs

commit 32d18289f2b34dd2384aa5a6688e7afe413b7de9
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Oct 27 23:19:10 2015 -0400

    move the object destructor to the objhead
    
    rather that duplicate it once for every object.

commit d066f6579e39e282437dc5423ea3d14ad4ed2175
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Oct 27 23:17:59 2015 -0400

    convert the per-child mmaps list to the object cache

commit 1d215167da8913f20466bdcfd7f6bffdaf4cc117
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Oct 27 23:04:07 2015 -0400

    remove unnecessary dereferencable field.

commit 1c772ad87ed1ea6c694cb377363d50e162e8de82
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Oct 27 22:26:01 2015 -0400

    add new objects to the tail of lists.

commit 1bd0cca9e60f680fa99a35a60f439f9e1ae3d63b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Oct 27 21:32:10 2015 -0400

    introduce a function to pick a random object

commit 7d9f5fe38c7e23316acca81d39e2b691b5509d9d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Oct 27 21:22:07 2015 -0400

    object list initialization

commit 3482d1f68f424f4365f65894af140853ab648fbf
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Oct 27 21:15:40 2015 -0400

    move the object type out of the object struct

commit 0e26f6bcbcf8891e7876699d271d3c236e378ab8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Oct 27 21:00:31 2015 -0400

    switch to just a single add_object function

commit f4de988b35073d1abe41b29eab84a1c68943c323
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Oct 27 20:53:53 2015 -0400

    maintain a count of objects in each object list

commit 97f4e4ec06ecdac7c8a07df66b3f88bcdaf5c36d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Oct 27 20:26:00 2015 -0400

    sketch out the basics for a generic object cache
    
    right now this does nothing.

commit 24b93efdf1ff4fc01d4eb6b3296325afe2550d00
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Nov 3 12:05:50 2015 -0500

    1.6 release

commit 40b872b7df17d1adda6114320908a73675d84334
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Oct 29 07:15:09 2015 -0400

    add a --dry-run argument to disable the 'do the syscall' path.

commit 53dd214b1f46309fab93a117c992c079d18c0a45
Author: Praveen K. Pandey <praveen@linux.vnet.ibm.com>
Date:   Thu Oct 29 14:26:13 2015 +0530

    remove pause param as it doesn't add any value any more
    
    Signed-off-by: Praveen K. Pandey <praveen@linux.vnet.ibm.com>

commit 88e8495155af7f9f3665a17fb48e7df810b651fd
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Oct 26 22:32:18 2015 -0400

    move check_if_fd proto to fd.h
    
    Removes the need for trinity.h (which gets included everywhere)
    to need syscall.h

commit d972b6279ecdcecab5bc3eea53333dac9b94fa9b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Oct 20 16:48:11 2015 -0400

    fill out destringid flags for request_key

commit 1d2df023acba43ed5b05da71b89e65dd48be9eeb
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sat Sep 5 21:08:05 2015 -0400

    beginnings of userfaultfd

commit 83d080dba735592ad1d19ef13aa6ab0df1a88734
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sat Sep 5 20:58:29 2015 -0400

    Add PR_CAP_AMBIENT

commit db7ee382816d578367843b4d07ff967dee49837b
Author: Alexander Potapenko <glider@google.com>
Date:   Fri Sep 4 17:25:26 2015 +0200

    Add /dev/kmem to the list of dangerous devices that Trinity should not access.
    
    Writing to it may cause kernel memory corruption.

commit 45e3e2f3885c892c2b4b80c701ea987ac6222757
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Sep 4 00:35:44 2015 -0400

    Add NETLINK_CAP_ACK

commit 7855d8ba2141d6e906cb86a5ced55e48fb9cf358
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Sep 2 00:08:22 2015 -0400

    update 32-bit x86 syscall list
    
    socketcall got demuxed.

commit 8841696832414bd62c8437588cb93744679ad3be
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Sep 1 01:11:44 2015 -0400

    update s390 syscall lists to enable NUMA syscalls.

commit 8f12a40b81662487fee49d33bcac16db8e2b2e6e
Author: Vinson Lee <vlee@twitter.com>
Date:   Mon Aug 31 16:40:51 2015 -0700

    socket: Fix missing-field-initializers build error.
    
    This patch fixes this build error.
    
      CC    syscalls/socket.o
    syscalls/socket.c:48:2: error: missing initializer for field ‘func’ of ‘const struct socket_ptr’ [-Werror=missing-field-initializers]
      [AF_ECONET] = {  }, // DEAD
      ^
    syscalls/socket.c:22:9: note: ‘func’ declared here
      void (*func)(struct socket_triplet *st);
             ^
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit e530c3fc02244c4d10612f7c12f1fdf0db01f77e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Aug 31 17:52:51 2015 -0400

    scope reduction

commit 75e03309ea79f447b04c3a8c05de18a605c87aa2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Aug 31 17:51:41 2015 -0400

    scope reduction

commit a71d7cd049bfe70f8a3123f8deeb83d52241f086
Author: Hangbin Liu <liuhangbin@gmail.com>
Date:   Mon Aug 31 17:18:37 2015 +0800

    fix build prctl.h header on RHEL6.7
    
    latest RHEL6.7 kernel headers only defined PR_SET/GET_NO_NEW_PRIVS, which
    lead to 'PR_GET_TID_ADDRESS' undeclared build error.
    
    Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>

commit 74b9f03a1eea19572150bec780e55a9dbf59e70c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 4 20:02:54 2015 -0400

    rand: add ':', ',' and '.' as possible separators

commit 9c674b08c58afb47ef009723e81266bdeea7a1ff
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 4 20:01:55 2015 -0400

    compile-time calculate the separator array size

commit f54f3b9be623293e2640c87b026d26e506af926a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 4 19:55:32 2015 -0400

    random: remove unnecessary local var

commit 0f88dffcd81247e13a52d8183464b68b87122604
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 4 19:53:43 2015 -0400

    watchdog: wait longer at startup
    
    if we're starting a *lot* of processes, it can take longer than the
    1000 retries in this loop when we only wait a microsecond each time.

commit ab9a8dec47ee51d07e9dc47fb314770b761f9fb2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 4 19:52:52 2015 -0400

    perf_event_open: use RAND_BOOL()

commit e8af8f5883ef313272d986905b8c789f21de35ca
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 4 19:52:21 2015 -0400

    munmap: use ONE_IN() and RAND_BOOL to make clearer what's intended

commit 7f68f0378bb115fee24c4e74de15529f400b4adf
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 4 19:51:57 2015 -0400

    move_pages: use RAND_BOOL()

commit 95f25fd7105ec5644f5d16bb3e5f9272f9c68629
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 4 19:51:04 2015 -0400

    remove unneeded includes

commit 606e7f79ceb1f32bf0515e421ed27d1916206c99
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 4 19:50:35 2015 -0400

    decnet: use RAND_BOOL()

commit afeb2ca1856ac9f3e0e02e5ce0393cb584a73358
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 4 19:50:16 2015 -0400

    if an fd provider fails, just skip over it.

commit 5259f945d76820ef553b99be8515978cff1a3229
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 4 19:49:31 2015 -0400

    TODO: check loadavg before sanity checks

commit 3933afb88099d022b40d948f285d19f70beceb1d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 4 19:49:01 2015 -0400

    decide at random at startup whether or we bind processes to CPUs

commit 7a3796c3e3d20d97e50d6e18ef00e39a06ea0fb6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Aug 3 23:19:39 2015 -0400

    re-add accidentally dropped netlink setsockopt's

commit 0bceaf65d0bfabe813a1ec86c986c1b35a511085
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 23 19:03:53 2015 -0400

    select: fix shifting out of range of datatype in FD_SET

commit 0bc95033e002a643b3193592c4c29b65c0569ede
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 23 18:56:11 2015 -0400

    if we've filled the socket array, don't try to bind another fd.

commit 4ae63b03513fd224e070ba57a16578546f7b120f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 23 18:53:36 2015 -0400

    fix out of bounds read in socket creation
    
    The actual bug here was the incrementing of nr_sockets in add_socket.
    When we get to the setsockopt, we've advanced past the end of the array.
    
    Also eliminate the nr_to_create var which pointlessly duplicates nr_sockets
    and gets out of sync.

commit a4bebb33c169c6780a2da3cb0c41ab1b89d6f48b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 23 00:25:00 2015 -0400

    mask out a random number of bits from the new seed.
    
    The new seeds are overwhelmingly large 32 bit numbers.
    This at least gives some coverage to the lower end.

commit 4f55491789f411546d2e4b5e9790fdd57841bf8d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 23 00:23:12 2015 -0400

    open urandom just once on startup
    
    We can fail to open it from the child later otherwise, and end up
    falling back, and in some cases, seeding with '0' repeatedly. Oops.

commit 08bcce0b86046dc150d3100a77152dff7d19083c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 23 00:13:27 2015 -0400

    some glibc's check the mode arg for O_CREAT & O_TMPFILE
    
    Make sure we fill it in, or we get killed.

commit 0f7305ea1ae4d630d14e7f6efc5cc93bfa89328b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 23 00:13:02 2015 -0400

    fix the "check for <8gb" test

commit f98a00ef0d810ef0d3ff11d3b2695e456b807d06
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 23 00:12:14 2015 -0400

    SIGABRT/SIGSEGV in main should be set to default

commit 75bcdf9523b540058b7163955718fa5bc9b40c62
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 23 00:11:41 2015 -0400

    make page_rand be a length we can return

commit b440b5d75a06552865ef1b05b3a1dcc8fde5202d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 23 00:10:57 2015 -0400

    add some more cases to the 'create a struct look-alike page' function

commit da08be88e75bfbd20efbc275782d2140d6b0171a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 23 00:10:24 2015 -0400

    socket: use struct initializers to save a loop

commit 711ccbf8e5040063fbff78b571324b86876aeeb6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 23 00:09:44 2015 -0400

    setsockopt: use struct initializers to save a loop

commit b9b5765972c9064ef0c6eb8e08bab99d1dca9762
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 23 00:09:09 2015 -0400

    if main crashes before watchdog has begun, we need to check for it

commit 112b58e51a4319b1c6e0f6f906b0e2440b9caba1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 23 00:08:55 2015 -0400

    check for memalign failure

commit 4816e49eaec16d4354756dac77cc9c7d41e0dbb7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 23 00:07:35 2015 -0400

    fix get_interesting_32bit_value shifting
    
    This is supposed to shift 0xff through 7 positions, not 31.

commit 11ccbad89e571a14f863b9f2c2b336b6644ac8d1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jul 22 23:59:55 2015 -0400

    flesh out AF_ALG some more

commit fc71ae9107b4e7a4e610c703b5bf4fc2b3f93523
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jul 22 23:58:10 2015 -0400

    fix undefined behaviour in RAND_32
    
    We were shifting past the width of the type.
    
    Make rand_32 simpler.
    If your rand() has some problem with the LSB, fix your rand.

commit 14bf4bc56fcad52ca9503ae626a3eeeb08774d48
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jul 22 23:57:16 2015 -0400

    only use fallback argument if argtype isn't defined

commit 7b6e6c5fe7337d10fc434d9f178c9ca3f7b07d6c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jul 22 23:52:04 2015 -0400

    don't loop over socket types, pick them randomly.

commit ca0de735e31ff6700d73a1e48dce79689948d6eb
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jul 22 23:38:50 2015 -0400

    fix gcc5 warning about trashed var

commit 95ec5ea3d0e71d031fc7352b49914a5ea384f782
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Aug 4 19:47:20 2015 -0400

    re-sort the getopt param string.

commit 1b9845ace0de65480f4e0229950af53658c84f0e
Author: Praveen Pandey <praveen@linux.vnet.ibm.com>
Date:   Mon Jul 20 14:29:16 2015 +0530

    fix trinity compilation (make) error in ppcle system
    
    Solution include termios.h in ioctls/vt.c
    
    Signed-off-by: Praveen Pandey <praveen@linux.vnet.ibm.com>

commit 2a37e77e0276d8a9b7c420e1332d39dbf0f31b44
Author: Vinson Lee <vlee@twitter.com>
Date:   Fri Jul 17 11:15:41 2015 -0700

    Fix build with kernel 3.4 headers.
    
    PR_SET_CHILD_SUBREAPER was added in kernel 3.4 but PR_SET_NO_NEW_PRIVS,
    PR_GET_NO_NEW_PRIVS, and PR_GET_TID_ADDRESS were added in kernel 3.5.
    
      CC    syscalls/prctl.o
    syscalls/prctl.c:33: error: ‘PR_SET_NO_NEW_PRIVS’ undeclared here (not in a function)
    syscalls/prctl.c:34: error: ‘PR_GET_NO_NEW_PRIVS’ undeclared here (not in a function)
    syscalls/prctl.c:34: error: ‘PR_GET_TID_ADDRESS’ undeclared here (not in a function)
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit d49c84ee568ee3900370d090cd90d88a0e61d1f2
Author: Vinson Lee <vlee@twitter.com>
Date:   Tue Jul 14 11:25:55 2015 -0700

    Add POLL_BUSY_LOOP symbol for older distros.
    
    POLL_BUSY_LOOP is not available until Linux 3.11.
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 1973119fc24bed6f28569b5295a4db5efd02cb84
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jul 8 19:22:34 2015 -0400

    resync aarch64 syscalls

commit 80c6bc84cde79831d92424c72ae975b7edc2dbf2
Author: Vinson Lee <vlee@twitter.com>
Date:   Tue Jul 7 14:46:06 2015 -0700

    Add POLLFREE symbol for older distros.
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 23982097aebc3940987f7c267ee128bfb6af6772
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jul 6 18:39:01 2015 -0400

    clear optname on each do_setsockopt invocation

commit 8816a902baf859cb0844b4802c0cfadb129a063d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jul 6 18:38:42 2015 -0400

    clear optval after freeing it

commit b3838221560d58fb3a69f106e539cd5a8e72994a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jul 6 18:38:13 2015 -0400

    Fix up a bunch of incorrect RAND_ARRAY uses

commit bcd0e418950e1ff8ab245966bae281ff5148222e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jul 3 12:47:40 2015 -0400

    fix up corner cases and unify table counting.

commit 0e4851dce759c737da02ad59b8dd17dbe0009a4a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jul 3 00:08:04 2015 -0400

    don't output "64bit: 0 disabled" when using -a64

commit 718af5d83525a8d598962cbf91f25312fc3c152f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Jul 3 00:02:21 2015 -0400

    rename enablednr -> enabledstr.
    
    Also reduce scoping

commit b8cafcb52d59490ec74e31f5406b8127605d8e77
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jul 2 20:58:48 2015 -0400

    pretty print the 'all disabled' case for 32bit.

commit dfd9160e95fc364960974697f479dceb25f1bab9
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jul 1 18:50:59 2015 -0400

    remove remaining duplicate code from poll/ppoll unification

commit 79d555c8a27beb8af4876a7269a1edffaa39b6a6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jul 1 18:50:10 2015 -0400

    replace open-coded freeptr

commit d92549fd9b40d9cc6cc805f7febd31536bfb0b1a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jul 1 18:49:34 2015 -0400

    make ppoll impossible to fail allocations

commit e332665dd21584c51c46abd4d3bdb761c2397e23
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jul 1 18:48:24 2015 -0400

    merge poll & ppoll

commit 568ce76bec014a4cbf76d7c66c83626caacf0a28
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jul 1 18:29:04 2015 -0400

    slightly improved sanitisation of ppoll

commit 9e5dd744d12b50253b435ea1fa425a68c6f72351
Author: Jiri Slaby <jslaby@suse.com>
Date:   Tue Jun 30 12:01:01 2015 +0200

    compat.h: define CLOCK_MONOTONIC_RAW
    
    It is not defined on older distros yet.
    
    Signed-off-by: Jiri Slaby <jslaby@suse.com>

commit 12f99bd190872b222c7b5e81a291258afc066324
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jun 29 20:56:22 2015 -0400

    fix warning about main not having a return value

commit 2899932b2ec78e29a19bb1e965b037c338e2bb9f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jun 29 20:55:02 2015 -0400

    unsigned var will never be negative

commit f87f53120f1793cd1096a5e50131311549a9f424
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jun 29 20:54:24 2015 -0400

    Increase length of string when prefixing with ./

commit 505c19cd59f1806f800cd39f57004895b9ef22cf
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jun 29 20:54:02 2015 -0400

    seconds won't go negative

commit 1b5fdcf946bec06c63f5b26297f9f61a566b1fb0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jun 29 20:53:43 2015 -0400

    reduce scoping

commit 1fba74f945e7e1b66ee15525fffbe6420ae6e94f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jun 29 20:53:22 2015 -0400

    printf signedness fixes

commit 33964ccba5b84403b0c5ca5326d432c2c4bf3938
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jun 29 20:52:24 2015 -0400

    printf signedness fix

commit b3372e20cc65acd57ea9cd0783ce162a40c1b9e7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jun 29 20:52:03 2015 -0400

    printf signedness fix

commit 14431582c11412675630e45e0699f12bb12aca37
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jun 29 20:51:38 2015 -0400

    nr_blockdevs can't be negative

commit 5674b35b3ff9491cd0992764b0c7bafabbbefd0e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sun Jun 28 21:45:30 2015 -0400

    fix build on 32bit

commit b36ad6cf295615ce6ee47ed3b35bc76c69f878a0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sun Jun 28 17:42:56 2015 -0400

    don't touch rec-> when sanity checks fail.
    
    If the checks have failed, everything there is suspect.
    Use the on-stack copies instead.

commit 0ac5645e6cdb2e2a9274d0e12ea1fe60b5f49e57
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jun 25 00:35:52 2015 -0400

    add IPPROTO_MPLS

commit 9dc919749d428fb84b598cd232fdb4d811a7b846
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jun 25 00:31:58 2015 -0400

    add NETLINK_LIST_MEMBERSHIPS setsockopt

commit 139ca8d4f00ee40e34f57d3bfb3f18ee36784335
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jun 25 00:31:15 2015 -0400

    add NETLINK_LISTEN_ALL_NSID setsockopt

commit 2f049aa5d78026b0c361a5ad2c0a33a4f56377b1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jun 25 00:28:46 2015 -0400

    add IP_BIND_ADDRESS_NO_PORT setsockopt

commit 9895c7ce93ebaabe26bfc5b47c4e3797eaf6daee
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jun 25 00:27:27 2015 -0400

    add IP_CHECKSUM setsockopt

commit b78ee9ec7752e81c8c9a5eeb65fa01a19c4b33e7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jun 25 00:23:08 2015 -0400

    add TCP_SAVE_SYN/TCP_SAVED_SYN

commit ecfef58f755bf3c9341ef833910fcb1c7826354c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jun 22 20:53:31 2015 -0400

    print the syscall name instead of the syscall number

commit 6b0dc33f57a963660d6c5e432b0ebb3a32ea5d0b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jun 22 20:17:54 2015 -0400

    don't mmap large pages if we're low on memory

commit d5e15eeda80eabdc5966e20be0be22105c9bcd98
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jun 22 20:12:17 2015 -0400

    redesign the GB/MB macros to take an arg

commit 28439694a89bc911128390a3199950ec21ab2879
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jun 22 19:38:12 2015 -0400

    use .init for mmap size initialization

commit a8d96630a7fd4907bf99e925af69f0287e4c3142
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jun 22 19:37:39 2015 -0400

    write protect the child ptrs after allocation

commit 219da87fa4bcd701e5fccc26358fd7c6a54a0f05
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jun 17 16:47:44 2015 -0400

    explain why we use alloc_shared for the childdata structs

commit 6fb41f61a39b2107119a4e6009f4f24ba0183c09
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jun 17 00:53:52 2015 -0400

    only do the shm write protecting in debug mode.
    
    This is quite high overhead.

commit 0ef58aee8e5c6acc5df8a2953ef76aef7bbfda89
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jun 17 00:52:49 2015 -0400

    mm: mark mmaps as writable before writing to them
    
    instead of just segfaulting if they've become read-only.
    
    Later, be smarter and split mappings when sub-regions get mprotect'd

commit fb0ba4c67c9db91fa60faec15d992683ff68372f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jun 17 00:51:22 2015 -0400

    wait for main process to start.
    
    There was a tiny window where we would crash if the main process
    failed to start for some reason. By checking that main actually
    starts up, we eliminate that problem.

commit 037e253942ba795ab8ef7bfed26290edd64869f3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 16 18:37:36 2015 -0400

    if we're not logging to an fd, we should return false

commit 1a0ddebe76f21eb186c9f133aa61b0e579ead160
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 16 18:02:34 2015 -0400

    prepare child logging for multiple logging types

commit 65498dfd1c2cf39d9eb7b92139f82b58f5598609
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 16 18:01:08 2015 -0400

    only log at child logfile fd's if logging to files is enabled.

commit 202dea7571c88eca0a8bcce78e6b7f6673fc8835
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 16 17:53:58 2015 -0400

    move file logging details from main()

commit 989bbaa014f463833e323e5668892ad334a4454e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 16 17:51:09 2015 -0400

    split log.c into generic logging, and file logging

commit 4be63b9537b9ad84dc9cb9dd585458cc6d3eedbe
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 16 17:43:23 2015 -0400

    preparation for different forms of logging.

commit 82f10d8ea04c60ce42964438313da04b2ea4a262
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 16 17:18:52 2015 -0400

    just inline flushbuffer

commit caf6991947fd2a49a67982c8e2f5f9b031768a48
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 16 17:16:49 2015 -0400

    Move output buffer rendering to log.c

commit ffbe903e230ffe7a8f62e8a1e2fbba1e6a403b1b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 16 17:16:32 2015 -0400

    refresh patches

commit 48c7cef34d5c4cfcea1c6fcb3199e6ae6afee52d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jun 10 18:54:03 2015 -0400

    page align the page that read_last_page writes to
    
    copying the similar change from 600a071dbdd0d6fbc9afe4ee61ae09d3977a8510

commit 17c657a34c113c4492136ab5d2db7822763166bf
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jun 10 18:53:55 2015 -0400

    scope reduction

commit bd1bb8662fc60d884012ec41b40c9326e9ecf0bf
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jun 10 18:53:41 2015 -0400

    scope reduction

commit 4e533d11f1664b0f24756330b7407d3851a36ac6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jun 10 18:53:26 2015 -0400

    signedness

commit a1bde5c1cdf34e97d74ca67b754ff772aecb4a3f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 9 21:12:37 2015 -0400

    if all children are stalled, panic, and start a killing spree.

commit 98742ebca1bdb822fac7dbbe0456ca781ee28b5d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 9 21:11:07 2015 -0400

    remove the watchdog wrap-check.
    
    It needs to be way smarter, and interferes with some later work.

commit 11348424891ae50550467db29e1b5ec61c6bbf41
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 9 21:09:35 2015 -0400

    don't print stall stats if zero

commit c32dd95ab1cd9a612f9cf5a51e785316ba4e22fa
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 9 21:06:33 2015 -0400

    change the priority of the watchdog process

commit e8646038d918ed5aedb9cd68e1ccdf36a81a75cb
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 9 21:04:48 2015 -0400

    add some extra sanity checks inside the child process

commit ecd565a115253905607f0c9c36e95468c015f167
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 9 21:03:43 2015 -0400

    add some functions to (un)write-protect the shm
    
    Use them in critical places

commit 06b82f16347475ff4c9d3e586f4841f8609ac274
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 9 21:01:52 2015 -0400

    remove redzones around shm
    
    again, pointless.

commit 08a9b0584ed03253e559e2350aa369ac9524e748
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 9 21:00:28 2015 -0400

    remove guard pages from single page mappings.
    
    They aren't really useful

commit e5e8997b9f018ff5ff27e9e1b9abc704917fc828
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 9 20:59:11 2015 -0400

    make sure we don't try to dump children that don't exist

commit 5b1994e7d0372cd992cb05824affc0abf57af703
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 9 20:58:11 2015 -0400

    give read fault the same single-page treatment we did for writes.

commit 921eae523d60c299e2a519a0099165d60d2e7ed1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 9 20:37:52 2015 -0400

    sort case statements

commit 04697e75f6404f2cc12cc9aa2730db78e15727c9
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Jun 2 21:06:21 2015 -0400

    replace huge compound if in munge_tables with function.

commit 4fd93d2b593db9a2dfd8149d0c118c32d47b8812
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:03:44 2015 -0400

    use RAND_ARRAY in mremap

commit 0aaec8ca58611ed70615999f1407995bfbc0f0fd
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:01:34 2015 -0400

    use ARRAY_RAND in proto udplite

commit 9def05e78cf7e4ecc167adc86b14a5941249cec3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:01:34 2015 -0400

    use ARRAY_RAND in proto udp

commit a2a79fb75a95ca79686d11691ddb0a537ba6dc57
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:01:34 2015 -0400

    use ARRAY_RAND in proto socket

commit 14fb61b8491d2b58f4d89e1d53ddc6cf159784bb
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:01:34 2015 -0400

    use ARRAY_RAND in proto sctp

commit cee4ab101c5bc5ea64f12cf48ae85b1ba492e63e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:01:34 2015 -0400

    use ARRAY_RAND in proto rxrpc

commit 8065cb11d0bd03ebe17185c5865112560840fd6d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:01:34 2015 -0400

    use ARRAY_RAND in proto rds

commit 77b132233590b24add8b566513f5162dd59abff7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:01:34 2015 -0400

    use ARRAY_RAND in proto rose

commit a86ef0a025a60a19cd2342b36e115abd80858282
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:01:33 2015 -0400

    use ARRAY_RAND in proto netrom

commit b40c9bcad27e28fa907c57cc318247169be77fc5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:01:33 2015 -0400

    use ARRAY_RAND in proto netlink

commit 4759137ccd9ea496886cc86b52f8b69077b59b7c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:01:33 2015 -0400

    use ARRAY_RAND in proto llc

commit dc8e1ccc00e41abf8a416edcb58aa4a582d60ae0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:01:33 2015 -0400

    use ARRAY_RAND in proto iucv

commit 8410403e2a78934fa8d685df30e58904c92d8619
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:01:33 2015 -0400

    use ARRAY_RAND in proto irda

commit 1845751216b520ded991fd1cb58ce86391df5fea
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:01:33 2015 -0400

    use ARRAY_RAND in proto icmp6

commit 00b919e5481ec88b854f6b9c25b9d8be8149c7b1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:01:33 2015 -0400

    use ARRAY_RAND in proto decnet

commit f01f5469b2f8a2cd930fc108a074e27eff7a490d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:01:33 2015 -0400

    use ARRAY_RAND in proto caif

commit 6acdcf0127fd722e8aabb35630af6bb63bb9526d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:01:33 2015 -0400

    use ARRAY_RAND in proto bluetooth

commit 56910ac9682e1a7424cab74b23a355132c7dfec5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:01:33 2015 -0400

    use ARRAY_RAND in proto ax25

commit df18cbb169102c04ba819c5ce56d9d334d1dbbde
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 23:00:35 2015 -0400

    use ARRAY_RAND in atm

commit 102d0991fd8aaf93a418ff613e3dc7ec3a9c77fe
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 22:55:05 2015 -0400

    use ARRAY_RAND in dccp

commit 8f784211c77e2cc4e12ae5b89d9bf30d4c5b4c33
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 22:54:36 2015 -0400

    use RAND_ARRAY in packet

commit 7092447417eaadf419023949b9996feeb3462515
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 22:54:04 2015 -0400

    use RAND_ARRAY in tcp

commit 2f41fd0de1d0bb37bf26bafd9ef4f99a751833d1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 22:53:24 2015 -0400

    use RAND_ARRAY in pppox

commit f64cb86db1058674a8df0071dd6435267810b6a4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 22:52:43 2015 -0400

    use RAND_ARRAY in tipc

commit fc6dc7ed4cdf1425410bb6eb819577d3e7549db1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 22:52:29 2015 -0400

    remove trailing ;'s from macros

commit 1462752b121e4eca5c3be656660fdee95674ebcd
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 22:44:11 2015 -0400

    use RAND_ARRAY in mmap

commit ddf6f76d56f80bb9e466c5fcdd12c76cb69dab67
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 22:43:47 2015 -0400

    add a RAND_ARRAY variant for unnamed elements

commit efa61c59d6e794a9e936ef071c16215d9c64c439
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 22:40:23 2015 -0400

    remove unnecessary define in mmap syscall

commit 32c72e9f45ff724acf387ef2e64a70cb99d4bcc4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 22:36:25 2015 -0400

    use RAND_ELEMENT in ipv6

commit fb2e288f6155cd01d8061fd6a56d7cd69388814b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 22:35:03 2015 -0400

    add RAND_ELEMENT for getting a random array element

commit 3b3dbcb07f7ed1e9ff1cd2bc76fc7c468471456d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 21:31:08 2015 -0400

    remove bogus spaces

commit 8729916dba0a2b6a36aec1a7098467e4e85b8fd1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 21:30:41 2015 -0400

    replace switch in pppox_gen_sockaddr with jump table

commit 8511850a1dd7c02955d9fcaedde4c7dddef90728
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 21:28:21 2015 -0400

    replace switch in pppox_PX_PROTO_OL2TP with jump table

commit 40305c3d4e5f8444e129921b0a4adc0690c1a1e4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 21:22:18 2015 -0400

    fix misplaced ifdefs

commit e0c597f627c3d5e7aea02f85f9dc779384d78ff0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 21:21:23 2015 -0400

    move the PPPoL2TPv3in6 case out to its own function

commit 20db02b380bc7de649fb42d7c529e45a1855253e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 21:20:09 2015 -0400

    move the PPPoL2TPv3 case out to its own function

commit 3f6aba8c27e77f4943c2f5f8cb345698711a8d11
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 21:17:19 2015 -0400

    move the PPPoL2TPin6 case out to its own function

commit 5145778e7d7b244aaa7b79a4800bc17fe992fa7c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 21:15:54 2015 -0400

    split the PPPoL2TP case out to its own function

commit 67e2488593e228b9127829aa6126eaa3f19301ce
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 21:13:54 2015 -0400

    split pppox_PX_PROTO_OL2TP out to its own function
    
    this needs splitting further

commit 6dc6276ecc0771e2959e60f713f1a18d4983229b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 21:09:53 2015 -0400

    split PX_PROTO_OE out to its own function

commit 5595dcffc91cd5c36ec2c44bd93d4ebd0576b5e3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 21:02:32 2015 -0400

    move random thing generators to their own subdir

commit caf9a4b386e305e809f294ea03e2326546f52850
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 21:00:01 2015 -0400

    move mm related things to their own subdir

commit a148f57388749eafc2ce252e50f820e2f45ffbec
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 20:57:08 2015 -0400

    move all the fd generators to their own directory

commit 8e10b5eb60e561d3fce462882b7a20b1b8785eba
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 20:48:55 2015 -0400

    update alpha syscall list to 4.1rc5

commit 158feb3188c524b40018a34d3b8a9f8553d3f153
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 20:46:30 2015 -0400

    invert test to reduce indentation

commit 63d554073c71810796f5a338f6dfc436a7b5008c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 20:46:01 2015 -0400

    move the child cpu binding out to its own function

commit 4bb2e6bb54c5074b1c1fd9bf1832d9b6cf993f24
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri May 29 00:18:42 2015 -0400

    page align the page that dirty_last_page writes to

commit cd609890a24f1811b84d54d558f27cd852ccaf48
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 28 00:05:16 2015 -0400

    sometimes clear a bit in the ARG_MODE generator

commit 1ae3701fee7900ebe9e3c6ffe8a72c8ba51ed66f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 28 00:03:33 2015 -0400

    remove weird bit setting in ARG_MODE generation
    
    also add in missing modes.

commit bdf9e2687636cf72a7968386f711c3c94b349211
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 28 00:02:22 2015 -0400

    use the right function table when choosing fault write routines.

commit e96f1d97b49642929021d7d0c19adc9fce1b2adf
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue May 26 22:50:17 2015 -0400

    improve syslog argument generation a little

commit 767f08d8e85f29fbdb32c7353836a8f2300dcd10
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 21 21:12:06 2015 -0400

    display how many children are currently not making progress

commit 74a6440c987a38f832e50fdfe571d3338458f095
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 21 21:03:30 2015 -0400

    special case ENOSYS in check_retval_documented
    
    Now instead of ..
    
    [child23:21426] getrandom returned an undocumented return value (38:Function not implemented)
    [child23:21426] getrandom (317) returned ENOSYS, marking as inactive.
    [child0:21403] execveat returned an undocumented return value (38:Function not implemented)
    [child0:21403] execveat (322) returned ENOSYS, marking as inactive.
    
    We'll just get the latter message.

commit f4374c72d827006154fd81e3eb8007e638736ef6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 21 20:57:17 2015 -0400

    don't try to bind to cpus outside the cpuset
    
    when running with more child processes than cpu cores, we
    tried to do stupid things.

commit 47783d18ec866af3ba6f2e5349cae36462d99907
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 21 20:39:48 2015 -0400

    delete_mapping is only called from post_munmap, so inline it.

commit d5ec101fe42a58d78113a34ef040306a6ee5e109
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 21 20:36:10 2015 -0400

    watchdog: tweak the wraparound detector.
    
    still triggers too easily when run with lots of children

commit eb36871656308e0117336245abcba66026aa0bb9
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 21 20:31:40 2015 -0400

    remove magic '500000' threshold from watchdog progress detector.
    
    On a machine with enough fast cores, we can trip this easily.
    Just remove the check, and replace it with a check to make sure
    it didn't go backwards.
    
    I'm torn between just removing this completely, and may do so later.

commit 6727b7b251934aaf296be07e84dcbfa927090dc0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 21 20:29:41 2015 -0400

    if we detect corruption in the child, behave like the watchdog does.

commit 99156bab580674fe8d69ef68186b7ee7af4092c4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 21 20:27:24 2015 -0400

    blacklist /dev/sd* from the file list on startup
    
    if we run as root, even with --dropprivs, we should disallow
    direct access to disks, in case we scribble over the root disk.

commit 2d9735e2132cf6e8fb7c57c9953ff725cd721b75
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 21 20:24:58 2015 -0400

    set exit code depending on what happened.

commit c3e2b975aa6dd56661869f37440db79b0142fdd6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 21 20:15:06 2015 -0400

    ignore RT signals from children

commit 16bf856d767cf6e92c43d82cd2df7f172e8f7d0a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 21 20:09:46 2015 -0400

    split the mapping write functions into single/multi variants

commit 841ff2ff46a691c3f5353df43b5bf168bf72bb8b
Author: Vinson Lee <vlee@twitter.com>
Date:   Wed May 20 14:15:04 2015 -0700

    compat.h: Add CLOCK_BOOTTIME.
    
    CLOCK_BOOTTIME is not available with glibc < 2.14.
    
    Fix build on CentOS 6.
    
      CC    syscalls/perf_event_open.o
    syscalls/perf_event_open.c: In function ‘create_mostly_valid_sampling_event’:
    syscalls/perf_event_open.c:1118: error: ‘CLOCK_BOOTTIME’ undeclared (first use in this function)
    syscalls/perf_event_open.c:1118: error: (Each undeclared identifier is reported only once
    syscalls/perf_event_open.c:1118: error: for each function it appears in.)
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 86d2cb280041ff8d3c0b5ae394f6706ca226abd5
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue May 19 23:27:50 2015 -0400

    remove hand-coded instance of sockoptlen in do_setsockopt

commit cef425da945201786747950e217ad22822935a69
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue May 19 23:18:26 2015 -0400

    set the pass/fail statistics correctly in the execve case

commit 5ee175d38d92faf208bcd6284c8ca28360eac557
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon May 18 21:29:32 2015 -0400

    make ARG_LEN return some common sizeof's

commit b4c44df7b67a745ab49a825aa75f064e40179928
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon May 18 21:04:28 2015 -0400

    rename get_so_len to sockoptlen

commit 19ebcc8398b82f6238b40eb0ef2b7b26f4d033bb
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon May 18 21:03:14 2015 -0400

    move the sock_option len setting to a common function

commit c53caec15c1e868f484f367f8aeef9662327e9ae
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon May 18 20:16:43 2015 -0400

    some of the ip socket options take chars
    
    instead of annotating them, just randomly choose between int/char
    for the fallback case.

commit d28dd666cd36097326f91ac31dfc9ac70c6f5ba0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 14 22:13:18 2015 -0400

    rename ip_option to sock_option
    
    Other non-IP setsockopt functions will end up using this.

commit 700516350852569e0c15c7f5591a0d1dcb61ed64
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 14 22:08:06 2015 -0400

    add TCP_CC_INFO setsockopt

commit 3dfaf2147fb3d628d21900a99c2ef67168eab8a6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 14 22:07:30 2015 -0400

    merge execve/execveat sanitise routines

commit 443983cfd9be3c2fb095a63481bdca331c2adc15
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sat May 9 20:35:49 2015 -0400

    some of the multicast sockopts can take two struct types.

commit 03382f3b7d34592edfe3d4b5255755c5a49c4aee
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 7 00:13:46 2015 -0400

    generic_fd_from_socketinfo can be called with a NULL si

commit 5b9353fc8297a2fa93bfab098adbe24ece619d14
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 7 00:12:44 2015 -0400

    fix get_rand_socketinfo offset confusion

commit 20c95f965a6bbf1e7d65e28572736d2df566a920
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 7 00:11:33 2015 -0400

    fix syscall_sendto 'fd' argtype. (it's a socketinfo)

commit 47a38167af727aeee22d17c77579358dd1c7cc13
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu May 7 00:10:02 2015 -0400

    fix SOCKETINFO arg-decoder.
    
    at this point, we only have the fd, not the socketinfo

commit 72b1c14e0f3ab2866e6ba6b5062220f052b7c958
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed May 6 22:23:04 2015 -0400

    stub in accept4 setup on sockets
    
    disabled for now, needs more smarts to not block.

commit 67f9d22dc9fbceb48fca15b925b538ed8b38f8b0
Author: Vinson Lee <vlee@twitter.com>
Date:   Wed May 6 12:27:11 2015 -0700

    net/domains.c: Include compat.h.
    
    Fix build errors on CentOS 6.
    
      CC    net/domains.o
    net/domains.c:59: error: ‘PF_CAIF’ undeclared here (not in a function)
    net/domains.c:59: warning: missing initializer
    net/domains.c:59: warning: (near initialization for ‘domains[40].domain’)
    net/domains.c:60: error: ‘PF_ALG’ undeclared here (not in a function)
    net/domains.c:60: warning: missing initializer
    net/domains.c:60: warning: (near initialization for ‘domains[41].domain’)
    net/domains.c:61: error: ‘PF_NFC’ undeclared here (not in a function)
    net/domains.c:61: warning: missing initializer
    net/domains.c:61: warning: (near initialization for ‘domains[42].domain’)
    net/domains.c:62: error: ‘PF_VSOCK’ undeclared here (not in a function)
    net/domains.c:62: warning: missing initializer
    net/domains.c:62: warning: (near initialization for ‘domains[43].domain’)
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 381f62c0d2cd75a5cc48dc6d3a8f85ffd30cc4b7
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Wed May 6 12:50:29 2015 -0400

    update perf_event_open support for Linux 4.1
    
    This patch updates trinity support for perf_event_open to the features
    added in Linux 4.1-rc1.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 361d083a3cc7487e5929ebea4073b1497a08d608
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon May 4 18:57:46 2015 -0400

    only show all warnings on gcc 4.9 and above.
    
    Now that wheezy has been released, I don't care about anything older.

commit b8cf152270a1a2c6e0035e36b0d8e499d11eee48
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon May 4 18:55:59 2015 -0400

    Turn off -Wno-maybe-uninitialized in release builds.

commit 97a9a71c94b0785fc283c2c60fae52b218cab020
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon May 4 18:30:11 2015 -0400

    do SO_REUSEADDR before attemping to bind to random sockets

commit 453ab5374433a55eab2a3e5251754ca8456b70ae
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 21:57:02 2015 -0400

    Add a check for the broken ipv6 headers to configure.sh

commit faed7cdd718923036233232fe0ba0f528c895159
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 21:47:09 2015 -0400

    Add some more lengths for ipv6 socket options.
    
    This is going to break some distributions until glibc gets updated.
    
      CC    net/proto-ipv6.o
    In file included from net/proto-ipv6.c:12:0:
    /usr/include/linux/ipv6.h:19:8: error: redefinition of ‘struct in6_pktinfo’
     struct in6_pktinfo {
            ^
    In file included from net/proto-ipv6.c:7:0:
    /usr/include/netinet/in.h:534:8: note: originally defined here
     struct in6_pktinfo
            ^
    In file included from net/proto-ipv6.c:12:0:
    /usr/include/linux/ipv6.h:24:8: error: redefinition of ‘struct ip6_mtuinfo’
     struct ip6_mtuinfo {
            ^
    In file included from net/proto-ipv6.c:7:0:
    /usr/include/netinet/in.h:541:8: note: originally defined here
     struct ip6_mtuinfo
            ^
    
    unfortunately, I couldn't figure a good way to work around this
    without cut-n-pasting chunks of the ipv6 headers into trinity.
    
    until glibc is updated for your distro, hand-apply the patch here:
    https://patchwork.ozlabs.org/patch/425881/
    to /usr/include/linux/ipv6.h

commit 8d1bebf94435aa0c95cad88893a2935ccbc0ae60
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 21:14:57 2015 -0400

    add sizes for the ipv6 multicast socketopts

commit f32a09b51938bad28e49806b9898edd562622d0a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 21:12:48 2015 -0400

    default to sizeof(int) in ipv6 socket options

commit 34fb809c3cc3da4a1484b57d0ce7cf340d999240
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 21:11:20 2015 -0400

    default to sizeof(int) in ipv4 socket opts

commit 64c456c896f06c29133a9d835bc0d2364ee1eb7b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 21:08:54 2015 -0400

    convert ipv6 to use ip_option struct.

commit e3432a9b4e3c46af43f63ea6264f26d70d3fc5fb
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 20:34:18 2015 -0400

    mark IPPROTO_ICMP .type as SOCK_DGRAM

commit 257006a44453c5a77827e0d8bf4c438ec17bfa89
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 20:23:15 2015 -0400

    add TCP_TIMESTAMP & TCP_NOTSENT_LOWAT socket options

commit 35fd22c8c131f1430761824787bce3e6e63a1193
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 19:59:15 2015 -0400

    associate inet option lengths with names.

commit 396aba78e43e85f2be6d3a18419cdc013f31af60
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 19:36:25 2015 -0400

    use the same socket generator for ipv6 and ipv4

commit d0c69fc1f865c2dfccdf16311365adf790901c65
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 19:28:42 2015 -0400

    use PF_ definitions instead of hard-coding numbers

commit 830d58cc1c96facbdc5c869a4f340b49b9c936c1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 19:07:22 2015 -0400

    make the PF_ prefix for domains optional.

commit 776954337e058580aa16bb3b1cad2ccb25c42c9d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 19:04:08 2015 -0400

    remove a bunch of pointless defines

commit afa4ba95cb05c65420e26a21f004c18d2897f97d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 18:56:19 2015 -0400

    merge net/proto-pppol2tp.c into net/proto-pppox.c

commit 72c660ed3272a50bb04bf7b19d30fc05a3186680
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 18:53:32 2015 -0400

    do IP protocols as one of the random fallbacks in setsockopt

commit abdeba8b25bf7fd6f4d44181f595016bb2eb1a21
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 18:50:19 2015 -0400

    move the setting of so->level into do_setsockopt and helpers.

commit a97bf2b81d94d2fd596b0eefeb2e1d5c5bc4390e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 18:20:46 2015 -0400

    typos

commit 9ce322f03b16f8e1ea1695aca8336b549ccdcda1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 18:20:15 2015 -0400

    once we've matched the protocol, stop searching.

commit 1cbc765da2aed954e43f370db5570c49feb4039d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 18:16:18 2015 -0400

    remove detritus

commit d179a249f7100d1545295ebb0804eaab97c41137
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 18:15:01 2015 -0400

    merge phonet/pnpipe.

commit 6530853d41b0d4a1135e9ece9d17485ad8fd8683
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 18:13:26 2015 -0400

    ATM setsockopt cleanup
    
    Keep it simple and just use one function for both pvc and svc.
    Also throw away the aal file, as it doesn't even do anything.

commit 25fdaf22b0b2b320c738d956304bf765b045c042
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 18:07:53 2015 -0400

    remove pointless NR_SOL_ATM_OPTS define

commit ea1ea34280cbd99d1e50017dcae0b736a549eb80
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 18:04:17 2015 -0400

    match AF_PHONET up with pnpipe_setsockopt

commit 429f29632ecd80744dafa74c1d74cd34552c578d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 18:03:44 2015 -0400

    match AF_PPPOX up with pppol2tp_setsockopt

commit 6ba2400209500f65b3984670634117654920ddd2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 18:02:02 2015 -0400

    Remove possibility to pass numbers as -P arguments.
    
    Who even remembers these ? Just use the names.

commit 0fbfcf2d441097994cead3c538e8245abe76297d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 10:05:37 2015 -0400

    add missing argument to format string in inotify fd init

commit b9cff29f65688a907a751262151e35432c2eccca
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 29 08:57:52 2015 -0400

    add '\0' as a potential separator in generate_rand_bytes()

commit aa46fa767d250c59ff2912dbd02fbf1542013c78
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 23:05:10 2015 -0400

    consolidate do_random_sso

commit 8ea020c32a79c8e71520206bad0b63e3bf79ea69
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 22:30:23 2015 -0400

    invert tests in do_random_sso for code clarity

commit 22dbd7923cd89a15bee2bf34afdef40e54876da7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 22:28:12 2015 -0400

    occasionally do the generic socket options.

commit 6750723acc69507bbfb9acb5ad65aad19b828c91
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 22:19:09 2015 -0400

    extract call_inet_sso_ptr, and make it search for matching protocol

commit 1f984da75a99d898b3f341d7d6b87286cedd2a02
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:59:09 2015 -0400

    switch getsockopt to use ARG_SOCKETINFO

commit 1eff09853ff7b9b8becf31143b369807a0dbbf5b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:54:06 2015 -0400

    check for timerfd_create returning ENOSYS

commit c473bddf5727844f4bce7ad1e9b3a45a94b7d16f
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:52:12 2015 -0400

    remove unnecessary includes

commit 163aecef35252b37ad34fdd259b2741df8a174e4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:51:29 2015 -0400

    remove unnecessary includes

commit 07c508814ca8e41508000726f2aeca6ab0a0a4b8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:50:56 2015 -0400

    remove unnecessary includes

commit b1883c977a91dd4eec5a4a5bb4f803fe6f3ca510
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:50:41 2015 -0400

    remove unnecessary includes

commit a6a4228ffbeef5cc4cbbfee7039706da1882d4a8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:50:28 2015 -0400

    remove unnecessary includes

commit 9738787b352e6264d68e6462a9e3126bfb490c05
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:50:04 2015 -0400

    remove unnecessary includes

commit 8649a69b29f228c99892fcf9bcf6dafd530b26b0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:49:37 2015 -0400

    remove unnecessary includes

commit ac746d7bb197cfb72b48e4edf8e7115be4fc686b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:49:02 2015 -0400

    remove unnecessary includes

commit 7e1f5aba54e8c21dcd13d41467c5cbc618f90edc
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:48:48 2015 -0400

    remove more unnecessary includes

commit c7c1a862749ea546c80b323bf2d603fb60dbb053
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:48:14 2015 -0400

    fix function name

commit a5435cc61e431c0506c6eac67b6e938e13b178f2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:47:51 2015 -0400

    remove unnecessary includes

commit 9ebc35e5e4c4f7a6bb0dc3fcc720e654f418a765
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:47:15 2015 -0400

    add inotify fd provider

commit 7fa2b78064eab277afababd9c571e66604e71b75
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:31:49 2015 -0400

    clear send() buff ptr on failure, so we don't free() junk

commit 6642bbc12cb1a256216d046504159550b8c40f26
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:30:44 2015 -0400

    switch send() family of syscalls to use ARG_SOCKETINFO

commit 1617fde5431c5c155f1ebf5a83b43bd1b1456677
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:27:22 2015 -0400

    switch recv family of syscalls to use ARG_SOCKETINFO

commit 937e7fcd6d0d3b304c66044a4d8aeeb3ada61645
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:22:51 2015 -0400

    switch getsockname to use ARG_SOCKINFO

commit f9f24270a05719f0ced752e335b753360d4d3746
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:21:35 2015 -0400

    switch getpeername to use ARG_SOCKETINFO

commit 2046709e041e636163112eefd1703c19b78f8949
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:20:26 2015 -0400

    switch connect to use ARG_SOCKETINFO

commit b99fcbcbb7d7cdbf71f26d3cdffba7318874e7c9
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:19:06 2015 -0400

    switch bind() to use ARG_SOCKETINFO

commit bab0f25bbe041f7d04007d8c1cc3782eee984120
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:17:39 2015 -0400

    switch accept4 to use ARG_SOCKETINFO

commit 50872b90f184ece622b51a49453d924e42ff7db8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:16:08 2015 -0400

    switch accept() to use ARG_SOCKETINFO

commit e0012330c1a621f9ab0fea75459469233c866818
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 21:15:29 2015 -0400

    add generic_fd_from_socketinfo()
    
    This will be used for networking related functions that don't
    care about the triplet, but just want an fd (most the time).

commit aa1177a5d70d674a57d5802885d667be46d47829
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 20:37:50 2015 -0400

    fold ip_sso_demultiplexer into do_setsockopt
    
    Special casing AF_INET this way will save us having to pass down
    the triplet to all the different setsockopt functions.

commit 8ef5e833cd29358fa1ca364a0a5ce1cd5503081d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 20:36:51 2015 -0400

    split out the calling of the sso function ptr to its own routine

commit 96b665c49f4e27d4603c63aefdede152eeeebd25
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 20:14:16 2015 -0400

    Fix up the version.h generator for when we have .git, but no git.

commit f211ae02167055ee8fb45c2eec65a15ab022a786
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 20:01:20 2015 -0400

    instead of looping in do_random_sso, fall back to the generic socket opts.

commit c4b18a00d00063acbb046af4bece3b45829c9561
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 19:58:57 2015 -0400

    add protocol definitions for the AF_INET setsockopts.

commit 5be90a32ea7cd963f481ddc578adb3fbf2c326d2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 19:46:19 2015 -0400

    still do the random setsockopt() function in same fallback cases.
    
    We can get to do_setsockopt with a NULL triplet, if we get passed
    an fd that isn't a socket for eg.
    
    Might as well also use this fallback if we land on an unimplented sso func.

commit ef0bed41d5c128811ed0920354da18a74e7085a6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 19:36:51 2015 -0400

    match the protocol of the socket to the right setsockopt function

commit 53edc03f3d6f31b0e1c599554e6984edeaa4ed9c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 19:31:51 2015 -0400

    reorganize setsockopt function pointer array
    
    - Add in the associated family, for later comparison
    - split out all the AF_INET socket types to their own struct

commit 8b082ef5a2aab3944ad692ecdbe6ddae347ea737
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 19:08:56 2015 -0400

    sort, and add placeholders for all socket families

commit a8b4334219cdc090a2b50a85918d711a59ed39d1
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 18:52:44 2015 -0400

    centralise SOL_* definitions into net.h

commit d3afbc05e28ffb05b6faf40f537693a05d6da0e6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 18:31:54 2015 -0400

    pass down the socket triplet to setsockopt
    
    We don't do anything with it yet, but this at least wires things up.

commit 2186621f7a6f0f148aa57fd1b419c257e8e86e9c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 18:29:40 2015 -0400

    enhance generate_rand_bytes number generator
    
    Sometimes generate negative numbers
    Insert separators (commas/spaces) at random intervals.

commit 48ce62ec7684d4438a0d0a65eadb6b91864a7c65
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 28 18:29:03 2015 -0400

    don't evaluate switch each iteration in generate_rand_bytes()

commit cf78a7f38ceb26109b7443e6cd8178afefac26e2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Apr 27 23:46:20 2015 -0400

    remove devel/release targets
    
    setting the DEVEL var is now done in the Makefile automatically
    as long as the version is set to *-pre

commit 23bec2ad975f2f1e2fe4fdcb9f779fbdec87113c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Apr 27 23:42:56 2015 -0400

    Enable -Werror & devel mode

commit a0f944a3d48cfb817805acf0ac2ca7c556d72799
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Apr 27 21:24:56 2015 -0400

    invert early error path in write_socket_to_cache

commit 46f8f4e8a607b8d4ad38d08e2e67c913bd34d10a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Apr 27 21:22:14 2015 -0400

    fds.c: scoping reductions

commit b96512b8d84decc88d154f369b32991ac357d9b7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Apr 27 21:10:59 2015 -0400

    cap the max length we copy into salg_name
    
    (shut up coverity).

commit 97bf8fb0eddfc4eef7682a97ee123eb650fd256d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Apr 27 20:51:08 2015 -0400

    add ARG_SOCKETINFO to arg decoder

commit e9313329df3cb8c64ab5d4e4ea1a03a45de8392b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Apr 27 19:02:23 2015 -0400

    in some cases, we'll still want to pass a non-socket, just for fun.

commit f6a499ccb5449d9589be7041f441fbd93daf65c3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Apr 27 18:58:52 2015 -0400

    introduce ARG_SOCKETINFO
    
    For networking related syscalls, passing anything but a network
    socket as an fd is a waste of time for the most part.
    
    This will also allow us to do things like match up the right protocol
    for things like setsockopt() with the right kind of socket.

commit d3ec8c34d6470bab23a0fb3cfde46623a8737ba0
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Apr 27 18:37:50 2015 -0400

    bare beginnings of proper proto-alg sockaddr generation
    
    obviously lots more to do here, for a start adding in all
    the other algorithms.

commit 8d2a617ef846e987606792019f6748fc2268ea4d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Apr 27 00:24:57 2015 -0400

    add FALLOC_FL_INSERT_RANGE to fallocate

commit 3f1661f3643d33e8219a2772f81c57f71037e1d7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Apr 27 00:22:31 2015 -0400

    split out the write-socket-to-cachefile code to its own function
    
    to reduce size of generate_sockets() a little.

commit 9ea4575bd8aa8f5e515eb5ece663b47ee214a996
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Apr 16 11:23:52 2015 -0400

    fix off by one in domain parser

commit b55dc24a2f6561e5894f7f8ce8febda86383b118
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Apr 16 11:22:37 2015 -0400

    add PF_IB

commit 247cf75c582f668630252e48f54f0831fcf388f7
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Apr 16 11:22:00 2015 -0400

    Add PF_MPLS

commit e4943a72be2b4fedc2d4959591188be2ed619283
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Wed Apr 15 22:39:20 2015 +0300

    DRM dumb buffer file descriptors
    
    Use the DRM_IOCTL_MODE_CREATE_DUMB ioctl to make "dumb" buffers; then use
    the DRM_IOCTL_PRIME_HANDLE_TO_FD ioctl to get file descriptors.

commit 8d969f5c3b61c583370858df72dd52910201cbf2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 15 09:06:45 2015 -0400

    replace grepping through cpuinfo with nproc invocation

commit 4fb493c9e29b88c5ab965aa5be66fa95bd69a2ff
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 15 09:06:26 2015 -0400

    silence some new noise in dccp.

commit 0a981f3c42a8ec2bc36e2fb7fc5e9d2af0efa5a8
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 15 09:06:14 2015 -0400

    s390x: add execveat

commit 7a80f08ba239ba43159d9a7ed84d6d9722d2018d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 15 09:06:04 2015 -0400

    s390: add execveat

commit a2a63e05024aff6b64b981fe5d643cce9b13a8f2
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 15 09:05:52 2015 -0400

    parisc: add execveat

commit 354a627379b002a902978dabd5b49c2634989c1d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 15 09:05:36 2015 -0400

    mips: add new prctl's

commit 678a6c2529a3d779e545968546ba27b966f794df
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Apr 15 09:04:58 2015 -0400

    remove .git repacking from tarball target

commit b46df7e77540bc3b05758c213dc329e51ffc9c1d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Apr 7 21:48:57 2015 -0400

    ipv4: just default to SOCK_RAW if unspecified

commit f05db51dd7a44e96338e7fc4f77e41b8e4959dbd
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Apr 6 21:08:53 2015 -0400

    beginnings of some code to match up protocol/type fields of socket()

commit 43f19e8cbee6541be664d05b941dcd1511c8f83d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Apr 6 20:52:35 2015 -0400

    remove another pair of pointless macros

commit e877b8c686ac3fa0e0f28e75329b29cb91bc4e76
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Apr 6 20:49:49 2015 -0400

    remove only-used-once define.

commit 79efd382d904984bca6610815afbb31e76892d7d
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Apr 6 20:48:55 2015 -0400

    add a bunch of missing ipv6 socket options

commit 7373b3d7c629acddac89bf4815a654b167150503
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Apr 6 15:31:20 2015 -0400

    netlink: nl_pid is a port id, not a process id

commit 3a220715e9b022542421c985b40b3389e7067c60
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 25 17:19:33 2015 -0400

    ipv4: avoid potential divide by 0 when using /32 masks

commit 3cb6c395cb292c7e76fbd3affccf1e756629e392
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 25 09:12:28 2015 -0400

    convert new_ipv4_addr to use a table instead of switches

commit c2aedd1d5bef967b784ec01b3467176f394adbab
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 25 00:46:35 2015 -0400

    split up the ipv4 random address generation

commit 32aec8d1d302ebefe58aeb07a7c388c1f5bcea91
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 25 00:38:25 2015 -0400

    ipv6: formatting cleanups

commit 8a7094ac44c7409c3e721dbc43ca3a7439403c08
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 25 00:36:53 2015 -0400

    ipv6: use arrays rather than nested switches of doom

commit d3c7d548605839739184dcb9849a80fbc5e7adf6
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 25 00:18:27 2015 -0400

    add support for stateless IP/ICMP translation (SIIT) ipv6 addresses

commit 7f0e3a19caa35b55e5252f391a8595f31341780e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 25 00:16:49 2015 -0400

    simplify v4-in-v6 using inet_pton

commit 5f443e7a4aa16f60b7ec1bed71550b94211f87f3
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Mar 25 00:13:15 2015 -0400

    use inet_pton to simplify code

commit 2cad6d3ab9d4cf577f5e6328c1214d24b972d3ab
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Mar 24 18:59:13 2015 -0400

    ipv6: only generate v4 addresses where necessary

commit 5f9d4f1cb4f7541f38729e3c20185bc5aa37701b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Tue Mar 24 18:58:22 2015 -0400

    generate some more ipv6 addresses other than localhost

commit 25e6db510cf4e5d5c36f2fe3aa5c11eca5c710b9
Author: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Date:   Wed Mar 18 09:58:46 2015 +0000

    vt.c: add missing include to fix building with uClibc
    
    "#include <linux/serial.h>" is needed in order to build trinity
    successfully with uClibc. Otherwise it would fail displaying an error
    message like this one:
    
    ioctls/vt.c:152:8: error: invalid application of 'sizeof' to incomplete
    type 'struct serial_rs485'
      IOCTL(TIOCGRS485),
    
    ioctls/vt.c:155:8: error: invalid application of 'sizeof' to incomplete
    type 'struct serial_rs485'
      IOCTL(TIOCSRS485),
    
    Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>

commit 6a7e8fd6d70f3ec87641490d8dc8c4d39fd84752
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Fri Mar 13 13:24:00 2015 -0400

    clarify the intent of an extra case in the rand64() limiting switch

commit 1932bbc98b9700a6a73b781533f6417ce62b7411
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Mar 12 10:47:31 2015 -0400

    signedness fix

commit 8e76beb70861af2e6ad26825503a457f5a8c905e
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Mar 12 09:54:52 2015 -0400

    ignore return from fcntl when randomly setting flags.

commit ca5410f776acddf1f65c0e0cb99bf0062ccc9e6c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Mar 12 09:53:57 2015 -0400

    proto-rose: Fix overrun pointed out by coverity.

commit c3516c50c2fb60c553652646cfe25f0a9e098131
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Mar 12 09:23:17 2015 -0400

    make 'make scan' use NR_CPUS

commit 65885f94d33ecb12944d90ed40f0fbcc8942f00c
Author: Tyson Smith <tysmith@motorola.com>
Date:   Fri Mar 6 14:36:30 2015 -0800

    Make RAND_RANGE() a macro
    
    Renamed to uppercase to match other macros. Removed swap call. This should
    be done outside of the call.

commit aa2d2ea8d58569676d457c1c5e2bcc65ed110a9d
Author: Tyson Smith <tysmith@motorola.com>
Date:   Fri Mar 6 14:35:38 2015 -0800

    Rename rand_bool() to RAND_BOOL() to match other macros

commit 607fbd4c644f9a98a4363e412ef82e6e6a70def6
Author: Tyson Smith <tysmith@motorola.com>
Date:   Fri Mar 6 14:35:14 2015 -0800

    Decrease the likelihood of performing mangling operations
    
    These operations provide value however if they happen too often it prevents
    the use of unmodified values that are generated that are more significant.

commit d8e95523b72cd7ad825c5b919c0e355312a2321d
Author: Tyson Smith <tysmith@motorola.com>
Date:   Fri Mar 6 14:34:09 2015 -0800

    Actually flip the sign of the value (1 -> -1)
    
    I believe this is what was intended, previously the high bit was just being set.
    Perhaps it was meant to be toggled? If that is the case we can add that too.

commit e22220c08b488e16cfbd882ce56330f047657abb
Author: Tyson Smith <tysmith@motorola.com>
Date:   Fri Mar 6 14:33:48 2015 -0800

    Cleaned up rand32() and remove __rand32()
    
    Removed the block at the beginning of rand32() that just basically |s and ^s
    rand() with rand() or a more interesting number to basically get rand again.
    This is just an expensive way of getting a random number. Merged the small
    switch case from __rand32() in to rand32() now that there is room.

commit 313f3dca5cacb06dcee753734637068fbdd9e579
Author: Tyson Smith <tysmith@motorola.com>
Date:   Fri Mar 6 14:32:21 2015 -0800

    Cleaned up rept8() and renamed to rept_byte()
    
    This function basically generates 256 different values. These values do not
    seem to have much significance and unless it has triggered results in the
    past it should likely be removed.

commit 85c3fa1ae70e1555cfa9ea825bc3abfa1f64048e
Author: Tyson Smith <tysmith@motorola.com>
Date:   Fri Mar 6 14:32:06 2015 -0800

    Added RAND_BYTE() macro
    
    This will prevent open coding rand() & ff to get a random byte and the
    off-by-one's that accompany it (rand() % ff or rand() & 255). Made
    substitutions where possible which led to fixing existing off-by-ones.

commit f38c3483f8d36b6ae6f127ddad1bd577e81914c9
Author: Tyson Smith <tysmith@motorola.com>
Date:   Fri Mar 6 14:31:21 2015 -0800

    Make rand_bool a macro
    
    Since rand_bool is used a heavily used operation simplifying it may
    shave off some cycles here and there. Also replaced %2 with &1 with the
    same intention.

commit 184329743bf536f2f2ccc8483f3534fe965b3cfc
Author: Tyson Smith <tysmith@motorola.com>
Date:   Fri Mar 6 14:30:39 2015 -0800

    Remove static from rand_single_bit()
    
    Added a check to verify input value is between 0 and 32 or 64 bits
    depending on __WORDSIZE.

commit e434b97b8230673915002898328e2fd12da08e6d
Author: Tyson Smith <tysmith@motorola.com>
Date:   Fri Mar 6 14:29:59 2015 -0800

    Remove rand8x8()
    
    Removed rand8x8() function since it is essentially the same as a call
    to RAND_32 or RAND_64.

commit 6ba7621cd71e0301b2ea103dc9abaf393d74a3ff
Author: Tyson Smith <tysmith@motorola.com>
Date:   Fri Mar 6 14:29:29 2015 -0800

    Created RAND_32 and RAND_64 macros.
    
    These macros will generate a random number between 0 and 2^32 - 1 or
    2^64 - 1 respectively. There is also a compile time check to verify
    that a full 2^n bits of randomness will be generated (depends on RAND_MAX).

commit 53f7cc204aaadcd0d7fc64a4e1dbecf7138f7524
Author: Tyson Smith <tysmith@motorola.com>
Date:   Fri Mar 6 14:27:22 2015 -0800

    Created ONE_IN(x) macro
    
    It takes a positive value and it can be used to express how often a branch
    should randomly be taken. Made substitutions where possible.

commit 884dbeecfd1be073d76f8e4240508ff9625abd77
Author: Tyson Smith <tysmith@motorola.com>
Date:   Wed Mar 4 13:12:37 2015 -0800

    Removed overlapping values in get_interesting_value()
    
    Added 50/50 chance of skipping 64bit modifications code block. This should
    likely be closer to 1 in 4 odds of entering this branch, perhaps less if
    these values haven't been triggering issues in the past.

commit adaae913f961910a6743c2b0a50622b5471f55ae
Author: Tyson Smith <tyson.w.smith@gmail.com>
Date:   Wed Mar 4 13:12:17 2015 -0800

    Simplify get_interesting_32bit_value()
    
    Removed values that overlap with values generated but the other functions.
    Added multiple of page_size case.

commit 75bb4146cf1e00f98fdc478bed9688a28e1cac32
Author: Tyson Smith <tyson.w.smith@gmail.com>
Date:   Wed Mar 4 13:11:54 2015 -0800

    Simplify get_interesting_16bit_value()
    
    Removed values that overlap with get_interesting_8bit_value() and others
    that are covered by the 2^n case and 2^n with plus_minus_two().

commit a115fe5cf6c144b285472164b65e3b63c0c0957e
Author: Tyson Smith <tyson.w.smith@gmail.com>
Date:   Wed Mar 4 13:11:18 2015 -0800

    Make get_interesting_32bit_value() static
    
    All calls to get_interesting_32bit_value() should now be made to
    get_interesting_value() to help simplify generating interesting values.

commit e253aa9acc614609b47b217560979ca7e18e19e8
Author: Tyson Smith <tyson.w.smith@gmail.com>
Date:   Wed Mar 4 13:10:54 2015 -0800

    Unnest the get_interesting_<Nbit>_value() calls
    
    This will help evenly distribute the calls to each
    get_interesting_<Nbit>_value() functions. It will also make the modification
    to the distribution easier in the future if tweaks are needed. Move call to
    plus_minus_two() from get_interesting_32bit_value() to get_interesting_value().

commit 08a11ab4309e7ba443ee2747fb38745f16772b6f
Author: Tyson Smith <tyson.w.smith@gmail.com>
Date:   Wed Mar 4 13:10:28 2015 -0800

    Simplified get_interesting_8bit_value()
    
    2^n case will now cover up to 128 where previously it only went to 64.
    Removed -128, 127, -1 cases, they will be covered by the 2^n and 0xff
    cases with plus_minus_two() call.

commit fa9ac9663678697872c48b1018779772c7e9abd6
Author: Tyson Smith <tyson.w.smith@gmail.com>
Date:   Wed Mar 4 13:09:55 2015 -0800

    Remove value modification decision from plus_minus_two()

commit 5995417e75d644f143d23287a69f70eed233c76b
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 2 11:08:20 2015 -0500

    1.5 release

commit 8ba528cde9123b60d0795eaa6d55de9ebcc16243
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Mar 2 11:07:11 2015 -0500

    Disable -Werror & devel mode

commit e66c85360abc440c887dc90165b8fc5120ac3e4e
Author: Vinson Lee <vlee@twitter.com>
Date:   Wed Feb 4 17:06:42 2015 -0800

    kvm: Add ifdef around IA64 ioctls.
    
    Fix build error.
    
      CC    ioctls/kvm.o
    In file included from ioctls/kvm.c:8:0:
    ioctls/kvm.c:115:8: error: ‘KVM_IA64_VCPU_GET_STACK’ undeclared here (not in a function)
      IOCTL(KVM_IA64_VCPU_GET_STACK),
            ^
    include/ioctls.h:53:15: note: in definition of macro ‘IOCTL’
      { .request = _request, .name = #_request, }
                   ^
    ioctls/kvm.c:116:8: error: ‘KVM_IA64_VCPU_SET_STACK’ undeclared here (not in a function)
      IOCTL(KVM_IA64_VCPU_SET_STACK),
            ^
    include/ioctls.h:53:15: note: in definition of macro ‘IOCTL’
      { .request = _request, .name = #_request, }
    
    IA64 ioctls were removed in Linux 3.19.
    
    commit 6b397158d07f885154b871a15f879d25b3de7579
    Author: Radim Krčmář <rkrcmar@redhat.com>
    Date:   Thu Nov 20 14:43:18 2014 +0100
    
        kvm: remove IA64 ioctls
    
        KVM ia64 is no longer present so new applications shouldn't use them.
        The main problem is that they most likely didn't work even before,
        because of a conflict in the #defines:
    
          #define KVM_SET_GUEST_DEBUG       _IOW(KVMIO,  0x9b, struct kvm_guest_debug)
          #define KVM_IA64_VCPU_SET_STACK   _IOW(KVMIO,  0x9b, void *)
    
        The argument to KVM_SET_GUEST_DEBUG is:
    
          struct kvm_guest_debug {
            __u32 control;
            __u32 pad;
            struct kvm_guest_debug_arch arch;
          };
    
          struct kvm_guest_debug_arch {
          };
    
        meaning that sizeof(struct kvm_guest_debug) == sizeof(void *) == 8
        and KVM_SET_GUEST_DEBUG == KVM_IA64_VCPU_SET_STACK.
    
        KVM_SET_GUEST_DEBUG is handled in virt/kvm/kvm_main.c before even calling
        kvm_arch_vcpu_ioctl (which would have handled KVM_IA64_VCPU_SET_STACK),
        so KVM_IA64_VCPU_SET_STACK would just return -EINVAL.
    
        Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
        Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 99beb663c8730e3bacfed9261cbabd3625871366
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Sun Feb 15 19:54:52 2015 -0500

    remove the taviso random function
    
    This came from an era when I was trying to figure out why iknowthis
    was finding bugs that trinity wasn't. It turned out to not be
    because of the different random function, but because we kept
    hitting other bugs first due to the differences in our random
    generation. Fix the other bugs, and run trinity for long enough,
    and you'd also hit the iknowthis bugs.
    
    (There were also a few ->sanitize improvements for some syscalls
     along the way).
    
    Either way, this hasn't actually been useful for over a year, and
    just adds pointless complexity to the random code.

commit df98727e460130e2f94a98bf0523447bff852202
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Feb 9 20:42:47 2015 -0500

    parallelize coverity check

commit bf644a67d18b20ed9a416f0b81f1b1e3adf6299a
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Tue Feb 3 12:01:23 2015 -0500

    update perf_event_open() for Linux 3.19
    
    This updates the perf_event_open support to match changes made with the
    upcoming 3.19 release.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit bb91cbb61fe2fd478f15c396723d8dad1e65d357
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 29 15:39:18 2015 -0500

    freeptr missed an indirection, and was freeing the ptr to the ptr instead.

commit 7feade032d56f74541a207e316257c2e08a2cf5a
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Thu Jan 29 09:11:55 2015 -0500

    generate random data into the buffer for write()

commit 9c1df256e75f03077360ab26cc264836cf0d563c
Author: Dave Jones <djones@akamai.com>
Date:   Wed Jan 28 11:28:59 2015 -0500

    write/pwrite64 can use a similar interface to what we're doing in send
    
    Might be worth abstracting this out further later.

commit f8550800c54b8f2fc72ab198928625c8464372c3
Author: Dave Jones <djones@akamai.com>
Date:   Wed Jan 28 11:19:54 2015 -0500

    sendto can use the same sanitise/post as send

commit 1353a3067f8c0daa57b5600ab81ae80cd7f63d1f
Author: Dave Jones <djones@akamai.com>
Date:   Wed Jan 28 10:59:25 2015 -0500

    don't allocate a whole page in send_sanitise if we need less.

commit 3dd4352603874633d38d994b38174a672d057b42
Author: Dave Jones <djones@akamai.com>
Date:   Wed Jan 28 10:57:39 2015 -0500

    helper function to zero out pointers after freeing.

commit 32f45d8947c74e9c5eeb59b404e8abfbbc09d059
Author: Dave Jones <djones@akamai.com>
Date:   Wed Jan 28 10:44:15 2015 -0500

    bare-bones fuzzer for send()

commit ab98c4088860496bdc83bba3fcdf82b429830492
Author: Dave Jones <djones@akamai.com>
Date:   Fri Jan 16 10:30:09 2015 -0500

    wire up execveat for Arm

commit 61732e0c056895a311b95199a41439315d989d9e
Author: Dave Jones <djones@akamai.com>
Date:   Fri Jan 16 10:28:42 2015 -0500

    Add new MPX prctls

commit 69300f9f7d3bea13495b590ea42eebc4d64d1a69
Author: Dave Jones <djones@akamai.com>
Date:   Fri Jan 16 10:26:40 2015 -0500

    fix up prctl options
    
    Sort into order of definition in header.
    use ARRAY_SIZE instead of hardcoding.
    Add missing TIMERSLACK options.

commit 6f2f6239d9b8ba986212db9f5dc9c1818357d0a4
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jan 7 13:14:30 2015 -0500

    TODO/bugs update

commit 68aa63a64b71ef164d1661dfdcbdd41facbdfe68
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jan 7 12:50:12 2015 -0500

    alphabetical order for params.

commit 7941ac23f1d4944cb584d0ee5121c787920527ca
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jan 7 12:10:22 2015 -0500

    sometimes generate negative numbers as strings

commit 1e937d184d17209bb0d97608b08e4481ad576157
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jan 7 11:21:39 2015 -0500

    add new SO_INCOMING_CPU socket option

commit d703710747c769f4492272c2fdf24ba2b113d283
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jan 7 11:19:39 2015 -0500

    placeholders for new s390 syscalls

commit 5533151648f51fda13c38fb8de69907524e8e57c
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Wed Jan 7 11:18:05 2015 -0500

    add sys_execveat support

commit 75c306e652c8d915310cdddfbe75d7c5e91f5573
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 5 12:08:13 2015 -0500

    update email address.

commit ad12a5147f599bb7c55f5a1b78125b772b87fa81
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 5 12:07:16 2015 -0500

    remove double include in generate-args.c
    
    Spotted by Harrison <hbowden@securelabsllc.com>

commit 0cd9276573b677f1fae518ec2f28db17653115bc
Author: Dave Jones <davej@codemonkey.org.uk>
Date:   Mon Jan 5 12:06:53 2015 -0500

    rediff the fuzz noise patches

commit e3c266161a803df90ba9e771f27e22edc4a33fc1
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Sat Dec 6 22:25:27 2014 +0200

    fix syscalls-arm.h

commit 2720d1393ea31ca80aad5aa58a748e7ce294696f
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 17 12:04:46 2014 -0500

    add bpf for parisc

commit 14cf2a7b26b8d62e87ae7da27e2de9323578be22
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 13 14:29:02 2014 -0500

    update patch for new fuzz noise offender

commit 066d6c35517d36b7e2e7a17e397f6d5b3038c4f0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 11 12:00:00 2014 -0500

    dm_sanitise needs a writable address

commit caa612860d89e25cde2667af8022e80f93e0b3b0
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 10 12:09:38 2014 -0500

    wire up bpf for sparc

commit ce7d8cb02e851df77c43340ded894bd26e30d5fc
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 10 12:08:28 2014 -0500

    wire up getrandom, memfd_create and bpf for mips

commit 40699bb02723f08f17d7d0b7bbe612c597d44c9a
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 10 12:07:09 2014 -0500

    wire up sys_bpf for arm

commit 69f8e7d3be823776e8a5e1a79adfef51a4e7a4b7
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 28 00:11:51 2014 -0400

    fix off-by-one in testfile generator

commit 0b5ed21d660e717e25a1c2f90cae40bad564557d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 28 00:10:39 2014 -0400

    move set_rand_bitmask to random.c

commit ca29382886c33af4bd2a0d00c24390a6bb2d9309
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 27 23:47:09 2014 -0400

    make open_testfile's open flag chooser use set_rand_bitmask

commit 96d741e7708e4980234fae295537dccf1b898f9c
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 27 23:42:34 2014 -0400

    make random_fcntl_setfl_flags use set_rand_bitmask

commit a2c63f65957d264ba48c13d5c6573f83fa4905db
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 27 23:38:43 2014 -0400

    use random fcntl after fopen in the proc/sysfs fd generator too.

commit 91a06d73d74eb353e6ab9344c64d853dce047f9d
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 27 17:21:56 2014 -0400

    the new rand string code doesn't need to be in one-line functions.

commit 6578168f07393751d4285643bc5b4f396dd2b634
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 27 17:20:26 2014 -0400

    make generate_rand_bytes sometimes generate just strings of numbers

commit f6ec99ce53347d6d96b7b54c906a6c17741e5645
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 27 17:18:29 2014 -0400

    don't do the coin-toss on every loop iteration

commit 3619d3cc389d955ac5b0d4ae03f135028dde029a
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 27 17:16:55 2014 -0400

    make generate_rand_bytes sometimes generate printable ascii

commit aaac8366edba3351a178bc9610c10134a9e88a0f
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 27 16:00:41 2014 -0400

    free execve strings if the syscall failed.

commit 03ab8448b0865f50a8823bc09301e9905e33800f
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 27 12:39:32 2014 -0400

    scope reduction

commit e74366a56ea860df1ee81bfe099009041aea0ff5
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 27 12:39:03 2014 -0400

    use random fcntl flags if we open testfile fds with fopen

commit f246642638479a358eafc4f3d0eb37a3891b57b2
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 27 12:36:22 2014 -0400

    factor out the random fcntl SETFL flag generator

commit 578e36550447973616bd3f676a57e92741511002
Author: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Date:   Mon Oct 27 14:37:30 2014 +0000

    kvm: ifdef KVM_CREATE_SPAPR_TCE and KVM_ALLOCATE_RMA
    
    Some PowerPC toolchains have an old Linux headers which don't define the
    KVM_CREATE_SPAPR_TCE and KVM_ALLOCATE_RMA macros. To workaround this
    problem we enclose the lines which use these macros into ifdef blocks.
    
    Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>

commit 25f450042658061b7c229863ff9fb67d0b7a6b78
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 24 16:05:35 2014 -0400

    update the get_map() FIXME
    
    it's not really a big deal to have to care about initial_mappings
    before we have child processes running. It's still a lot better
    than when we had page_rand

commit 036ac27faf7325af95d1da2be69b212919d90b58
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 24 16:02:12 2014 -0400

    after setting mapping to MAP_RW, update its prot in the map struct.

commit 5dc2de29ac75b0d850d53c4689b775e20e129279
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 24 15:58:53 2014 -0400

    make perf_event_open allocate perf_event_attr struct with zmalloc
    
    and free in ->post

commit 475c29d47dcc61ff1443db0d5bef55b032182ee0
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 24 15:52:20 2014 -0400

    make setsockopt use zmalloc/free rather than get_writable_page
    
    This removes one case where get_map() needs to care about non-child mappings.

commit 3fca55ba63c780438ebf34f1ef21549497ab2e3e
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 24 00:11:28 2014 -0400

    partial revert some of the get_map() cleanups.
    
    Turns out we _do_ do get_map from main pid context.
    As the comment describes, the setsockopt code uses it when
    we do socket creation on startup.

commit 6a2d24ace1e2aa830dbf5c5e044f23274e95d495
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 23 23:55:50 2014 -0400

    header comment for maps-initial

commit a8d24232d43d6e21bb5c03378c96186efedcb497
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 23 23:53:56 2014 -0400

    add header comment for dirty_random_mapping()

commit ad27a7d0634902745794bbd8cc82c24a8619ff2c
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 23 23:52:23 2014 -0400

    fold the remnants of maps-fault back into maps.c

commit 201a2a07e46db32dc3204ca3f7229a0280b18c62
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 23 23:42:44 2014 -0400

    rename TRINITY_MAP_LOCAL -> TRINITY_MAP_CHILD

commit c979031c073a11694a74428b065b9c9b13e6f6a3
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 23 23:42:20 2014 -0400

    When we copy the initial map list, leave the ->type as MAP_INITIAL

commit 013a45e3b813a167fa68f277edba344a7ad2c55b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 23 23:38:51 2014 -0400

    rename TRINITY_MAP_GLOBAL -> TRINITY_MAP_INITIAL

commit 1d915b5805916d11788ecf42cfe8ea5b14136dc4
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 23 23:37:58 2014 -0400

    fix up some comments missed in the global->initial renaming

commit 97d2353a22dd73f564645291247783d07a25decf
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 23 23:36:40 2014 -0400

    rename 'shared mappings' to 'initial mappings'

commit f48479706592b1a5d0a3b3ed50e2e0a8931ded5f
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 23 23:33:51 2014 -0400

    fold __get_map() into get_map()

commit a754cebe97eb30f4c08bbb52d2e315340ee59887
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 23 23:31:07 2014 -0400

    remove bogus includes that crept in while debugging

commit 57b8a32b049c615ad4f0bb2be6ff21d6706b54a0
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 23 23:29:27 2014 -0400

    introduce init_child_mappings
    
    Make a copy of the global mappings on starting a new child.
    Any manipulations that happens on those mappings will happen to
    the copied map structs.
    
    This has the nice side-effect of simplifying get_map() as it now
    only has to ever iterate the one list.

commit d2b61d9768ce3ec5bd8987881dfaf1f30e5b616d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 23 23:02:10 2014 -0400

    move delete_local_mapping into delete_mapping

commit a2530dcd6780af309baf37843b979126aac48669
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 23 22:59:19 2014 -0400

    get_map is no longer called from main process.

commit b6ab5aaf219453c40b651b64043c6a5a6e209690
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 23 19:52:14 2014 -0400

    periodic dirtying of mmaps.

commit ed3e476e1e86437338f3ecd538f6172aa79c32bf
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 23:18:04 2014 -0400

    split up mapping read/write fault routines

commit d7a258d8808056f924e1721e5d22d7977ee8b253
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 19:24:57 2014 -0400

    Removal of page_rand and all supporting code.
    
    We now only use mappings for random pages.
    There needs to be more work done to periodically dirty mappings,
    as we only do that on mmap/mremap right now.

commit 7d63be970871beeed42b1962d2c5fcab622aa2b1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 19:20:37 2014 -0400

    move generate_random_page to maps-fault
    
    This is the only useful part of random-page.c, which will be going
    away when page_rand gets killed off.

commit 0b0b120ab27346bb5c00e60dcfbaa421a9f95c14
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 18:48:52 2014 -0400

    add missing include

commit 48ed156a083cdfb13a4380d3cb9144784dab0bc9
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 18:42:58 2014 -0400

    convert execve to use generate_rand_bytes

commit 29dea25a7f6295c9e5dcd0d714476c87250477f1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 18:39:19 2014 -0400

    move scsi ioctl away from page_rand

commit 2d127f0ae0327f406be46f99dcb1f8c225fd3769
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 18:38:34 2014 -0400

    move device mapper ioctl code away from page_rand

commit 26a6e2363dc2b020202a3861f6b02b8bcf724729
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 18:37:45 2014 -0400

    move autofs ioctl code away from page_rand

commit 5543f3ac0ea1521b27103669010ade8c6d2ab545
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 18:36:23 2014 -0400

    convert rose_gen_sockaddr to use generate_rand_bytes

commit 8945b88c17394c91b77de26e9a1d61ad198e11ba
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 18:35:00 2014 -0400

    convert packet_setsockopt to use passed down optval instead of page_rand

commit 0c52b133bea7a80082fe43dc2460cda55c696704
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 18:33:04 2014 -0400

    use generate_rand_bytes in ax25_gen_sockaddr

commit d5e8ccad10c9ca4e22f0ce4c2f223cd92a18a9c8
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 18:31:56 2014 -0400

    remove unnecessary maps.h include now we don't use page_rand

commit aae1cdd2220403e832dc283c928eb695b7ce0666
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 18:31:37 2014 -0400

    use generate_rand_bytes in x25_gen_sockaddr

commit 68f2aa7e40e7c1b21e2060059c48a27ec26ebdb6
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 18:30:41 2014 -0400

    use generate_rand_bytes in unix_gen_sockaddr

commit 3f7fc78d1b299b54fe5d9b33f6b78bdef3a2b431
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 18:28:49 2014 -0400

    add generate_rand_bytes function to generate a stream of randomness

commit 84d37c707bf955b77bc5b64f669938ed46407e32
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 18:05:17 2014 -0400

    sockaddr in post_sendmsg is now always freeable.

commit e8baa5224b6ecfcd7a3aa3b1c39725410e4779e0
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 18:01:07 2014 -0400

    fix generate_sockaddr allocation.
    
    This needs to be an actual allocation in the case we don't have a proto
    definition doing an allocation already.  This is because open_socket()
    will try to free it on startup.

commit af5f9a594dce06f5cca0a0f610f2b3885a4e8907
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 17:43:06 2014 -0400

    kill off ARG_RANDPAGE

commit e1c474ed3fe6125ff3c584a156602296c780017c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 17:38:36 2014 -0400

    make ARG_UNDEFINED sometimes return a mapping

commit e55cb92726a3d22cf5cefda4f448c0507f9af560
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 17:18:25 2014 -0400

    simply get_writable_address to just use mmaps.
    
    The leaking of zmalloc'd pages was pretty crappy, and now that
    we have better mmap support, this should give us good enough
    results.
    
    Also, it's another step towards being rid of page_rand

commit a0a34183c56425ef2e0872d0b3a5deaabb91b4ba
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 17:08:28 2014 -0400

    make x25_setsockopt use the optval passed down

commit 7211c7790df24382c09adac5cc095e487b527cdb
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 17:07:45 2014 -0400

    make udplite_setsockopt use the optval passed down

commit 6e246375a707fd38fc67cdee09dfe0fa73d9d9dd
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 17:06:59 2014 -0400

    make udp_setsockopt use the optval passed down

commit a74156728e0a444d2fce3eee73bcf419bc6a0dda
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 17:06:15 2014 -0400

    make ip_setsockopt use the so->optval passed down.

commit f2593807addf4a28cea3cf405100732450eca450
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 17:04:39 2014 -0400

    move generate_sockaddr away from page_rand usage.

commit ecf45c25830102e25107368c48aa0e7413107f07
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 17:02:40 2014 -0400

    setsockopt: so->optval needs writable memory

commit 9c76258ee5bf6b0bc0f3580151458026a1ea84fa
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 22 16:17:07 2014 -0400

    wire up sys_bpf for s390

commit 2b0d6280ee1462b1110333cfb5dbe4c03dbeb6c0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 21 23:30:39 2014 -0400

    mark some TODO's for page_rand removal later.

commit 4f330e5f5cca13fb0aa7e8431a08726e865932eb
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 21 23:23:58 2014 -0400

    fix optlen/optval confusion in proto-tipc setsockopt

commit 7611ceb81adafc82a3bc15bb8910e8e7b8678967
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 21 23:22:52 2014 -0400

    remove double inclusion of maps.h

commit 0ac1ad2dc0fbd5d0bc829b42f16e6c5e955ba9cc
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 21 23:22:01 2014 -0400

    fix optlen/optval confusion in proto-pppol2tp setsockopt

commit d5be94c4220292f85cd8dd416fcd16decab9d07f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 21 23:21:08 2014 -0400

    fix up optlen/optval confusion in proto-iucv setsockopt

commit b00d8f280c2b3a1bde55da5371be419a5f2ac18b
Author: Vinson Lee <vlee@twitter.com>
Date:   Tue Oct 21 17:04:03 2014 -0700

    ioctls/drm: Fix build errors due to undefined symbols.
    
    This patch fixes these build errors on Fedora 22
    3.18.0-0.rc0.git9.4.fc22.x86_64.
    
      CC    ioctls/drm.o
    In file included from ioctls/drm.c:20:0:
    ioctls/drm.c:274:8: error: ‘DRM_IOCTL_EXYNOS_GEM_MAP_OFFSET’ undeclared here (not in a function)
      IOCTL(DRM_IOCTL_EXYNOS_GEM_MAP_OFFSET),
            ^
    include/ioctls.h:53:15: note: in definition of macro ‘IOCTL’
      { .request = _request, .name = #_request, }
                   ^
    ioctls/drm.c:275:8: error: ‘DRM_IOCTL_EXYNOS_GEM_MMAP’ undeclared here (not in a function)
      IOCTL(DRM_IOCTL_EXYNOS_GEM_MMAP),
            ^
    include/ioctls.h:53:15: note: in definition of macro ‘IOCTL’
      { .request = _request, .name = #_request, }
                   ^
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit f81da69ef37e223c39d8922ae376454da3738d29
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 20 18:01:42 2014 -0400

    commentary for reap_child()

commit 0d90705b610936a7a824199407b9607b50316742
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 17 16:46:50 2014 -0400

    watchdog: fix off by one in adjtimex compensation.

commit 4e1682b566f84871d1bd0d3f300ad6c2a5761fb4
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 17 15:19:22 2014 -0400

    switch rand-page dumper to use pid in its filename
    
    mkstemp was only used to generate something unique. But we only
    really want one per process, because we'll keep hitting the
    same corruption, and dump it multiple times.

commit 9a823c8972e98573b3d7c3fa982bf4620ff3ae1f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 17 15:14:46 2014 -0400

    micro-cleanup in page-random
    
    use 3rd arg of for statement instead of incrementing in the loop

commit e90484b628c409c9c62bd82ab8edd0af3396f3f1
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 17 15:06:45 2014 -0400

    make the child reap itself before it exits.
    
    This seems to fix a bug I was seeing when running with _MALLOC_PERTURB set.
    The symptoms looked like a use-after-free in the child struct,
    which is puzzling as they are allocated during startup and never freed
    until we exit.  I don't fully understand it, but this is a nice
    code cleanup while also making the problem disappear.

commit f9b29192dee615852f72bd209cf942a63b7c23e7
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 17 15:03:32 2014 -0400

    watchdog: sanity check that shm->running_childs is correct.
    
    If we end up with nothing but empty pid slots, and somehow
    shm->running_childs isn't right, correct it.
    
    It would be nice if shm->running_childs was never wrong, but
    on occasion things go awry.  There's probably a bug or two
    in the child handling somewhere, but this works around it for now.

commit 0d7e928f54021aa5d132240eabfddba36192f726
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 17 15:01:52 2014 -0400

    watchdog: reap dead kids in the kill_all_kids loop
    
    Again, because we're spinning here, we may need to do
    periodic cleanup, like the watchdog main loop does.
    
    [ A future cleanup might be to split out the common parts
      of the two 'spin indefinitely' loops to a single function. ]

commit f23ca35ff6bb7cdcaf5c592af8400d879dd4bfc0
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 17 14:58:37 2014 -0400

    watchdog: check dead children aren't holding locks.
    
    As well as the main watchdog loop, we need to do this
    in kill_all_kids() as it can spin indefinitely.

commit 63d9dd2a15f83273129212924a315f26bfbf5f11
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 16 14:51:50 2014 -0400

    detect when syscall emulation isn't present
    
    I noticed that when CONFIG_IA32_EMULATION is disabled, trinity
    continually tries to 32-bit syscalls, which is pretty wasteful.
    After failing N times, we now disable all 32-bit syscalls.
    
    For now at least, n = number of pids * 2, which should give
    a pretty good chance of _one_ of those succeeding if emulation
    is present. We only need one success to note that we don't need
    to keep checking.

commit b66b48239b5223d15486cd736e26656fa259bf89
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 16 13:25:54 2014 -0400

    inline __syscall_return in the 32-bit path.
    
    Also update maximum errno

commit c1970e80f221df4914dee9bbd74d2adf241b6769
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 16 12:45:54 2014 -0400

    save r8/r9 across 32-bit syscall.

commit 186f0db4bea334c7ce57cea22cb2e668c43c18c2
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 15 12:03:20 2014 -0400

    wire up sys_bpf for ia64

commit 6b6400e973d96c02e297b6367a6941654f59e3f0
Author: Samuel Martin <s.martin49@gmail.com>
Date:   Sat Oct 11 18:15:18 2014 +0200

    scripts/gen-versionh.sh: fix VER definition
    
    Correctly check that the git program exists and the sources actually are in
    a git repository.
    
    Also simplify/invert the the logic git check logic.
    
    Signed-off-by: Samuel Martin <s.martin49@gmail.com>

commit ed5af7843d19056ebad954ba175c90a0f5c4781a
Author: Samuel Martin <s.martin49@gmail.com>
Date:   Sat Oct 11 18:15:17 2014 +0200

    scripts/gen-versionh.sh: rework git check and use
    
    Signed-off-by: Samuel Martin <s.martin49@gmail.com>

commit b67af24b58252ecb7b9e3eb2df68b27e3b55bc7f
Author: Samuel Martin <s.martin49@gmail.com>
Date:   Sat Oct 11 18:15:16 2014 +0200

    scripts/gen-versionh.sh: fix indentation
    
    Signed-off-by: Samuel Martin <s.martin49@gmail.com>

commit 59044ee4cfdb68c3f9d38537042c4eb55e8dee13
Author: Samuel Martin <s.martin49@gmail.com>
Date:   Sat Oct 11 18:15:15 2014 +0200

    scripts/gen-versionh.sh: rework VER definition (less programs spawned)
    
    Signed-off-by: Samuel Martin <s.martin49@gmail.com>

commit 7f648e4f8f63e267baa958c0f2b5cc9fa28390bb
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 13 12:23:16 2014 -0400

    remove no longer needed hunk of silencing diff.

commit 3d7d34227c7c2a44056d2ad2c91d44d71bc18834
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 10 12:25:11 2014 -0400

    add basic sys_bpf fuzzing.
    
    Lots more to do here

commit b4950ea3749d2a293543e80ae6dfd74f5599895a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 9 11:05:28 2014 -0400

    clean the version file from the right dir.

commit fa98e5ffb393e4172db652934be4b1b5afbacd14
Author: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Date:   Thu Oct 9 15:54:02 2014 +0100

    Add O_LARGEFILE definition to compat.h
    
    Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>

commit b13badf3c3fc046796594f0a6136577b15570da8
Author: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Date:   Thu Oct 9 15:32:19 2014 +0100

    Add SCHED_IDLE definition to compat.h
    
    Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>

commit 62eb18a2d776a99313b160de861e7b1004e77b42
Author: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Date:   Thu Oct 9 12:22:01 2014 +0100

    Add a check for rose headers
    
    Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>

commit cbcae80bf081e1af998d04c4b36cd0b6e6b89b59
Author: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Date:   Thu Oct 9 12:22:00 2014 +0100

    Add a check for netrom headers
    
    Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>

commit 67c689bb003615c25e02d40206d1c645a3050a73
Author: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Date:   Thu Oct 9 12:21:59 2014 +0100

    Disable ipv6 if there is no ipv6 support
    
    Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>

commit 3df2c224b8f9cfb7578312947dd97604d38e0764
Author: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Date:   Thu Oct 9 12:21:58 2014 +0100

    Disable icmp6 if there is no support for ipv6
    
    Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>

commit 28de63bf75e224fc90e9e6c02411b4a090f0674c
Author: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Date:   Thu Oct 9 12:21:57 2014 +0100

    Make ipv6 support configurable
    
    The user can enable or disable the support for ipv6 by setting the IPV6
    variables to "yes" or "no". If it's not used, then it defaults to "yes".
    
    Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>

commit dc8826f0a6aa3488d5fa66d18bc07ba9f878ff6f
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 8 13:24:35 2014 -0400

    add a check for appletalk headers

commit 292cdaad16071e24baf9873caffd43656cfb85bd
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 8 12:18:51 2014 -0400

    uclibc can be configured without backtrace support, so check for it.

commit ffaf4230b24c3f16f1821a0ea6dfc7cce683b40e
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 3 15:10:26 2014 -0400

    use mkstemp for pagerand dumper
    
    Otherwise we fail to write over the existing file if we've already dumped

commit 01ce8e378d00843825231f0d126df91096a8d56f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 3 12:07:13 2014 -0400

    parse argv in analyze-sockets

commit ce5d66629555ecfc5391005057f98b5a3d4e4288
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 3 10:42:07 2014 -0400

    check the whole page_rand redzone, not just the first byte

commit 502fe8e7eab7ca4d86392852fa1101104176464d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 3 10:29:14 2014 -0400

    dump the random-page to /tmp
    
    because cwd might be somewhere unwritable.

commit 1fbbc40d0d3920af9272cf9d6c803d73f5a473b9
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 3 10:27:16 2014 -0400

    show a backtrace if we hit the BUG in exit_main_fail

commit 2976de7ab9e073885dc8456b2d8a7ca5d82c6ded
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 3 10:26:06 2014 -0400

    remove unnecessary include

commit 3247648119a1416732d468365ec3352ca3730061
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 3 00:29:42 2014 -0400

    fix up warning in analyze-sockets
    
    This will be removed when I add argument parsing tomorrow

commit 811b424da18627f6d53a21616f1366eb593d1a92
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 3 00:29:36 2014 -0400

    add makefile for tools dir

commit d0cbf4b028c1f8dc9cd4843d7a80b02db89dcb77
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 3 00:25:07 2014 -0400

    ignore generated files

commit bcaade7cfec436b4b4e04028f46e5b4396a67294
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 3 00:23:49 2014 -0400

    add dumb tool to dump socket cache contents for debugging

commit 2d557f904ed1b18254cc0347fec8838f0a8be487
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 3 00:15:48 2014 -0400

    rename 'proto' to 'domain' where necessary
    
    To avoid confusion with the actual protocol argument to socket()

commit 5a96bedec59a6830de6ea8a502a704f95778bca2
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 2 23:02:40 2014 -0400

    remove unneeded include

commit d0d9f4e91e542de4c1105e7f435ed0455e37e328
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 2 10:57:48 2014 -0400

    ipv4: pick more random protocols in socket creation.
    
    the proto/type switch was inside-out, and should pick protocol first,
    given the much wider choice.

commit 9d6a0c91582ed85bf20de5582a74a79b1f549df0
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 16:57:30 2014 -0400

    sanitise support for poll()

commit 122b74699dbe92a9a322ec4fd9b5b615ba2e2a7e
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 16:13:30 2014 -0400

    make sanitise_mmap use set_rand_bitmask()

commit 603e638e9fccbc4b1171b17a85ef12b18052769b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 16:02:28 2014 -0400

    smarter fuzzing for select()

commit 19adeec66f8e289e29e77d06b03eb2a50b39b615
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 14:08:17 2014 -0400

    fix --group
    
    Syscalls weren't actually being enabled.

commit c437fb273ae62328cadba25a30e48672884b1e25
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 13:51:41 2014 -0400

    update -g help to mention vfs group

commit 45b647662f904bfff34ceaf3e671c70eb11cd4f5
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 13:41:10 2014 -0400

    remove all the 'no_files' code.
    
    This was a mess, and never really worked as intended.
    With the arrival of dynamic fd providers, it got even more broken.

commit ae97ba4cfb661d08439decc4e7227c52920a7f40
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 13:35:21 2014 -0400

    update the README a little.

commit e47462ece6f4b8d8f8171ca96c462ad02b00950e
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 13:26:00 2014 -0400

    move ia64 specific getunwind syscall to ia64 dir

commit 5e6b5a88a1fca970c904fc6a96a2ed7b4f60bc95
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 13:14:45 2014 -0400

    describe usage of poorly named 'BUFSIZE' define

commit 17980e6e37d80edf78d8a851981bc06e0f72201e
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 13:11:40 2014 -0400

    remove len argument from strip_ansi

commit 7719c87fcbd411a41bd0dd555801625d0ff68ab6
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 13:00:19 2014 -0400

    rename __output_syscall

commit 8d6521c235deead04d1af9459d5226a96437a84d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 12:54:41 2014 -0400

    add an --enable-fds option
    
    This is shorter than really long --disable-fds command lines

commit 8fca185e70fc177691396b1a513b0d319d81ff03
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 12:29:36 2014 -0400

    MAX_PIPE_FDS now accounts for both fds

commit 506c74cc7f67563921e54c8b41e10e174272f6d7
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 12:28:14 2014 -0400

    pipe fd creation improvements.
    
    switch to pipe2, and pass in variations of possible flags.

commit 9fb414719213ad38b6efe06024ee5c6f1fa34057
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 11:54:47 2014 -0400

    move sso_socket to fd-sockets.c as it's now the only user.

commit 601957afb21e45073e781bd093478276b31e6705
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 11:33:51 2014 -0400

    save child-local render buffer allocation for subsequent use

commit 56bf05dcfeba099d8b4eb25a8dac8cb3afbe561b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 11:30:07 2014 -0400

    render arguments to child local buffer first, then copy to shm.

commit ff6e0e4e1b9390a26b7d50dd05ddb0131fe7b77f
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 11:12:22 2014 -0400

    prebuffer sanity check across system calls

commit dcbfd5a1175ca655ba5bc4644bdc7d13bfca4a85
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 11:07:05 2014 -0400

    split argtype decoding out of render_arg into own function.

commit 20b2bd6a1c3845009a552b33c49860a87d8bbabe
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 10:35:48 2014 -0400

    simplify no_syscalls_enabled

commit b20658a636e00031fe94d3f38d09b36d41110576
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 10:18:54 2014 -0400

    update comment

commit 899150775f4afd8dd137e7cf369a57265b0f61ab
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 10:09:12 2014 -0400

    introduce for_each_arg helper, and use everywhere

commit e22ecdc71527b08437626ed7534f24d8046ba83d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 1 10:04:58 2014 -0400

    simplify

commit 17d2eb0ace1758dd08cc11ed41da5f97403ef700
Author: Dave Jones <davej@redhat.com>
Date:   Mon Sep 29 19:52:56 2014 -0400

    Revert "allocate the output buffers on child startup."
    
    This reverts commit 10e37a39ed8686cb1801adc42bfbb67ce39975ce.
    
    While this had the effect of avoiding the occasional corruption
    to the child structs, it had the unwanted side-effect of breaking
    the post-mortem code, because it could no longer dereference
    the buffers from outside the child process.
    
    I still need to spend some time hardening the output functions
    to not overrun these buffers, but since 1280b9854b, we should
    at least not hang when we hit corrupted states any more, which
    is a good enough stop-gap until I fix things up properly.

commit f4b1a920418c7112ec4ab57f00c32d9951feb476
Author: Dave Jones <davej@redhat.com>
Date:   Fri Sep 26 16:35:46 2014 -0400

    watchdog improvements in the face of corruption.
    
    There were a few cases where corrupt state would cause the
    watchdog to hang, looping sending kill(0) to child processes,
    never sending SIGKILLs.
    
    Instead of bailing kill_all_kids when we encounter corruption
    detected in reap_dead_kids, we do things in two phases.
    Reap the dead children first, and then if we continue to find
    corruption in the pid list, just ignore, and send kills to any
    children in the list that look real.

commit b4bc8b38aac2a800aa136069b0266995ba3d1c62
Author: Dave Jones <davej@redhat.com>
Date:   Mon Sep 22 18:13:35 2014 -0400

    remove some stale info about outstanding bugs.

commit 3fbb4131d24e8186923b4504776b24a854df1bf8
Author: Markos Chandras <markos.chandras@imgtec.com>
Date:   Fri Sep 19 15:11:55 2014 +0100

    maps: Rename MAP_{GLOBAL, LOCAL} to TRINITY_MAP_{GLOBAL, LOCAL}
    
    The MAP_LOCAL flag is actually a valid flag and some architectures
    make use of it according to the mmap manpage. More specifically,
    MIPS and Xtensa define the MAP_LOCAL flag in their userpsace mman.h
    header leading to build problems like this:
    
    In file included from syscalls/mprotect.c:6:0:
    include/maps.h:8:0: error: "MAP_LOCAL" redefined [-Werror]
     #define MAP_LOCAL 2
     ^
    In file included from syscalls/mprotect.c:4:0:
    /toolchain/mips-linux-gnu/libc/usr/include/asm/mman.h:37:0:
    note: this is the location of the previous definition
     #define MAP_LOCAL 0x080  /* Copy on fork/sproc */
     ^
    
    Therefore, we rename the definitions in trinity to avoid conflicts
    with the linux headers.
    
    Signed-off-by: Markos Chandras <markos.chandras@imgtec.com>

commit 2f71f52c25db19d65547ec352e1ffceac9d05189
Author: Markos Chandras <markos.chandras@imgtec.com>
Date:   Fri Sep 19 15:11:54 2014 +0100

    configure.sh: Respect CFLAGS from the environment
    
    Currently, configure.sh does not inherit any cflags from the
    environment. However, for MIPS, it's necessary to pass additional
    cflags to decide, for example, if we want a big endian (-BE) or
    little endian (-EL) build. As a result of which, we modify the
    configure.sh script to respect additional cflags passed by the
    user.
    
    Signed-off-by: Markos Chandras <markos.chandras@imgtec.com>

commit 39bcd64dc1f11978d9e94bf82586e9fb750787f4
Author: Dave Jones <davej@redhat.com>
Date:   Mon Sep 22 09:06:46 2014 -0400

    update parisc syscalls

commit 8b4044c7012f664bd205258c8875a4ddd99c8087
Author: Brian Haley <brian.haley@hp.com>
Date:   Thu Sep 11 21:35:37 2014 -0400

    Add missing EPOLLWAKEUP define to compat.h for older distros
    
      CC    syscalls/epoll_ctl.o
    syscalls/epoll_ctl.c:14:2: error: ‘EPOLLWAKEUP’ undeclared here (not in a function)
    
    Signed-off-by: Brian Haley <brian.haley@hp.com>

commit 1c14d5bc9f13b49cb93382b499ff004ea430835b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 10 17:20:48 2014 -0400

    put a random fd in the epoll struct too

commit 7e8d298a5630047fe7b6585df23b5c94a1d3bc50
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 10 17:19:53 2014 -0400

    flesh out generation of epoll structs a bit

commit 1c12821078c0a6dda9a6daf6d0ba76e70935ec38
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 10 13:50:49 2014 -0400

    remove pointless chmod
    
    Any permissions problems will be on the directory, not the filename,
    so modifying the perms of the file before we unlink it is pointless.

commit 3e1813f26ca67dde74e319a7d7db31cf0b39167d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 10 12:18:04 2014 -0400

    fix mmap2 definition in ppc syscall table

commit 7acaf1f9d89943a5de6c1eb907f97ea8ad7ea682
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 10 11:41:35 2014 -0400

    if testfile creation is spinning without progress, output errno

commit d39f86b133917fe54ca7c817fa08ff41b4731cd9
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 10 11:38:48 2014 -0400

    if testfile creation fails, retry instead of failing hard

commit 3a3a9abdcadc5820d6a2e21b241644ee5b4189c6
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 10 11:37:01 2014 -0400

    remove old testfiles on startup if they exist

commit 3efd51ea9953909574a4491cef99e38b19442e71
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 10 11:35:49 2014 -0400

    re-use testfile filename buffer instead of re-allocating it

commit e98b7af8e6c4814a458f66998d83e8ef3352e2f9
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 10 10:45:40 2014 -0400

    Fix mbind flags definition.
    
    What we had in the 'flags' field should have been OR'd into the 'mode' field.
    Move them to the right place, and then add the actual flags that should have
    been in the flags field.

commit 43b069eaf2a887a3e55857dd12d0ce0054b3ce18
Author: Michael Ellerman <mpe@ellerman.id.au>
Date:   Wed Sep 10 14:26:54 2014 +1000

    Fix powerpc build on Ubuntu
    
    In commit df1508cd4151 "move rtas syscall definition to ppc specific
    directory" the rtas syscall got moved to syscalls/ppc, and we added a
    rule in the Makefile to pick that up based on the output of
    gcc -dumpmachine.
    
    That rule only looks for "ppc*" though, on Ubuntu gcc emits
    "powerpc-linux-gnu", or "powerpc64le-linux-gnu".

commit 2d8df308b5e19e13f39bc79e7ec572d8aab6cb45
Author: Michael Ellerman <mpe@ellerman.id.au>
Date:   Wed Sep 10 14:26:53 2014 +1000

    Fix build break when SYS_memfd_create doesn't exist
    
    In memfd_create() if SYS_memfd_create is not defined then uname and
    flag are unused and gcc complains.

commit 2f0ddf9f8b578f3af59a0c7f41903d4c22a2c01b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 2 13:51:43 2014 -0400

    for the char's we can go further, and just use rand()

commit d265eaaa56c388002d96b1bedea5788cbf0ba2fd
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 2 13:49:50 2014 -0400

    when creating random ints/chars, we only need rand32() output

commit 1e980b154c6903444c592feabb7372b7d79b1864
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 2 13:34:14 2014 -0400

    update alpha syscalls list

commit 9549da0e9e2f3dd5bde364c50ef4f1fa55063038
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 2 13:32:40 2014 -0400

    abort if we fail to open logfiles.

commit 1905108c489826c34e127567607652d3bd176280
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 28 12:13:24 2014 -0400

    update s390/s390x syscall lists

commit a2d78ab0e374b5f4ba38bc3d9eb75e271e6c77c1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 27 11:36:34 2014 -0400

    change sizeunit to print bytes when passed a non KB aligned size

commit ff2b7ffa8c02fc3428402982005ef3fc10ffb5ca
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 27 11:35:58 2014 -0400

    fix possible off-by-one when printing units of bytes

commit 2a73982d860decc7e38eff7496aab97c5060548d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 26 14:55:17 2014 -0400

    pass a child struct into init_child

commit 2e3d256a26b615551b38b6177de4eb8fd3fc7114
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 26 00:08:21 2014 -0400

    watchdog: if main died, clear its pid.

commit 3f05e97b3b3cc86009ee03bd4806faf5d4370c87
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 26 00:07:30 2014 -0400

    munge the size in common_set_mmap_ptr_len

commit 5c925fc1e64b047d4c87b0e199c590a02f188d18
Author: Dave Jones <davej@redhat.com>
Date:   Mon Aug 25 21:46:27 2014 -0400

    move the debug flag into the shm.
    
    This allows us to gdb to the mainpid for eg, and enable the flag
    when for whatever reason we can't gdb to a child process.
    
    .. or when we just want it enabled globally rather than just within
    one process.

commit 050a1349ce706685bcb90bac8e7cf93cf36adf13
Author: Dave Jones <davej@redhat.com>
Date:   Mon Aug 25 15:53:54 2014 -0400

    add memfd_create for ia64

commit f341fb3d6566014db5e8d46b80c5f8d49edb5c54
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 21 23:56:39 2014 -0400

    add list_move from the kernels list.h

commit 7d7cff7a97b942bd527336d841b8ee5c863d7645
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 19 14:30:30 2014 -0400

    simplify get_random_fd
    
    the "25% chance" over-ride is kind of pointless given we give fd's
    a random lifetime.

commit e5b3b4a82e88050c47aff606a669e155df899df3
Author: Brian Haley <brian.haley@hp.com>
Date:   Fri Aug 15 17:02:39 2014 -0400

    Fix two shadow compiler warnings
    
      CC    fd-files.o
    fd-files.c: In function ‘list_to_index’:
    fd-files.c:216:15: warning: declaration of ‘index’ shadows a global declaration [-Wshadow]
      CC    fds.o
    fds.c: In function ‘process_disable_fds_param’:
    fds.c:164:38: warning: declaration of ‘optarg’ shadows a global declaration [-Wshadow]
    
    Signed-off-by: Brian Haley <brian.haley@hp.com>

commit cdd044e1d8ae97ed831946fbcd72a333d9e88010
Author: Dave Jones <davej@redhat.com>
Date:   Fri Aug 15 15:35:59 2014 -0400

    move memfd_create syscall definitions to the memfd header

commit 5c2a09dedd5f4b7d96f879250338917fd1a4249d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Aug 15 15:34:48 2014 -0400

    fix compilation for arches that don't have memfd_create yet

commit 9c8747b2c6e3a9119e727592493ffad65dd8d658
Author: Dave Jones <davej@redhat.com>
Date:   Fri Aug 15 15:33:31 2014 -0400

    add memfd_create for sparc

commit c2717f4fcda15d5ef7835987458254c50a864adc
Author: Vinson Lee <vlee@twitter.com>
Date:   Thu Aug 14 15:13:11 2014 -0700

    proto-phonet: Include compat.h.
    
    Fix build error when missing PF_PHONET with glibc < 2.10.
    
      CC    net/proto-phonet.o
    net/proto-phonet.c: In function ‘phonet_gen_sockaddr’:
    net/proto-phonet.c:17:19: error: ‘PF_PHONET’ undeclared (first use in this function)
      pn->spn_family = PF_PHONET;
                       ^
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 5fdd279ec3799e7197ef6bdca4c4f3f6a985b1ad
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 14 12:05:36 2014 -0400

    add seccomp and getrandom for sparc

commit 0dca92441523cd719c89bf9e34ff051c4c1e5bd1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 13 19:53:12 2014 -0400

    constify fd_provider structs

commit 89cfcf899908b6b0c215fc93d63730407387ca9c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 13 18:37:34 2014 -0400

    shut up a coverity warning about uninitialised var
    
    It's not possible to get here with flags set to anything other
    than the three cases in the switch, but it doesn't hurt to silence it

commit 2d1c22436cc77a72c521d279ed6e12696b1f3065
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 13 18:33:50 2014 -0400

    strcpy->strncpy

commit 175a8c096afb2d7710422e23087262a282883851
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 13 18:16:12 2014 -0400

    add a --disable-fds parameter to disable certain fd providers

commit 8cc98e73d4dfacac6d1b1a2bbb18175e20726f13
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 13 17:55:39 2014 -0400

    add a ->enabled flag to the fd_provider struct

commit b2d2006585eb89b3a31151894e45a0df3cf6534c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 13 17:38:30 2014 -0400

    add a .name to the fd provider structs

commit d07cfc61a56beaa0720464b4d55565b326656431
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 13 17:31:54 2014 -0400

    rename setup_fds to open_fds
    
    rename fd-files local 'open_fds' to not clash with new global.

commit c6c96b83e2c1fa5b4997e66adfb9f7b819c68d62
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 13 17:28:34 2014 -0400

    split the fd provider list creation & fd open functions.

commit 20affc48e2fb75832bc89a49c182c770896a6035
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 13 17:26:18 2014 -0400

    fix off by one in loglevel checking that meant [init] messages were skipped

commit ece92c4326f35f64af3494de728c52edd86cd72b
Author: Hangbin Liu <liuhangbin@gmail.com>
Date:   Thu Aug 7 16:56:38 2014 +0800

    trinity: add connection support for AF_INET and AF_INET6
    
    Trinity is a syscall fuzz tool for local host testing. But it would also be
    very useful to fuzz during established sessions over the network. This patch
    add simple connection ability for IPv4/IPv6 so user can use it as a
    client/server and simulate as a real environment.
    
    Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>

commit 2b457b296ab87dd45614d4f4cec0534e75173228
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 13 10:31:51 2014 -0400

    allocate the output buffers on child startup.
    
    by doing this, when these buffers overrun, we'll crash the child
    rather than spewing over the shm and destroying trinitys state.
    
    The better answer is to fix the overruns, but that's a longer term FIXME

commit db79cd1a4a1d148e8fb68283b1a765ecbbb77d87
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 13 10:14:30 2014 -0400

    remove out of date comment

commit ddac434c9f352a86e28625552042ccb5dcec1298
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 12 13:44:59 2014 -0400

    fix up some syscall list headers

commit 8f588bec8aa5e76a8c0961ea82412e1c05416225
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 12 13:44:46 2014 -0400

    update mips syscall list

commit 85f339a4e41604118fd67f8275ae1744b9628fca
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 12 13:44:34 2014 -0400

    update tile syscall list

commit 02e06cc7e9a36c7c573cfd345bb4f88e038293c0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 12 13:44:22 2014 -0400

    update sparc syscall list

commit f5fa13ef43d6d12b0d21abb6ac4fde9626fd0c64
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 12 13:44:09 2014 -0400

    update parisc syscall list

commit 59386a3218cfdba05ec9024c4721028d9bfacf14
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 12 13:43:51 2014 -0400

    update ia64 syscall list

commit 60293c57683db18f46348c33663f7ffc740fd54e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 12 13:43:37 2014 -0400

    update arm syscall list

commit fc612caf942d6af1489e0cec72fa911d541540e5
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 12 13:43:22 2014 -0400

    update aarch64 syscalls list

commit 655240f7a69a2e2fc4f9ec280ac6e23db89ee932
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 12 11:06:46 2014 -0400

    add kexec_file_load

commit a07d53f161b4bd26865bb16a052eef4e9026f7b8
Author: Dave Jones <davej@redhat.com>
Date:   Mon Aug 11 15:21:50 2014 -0400

    add memfd as a fd provider

commit eb370e311e5f59e7b09b7ee7de2518530f7dff1e
Author: Dave Jones <davej@redhat.com>
Date:   Mon Aug 11 14:30:46 2014 -0400

    initial memfd_create support

commit 9a2ae460eb2c654561e9f881c6f8685adc4b2aa8
Author: Dave Jones <davej@redhat.com>
Date:   Fri Aug 8 13:43:56 2014 -0400

    remove commented out bind/listen error checking.
    
    This all needs to be rewritten someday anyway, and it won't look
    anything like this.

commit 452997fb09d2e2cf3391d0e75495f2d2739e7c6f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Aug 8 13:43:42 2014 -0400

    reduce var scope

commit 76e36192ecc40c999ba5e4907b8975ec01fd65a5
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 7 12:08:22 2014 -0400

    add the flags and special cases to seccomp sanitise

commit c8bb47212235931fa7458da20418c4c466d9a8fa
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 7 12:03:00 2014 -0400

    add getrandom to i386 syscalls list

commit f5e39e7a28e6eaa992b18eff3a4d6a2ab3272cc3
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 7 12:01:49 2014 -0400

    add new seccomp syscall

commit 537f1c3f1f94d52c1a1745d2244ef368668d8308
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 7 11:56:30 2014 -0400

    wire up sys_getrandom

commit bd42a545d78c088b8fa52754724a2928cf0516f2
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 6 23:51:52 2014 -0400

    silence sctp noise

commit 72291d0c5fb34ec0c53a4fe6847a074c447f97b7
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 6 19:46:43 2014 -0400

    correctly terminate mode array
    
    we were always clearing mode[2], but mode[1] might be non-zero
    in the 'r' or 'w' cases.

commit 5d81bc714fb7f76db0495c76b2508d3d01c4a23d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 6 16:04:42 2014 -0400

    re-add the logging of creation of the testfile fd's

commit 75461537b79e1b51f83a912310801b771814b897
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 6 15:55:33 2014 -0400

    add back the filemode to the open(O_CREAT)
    
    somehow inadvertantly dropped this.

commit 671c2a39b2ee8c5176ca509bd473782c57607e01
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 6 14:39:48 2014 -0400

    rewrite synclogs() to only operate on a single child.
    
    As logfiles are local to a specific child, we shouldn't try to
    flush logs from other children any more.

commit 673ec13447cef9d9ed1f75241a2679b22c2ef5be
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 6 14:38:54 2014 -0400

    don't try to flush logs from watchdog context.
    
    Since we made log fd's be child local data, the watchdog
    can't actually see those fd's, so trying to flush them could segfault.

commit db79ac0ab7c5dd1fa3daa54120e6a406497b5d86
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 6 14:30:39 2014 -0400

    Change the sysfs/proc fd provider to sometimes use fopen too.

commit bcc4b57c3d46672fa3ddd6ab5051ad061c55ab70
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 6 10:52:04 2014 -0400

    remove index from open/fopen helpers
    
    just use the filename passed in instead of recreating it.

commit 3bbdeb4f9cd959c20a0c43dd592be427b1298105
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 6 10:28:38 2014 -0400

    create a separate function for random 16 bit numbers

commit a39ade5178a0d49c2af886bd337023c0c99fbfe2
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 6 10:19:16 2014 -0400

    create a separate function for random 8 bit values

commit 6c714e9c755e5f68ae897c7e889675d1cad4239f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 5 11:45:51 2014 -0400

    sometimes create testfiles with open, sometimes with fopen.
    
    Turns out the btrfs oops I'm currently seeing only happens with file streams

commit d7b1982d11f6e3223d1c2ee09f19e026ea5388e2
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 5 11:36:08 2014 -0400

    use some random O_ flags when opening test files.

commit 948f43c387e695bcd3c2ff7bef9980e2466ee33b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 5 11:19:23 2014 -0400

    rewrite the testfile creation to use open instead of fopen

commit f80d1fa6d636a4d364b73b9cbd0fc1772bd006bd
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 5 01:49:09 2014 -0400

    add another missing break

commit 2426fa3f888414fdfcb97a6f638bae36c2e6a51b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 5 01:48:32 2014 -0400

    add missing break

commit e4eb2fa09c7edb43382e645c789d4d30c3de3496
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 5 01:47:16 2014 -0400

    if opening one testfile fails, don't keep trying.

commit 8a6544352ff9876270f8fa41693b0c6c5442b181
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 5 01:45:25 2014 -0400

    fix potential null dereference if the open fails

commit 81d9795e26e2044ae312986055c8c269d39765be
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Tue Aug 5 00:08:04 2014 -0400

    perf_event_open() Linux 3.16 additions
    
    Update perf_event_open support to add new features from the recent Linux
    3.16 release.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 73dede62b80811e069b1b5d65b51ae78088e3fd7
Author: Dave Jones <davej@redhat.com>
Date:   Mon Aug 4 21:16:55 2014 -0400

    add missing header from previous commit

commit 0e5f76be98776240bf75dff8cb914595d7058c21
Author: Dave Jones <davej@redhat.com>
Date:   Mon Aug 4 19:55:17 2014 -0400

    begin some infrastructure to use a bunch of test files for fsx like ops.

commit 2d4911449bd2de4f25ae3a9ffa6e4d6c9ab6e85b
Author: Hangbin Liu <liuhangbin@gmail.com>
Date:   Fri Aug 1 09:57:14 2014 +0800

    net: add inet/inet6 rand_socket sctp support
    
    Acked-by: Daniel Borkmann <dborkman@redhat.com>
    Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>

commit 34d5c36523d7320e977c049c2b62acb0e75859da
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 25 12:29:28 2014 -0400

    put longopts array into alphabetical order

commit b4c3408ee2028c6eb44fddb194439f72d0b8b9af
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 25 12:22:34 2014 -0400

    rename dev -> bdev

commit 674bd25aa4963a7f53b1a05d9c39f9051a68e22a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 25 12:21:27 2014 -0400

    Add --bdevs parameter for adding to block dev list.

commit 434032b6e25a072b1aca14a687275b9b31b44a0e
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 25 12:21:06 2014 -0400

    remove unnecessary local variable

commit 7ba982e353a3760bc1dc703648760432721ce665
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 25 12:11:43 2014 -0400

    Add some skeletal code to maintain a list of victim block devices.

commit 0eb583ee9b72242379e4d6ae4333fbdb8e6cae59
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 24 20:08:09 2014 -0400

    move statistics out to separate structure

commit a52dc86715e93349ebba1060f5d32c95bccf9f39
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 23 14:31:27 2014 -0400

    once we've picked an fd provider, stop searching.

commit a6859b61c88ea75d15b6e985aba12c491befa3f4
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 23 14:07:44 2014 -0400

    fix cut-n-paste error in definition of eventfd fd's

commit b9f465ba892895bd3ac4fc70d9bd796b17916864
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 23 14:06:17 2014 -0400

    add timerfd creation to the fd generators

commit 09173f0e9ae2f6e8fb1352b0dab34e6b74fc06d2
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 22 14:03:40 2014 -0400

    move handling of child logfile errors into find_child_logfile_handle

commit 58deb383a49397df5a6210a583d0d4411852f89d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 22 13:51:28 2014 -0400

    rename __find_logfile_handle > find_child_logfile_handle

commit 11274cda4548ae0b4bd7f1e15c1ddca3c4ded6a2
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 22 13:49:50 2014 -0400

    Pass __find_logfile_handle the pid we're looking for

commit 32f926c151ef82918c2702734e05fe4f92a81c2d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 22 13:45:15 2014 -0400

    rename robust_find_logfile_handle -> find_logfile_handle
    
    old find_logfile_handle -> __find_logfile_handle

commit 7fa7ef633847cacd2a06c8e55582c9f3573ac84a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 22 13:42:10 2014 -0400

    remove pointless ARG_RANDOM_LONG
    
    This is the default case when we don't have an ARG_* set, so just
    use the fallback.

commit 030922dc9ba3eecf838a01fd229ea3b54775be59
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 22 13:37:45 2014 -0400

    un-number argtype enum

commit 2f1ba2c835099976b561d6151598b8e76f3473b1
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 22 13:32:10 2014 -0400

    remove unnecessary includes

commit 3270a153d2b332bd0a6e275816ec10d9ecbc1573
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 21 19:28:13 2014 -0400

    add beginning of forthcoming getrandom syscall
    
    not wired up to syscall tables yet, until the syscall no is confirmed.

commit 4af9a0f219ff739fafce7416e034c6c56ace67e7
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 18 13:52:23 2014 -0400

    replace open-coded rand_bool()

commit 2072f570f0743de1923d294a2fae4355cd7fe112
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 18 13:20:46 2014 -0400

    move rtas syscall definition to ppc specific directory

commit 9c59dafac1229d58c30c66126081816776a0573a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 18 13:17:47 2014 -0400

    merge all the dup variants into the same file

commit 8121ff9191e24ae612bb3ef161dfcfd096cb180f
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 17 23:29:50 2014 -0400

    Allow set_rand_bitmask to return without setting any bits.

commit 9becc054301825c03a564cd0c142750fad58e67f
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 16 13:01:30 2014 -0400

    socketpair: get the pointer to free from the right register.

commit c3c4d9199b97d3283e1272c8517d9da529327a28
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 16 11:41:24 2014 -0400

    sanitise/post for socketpair

commit 68a1d002d3b6d6a6ca9001f4f8d58bad65151af5
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 16 10:34:38 2014 -0400

    merge both eventfd variants into same file

commit f430b99f5e4bd077586241bd6b87c1e9f776bcc7
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 16 10:30:41 2014 -0400

    mrege the two epoll_create variants into the same file

commit 1496c42382339b37b29b924f11e4e896f1abd0b8
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 16 10:23:13 2014 -0400

    replace open-coded rand_bool

commit ef3488fff4435134140432a9e90e3a046a0cd52d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 16 10:19:37 2014 -0400

    merge all the write syscall variants into one file

commit e340bb75370150ef1e5998a81701b3acb007b4b1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 16 10:17:15 2014 -0400

    merge all the read syscall variants into one file

commit 494fad27a2e9dea8f7386b4e5f42183db4f6a787
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 15 17:10:06 2014 -0400

    scope reduction

commit f89e3bc32292341fe6e676a447920c9116069fef
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 15 11:44:29 2014 -0400

    move debugf to debug.c

commit 01e966ce0a27b4b6c1e6454d34c047edbca1421a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 15 11:42:02 2014 -0400

    move set_rand_bitmask into utils.h

commit 6dd22fe36b1337dae2512ccfca2856135aa90912
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 15 11:38:53 2014 -0400

    rename variable in handle_arg_op
    
    also add commentary

commit d56408b867237173c0b925aab76c90d581538eae
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 15 10:12:26 2014 -0400

    remove the unicode generator.
    
    This isn't really useful in kernel context, and hasn't seen any
    real activity in a while. It just causes a mess of mangled filenames.
    Interesting for finding userspace bugs, but not so much for the kernel.
    
    I'll do something with it as a standalone project some time.

commit 14a9b6f3b270d4baeea54c7b1a90dbabe452b9c0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 15 10:10:47 2014 -0400

    add helper function for ARG_OP/ARG_LIST parsing
    
    These types are similar enough that we don't need to duplicate
    all this code to extract the struct entries.

commit bccc851b43ae19085ed4f67bcc247f12e31bf9ba
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 15 09:59:19 2014 -0400

    reorder functions

commit d79a876175cf9862cc1f2e20605385e609ffb52e
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:37:14 2014 -0400

    poison alloc_shared allocations to make sure things get initialized.

commit e1b970655a74702b633256a86716aa2dbdd9c8b0
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:36:39 2014 -0400

    initialise random seed in init process.

commit 1a53d2b8e1d140fd7be93a9f758972ab40afc447
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:36:19 2014 -0400

    socketcall: zmalloc conversion

commit e9a2511fe9896391211bd77a43f057ecf91ed841
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:35:35 2014 -0400

    send: zmalloc conversion

commit 0a9ec3d8aab29424303a3d303ea28ee6979d9e86
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:34:05 2014 -0400

    remap_file_pages: better sanitize for start/len/offset args
    
    Allow start to be something other than the beginning of the mapping.

commit 11c23033cc1cdabe7aa16c76e0576d0b21540273
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:32:39 2014 -0400

    pread64: set size to be at most 1 page.

commit 179cc47c25362437fdecec94042d8f2d408efc06
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:32:17 2014 -0400

    modify_ldt: zmalloc conversion

commit 3084c7278e47d13b129022276fa05bec511ec82e
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:32:05 2014 -0400

    mincore: zmalloc conversion

commit 8d62a461160ec7c776e1df1081142200edacb8ec
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:31:22 2014 -0400

    execve: zmalloc conversion

commit 9df57fee7af9ed3abcbfcfe0afc48bf9a95ec1b8
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:30:39 2014 -0400

    explicitly clear the logfile ptr on child init

commit 35deacd58fe52bd906d5b4cb508f847421dbe3c9
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:28:55 2014 -0400

    the array of child struct ptrs doesn't need to be a shared map.
    
    having them just be COW'd is fine (in reality we never write to
    them again after shm setup).

commit 41ab822efcae9455f5c5283d813920ee9bbeeca9
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:27:58 2014 -0400

    use memcpy to copy syscall record to previous

commit 196700df5ce05c0973560275f25915b3502f37a7
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:26:26 2014 -0400

    Allow seeding from the init process.
    
    Right now we don't do any rand uses in the init processes, but later
    patches will, so will require this.

commit 30be68bd315a64462ec4b91bacf003ad22ba7c7c
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:24:19 2014 -0400

    remove the 'put junk in pathnames' code.
    
    it doesn't find bugs, and just causes a mess.
    It also occasionally causes trinity to corrupt itself due to lack
    of bounds checking in pathname users.

commit ae50273597b61a4a8d4ea2643bd940b9a9e8e22c
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:23:05 2014 -0400

    convert alloc_iovec to use zmalloc

commit 90fd80395203d1edc884d8a4630c4be13b470591
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:22:39 2014 -0400

    proto-x25: zmalloc conversion

commit a2d684e2d54786a9fe9850d5d9dbfb05eebad718
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:21:54 2014 -0400

    proto-unix: zmalloc conversion

commit 4fe9db4795199f3b6e1625560f5e9bfd3d560f74
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:20:54 2014 -0400

    proto-tipc: zmalloc conversion

commit 42988029a60945c807aa78ac3deb39b6f1c78929
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:20:38 2014 -0400

    proto-rose: zmalloc conversion

commit 1171627be518771a68ab982b579ff48c70796d24
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:20:23 2014 -0400

    proto-rds: zmalloc conversion

commit f206dc66ce2a649ce07a2e8755885a91d4cb639e
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:19:52 2014 -0400

    proto-pppox: zmalloc conversions

commit 0ad0e26af2345b95a206feb8e251ee50c85d1a4b
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:19:16 2014 -0400

    proto-phonet: zmalloc conversion

commit 729611a39d69808272fd24935cc3eb4daeefe4a0
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:18:59 2014 -0400

    proto-packet: zmalloc conversion

commit 10702f31bc5eea46801dd844335cf5b4ff9c797d
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:18:37 2014 -0400

    proto-nfc: zmalloc conversion

commit 0520d3c6d46947733b4d77e1c5940799f8c08c03
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:18:18 2014 -0400

    proto-netlink: zmalloc conversion

commit 4e523b7b24e4ae8a2df0a53309673c9719f5ba0b
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:18:03 2014 -0400

    proto-llc: zmalloc conversion

commit 5e52be5fbdd4fa314930a1bb18cd80b8c3e47db5
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:17:40 2014 -0400

    proto-irda: zmalloc conversion

commit 491344909c2d11461c30765d6150f8f17845c9e9
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:17:23 2014 -0400

    proto-ipx: zmalloc conversion

commit aa19609f68ba446fbb0f94a627a8a7c941be4bc3
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:17:05 2014 -0400

    proto-ipv6: zmalloc conversion

commit ff94603d1d961a4e5330696b6caa7dc8d4c3da35
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:16:46 2014 -0400

    proto-ipv4: zmalloc conversion

commit e296e2c0cc1f979b4eeded00c97f1b3376d6563b
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:16:13 2014 -0400

    proto-econet: zmalloc conversion

commit 3afad4fd8c070803c6e48a2790bbbec688f51377
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:15:44 2014 -0400

    proto-decnet: zmalloc conversion

commit 1904ed5e600c08329001cdd30f2d747d81bed4a4
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:15:20 2014 -0400

    proto-can: zmalloc conversion

commit 0ed075afda35fd04940f1d9ec88798f9fccdefbe
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:15:00 2014 -0400

    proto-caif: zmalloc conversion

commit e3d5b647e781a1314fbf5ba0ea2dc51c4e7d1fb1
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:14:40 2014 -0400

    proto-ax25: zmalloc conversion

commit f2dbc79c441133e267f42730928756969c780ac9
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:14:19 2014 -0400

    proto-atm: zmalloc conversions

commit bc0a81d4c6f5d0e5e6f78c61eeb765d2960e8383
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:13:45 2014 -0400

    proto-appletalk: zmalloc conversion

commit 3ad0d29e57bfa80221c1ed7549d4132a02fff350
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:12:59 2014 -0400

    proto-alg: zmalloc conversion

commit 483b1c7ff4b28117b86de5a40841b806cd1bab4c
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:11:36 2014 -0400

    net/bpf.c: zmalloc conversions

commit 734bab99c9e50d0109c8321a402cd4fcd4709fc7
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:07:42 2014 -0400

    convert alloc_zero_map to use zmalloc

commit db363858085ca26f75e49a338d5036a3e4624ded
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 14 18:05:07 2014 -0400

    output pid of exiting child as well as childno

commit a3fc826ca5e38822363edd59d73918012f2fe504
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 11 10:37:17 2014 -0400

    fix version.h generation for the Makefile case

commit 0b98e5fa5319003d290caa08ef990b3b1c82fac4
Author: Vinson Lee <vlee@twitter.com>
Date:   Thu Jul 10 16:24:54 2014 -0700

    scripts/gen-versionh.sh: Explicitly use bash.
    
    Fix build error on Ubuntu.
    
    scripts/gen-versionh.sh: 34: [: 1: unexpected operator
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 0e136a0122c7401abb8ceff7af34102f7a61f36e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 10 12:42:43 2014 -0400

    replace config.h instances with variable

commit da16cc251883586d9f34d786ce6e132cf3061a4b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 10 12:41:04 2014 -0400

    move version.h to include/

commit 9824228cb464cd924f4556334ddb63af98861fc6
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 10 12:38:03 2014 -0400

    only run sed over config.h if there's something to delete.

commit ba84ce03c8e99cb3128541a999183c46e9697981
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 10 12:25:20 2014 -0400

    delete the old VERSION definition from config.h
    
    this should help transition to version.h for people building
    from git.  After 1.5 is out, we can remove this.

commit 8b383d47f337d8c5b27f6ec4eeb4c63b180abbe3
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 10 11:21:26 2014 -0400

    don't regenerate version.h if it hasn't changed since last time

commit 24407153e8a3fd1cf0481dc348713bf87de6beeb
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 10 11:06:12 2014 -0400

    generate the version string at make time, instead of configure time.

commit 4688e45592b89d10eac4b9491c7f1e52faa0d1b6
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 9 19:02:31 2014 -0400

    replace the crappy pagerand dumper with a write to a file.
    
    feeding this to hexdump/strings etc is more useful than
    reimplementing the same thing within trinity.

commit 8df02f7c73b7931bc78e46fd1610693b787f64ac
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 8 16:59:29 2014 -0400

    decode_exit will only ever take shm->exit_reason

commit 0ab06dd928b9bd6ecdb82c48c292dc3ee3124738
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 8 16:57:49 2014 -0400

    dump reason why watchdog exits

commit 89a0cc4bf14947a50acccce3310a77b0c0bea040
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 8 14:34:00 2014 -0400

    reintroduce sleep in watchdog startup.
    
    5bff6ff1e011236e495a124a9a6e78daf8cddba7 was actually necessary,
    otherwise we seem to just spin there for the entire time the watchdog
    process runs.
    
    We can get by with a much smaller sleep however.

commit fd0de03f5d510df34572e880a508afdb30c26a61
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 7 17:10:13 2014 -0400

    make stuck_syscall_info check the rec->state first
    
    In any other state, we can't possibly be still blocked on the fd.

commit 201df2381d2171a7a3128e3fa15e6aa8290cd42f
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 7 17:07:26 2014 -0400

    move the check_if_fd lock-taking up into the caller.

commit 48ce9db6066146d1b582b5eb0a2bfb377a9f2e66
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 7 16:59:13 2014 -0400

    remove unnecessary sleep
    
    we never stay here long enough that this is a big deal, and the sleep
    actually means in some cases the children start running slightly
    before the watchdog starts its main loop.
    
    In practise, not a problem, but the text output looks odd.

commit 37240a06bef205d35bb7b1056f8a0d5d30f3332d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 3 13:17:13 2014 -0400

    wholesize is always multed by page_size, so just do it once

commit dace2e6f56c83fcda8fa1b4408534042a0563140
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 3 13:16:31 2014 -0400

    rewrite the shm redzone setup

commit 4051176a4220738e2409287dfd05ce05fee8e15e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 3 10:27:37 2014 -0400

    if we double lock, tell other children to die too.

commit 218a44113292a0e797d065738a8f34a02e7909df
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 3 10:25:39 2014 -0400

    if we double take a lock, we should fail hard

commit ba1b8a9cbc51aa1ac966cacd45e2a6cd0afbddf3
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 3 10:24:16 2014 -0400

    move the dontkillme setting into show_backtrace
    
    I still don't know why sometimes show_backtrace hangs for several
    seconds. My guess is some fuzzing has corrupted the symbol table,
    and the unwinder goes crazy. It always seems to coincide with
    getting an empty trace.

commit 1d85b31931e2ab9b16d26a88ccc15e5f57eaf5f3
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 3 10:21:50 2014 -0400

    remove the sleep from after the log debug

commit b44afbcbbc49d91560f8efd9c1fe76b327a4fce5
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 3 10:20:52 2014 -0400

    when acquiring the lock, do the state transition first.
    
    There was a tiny race here where two pids could get the same lock.

commit f9f95aab262307bcc4b80bc72d7b673a8d4f9f11
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 2 20:21:48 2014 -0400

    if a child gets reparented to init, exit.

commit bd553e0557f5c7f511af504f25d89a3bb0d9c7b3
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 2 16:50:46 2014 -0400

    exit immediately when reparented

commit 05139b04b356c0a6e583923b4b0ce4630fd0726c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 2 16:50:31 2014 -0400

    remove leftover output from reparent debug

commit bea7149c2482be79f816f5cf9bb820bd8de03774
Author: Vinson Lee <vlee@twitter.com>
Date:   Wed Jul 2 10:28:56 2014 -0700

    Add SEEK_HOLE and SEEK_DATA symbols for older distros.
    
    This patch fixes these build errors on CentOS 6.
    
      CC    syscalls/lseek.o
    syscalls/lseek.c:18: error: ‘SEEK_DATA’ undeclared here (not in a function)
    syscalls/lseek.c:18: error: ‘SEEK_HOLE’ undeclared here (not in a function)
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 29e1d566da6b216c6b5970bc558aacce23ee8d1f
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 2 16:07:32 2014 -0400

    wipe the child->syscallrec structs with memset instead of by hand

commit 708b8eaf91349ca6981706071e52aa643488c687
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 2 16:04:54 2014 -0400

    fold create_child_structs into init_shm

commit 812acb179096375bf653d748ec6ae7699d541fc0
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 2 15:52:33 2014 -0400

    increase the lock steal threshold

commit fe476894e8c9e562dd2f2caba4032877c58b7efc
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 2 15:52:12 2014 -0400

    do the sleep /after/ the backtrace, not before.

commit a571afeea7469af807a8c4d31b55a91ff402ecac
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 2 15:51:07 2014 -0400

    remove unused vars

commit dbb7ebea72936ed2d8d847ad19a687657ceb23e1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 2 15:49:54 2014 -0400

    remove the 'dump previous syscalls' from check_parent_pid
    
    we only call this periodically, so the last syscalls are likely irrelevant.
    For the same reason, remove the TODO about postmortem

commit c908ef4eae8d20f36bf5f65a03c99735e7eb52c6
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 2 15:48:20 2014 -0400

    output whether or not main is still alive when we get reparented

commit 16fca022934ed6ac7313a00b2a6fa9152657f271
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 2 15:47:00 2014 -0400

    handle some common reparent cases.
    
    Screwing around with setns and friends can cause our ppid to change.
    We don't handle this well at all, so we'll have to just exit the child
    for now until we come up with a better plan.

commit aeaceb205b9375d412129209ca4fdb2feffa7cb5
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 2 15:43:31 2014 -0400

    output the parent pid when we get reparented.

commit c31db880cee1d36b4d7bd353b56c0747d8c35d9e
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 2 15:37:53 2014 -0400

    watchdog: don't do fileno on child logfile fd if it hasn't been set yet.

commit 26e1c7f7533b0f61a355d523c0ebb175ad01977a
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 2 15:22:53 2014 -0400

    don't try to fclose the logfile if it hasn't been opened

commit 2c98d62a90ce6680d0beef1dd8a776e9844e212d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 2 12:31:43 2014 -0400

    export dump_syscallrec

commit 06d42ec6e788d739dc65958cdec3f6a05940539f
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 2 12:05:01 2014 -0400

    fix unreachable code in watchdog
    
    the >40s code was never executed.

commit f1044f3b9a8f4e9c99400ea5c473e21e20decbe2
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 2 12:00:14 2014 -0400

    remove unnecessary pid check.
    
    if pid becomes non-zero we bail the while loop and return anyway.

commit fbebd53ecbdc694d3b4022aadaab400fc7c23eb7
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 2 11:02:15 2014 -0400

    add missing printf parameter to locking debug statement

commit fa3e1acb8b3ab1ca326033073efcab8f04bb2f9d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 1 21:59:03 2014 -0400

    Fix 3rd argument of lseek.
    
    Its name is whence, and it takes one of 5 values.

commit b8f4704e2b958bf45d81db7e7dd29b5a6c6d1b1b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 1 18:32:54 2014 -0400

    dump childnos from watchdog when sanity check fails

commit 3f46b9d32c726ee1c16d76a5be196eb604911787
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 1 15:28:42 2014 -0400

    only do the ENOSYS check in AFTER state.

commit b7a8165222440385286903d8f45bf8545a7e7c61
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 1 15:25:57 2014 -0400

    simplify, by breaking up if and inverting checks to reduce nesting

commit 504cc673d7e5671bb1aaf86f3baf2ceafa732a71
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 1 15:23:03 2014 -0400

    remove duplicate -1 check.

commit aad21d911eac2128b89bc0dac13374f86c4153fd
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 1 15:21:34 2014 -0400

    move the ENOSYS check out to its own function

commit c3a597320b2398ad8caf742dd6551d843b574edc
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 1 15:18:07 2014 -0400

    rename retvals to errnos

commit 506b78b02a40a902893747e3250415937544fa4c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 1 13:03:28 2014 -0400

    fix "not filled out yet" check on return code checker.

commit 470f7c4a001855598b77fdca8e61e7086f3030ff
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 1 12:54:10 2014 -0400

    an experiment to check expected return values.
    
    unsure if I'll keep this, but maybe it'll be useful at least for
    finding gaps in man pages.

commit 96a07415292595c7ee115b29de883014be7d5609
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 30 20:51:04 2014 -0400

    don't pass LDFLAGS to every compilation unit, just the final link.

commit 1201c843bc0d8d9e27eccacb8b6e552d8d5cb095
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 30 20:50:43 2014 -0400

    -rdynamic should be an LDFLAG

commit 85d5662cd9c847a5e491638ed95203eea5e029fe
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 30 20:50:03 2014 -0400

    make the gcc/clang specific parts automatically work based on CC

commit f561dfde3d5754421b33cf18ba5bcc624243d313
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 30 20:28:20 2014 -0400

    remove highest_logfile()
    
    Now that the logfiles are opened by the child process, this needs
    to read the fd number from the child struct instead.

commit 18de43a56b26d90c5b01b9bcadaa0a6d8ecaf6fa
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 30 19:48:52 2014 -0400

    clear the guard page after the rand page, and check it.
    
    it's sort of lazy right now, and only shouts if the first byte
    is non-zero. In most scribbles historically though, that has
    triggered.  It seems pretty safe right now, but I want to keep this
    around at least during development.

commit ec2601000c6c31c64f29c1ce3d7ad1465ca3e084
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 27 15:54:23 2014 -0400

    move the opening of child logfiles into the child process.
    
    Instead of using up fd's in main, and inheriting them on fork.
    If we ran with a large -C, we'd eat up a lot of fd's in every
    process because of this. Now we just have one logfile fd per child.

commit f8d1708c5984fdaa6d1f588f1282eadd5eb37c76
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 26 18:37:24 2014 -0400

    only re-output the "still running" message if it's changed since last time.

commit 206e1d1b8b9caa7504956b87d3ab943c847684fe
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 26 18:35:55 2014 -0400

    only spew the sanity check fail msg once.

commit 83056b5872a4ca09404855ab4d635558397bac91
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 25 20:48:11 2014 -0400

    always skip over EMPTY_PIDSLOT, regardless of syscallrecord contents

commit 6d11e1b3411f5ac8c2a6ce1a429ba639f0175fb8
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 25 20:25:28 2014 -0400

    TODO updates

commit a451594218ffb26f49897622ccf01fcd45f744a1
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 24 15:59:35 2014 -0400

    document what each lock in the shm is for

commit 7ae4e683fca3dbe21cf70c505380e7a42a3021c4
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 23 22:17:27 2014 -0400

    add debug code to dump a syscall record

commit f47d16db5f8864fcf954ecea2e91c3eb5050a732
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 23 22:15:48 2014 -0400

    cache childdata ptr.

commit ae2546c20a1763166d65aa7a336f15bcb538b761
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 20 20:06:26 2014 -0400

    don't bail early if we're exiting because EXIT_REACHED_COUNT

commit b5bc327b38e7978360ad2b419f83a54e0d3c0e06
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 20 20:05:32 2014 -0400

    the 'something bad' path is only for child processes, not the watchdog

commit 721753fb9335d2a53a6f88f7193fd7065957efef
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 19 23:44:56 2014 -0400

    reword child creation text

commit da4bf08e46f429919579cf71f7f359c0fc742ae8
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 19 19:19:09 2014 -0400

    when we panic, set the flag to stop main spawning new pids.

commit bd76fd42ebcf5298f1c44581616fb41b5360273a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 19 19:17:06 2014 -0400

    introduce wrapper for setting shm->exit_reason and use it everywhere

commit dea56c081813a5df17a88000da0ef484db0db652
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 18 15:11:02 2014 -0400

    make the mremap 'new' address shift higher.
    
    Less chance of it colliding with something mapped low, which on
    occasion causes random corruption of the shm.

commit 14bd6d827bab26500c2111ef03666bf830dc0e5b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 17 22:04:59 2014 -0400

    dump some info about the child when we find a weird pid.

commit aab8c59ec4033bd4f63132dd790c858855859076
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 17 12:31:34 2014 -0400

    ugly hackaround for when children spin forever when watchdog/main crashes

commit 7c5428d1e379b4f9799508b21e2133e4e382a540
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 16 14:32:01 2014 -0400

    if the subchild fails, we'll be in state 'AFTER' rather than GOING_AWAY

commit 1c872b16402c4b60183a259054121585d15fb873
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 17:42:27 2014 -0400

    sub-childs need to update the stats too.

commit d25356a1f2a9e6951236014e4ab0c2d292360760
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 17:34:58 2014 -0400

    split the extrafork code out into its own function.

commit a48aec233d7c9f3c5f36ca23020485f1d64708db
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 17:29:39 2014 -0400

    add subchild waiting to the extra fork case, and unifdef it.
    
    It seems to be doing the right thing now.

commit 18e37bb3665640683e089ea834f85f8069e600d1
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 17:21:56 2014 -0400

    set the subchild process name

commit 928c588a46e95bdd68b930552b24729874717ee3
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 17:14:50 2014 -0400

    compile fixes for the commented out execve code

commit e13d08bb9cb63f70e302c8240f922792aaed0912
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 17:11:25 2014 -0400

    in the child pid, wait for the sub-child to set state before killing.

commit d168162cf3e9355227e42efbc430bd585bec60eb
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 17:08:52 2014 -0400

    set GOING_AWAY rec->state before we do the execve call
    
    as we'll never return.

commit 1728d720d6daab0969b5ad69492f498794aba8fb
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 17:02:40 2014 -0400

    rename generic-sanitise.c

commit bd08fcaeb50cfe429c06a3bd25e411e07bcd1875
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 17:01:40 2014 -0400

    move the this_syscallname prototype to the right header

commit 2c9d9c20815cdad852852a2ac28019efcaaaa1e6
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 17:00:33 2014 -0400

    move generate_syscall_args to generic-sanitise.c

commit 8b94de8a061844f41cb39466c39f784301b6726c
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 16:57:28 2014 -0400

    move mkcall innards into child_random_syscalls

commit b90554fa677798f3b4bc1ad1ca964b00f74e6c4d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 16:52:13 2014 -0400

    move this_syscallname to tables.c

commit 2faa89c98c49e8c7771000771b22141fb9e0d522
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 16:48:53 2014 -0400

    break up mkcall into several functions.

commit 00de2281818c06fc83a2fea5e6bc06a49116106f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 16:13:39 2014 -0400

    reorder the checks in check_if_fd so the argtype check is first.

commit 443c879d4ceb0a408947b8a6e1465b393b0ae357
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 16:09:19 2014 -0400

    sort the time checks.

commit 63fee579114da1c6e4b1c8885aa0b8f1902891e7
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 16:08:16 2014 -0400

    remove unreachable test.
    
    This is leftovers from before we had the >2145 test

commit 37942bcd67193823f62c5941546577f0213dbdce
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 16:07:23 2014 -0400

    create common kill path for stuck pids

commit 6546651aa88d47f26ffeb5a2774b95f67101da1d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 16:01:53 2014 -0400

    change stuck_syscall_info to take a childdata ptr instead of a child no

commit 75420f430b46e07e9919490951605c92511ea402
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 15:58:25 2014 -0400

    add a special case to the child progress checker for the 'normal' case.

commit 4b97229e1ec466a847477b9ccaf7f8fc23b27334
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 15:56:35 2014 -0400

    convert check_children to operate on a single child.
    
    This removes one for_each_child loop by repurposing the one in
    the watchdog's main loop.

commit 4c33c09405b1b39cd5fc7bb1e989db38ace927d1
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 15:46:10 2014 -0400

    move the out of range pid check into reap_dead_kids
    
    removes duplicate code, and simplifies check_shm_sanity.
    Rename it to shm_is_corrupt while we're at it.

commit a7220b93e87ee93e5be30ed0fc19e346908c6180
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 14:48:12 2014 -0400

    make kill_pid able to be called from watchdog context again.

commit 4698dfb05d13d837b8aafed6c10d15d118abd1d1
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 13:30:40 2014 -0400

    silence the param parser when passing -qq

commit a48ce8ca20783a17c0bf56e3c2da54f09538587a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 13:25:50 2014 -0400

    after freeing the ldt, destroy the pointer, so we don't double free.

commit 3eabe4e79d5989c7e62e2e1686ab0d91a230e741
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 12:41:45 2014 -0400

    set the dontkillme flag while we spin in BUG

commit e5f3bc7520c23b827aa7cbeaff45522d9dc1ff95
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 12:36:04 2014 -0400

    handle the case where git isn't installed, but we're building from a snapshot

commit 35d3bf0b90f1c26b47781d54cc433ef183ebabeb
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 12:27:49 2014 -0400

    move BUG() to debug.h

commit a7fbfa19fc3b0d189aa9a58e00077c58b7dda40a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 12:21:44 2014 -0400

    move the backtrace code to debug.c

commit 65249b21877731071e9d9a365805d15582b2773f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 12:18:58 2014 -0400

    demacroise BUG

commit 9e4c91ed3a444df1e387c99471a75618e16b5b50
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 12:06:03 2014 -0400

    print a newline after the BUG: message

commit 949786a84c381340d7d69220c427194aea8e7295
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 12:05:07 2014 -0400

    remove BUGTXT

commit 07503407b075ef28a13d2a10396ba5e6899e522a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 12:04:16 2014 -0400

    make BUG_ON print out he condition it hit.

commit 807dea7e0b57f447970c1041310e4bb54fd3ad1b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 11:57:09 2014 -0400

    log.h doesn't need params.h any more.
    
    fix missing include in syscall that the indirect include was papering over.

commit a347c336810d358b819668585ccec67527b3c6c2
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 11:56:03 2014 -0400

    Show a baccktrace when we BUG

commit 9d3474d42a05bc824fb856da5e84d19e997f6330
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 11:53:52 2014 -0400

    taint_tv doesn't need to be in the shm
    
    it only gets used in the post-mortem code, which is run from
    watchdog context

commit f955dc64f318ddeb8fb6d570361e25e1260654ec
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 11:47:09 2014 -0400

    have the children wait for post-mortem to be done before they exit.

commit 9f415fe33c403a32f1d3757406a5fb080b19d44a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 11:41:18 2014 -0400

    remove dead prototype.

commit bd0b0f086a97935872a5fab2cb5a987600142721
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 13 11:40:56 2014 -0400

    make set_dontkillme available outside the locking code.

commit aaf7bab741f736c00be4f2ee6a4d83ae741faab3
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 23:35:12 2014 -0400

    add a bunch of commentary for the child functions.

commit 291c717959ea27971a39aeae65a1d720bf7b77af
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 19:21:31 2014 -0400

    fix signedness of variable initialization

commit f36c033f059f017ffb1a6e1cceb2b343d556cf80
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 19:19:20 2014 -0400

    argument numbers are always unsigned

commit e1e41ec2f8ba9e3ff17c618be26fc3e1c4960d36
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 17:55:44 2014 -0400

    use the git-descibe output in the version number.
    
    caveat: needs to re-run configure.sh, which kinda sucks.
    Maybe I'll revisit it later and have Make do it instead.

commit 840e8b69ebc1322ee584567b36350b454c457e8a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 14:29:28 2014 -0400

    check we're in child context before dereferencing ->child

commit 1132271d288008ad83fdf9b6c831ba06c0a5fc45
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 13:15:25 2014 -0400

    remove duplicate variable

commit cf37c5d08b3bd8fdf93ef91846a7465120e37b99
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 13:14:15 2014 -0400

    pass the syscallrec to render_arg

commit 5226568f2771a26da344e3c36624edb097b1351e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 13:12:27 2014 -0400

    pass the syscall record down to generic_free_arg

commit 0b23c8510ab1a7e2972b6989a37cdb068d92341c
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 13:10:30 2014 -0400

    remove some unnecessary includes

commit ce0d1503db71239fbeb447aa5c20c60a2a9e17bf
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 13:07:15 2014 -0400

    move __syscall_return to syscall.h

commit 519b505ea393c69acb52cda75124f6e1bf1b81b2
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 13:03:53 2014 -0400

    pass the rec struct down to the log renderer routines.

commit e63aad58cdef2f324248f85583cb9cb62d7ab4c0
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 13:01:04 2014 -0400

    make render_syscall_postfix figure out where the buffer is

commit f9e3c20f2c4a45ad23975766c32776f89748d3d5
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 13:00:17 2014 -0400

    pass the rec struct into render_syscall_prefix

commit 87b2c4eab3cbd90c9cc6714ae152e871b4c98005
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 12:41:03 2014 -0400

    make sure lock only sets dontkillme in child context.

commit e8611d24dfa4de9450657e8cbb90c5584df0b021
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 12:27:52 2014 -0400

    remove childno arg from child_process()

commit a9a61020a96dc0d4835c5e67b5d02fbbcb6cad13
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 12:26:42 2014 -0400

    make set_seed work on a childdata struct instead of a child num

commit 195ed387ee6e4f7d7992461165d146f4c3b175b3
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 12:23:37 2014 -0400

    this_child conversion in child_process

commit cedbec79c1d3caf6f7abb32a473e4fd82ccd6fe0
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 12:22:42 2014 -0400

    don't pass childno down to the child_funcs

commit a0742a02a0e3c2d37879907b4f96b1fd29f6c799
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 12:20:28 2014 -0400

    remove childno arg from mkcall

commit 5d7c90cffe6658ffd63cf5bb080e0491a6040006
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 12:19:22 2014 -0400

    make output_syscall_prefix use this_child

commit da8a88b59d30ac88aa189ba8e775720e556f36ce
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 12:14:32 2014 -0400

    stash the child number in the child struct
    
    eventually we'll move to place where we don't pass this around, but
    there's still a few places we will need to know it.

commit 058804e77da3441e70c95d16cc56e3dacd009555
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 12:12:01 2014 -0400

    convert generic_sanitise to take a syscallrecord

commit 6dfb0ea6a5a4cc5121198ec4452d99a963e727d3
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 12:10:32 2014 -0400

    convert fill_arg to take a syscallrecord

commit 29cc7026550cff49761c91738595a8ac7a6a7188
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 12:08:18 2014 -0400

    convert ARG_ADDRESS handling from childno's to syscallrecords

commit 80362cb5f59e9033a45b258d61c732b40ccc9d8e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 12:04:01 2014 -0400

    remove childno use from commented out code in bind

commit 4c42a05ed7f325dd54e433f48256cabdf13698af
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 12:02:59 2014 -0400

    remove childno argument from ->post

commit f2619da43365275a742ed3169dc6114fe6e2473f
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 12:02:00 2014 -0400

    convert post_mmap to use this_child

commit eee56201e37159143d17f1c64b4be3b9cc104070
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 12:00:35 2014 -0400

    remove childno argument from ->sanitise

commit 94d3277fa6033f5abf03640e1f5f3f4cac007561
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 11:55:51 2014 -0400

    make the ioctl code work with syscall records instead of child numbers

commit a2d52c517cdd08cd5c3ae5959ef9226bd53901c7
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 11:44:18 2014 -0400

    remove childno argument from common_set_mmap_ptr_len

commit f292cf218bfeec8446f873a69502d464a1595433
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 01:34:36 2014 -0400

    convert output_syscall_postfix to use this_child instead of needing a childno arg

commit 61544ee780b9564d0935b15165d89185159ab395
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 01:31:37 2014 -0400

    convert render_arg to use this_child instead of needing a childno arg

commit 1407845efa6d261703d23d980ca8b86f5a7d7a80
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 01:30:23 2014 -0400

    make ->decode take a syscallrec as an argument instead of a childno

commit 70bed8ec473c69701532f9a367cdd976beb9c496
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 01:16:59 2014 -0400

    convert check_if_fd to work on syscallrec's instead of childno's

commit 860dd39b3b0f15b65df9ff5bd6d118302f31715d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 01:14:23 2014 -0400

    convert handle_sigreturn to use this_child

commit 766c0185cad176b733bc5702720184da30d2c436
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 01:10:26 2014 -0400

    remove duplicate clearing of log dirty flag

commit 402a8aa7393c71bab36a4031ba890dba27e313fd
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 01:10:02 2014 -0400

    remove childno usage from truncate_log

commit d2dc640a8472b4c4584e7e1347182ab2125deb7d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 01:07:43 2014 -0400

    convert get_argval from childno usage to this_child

commit b54a3b546592f0d0a9930aa2a4a15d09f3a11418
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 00:59:11 2014 -0400

    remove childno use from this_syscallname
    
    Only ever called from child context, so can use this_child

commit 55bd8889463c5bf94592395a767a7b8566d034fb
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 00:56:20 2014 -0400

    missed a this_child conversion in the do_syscall path

commit 2c6e840d17af5b83f2d06d701a97248ac550c01c
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 00:54:39 2014 -0400

    remove childno use from do_syscall
    
    This is only used from child context, so can use this_child

commit 408c81eec0869e520adaf0dc242a847c8a5b16ae
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 00:51:50 2014 -0400

    remove childno uses from delete_mapping
    
    We only ever call this from child context, so it can use this_child

commit ea820e16e9a7dbc7160457772392ee3a3398f3b8
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 00:49:38 2014 -0400

    change this_child to be a childdata ptr instead of a child number.

commit 5dc79ff55536b09366f1693957deae1f053e2e87
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 00:42:45 2014 -0400

    don't bother zeroing out the mappings ptr on reinit
    
    we initialize it right afterwards.

commit abdd3516e4b3f6759ed12bd3633a2eee40dcc0e2
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 00:41:26 2014 -0400

    cache childdata ptr to simplify code

commit 8dfa174148a1a8a506d3de47b72ad9ffbd825f90
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 00:31:38 2014 -0400

    remove the ->scratch pointer completely.

commit 2d5e126b0916876c7ba39f56bf37d92b2bcb732a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 00:29:36 2014 -0400

    remove usage of ->scratch from send syscalls

commit b7398569d8924f48c8e6b4250cd9464bc028ffde
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 00:27:08 2014 -0400

    remove usage of ->scratch from munmap

commit 88d856cab5763ce7afda11fc6b1b57287a36e943
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 00:26:24 2014 -0400

    make mprotect's map variable static

commit 191b2ea4061344317984d5d51c762077d9c4613e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 00:25:58 2014 -0400

    make mremap's 'map' variable static

commit f2c0e26c9493435cf1bca11b9ab44c273a149edd
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 00:25:23 2014 -0400

    remove usage of ->scratch from mremap

commit 76025c054e7c7e70f33f400f782c2c388f39d0ae
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 00:23:41 2014 -0400

    remove usage of ->scratch from mprotect

commit 67c201e74315eee5fef98210d9ae1357b82d5df9
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 00:19:49 2014 -0400

    remove use of ->scratch from modify_ldt

commit 070e959f540ad278e96c67fd28fdf5afeb5fcee9
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 00:16:31 2014 -0400

    remove usage of scratch from move_pages

commit fd932402302ed86e7859a96dd8321e5a67ef3852
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 12 00:10:01 2014 -0400

    remove unnecessary null check

commit b199b6c98ef5c755e8ccfdc81be0628d5a305c20
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 23:25:11 2014 -0400

    don't use mallocs as struct members in move_pages ->sanitise

commit 04c3a5ff4e81693d9db7a43e6d4121c68552be43
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 22:20:16 2014 -0400

    handle SIGBUS from child processes

commit 40624c748376f3aca162e6872b67de06c5e1ffde
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 21:01:32 2014 -0400

    remove duplication in the child signal handling.

commit e929c05200036e450d683f69989183922178178a
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 18:49:39 2014 -0400

    allocate per-child arrays separately instead of as one big array.
    
    No real functional change, but the difference in allocation patterns
    made it easier to track down the corruption that the page dirtying code
    was doing.

commit 11658c79a44952ba4362a736cc5a1e302a3c5ad3
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 18:08:32 2014 -0400

    fix read past end of buffer.

commit 207c4dfcba9da73343370695128fca0f631efb41
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 18:08:02 2014 -0400

    do the same simplification to read_every_other_page

commit 7aac4d695e7ab7b75d2b9293598b7519d7dbbbea
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 18:06:42 2014 -0400

    fix another write past the end of array

commit 32ff55e9a5cb9194d626c165d9e2b0fba5b52909
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 18:04:25 2014 -0400

    greatly simplify dirty_every_other_page
    
    Also fixes an overrun.

commit 09d9572c987f5123fa9a241fa2d3169a2eaa8b84
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 17:09:53 2014 -0400

    fix incorrect rand_range parameters

commit 9fa144baa0347f21aab2b4b014dfe513901083d4
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 15:34:44 2014 -0400

    if we find an invalid pid, skip over it instead of bailing out.

commit bf259d1038a7ea9bfeea8d9c7a451a3e993e49aa
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 15:33:43 2014 -0400

    output valid pids as unsigned values

commit 9427f7d3b6cd0acfe8991dc1d48c9d4bc0abb8ec
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 15:33:23 2014 -0400

    when dumping empty pid slots, don't also dump the pid number.

commit cac5242832e0c216462926dba66175d2f5eec2e5
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 15:32:12 2014 -0400

    output invalid pids in red too.

commit b040cc6c2350be484386b445eec81d8946e8665d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 15:31:26 2014 -0400

    special case the dumping of empty pid slots

commit 6d12d5f0689b2aee3bbe3120c424b1b385b368ae
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 15:14:33 2014 -0400

    remove an unnecessary getpid call

commit d92cb891dc4d96caeee9fa9b4016ca211ced8fb5
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 15:13:15 2014 -0400

    make sure we wait for our child pid to be set before we do anything.

commit 994bfe12001e27506df9747ad54638492cbcbb51
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 12:04:24 2014 -0400

    don't try to reap invalid pids

commit 345964dccab7d38abef967bb175b338b1dd0d530
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 12:03:23 2014 -0400

    make the post mortem code also dump the previous syscall rec.
    
    and separate each child with an extra newline

commit 54aa0736af4ace938789f5a3223ecbc222a181ea
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 11:16:08 2014 -0400

    output pids as unsigned values

commit 2a11cb6bf6c89025adb7efdd14096a8f59c22b2a
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 10:38:41 2014 -0400

    move the pid_is_valid output to its caller

commit bbe0151dc623e08c9f9a41ed53cda79fa8b841f3
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 11 10:23:00 2014 -0400

    output uid etc as unsigned

commit 7c2c2a2b34da8c5f279f88a031954ade639dedf5
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 10 09:53:35 2014 -0400

    Fix up the post-mortem code to handle the syscallrec buffers

commit 724129d8318637733e27a5acb40b42b295d824bf
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 10 09:52:13 2014 -0400

    remove the code that removes /'s from path names.
    
    It never found anything interesting. I'm not even sure what I was
    originally intending to do here.

commit 3244e29e8080c85b7f5c6259252ef3ac4fc57fc5
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 10 09:42:40 2014 -0400

    overview of how the reseeding works.

commit 0ee1368391bac88d90cf292ea879e0039faa0d71
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 9 18:13:06 2014 -0400

    generate_random_page should work on the page passed in, not page_rand

commit 9fc963bb4efc9b8d9c0fbcabb61d6e642b09fe48
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 9 13:01:04 2014 -0400

    if we detect corruption, don't wait for children, just get out.

commit ae46619d23ed2b3a0f9e357eaec407585f6534d2
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 15:54:56 2014 -0400

    rename SIZE to BACKTRACE_SIZE

commit 451ba14c3469d80824e56071cd1b21a69e1ad329
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 15:18:47 2014 -0400

    throw away unneeded sprintf return values

commit 3b1d01b09c5546036e605e7cb53d038af06b44aa
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 15:17:30 2014 -0400

    remove duplication of ANSI code generation

commit c2bf8d599fa8d45d0eb05be802efa9365f8cd1ed
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 13:19:31 2014 -0400

    add comment explaining why we don't do anything about bad pids in handle_children

commit 70269dd01e0453c2c92a944151c5197a9accb2fa
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 13:17:55 2014 -0400

    remove second call to getpid in lock()

commit 3575f6d791d8901c8ea202ef9d686a1267ef98fa
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 13:16:29 2014 -0400

    change wrap calculation to factor in possiblity of adjtime

commit e909bf577fc88abc4701d8402a28bd1ba9852b5d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 13:14:24 2014 -0400

    watchdog: calculate diff correctly if we wrap.

commit ac168b976931c7958d6b3b1515e5de6763d9753d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 13:13:05 2014 -0400

    watchdog: replace ptr to timeval with a ptr to the syscall record.

commit c4b653dc71d1772f1890fb0948c7c08cf988ef73
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 13:10:40 2014 -0400

    clear out child data from previous instances of the same child.

commit 74a224b7d9129623baef966423f28641ca9c0e99
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 13:07:09 2014 -0400

    disable child killing when we're dumping backtraces.
    
    in part because of the sleeping we do here, but also because sometimes
    it takes a while for backtrace() to gather all the symbols.
    It's a mystery to me tbh.

commit 44f0e91280c08e1a3b37cd2185e28f9215f88061
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 13:06:50 2014 -0400

    provide a way for children to become immune from the watchdog killer

commit 899d8d6db92b4bf46e9dc3b0c05b56001052ef16
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 13:03:08 2014 -0400

    dump backtrace if we detect a double lock.

commit 571305327db055f0862c199c0bb890276a0a32ac
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 12:59:49 2014 -0400

    the backtrace functions need -rdynamic to be passed to the compiler

commit c057dfe0360c48fc636e460a128dbf1ca13f924a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 12:58:25 2014 -0400

    resurrect the old backtrace code.
    
    It's not always useful, but from time to time it does actually work
    and can be useful for tracking down weird cases.

commit 6624be4b294847a7bce33ff915e39d1adfc1f19a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 12:56:19 2014 -0400

    only increment total_syscalls_done after we completed a syscall.

commit 4c803f5287e8b8e6a8b48539585fe24535eb36f7
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 12:55:39 2014 -0400

    update the syscall rec timeval under the record lock.

commit 34d4d87bf1d26c8c8c388d317fdd1e4cd19b2464
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 12:54:05 2014 -0400

    handle the case where we double acquire a lock.
    
    Shouldn't happen now that we do the bust_locks() after the signhandler,
    but just in case, add some debug code to catch any stupidity that might
    creep in later.

commit bca54bd0fcab2c0417fa43e8d28c68bddeb79920
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 12:51:55 2014 -0400

    fix a case where the watchdog code would hang on a lock

commit 86d74b6b8b00c10039f7c22e6bd29cd59b1f1b5e
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 12:49:44 2014 -0400

    function reordering in lock.c
    
    (to make subsequent diffs easier to read)

commit 5e1287521598e11df23d1685f7997ae4a9bd0729
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 12:47:56 2014 -0400

    on return from signal handler, drop the syscall lock if we had it.

commit 7612e63f7a8dff4eaba85bbcba00fc340339f98f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 12:47:01 2014 -0400

    add a bust_locks function.
    
    As the comment implies, this isn't going to be generally useful,
    and should only be used in rare circumstances.

commit 25039b27328770e38d450130abd0a94abdc8108e
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 12:43:20 2014 -0400

    move the 'block on fd' check out of the signal handler

commit fb2357f45507b1e6483b0a5965ebd63cdfeb23a4
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 12:04:16 2014 -0400

    fix another case where we need to make sure we have room in the buffer

commit c048c606e2e1678ef97b9ce7ecd980c3670f45de
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 12:03:26 2014 -0400

    make sure there's room in the buffer for pushing the extra 2 chars

commit 75b2a5e2ba69311e457c11f645cbf38a01cf7892
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 12:00:42 2014 -0400

    clamp the range of the pathname to the allocation

commit 775833efbc290413390f68c2c7b74a75117c857b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 6 11:58:19 2014 -0400

    remove the empty string case in random-pathname
    
    it's covered in generate_random_page

commit 4be4c70926d1d298e108b0ff26835d28ba67eeea
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 5 20:29:23 2014 -0400

    add something that I keep using for debugging when things go awry

commit 677fccdd2f454b49d0bdc9fafecf576b4939e207
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 5 11:14:40 2014 -0400

    make check_for_argtype operate on a syscallentry

commit 74a13faf4614ecc5d3f4ee76ae029ef8795c029a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 5 11:09:57 2014 -0400

    clean up a lot of duplicated comparisons in find_previous_arg_address

commit 565c28477052628c9872d58ea6ecfe70d2f59f5a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 5 11:04:45 2014 -0400

    remove outdated comment
    
    Now handled with IGNORE_ENOSYS

commit 07abae1e0763363249c920051f449b230be9b0f7
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 5 00:40:33 2014 -0400

    scope reduction

commit 5e90c202510b6f8bcfc498b77e4f9d2d08bb774d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 4 23:53:53 2014 -0400

    use childdata ptrs in watchdog.c

commit 3de9bb3d2d7c9f60c1f85f9052a8efe906240d4a
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 4 23:46:06 2014 -0400

    init_shm: use childdata & syscallrecord struct ptrs

commit 6e1eb371640d4da8da3a909436cda4d8b7ba0935
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 4 23:37:59 2014 -0400

    dump_childnos: convert to use childdata ptr
    
    This removes the repeated index calc in a better way.

commit 5b2bbff00fdde83cf53b8e4d43a9c8bd7121a4d4
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 4 23:35:58 2014 -0400

    dump_childnos: cache the index calculation instead of repeating it.
    
    gcc is probably smart enough to figure this out, but it also makes
    the code read easier

commit 1e2612aef4eea9dd09f39b19e2e44ac770781997
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 4 23:33:40 2014 -0400

    use childdata ptrs in main.c

commit fe1d20991acec1dd35aff6828697b009b81dbf2f
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 4 23:11:07 2014 -0400

    use local var for pid, seeing as we only just stored it in the shm

commit a52e275c0d72b5c9e099817b4a800f749acfd6dc
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 4 23:08:04 2014 -0400

    fix incorrect index conversion.

commit 06db2569ae149179b682c2029170e188b63fbdbb
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 4 23:07:00 2014 -0400

    use childdata ptrs in log.c

commit 28b6d7cd831ade5c8354d6082b3e0c3ea9d8c13c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 4 23:00:45 2014 -0400

    use childdata ptrs in child.c

commit 707252610462f4b0bc64685a9fc6e59a63a84cf5
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 4 22:46:24 2014 -0400

    use a childdata ptr in render_syscall_prefix

commit 17519f6706944b2f28ea57cb2d6d3b93e1774334
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 4 22:21:00 2014 -0400

    put all the per-child data together in a single struct.
    
    This simplifies the allocation, and makes things more obvious what
    is global vs local to a child.
    
    Accesses look a bit uglier right now, but will get easier to read with
    creation of some local struct ptrs which can be added incrementally.

commit 550d31329e3e52f448216441879c7e5b425a79b2
Author: Michael Ellerman <mpe@ellerman.id.au>
Date:   Wed Jun 4 16:05:29 2014 +1000

    Ignore ENOSYS from sys_msgrcv() and sys_ipc()
    
    sys_msgrcv() and therefore also sys_ipc() can return ENOSYS when
    CONFIG_CHECKPOINT_RESTORE=n. See prepare_copy() in ipc/msg.c.

commit b798333a40248e848235dcd0f3b5fd90d499b423
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 3 23:01:18 2014 -0400

    switch destroy_shared_mappings to use list_for_each_safe
    
    instead of mucking about with ->next by hand

commit 5cd35de838dd0292eb88a64e66329d590fa7a753
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 3 21:36:32 2014 -0400

    make rand_pathname only generate a max of 4k
    
    Otherwise, on systems where page size > 4k, we'd still have the buffer
    overflows when trying to print these buffers in output()

commit 9f57f23aac0c062e89182de41582dcb551bfebae
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 3 16:59:27 2014 -0400

    increase the prebuffer size.
    
    We have potential here for buffer overflows.
    Imagine for eg, we have a syscall that needs a pathname as an arg,
    and we end up allocating a 4k buffer of garbage to pass.
    The output routines try to render that 4k and happily scribble past
    the arrays.  They should probably be more mindful of the lengths
    being passed in so that they don't overrun, but for now just increase
    the size of the arrays. We do it * 6, because in theory we could pass
    junk pages like this in every register (though unlikely in reality).

commit 8553a747eddc2774e95c224cc5eab70d659409d1
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 3 10:29:38 2014 -0400

    use ARG_PATHNAME instead of random addresses in rename syscalls

commit 97658fa1eeb7e72b47384e8bd71eac55117d4672
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 3 00:14:21 2014 -0400

    split common code out to __output_syscall

commit 883827b4b46489ca28c386de96df5cfee282b4b1
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 2 23:36:46 2014 -0400

    make strip_ansi aware of the length of the buffer it works on

commit 56c1bf8df0c2138128081f197b68df124cf63e9f
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 2 23:32:10 2014 -0400

    strip ANSI codes before writing to log files again

commit 81e4a946eb94f5f3e133f2e78d4ce07fddafce76
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 2 23:02:41 2014 -0400

    remove obvious comments

commit a525a8a3de10c4a3f0843705024258f9a4eb24bf
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 2 23:01:02 2014 -0400

    collapse the two render_syscall_postfix cases into one function.

commit 78278780dfb0acb88b754aafbb934dceec5738cc
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 2 22:56:35 2014 -0400

    remove unnecessary brackets

commit 9a3a3cf86f03e14f56bf9af4a0a53d6f1a37204d
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 2 22:55:18 2014 -0400

    replace the various colour macros with inline sprintfs

commit ee6c5bcf3a3ce0a082fdc8e0edd109f298a775a1
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 2 22:47:18 2014 -0400

    runtime disable ANSI codes if monochrome is passed.

commit 7434c23e2a1f5f93855d3ba4f3b982790e3e74e3
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 2 22:22:10 2014 -0400

    convert the arg decoder to operate on buffers instead of fd's

commit 0d7a12232dfdeba897e55580a74f2c08980df464
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 2 20:29:53 2014 -0400

    split out the "decode a set of args" routines from the logging code

commit 4c55105df7a6da300077ae3d41af07ddc077d1ae
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 2 18:16:42 2014 -0400

    remove accidentally committed test code.
    
    sigh, I'm having a bad day.

commit eb36b2a0bf80f447ca460eaf45096ae83e244181
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 2 18:07:27 2014 -0400

    comment out the post mortem code again.
    
    This is going to need a total reworking, as this way of doing things
    isn't going to work.  If a child does something like allocating
    a page to mangle a pathname, we can't output that pathname from
    the watchdog context because the allocation is child local.
    So right now, we pass invalid ptrs to print as %s, and things segv.
    
    Not sure what the solution looks like, but this isn't it.

commit f2545e2af52fd0fb899ccb707a88006a92f9c020
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 2 18:05:52 2014 -0400

    rename opcount to op_nr

commit ff3cc6dd71a1fb6d1c6f27cf1a0437c471dc6bb2
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 2 17:25:15 2014 -0400

    move the op count into the syscall record

commit 551d24e08e75cc9ce94d91055f2177cc4377cc86
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 2 17:15:27 2014 -0400

    just remove the 'syscall out of range' check.
    
    We shouldn't be passing down numbers larger than those defined in
    the syscall tables.

commit cc5108ff851d38db7da31dd9c0d5ee259ae51633
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 2 17:14:02 2014 -0400

    make get_syscall_entry take SYSCALL_OFFSET into account.

commit 4fb375094ccc866add3ea62f140bf6085e8a9fe2
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 2 16:41:07 2014 -0400

    add renameat2 to ppc syscalls

commit 4a5a20d467b32eb2cc2b1757e7bf8bef54ba1f0b
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 2 16:16:42 2014 -0400

    make output_syscall_prefix_to_fd use get_syscall_entry
    
    This way we work out the correct table to read from.
    In doing this conversion I realised that the syscall rec only has the
    actual syscall number in it, and we need the offset here.
    
    One way to do this might be to make get_syscall_entry not care,
    by having it have knowledge of SYSCALL_OFFSET. needs more thinking.

commit 46eebfb3a43569083bb098a55807225958c946ca
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 2 16:06:52 2014 -0400

    compile fixes for the commented out post-mortem code.
    
    it's broken right now, but at least builds now.

commit 06585fb8bbd6039705cc4c6bd1ac0f72d8e643f8
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 30 13:50:28 2014 -0400

    make handle_sigreturn use a bool.

commit 8e9e8582ec14bac816602ac165706aff8f096850
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 30 13:45:49 2014 -0400

    only set EXIT_PID_OUT_OF_RANGE if we haven't got a pre-existing exit reason

commit f2bc9e3dc97321bb38ab589b70288f6f2e9b1b4d
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 30 13:40:45 2014 -0400

    push the EXIT_NO_FILES exit reason down into fd-files.c

commit 22e75e6d22bdb4a099e01d2e330ff3d2c0641d95
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 30 13:36:09 2014 -0400

    remove unused leftovers from when we used to regenerate fd's.

commit 94caf5ecb59e8e6211307a110557a77357cb47e0
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 30 13:34:03 2014 -0400

    add locking so only one child does the reparenting spew

commit 46011de7b964e87c524966f6d79cadd1c02285f6
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 30 13:19:48 2014 -0400

    move post-syscall record updating into do_syscall

commit 02a896d22520d8779ccca42db72549370e23051e
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 30 13:16:52 2014 -0400

    remove unnecessary local vars

commit 07722dde82f3ca77ef11dcf0994e333c79118676
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 30 13:07:06 2014 -0400

    remove no longer needed includes

commit c1a7fb9df4b1b19d4cba8d82f4db1e61cfa86d39
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 30 13:05:49 2014 -0400

    move the timestamps from the shm into the syscall records

commit d4aa4aac72cf66aeaaeddbdb479a342041afb3a4
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 30 12:55:24 2014 -0400

    cache the needalarm state

commit 12afe2107cf14a77402e6f387e308df9134ae4d3
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 30 12:43:34 2014 -0400

    output whether the syscall being disabled is 32-bit or not.

commit c20071d6ef529129210c3dd6d217f5b443a1ba8e
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 30 12:42:47 2014 -0400

    fix the "is syscall active" check

commit 3516340af9e3e9d6f9a10dd6ecbf020632dbff11
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 30 11:15:24 2014 -0400

    regenerate random page every 100 iterations of the child loop

commit bffd1b6c26dcd70768fbec567da72f63808f69b4
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 30 11:09:25 2014 -0400

    transform check_parent_pid into a generic function to delay work.

commit 594973647e6c5d3339110eb3e4762950b8cb2da3
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 29 22:42:39 2014 -0400

    remove the word/long rand page cases

commit f879e1c8a84460408217371b1b74e40882b068ae
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 29 22:40:13 2014 -0400

    remove duplicate memset 0 case
    
    Make it pick a random byte to fill a page with instead.

commit e42dad80dff52fe32afc0128e5abcc739eb2f0e1
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 29 22:38:42 2014 -0400

    remove untrue comment

commit a575181727d9e3f573a3233a251b5ef8dc48518a
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 29 22:37:15 2014 -0400

    move init_page_rand prototype

commit 1cf92a1e4e181846aa4977aeff0575b81ca28b39
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 29 22:36:18 2014 -0400

    fold the page_rand code from maps-static.c into random-page.c

commit ce77c771d461c61e6957df31dfa20a2a39b16932
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 29 22:34:19 2014 -0400

    kill off page_zeros/page_0xff
    
    just use page_rand to generate them randomly.

commit 53eb73c9284607a14aa0ef9d99ecae0aebce653e
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 29 22:12:15 2014 -0400

    remove text about oom killer.
    
    why they disappear is a mystery, but it isn't (usually) the oom killer

commit 71bf3fcceee3141fcb87b7d2d63434a1d3e760d7
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 29 22:11:17 2014 -0400

    move locking related stuff out of watchdog.c to locks.c

commit 7d0faa49655970b6660b6d2c7cc1a9073218f82d
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 29 22:06:23 2014 -0400

    replace open-coded for_each_child loop

commit 859db5ebf05b6ad04cfc9b2df4744fb3f9be0075
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 29 21:52:38 2014 -0400

    reorder decode args so childno comes first
    
    like all the other syscall entry functions.

commit 07c23f9fc63d2dc6f42b663e82b01624cfd81b21
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 29 21:37:49 2014 -0400

    shm reorg and commentary

commit 1815a6eca76380506a249e2825e8f0cc684e6b3b
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 29 21:31:16 2014 -0400

    remove stale comment.

commit 429960a898b1bf9b6a1af6be5d196ea0a09c0c7f
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 29 21:29:55 2014 -0400

    make output_syscall_prefix_to_fd use the child pid in the shm
    
    getpid() will give the wrong answer when we call it from the
    postmortem routine.

commit b96b7681ba458ccc61cf49d1072e15c73dfeaa12
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 29 21:07:22 2014 -0400

    introduce a DONE state for use in the previous syscall record.

commit a71536f95366acc516e200a1fd7eb358a5c45d68
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 29 21:04:39 2014 -0400

    make dump_syscall_rec take a record arg
    
    now we can pass both the previous, as well as the current record.

commit a2174ce74910c4a8fb42f7ae58f99d3609dc447f
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 29 19:48:12 2014 -0400

    don't allow multiple children to update syscall entries at once.

commit b29f7135c8c4ee65a4bc08499e856a3e51445d2b
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 17:53:53 2014 -0400

    export output_syscall_postfix_err and output_syscall_postfix_success
    
    for use in the syscall record dumper

commit 44649b23d54ef278cbf92a95624b57c180a0ad93
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 17:48:01 2014 -0400

    export output_syscall_prefix_to_fd for use in the syscall record dumper

commit 16309364e2e8d6febe7eb6015c65d371f435011f
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 17:45:03 2014 -0400

    pass an fd to the syscall record dumper.

commit 0b1ad2fcc76c7627d0fca5e98bbb2581309a4573
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 16:14:04 2014 -0400

    slot -> childno variable replacements

commit 441fa80623a3016e7bb3f781262e79fa3df822db
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 16:10:54 2014 -0400

    EXIT_LOST_PIDSLOT -> EXIT_LOST_CHILD

commit bf8ddd09c1885ad9ea1a5c338095b63a42f6f660
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 16:09:15 2014 -0400

    PIDSLOT_NOT_FOUND -> CHILD_NOT_FOUND

commit afbda8e316f91674aa9929490654e94f2be99262
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 16:06:11 2014 -0400

    more pidslot -> childno conversions
    
    find_pid_slot -> find_childno
    dump_pid_slots -> dump_childnos

commit 0f94f03754f197a801495b32ad9992147906684c
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 16:04:41 2014 -0400

    rename pidslot vars to childno

commit 2864bcada0b5a5a414d87c21cf508850c94fc9c6
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 16:02:43 2014 -0400

    rename for_each_pidslot to for_each_child

commit 9c9651a28e70b1fa9dfd0d1e080db12a45f73745
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 16:00:35 2014 -0400

    skeleton of the dumping pidslot work
    
    ifdef'd out for now until I figure out how I'm going to make
    output_syscall_* output just to the post mortem file.

commit 1016f2c08a7c99aa2ccb549b94ff45d9dc724ac6
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 13:46:21 2014 -0400

    remove knowledge of biarch/syscall tables from watchdog.
    
    move into get_syscall_entry function

commit 6caa4092f90320ffdc39936d9a43d04dd4c46670
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 13:29:58 2014 -0400

    remove potential usage of -1 as an index

commit 87b2171452e45528e228608a0d100783cfc6dcde
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 13:28:19 2014 -0400

    syscallrec -> rec

commit 1907eb0bf597ddd58cbb53eed932e8fa4cb36e1a
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 13:27:20 2014 -0400

    syscallrec -> rec

commit 97f1f7c90cb5492d9adc10411c204c50170846cc
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 13:25:31 2014 -0400

    syscallrec -> rec as elsewhere

commit bbd26a0e7a6246c539b7e08b7b102d992f043af7
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 13:02:38 2014 -0400

    make toggle_syscall_biarch_n static
    
    (also, holy shit that prototype is awful, could it take any more args?
     Something to revisit one day)

commit 14c9ecd47e649bec949e3d140f9e3f85736cca62
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 11:50:59 2014 -0400

    only fsync log files that have been dirtied

commit 67391f33ad6d492b3d5626068cff20afcf13e7a8
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 11:33:21 2014 -0400

    remove old debug code
    
    this issue got fixed when locking was added to the syscallrecord

commit e641d3a9ce4d8a308450aefda21405657304313d
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 11:30:10 2014 -0400

    syscallrec -> rec, as elsewhere

commit b7ee3ef285a040d9bd25f8c521453b2bbcca112b
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 11:29:18 2014 -0400

    close small race between reading fd and other related args

commit e2561c3003f6062baaa2df3059ef7e36669db580
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 28 11:26:27 2014 -0400

    move the post mortem code out to its own file
    
    it will grow some later when the syscallrecord dumping is added,
    and watchdog.c is getting a little cluttered

commit ce71f3e8f65fbb71bc738439b6e541835118e0ad
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 27 19:26:30 2014 -0400

    childno can be signed.

commit 1d36d2a8f109ea6ba055d177ecd235663bba0091
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 27 19:05:31 2014 -0400

    remove syscallnr param from output_syscall_prefix_to_fd

commit 0460574e7aa63f9344e79eff1c8f6049e406198e
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 27 18:55:31 2014 -0400

    introduce a state field to track syscall records
    
    This will be useful later when dumping them in post-mortem situations

commit 83062aa3c3bed1d070e3c15784a0753ce40334e2
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 27 18:44:23 2014 -0400

    shorthand. syscallrec -> rec, like we do elsewhere

commit 77fbef76f8cfac44223c064750d603c8b607b7f3
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 27 18:42:28 2014 -0400

    remove ret param from output_syscall_postfix

commit 4a31e7ac835e1821d5ffaebb34bfb4e311e096d5
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 27 18:40:34 2014 -0400

    store the result of errno post syscall in the syscall record

commit fe2159aae80266a2779c19138f2e22f99d818a43
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 27 18:33:07 2014 -0400

    remove err param from output_syscall_postfix

commit e99540adad51bf5a9af1771d0a4af5727dc8af75
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 27 18:28:42 2014 -0400

    remove syscallnr param from output_syscall_prefix

commit f567bc3aa7003db74b35fe6f68f51d8f865ddcde
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 27 18:21:01 2014 -0400

    remove pid arg from output_syscall_prefix_to_fd

commit ce2c0f845324a8ef02d3606fc6644e5cc0d0d421
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 27 18:16:40 2014 -0400

    commentary around the execve special casing

commit 6b4c5a4e875e4f7784a101ca1817b2def13ec18e
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 26 20:44:18 2014 -0400

    fix another buffer overflow
    
    now that we handle >64 pids, we can overflow the 512 byte array here
    quite easily.  print out everything one line at a time instead.

commit 5ee08e617eddd67dfc689f69c80d70bfdd105eaa
Author: Michael Ellerman <michael@ellerman.id.au>
Date:   Mon May 26 22:32:39 2014 +1000

    Update powerpc syscall list to match v3.15-rc7

commit 8b2130c9064f6a787e3c469dc1aee17678dd398a
Author: Michael Ellerman <mpe@ellerman.id.au>
Date:   Mon May 26 22:32:02 2014 +1000

    Mark perf_event_open() with IGNORE_ENOSYS
    
    At the very least specifying PERF_SAMPLE_STACK_USER will return ENOSYS
    on arches that don't implement it.

commit dd460eda355bfed3478b71fb0e3de55dd5a6d302
Author: Michael Ellerman <mpe@ellerman.id.au>
Date:   Mon May 26 22:32:01 2014 +1000

    Add an IGNORE_ENOSYS flag and use it
    
    Some syscalls return ENOSYS depending on their arguments. We don't want
    to stop calling them just because we hit one of those cases. Add a flag
    to specify this behaviour so we don't have to keep special-casing those
    calls in mkcall().

commit fdb3d3a62fb23d1a71554780395069482a27bb33
Author: Michael Ellerman <mpe@ellerman.id.au>
Date:   Fri May 23 19:09:25 2014 +1000

    Compat fixes for DRM

commit f548c8b396965eb81c7489a64c337ad2bc8ac2f1
Author: Michael Ellerman <mpe@ellerman.id.au>
Date:   Fri May 23 19:09:03 2014 +1000

    Fix buffer overflow in output() when pid slot is not found
    
    In output() we sprintf() the result of find_pid_slot(). We print the pid
    slot to the buffer with %u and have space for two digits of pid slot.
    find_pid_slot() potentially returns PIDSLOT_NOT_FOUND (-1), which when
    printed with %u is 4294967295 - ten digits.
    
    Fix it two ways, use snprintf() - truncated output is better than a
    buffer overflow. And allocate more space in the buffer, 32 bytes is a
    nice round size, and gives us space for everything.

commit d7e1900169dcc2adb0163ac3e5370602a3141913
Author: Vinson Lee <vlee@twitter.com>
Date:   Thu May 22 21:16:10 2014 -0700

    Fix build errors on Ubuntu 12.04.
    
      CC    syscalls/prctl.o
    syscalls/prctl.c:33:13: error: ‘PR_SET_CHILD_SUBREAPER’ undeclared here (not in a function)
    syscalls/prctl.c:33:37: error: ‘PR_GET_CHILD_SUBREAPER’ undeclared here (not in a function)
    syscalls/prctl.c:34:23: error: ‘PR_GET_TID_ADDRESS’ undeclared here (not in a function)
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 23afb71392f1b7761f72d0de7614b2ad5438003d
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 22 16:59:43 2014 -0400

    remove DIPC as a possible op for sys_ipc
    
    The kernel will ENOSYS things it doesn't understand, and as this
    syscall is obsolete it's unlikely to be extended.

commit 94a56465b23df6fd5332b226ac37a7057d8adad4
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 22 16:48:45 2014 -0400

    fabricate_onepage_struct can now just use rand32

commit 20997240cb2795794020770bd2331de3fb7923be
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 22 16:47:38 2014 -0400

    replace open-coded PAGE_MASK with use of define

commit 62f3a40e1f2aa93256ea7c6b84ea8e0de342c169
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 22 16:46:38 2014 -0400

    add masking of PAGE_MASK in rand32

commit 7bea4b7673763004b8651860b417c81d3bf74af4
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 22 12:48:51 2014 -0400

    make random page generation care about alignment.

commit 072ef5c210760658c6bd7822b8e14c1a4b8dcb7a
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 22 12:26:23 2014 -0400

    fix some shifting errors in mremap fixed addr calculation

commit 306c9b9eaa9689dec126d8de329cf47b54f13919
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 22 12:04:35 2014 -0400

    make rand_range operate on unsigned ints.

commit 13207f0b0b09fcad47f480ed5a878f54cb71f87e
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 22 12:03:54 2014 -0400

    fix yet another sign bit error

commit 6345b7222b4c9f763465bd22ceaedb3bba661458
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 22 11:59:25 2014 -0400

    fix another sanitize=undefined error

commit ba1795e805809d74882f484e7003d538ac1a2de6
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 22 11:58:52 2014 -0400

    fix sign bit errors picked up by sanitize=undefined

commit 1f9cd82112f6e3e4b010caf97173d05fcdd2755e
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 21:58:35 2014 -0400

    replace open-coded min with min invocation.

commit 3787090a79909e840be0d9eeb99ca5701f04f873
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 20:56:46 2014 -0400

    returning from this impossible situation is probably a bad idea.

commit 4fef313624cc28c259c11023118a55f210a30504
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 17:57:58 2014 -0400

    dump the O_RD|O_WR flags too

commit 8b43da94d33c2a4092750910b675d2c8b3299797
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 17:25:44 2014 -0400

    unify files_added / files_in_index variables

commit 307e647a0b6e4356e8ac5554fc2605735e5dc7e3
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 17:21:45 2014 -0400

    factor out the routine to split list into an index of filenames

commit eca35e8fd20d9da066def925e5c6e35fbd91f140
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 16:59:08 2014 -0400

    remove unneeded includes

commit a569806f9556b1ca964c483d473b2e277466b63f
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 16:55:24 2014 -0400

    remove another unneeded local var

commit 451b2bf178e2377b22c2d7f7c6247284bef37c95
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 16:54:37 2014 -0400

    remove unnecessary local var

commit d99127038772b67b76145f297f66eac53a1e47d9
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 16:46:23 2014 -0400

    separate fd provider list create/add.
    
    We need the head node to be empty

commit 47bd57ec541415855858be29a846613c7e1afc38
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 16:45:16 2014 -0400

    fix inverted test which meant we never init'd the socket fd's

commit e8f38829b1a70ab6e792cc1b8b905a1853ccb063
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 16:42:39 2014 -0400

    Revert "special case adding the first node to shared_mappings list"
    
    This reverts commit 4401f6d0f0bfdeb92595520dc3be23dee32efc77.
    
    We need the list head to be untouched.

commit b05fa0c32c6332e098f13dcd5a1fcad303470bec
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 16:13:14 2014 -0400

    special case adding the first node to shared_mappings list

commit df5ef0476183d916a833811d1d5bc9f11ca4e760
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 15:31:18 2014 -0400

    dynamically register the fd providers
    
    Split the structures out to their associated files.

commit 972e2f07791888a7f153f07f532bf9b20ea1b2d5
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 13:19:52 2014 -0400

    code reorg

commit 79a874eeae95a802a9b62dec7691b9d6a8c2b3a5
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 13:13:47 2014 -0400

    convert get_new_random_fd to use the ->get operations

commit 4c7afd7ac40ff9ed6e92b314b1cfe48631f4daed
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 13:09:29 2014 -0400

    remove the specific_proto == always use socket fds assumption

commit 1a34c2c4ea063775cf648c17b72112108db84bb4
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 13:09:10 2014 -0400

    move the files fd failure checking into the ->get function

commit d077df6b5d5d636aceaecd31e113dbe13a5e5b31
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 13:06:53 2014 -0400

    move socket->get failure checking into the ->get function

commit c90310f8acb912e4a380cac4d194f0c124a53d13
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 13:05:39 2014 -0400

    move perf fd->get failure checking into the ->get function

commit 5816588264dff25a3cd5b2f2b6278354655c22e2
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 13:01:54 2014 -0400

    convert files fd to use ->get

commit b100e9c05115a37ea80f407dbd2b88ef8806a0f4
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 13:00:48 2014 -0400

    convert eventfd fd to use ->get

commit dd405640a322c1e55f11f476c2b07de273414d97
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 12:59:43 2014 -0400

    convert epoll fd to use ->get

commit 5bf661bbb6313461d9aa9d1fc3d772ad9bd26644
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 12:58:49 2014 -0400

    convert perf fd to use ->get

commit edfcad714573d0b2bdbfb40caff1a8133fe127d1
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 12:57:11 2014 -0400

    rename get_socket_fd to follow naming scheme

commit eecb6f49152281984e0e0249cd1679d2a103136d
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 12:56:00 2014 -0400

    convert pipe fd to use ->get

commit d4eb82160302a51a6c248c88fe91c254449e6dc5
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 12:53:07 2014 -0400

    initial pass at fd_provider->get
    
    add a provider for sockets.
    
    We won't be able to really clean out get_new_random_fd until the
    dynamic conversion is complete.  Until then it might have to grow
    a few more special cases.

commit a8a64299e09cbc634d9ec3ced060431bc2cbc623
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 12:47:20 2014 -0400

    convert open_sockets to use fd_provider struct

commit 767e065857c2f749a0ab5be9613c4f01b99fcbe6
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 12:44:09 2014 -0400

    convert open_files invocation to use fd_provider struct

commit 8c1ea4a75ff6e1fb45d041f0d29744669799d5ac
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 12:42:52 2014 -0400

    make fd generator ->open's return whether they succeeded
    
    and propagate that failure upwards

commit c278d8eb76e1b101184eec93a1d82b646657ef0b
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 12:32:14 2014 -0400

    create an fd provider struct and iterate over it
    
    baby steps towards dynamically creating fd tables.
    Eventually this static struct will be allocated, and the various
    fd generators will register their open/get operations.

commit 2e22ec49b30cdff24704b2aa9714eda3a83e56e5
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 10:24:43 2014 -0400

    test-vm.sh: small updates to match other recent test scripts

commit 8bbaddbf5596927754d8677d5471e07fbe6c1df1
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 21 00:08:00 2014 -0400

    fold rt_sigaction contents into sigaction file

commit 07e298310fe3fc4a64c1b20c15c63da7a28e5ba8
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 20 23:13:21 2014 -0400

    move perf specific prototype out of sanitise.h into perf.h

commit 3890969278a618897fec02cd8ac919a7a8ab1a02
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 20 23:09:06 2014 -0400

    remove dead function

commit cb110ce6364edaec77a7a74a1d537ff27b67885f
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 20 23:08:28 2014 -0400

    scope reduction

commit 5d4087e949beefaebc3f6aa27ff070d9315e7516
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 20 22:59:51 2014 -0400

    scope reduction

commit c4d6aa96fe89f2ceb044f73b7409b8a8ff8c7139
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 20 22:57:19 2014 -0400

    scope reduction

commit 04e82d36d875dd53b8d6d40105548db92a2f36f8
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 20 16:07:08 2014 -0400

    move splice definitions to compat.h

commit 2523a5b2dacec7783c28b98cab78eb8d0d3699f4
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 20 16:02:30 2014 -0400

    make ->sanitise take a ptr to the syscallrecord
    
    most of them will need to reference either the args, or the retval.

commit fa4307ef4d031afa86deece443e2a0af291b3e31
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 20 15:09:54 2014 -0400

    make ->post take a ptr to the syscallrecord
    
    most of them will need to reference either the args, or the retval.

commit a4583f1ad2cc5068bc6d594284e9a4e3db6282cc
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 20 14:36:50 2014 -0400

    use syscallrecord struct in input ioctl sanitize

commit a49a226272f8eb9c886cc160222cfba8ff886325
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 20 14:35:19 2014 -0400

    add missing include

commit d18c6317b7d3a96c999e9cf73bd7ee03d8fb26cc
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 20 14:34:48 2014 -0400

    use syscallrecord in autofs ioctl sanitize

commit d0f8183b543c139e1a1001e77968a61b135eceda
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 20 14:32:38 2014 -0400

    use syscallrecord struct in mbind sanitize

commit ecf244c166b6ea0ccca1a6e99303c048f685c04e
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 20 14:31:49 2014 -0400

    use syscallrecord struct in mlockall sanitize

commit 6dd91cc7fc7c37b09f8703f66b78abaaa001bae7
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 20 14:30:54 2014 -0400

    use syscallrecord struct in ioctl sanitize

commit d870b8fa626eab36d46383fc9ac711c5ad156426
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 20 14:26:55 2014 -0400

    use syscallrecord struct in remap_file_pages sanitize

commit f37c1adc83d2ee1f6bfe1917ee3e041c703a03b2
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 20 14:26:03 2014 -0400

    use syscallrecord struct in rt_sigaction sanitize

commit eccc858a63ce7b7b3f7051c449fe53a8f12c7f9a
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 20 14:25:04 2014 -0400

    use syscallrecord struct in socket sanitize

commit 5e795e566a32fe6c0a4097540b27370061ffe20d
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 20 08:44:47 2014 -0400

    use consistent uint type for argnum

commit c2dc5c5e9dd4fcf48b62181f0320a8786811a2ed
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 19 20:47:33 2014 -0400

    make a bunch of char *'s in perf_event_open const

commit f565c38f716844addbb5479a904cca22a91242d9
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 19 20:45:49 2014 -0400

    use syscallrecord struct in perf_event_open sanitize

commit 971f0e853cfebcfc76b7227551828e783ec230e3
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 19 20:38:53 2014 -0400

    use syscallrecord struct in splice sanitize

commit dc8d2c89ba84ec2013366e7f33f57db1a55ea154
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 19 20:37:29 2014 -0400

    use syscallrecord struct in setsockopt sanitize

commit 48bf588e3cc804f9107553ed04b6df52afb2e454
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 19 20:34:57 2014 -0400

    use syscallrecord struct in socketcall sanitize

commit 692528dbdeb19841f53f3fac04f87a78e577b6e9
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 19 20:30:39 2014 -0400

    use syscallrecord struct in prctl sanitize

commit 2ca505f3f3dc3298368abb9b5e6e69e24af88e37
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 19 20:28:21 2014 -0400

    use syscallrecord struct in sync_file_range sanitize

commit b7b6340c5b63a3396b02c4b161818c33a7b08ca0
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 19 20:27:29 2014 -0400

    use syscallrecord struct in mremap sanitize

commit 17832c324eb7c03fe1b1ac322405a0aeaee95d5a
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 19 20:25:56 2014 -0400

    use syscallrecord struct in move_pages sanitize

commit 89e7f04250806d4b80dc7980d50071f8d3a5654c
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 19 20:24:25 2014 -0400

    use syscallrecord struct in munmap sanitize

commit 6a0a29020d9467f40972cccb967b65e43fd549f0
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 19 20:21:26 2014 -0400

    use syscallrecord structs in mmap sanitize

commit 7de1f6ec19692a9c3096e5a1c806d4e919db3be3
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 19 20:17:11 2014 -0400

    use syscallrecord struct in fcntl sanitize

commit ea9ae1cdb7f9e56dd79f9e7c066d3bbbae2c35cc
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 19 19:01:41 2014 -0400

    use syscallrecord struct in common_set_mmap_ptr_len

commit 8e9dd6c288b6b07af071beb17151e27779a52619
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 19 18:57:17 2014 -0400

    use syscallrecord struct in child_random_syscalls

commit 047ee482874bedbd1781996d44673b69d2cf4562
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 19 18:53:09 2014 -0400

    just pass handle_arg_sockaddr what it needs

commit 239be37f046006cb18a40c10c6b9e3cd099d37f2
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 19 18:51:32 2014 -0400

    just pass handle_arg_iovec the args it needs

commit 187bfe34b8d3eaf8b022a3a87774895dc0ddab91
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 19 10:29:46 2014 -0400

    add some missing prctl flags

commit 89009e4749f2b8d949fe3708cead0e88c6696219
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 16 16:17:47 2014 -0400

    don't bother calling find_previous_address if argnum == 1

commit ff2c56cef2428bd984e221287ccb6851e33fb9b8
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 16 15:55:36 2014 -0400

    rearrange find_previous_arg_address args to match handle_arg_address

commit 09c354b47b4d2494b4f691e081012352136d3eab
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 16 15:53:28 2014 -0400

    handle_arg_list really wants the syscallentry so pass it instead of call

commit d6f2dffe0a28ac9502d54e4b3481e6da3a58a008
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 16 15:52:37 2014 -0400

    handle_arg_op really wants the syscall entry, so pass that instead of call

commit 8f0b052925c33079d5be933f68093defce6a13c8
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 16 15:51:42 2014 -0400

    handle_arg_range really wants the syscall entry, so just pass it

commit 94383d223404584c39b53030296e417897ff23cf
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 16 15:48:33 2014 -0400

    fillarg doesn't need the call param
    
    it too can get it from the syscallrec

commit 61ef5676f0836e15e4068842d2543c92d82822b5
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 16 15:45:50 2014 -0400

    handle_arg_sockaddr doesn't need the call arg
    
    get it from the syscallrec instead

commit 2468eb45f80341f50f51e131a0b26ea58e6e40bb
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 16 15:42:15 2014 -0400

    remove passing call number around to handle_arg_address & find_previous_arg_address
    
    just read it from the syscallrec

commit 216add1ba60c1932daed6d0f1ac21d5f743ec42a
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 16 15:39:11 2014 -0400

    call numbers are always ints

commit 9d6df52daf0fc5afc91b068dec6223d6f8c039b3
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 16 15:35:49 2014 -0400

    remove call param from handle_arg_iovec
    
    we can get it from the syscallrec

commit dde69164c70e0dff256d463e8420e902ff9ab095
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 16 15:33:16 2014 -0400

    use syscallrecord structs in generic sanitise

commit a017b18ad2df369f90ac1a0b009bbbd49ed9b310
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 16 15:23:36 2014 -0400

    use syscallrecord structs in mkcall

commit b2119ea6b69a06d7ca63bdfa6708ab1b59c5a519
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 16 15:20:22 2014 -0400

    use syscallrecord structs in watchdog

commit 5b926550312412a29cb46a7ee2f963ed5c63bab6
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 16 11:40:12 2014 -0400

    partial revert of previous commit
    
    While log.h was missing the params include, we shouldn't make .c
    files start depending on implicit inclusion.

commit 19b662fc1286c307c402e1ffeb8ba49bba5a5474
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 16 11:29:38 2014 -0400

    add missing include dependancy in log.h
    
    remove unneeded include in pids.c

commit cf69be3193061fcdfd70f043dd60ec9684d7f05d
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 16 10:53:47 2014 -0400

    simplify do_syscall by using syscallrec ptr

commit 4bc9e6b674326f215796e4a11c1a91ebbc4786a7
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 16 10:47:44 2014 -0400

    remove x86-64 specific commentary

commit c933ecd48996be1b02fbcbc40f748ba757e5837a
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 16 10:45:52 2014 -0400

    move "do 32 bit syscall from 64 bit code" implentation to arch header

commit c61d07cd0c5fbcab59b0808995cd8c76e5c239fd
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 15 17:27:41 2014 -0400

    do the TRINITY_TMP chmod after a trinity run has completed, not before

commit 0dd7aa029a1006576ba4b9e663d952091a52a003
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 15 17:26:19 2014 -0400

    bump up nr syscalls to run each loop, and avoid execve
    
    a million execve's is pretty boring and takes a while.

commit c6e59851212874d8a13f2cba2760a9968c0468d2
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 15 17:08:16 2014 -0400

    remove the outputting of the reseeding.
    
    This causes a *lot* of the spew from trinity, and will only get worse
    as child processes start segv'ing more often.
    
    now that the watchdog dumps the last known seed when it detects taint,
    that is hopefully good enough.
    
    The one corner case is where we crash so badly that the watchdog
    doesn't get time to write the last seed to syslog.  In that
    circumstance, we used to be able to look back at the last known
    good seed.
    
    A better solution might be to dump some state periodically over
    the network to another system.  I'll experiment with this at some point.

commit 3c9140f5f6cede77fdd2b0a90c03e2e1bcdae493
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 15 17:04:48 2014 -0400

    When we're tainted, syslog the last seed.

commit 478164dfe63fbe5f6bebb2f6629a7730f9f9022f
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 15 17:02:59 2014 -0400

    move "we're tainted" code out into own post-mortem function

commit 172ec766076c66628ba9faec84b73802cf3c2be2
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 15 15:56:20 2014 -0400

    use a larger buffer for debugf

commit d360a190bbbcce74d1b358735dcc039163173e07
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 15 14:34:35 2014 -0400

    update ignore gz -> xz

commit 9488a354926076a515fc5a7006d4c09019d52ab4
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 15 14:34:03 2014 -0400

    Add some kernel patches that silence a lot of noise while fuzzing

commit af527f9d6b08cda185b08315ad6fb787c2d1a117
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 15 11:18:04 2014 -0400

    missed one place where we need SYSCALL_OFFSET
    
    Pointed out by Markos Chandras <Markos.Chandras@imgtec.com>

commit 09b68a379f9ba616d1a3f40b5b868a8aa4ca7ab9
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 14 21:05:41 2014 -0400

    annotate ipc syscall 1st arg

commit c03ddf4be26c46c36e3c93a8f2e5848074d78c73
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 14 16:10:44 2014 -0400

    fix incorrect use of git version macro

commit 2c00da102747377b7bf1352f8d94cbee0b839410
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 14 14:34:06 2014 -0400

    trinity.h inclusion cleanup

commit 7428f9b5dc051aa9e56d33a9032091105d4b14e9
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 14 12:10:44 2014 -0400

    only do_syscall needs to care about SYSCALL_OFFSET

commit 77feebc059f699776068d1016ee00a772a82c773
Author: Markos Chandras <markos.chandras@imgtec.com>
Date:   Wed May 14 12:35:38 2014 +0100

    syscall: Take SYSCALL_OFFSET into account
    
    MIPS and IA64 have their syscall tables starting at non-zero
    offsets so take that into account when executing a sycall
    otherwise trinity just fails with ENOSYS error messages.
    
    Tested on MIPS 32/LE system.
    
    Signed-off-by: Markos Chandras <markos.chandras@imgtec.com>

commit c6dd22e2fefa357784ca4b132a2a08ecd8738384
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 14 11:58:28 2014 -0400

    include trimming

commit b65337218e7bf28317917044a490c0d810f7b103
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 14 11:51:31 2014 -0400

    remove pointless alarm call.
    
    The main pid doesn't use alarms, so this is redundant.

commit 40569bc796880010e6c0ae6c0cc99e49d2cbb7e9
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 14 11:50:10 2014 -0400

    move child log truncation out of main loop into child init

commit 4721d63c2c9ce89df25fa2b79abbf2745cfe4574
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 14 11:44:12 2014 -0400

    Make fork failure only fatal if we have no running children.
    
    This is an experiment, and one I tried before, but let's see if
    we can at least do better than just exiting the first time fork() fails.
    
    Last time I tried this, I ended up with a ton of fork failures,
    so we might still need to do something else but that was before
    I did a lot of the VM work, so things may be better now anyway.

commit ab95169e3f28b5d22cb19bd7f77a352421be79b9
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Tue May 13 14:51:35 2014 -0400

    export pmu data from perf_event_open.c
    
    export the perf_event_open() pmus structure, that way perf_fuzzer can
    pretty-print the pmu names.  It saves a lot of code duplication.
    
    I could always just make this change in my local tree, but it's nice
    having perf_event_open.c the same in both projects.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 69bb556c8b26adc3cb6791c5b3cb140c0416d43c
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Tue May 13 14:45:28 2014 -0400

    make perf_event_open() create system-wide events
    
    This patch adds a create_mostly_valid_global_event case to the
    perf_event_open() code.
    
    Without this patch it is extremely unlikely that a global
    (not per-process) event would be randomly created.
    
    Most users will not notice a change, as you can only create global events
    as root or else if /proc/sys/kernel/perf_event_paranoid is set to < 1
    (the default is 1).
    
    With this change some of the more obscure event types such as uncore,
    northbridge, and RAPL power events are occasionally properly created.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 1b437a056af8df9a2f952171c28aefa5d71ce21f
Author: Wang Sheng-Hui <shhuiw@gmail.com>
Date:   Wed May 14 13:42:48 2014 +0800

    include tile headers
    
    Include tile headers to add in the tile support.
    
    Signed-off-by: Wang Sheng-Hui <shhuiw@gmail.com>

commit cbbca384a0e15c0853449afc7645cc0a1d3f68ad
Author: Wang Sheng-Hui <shhuiw@gmail.com>
Date:   Wed May 14 13:42:05 2014 +0800

    tile: add arch-tile.h
    
    Add Tilera constants definition.
    
    Signed-off-by: Wang Sheng-Hui <shhuiw@gmail.com>
    
     create mode 100644 include/arch-tile.h

commit 8048140b870fd8e28b2e3692e911e17d6d0cc154
Author: Wang Sheng-Hui <shhuiw@gmail.com>
Date:   Wed May 14 13:39:02 2014 +0800

    tile: add syscall tables syscall-tile.h
    
    Define the tile syscall tables struct syscalltable syscalls_tile[]
    
    Signed-off-by: Wang Sheng-Hui <shhuiw@gmail.com>
    
     create mode 100644 include/syscalls-tile.h

commit 3402cd886074502c6f7f5db0e40d04cfb2b6ee99
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 13 18:22:49 2014 -0400

    fix off by one in page size masking.

commit 3f12c5cf055aba4e52cabbade535ab44128bf671
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 13 12:00:08 2014 -0400

    add an exit_main_fail() function and use it.
    
    This should clear shm->mainpid in all the places that matter before the
    main process exits.

commit 4cfb489e26013b9b68db943075e4b2ad300053c7
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 13 11:46:53 2014 -0400

    Clear main pid before we exit.
    
    Reported and diagnosed by Michael Ellerman.
    
    If we don't do this, the watchdog will spin forever.

commit 13294bb73fc81cda04dced70627189603732638a
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 13 11:46:24 2014 -0400

    silence a command that I missed.

commit 7ddbd6e5988ba49ec26dabebe4ebbaa7444a2406
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 13 11:46:02 2014 -0400

    use a 'v' in the tag name

commit 3cc11a5c1c14a0f7372948b1146eea99fc902052
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 13 11:45:44 2014 -0400

    silence some Makefile commands

commit f3b81589ce1e8a636a6be2f1f62dfb0b359d91cf
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 13 11:44:39 2014 -0400

    Use xz for the coverity tarballs too

commit 1b467cc0c67001047e07de0ce35addbc987822ad
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 13 11:44:16 2014 -0400

    use .xz's when we create tarballs

commit b053e344f2b9847cd608aa70114b27d801f4ae7d
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 13 11:43:26 2014 -0400

    Enable -Werror & devel mode

commit 0ebd6fac77aeb9611b97dd3eeb3814711de5cbe9
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 13 11:43:12 2014 -0400

    begin 1.5 development

commit 535797e2cfb353001e16c3f4fad60528507ad3ef
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 12 13:00:12 2014 -0400

    1.4 release

commit 0ae4052e0cf767d30494913a7a93d9ad2d261bc4
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 12 13:00:05 2014 -0400

    Disable -Werror & devel mode

commit 9b862065a3370b17661fe160e98389117e53ddca
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 12 12:50:24 2014 -0400

    fix shifting error in mremap

commit 73153719f0045d51b2a1bc323f9f3cdfa09290d0
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 12 12:48:37 2014 -0400

    always close sockets opened so far before regenerating.
    
    Otherwise we overflow the fd struct.

commit 9dd5a3a2ef3009cb7de0f21dff0d35cb8f51dc3b
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 12 12:44:31 2014 -0400

    we don't care about return codes of socket shutdown
    
    it's not like we can do anything about them if something
    goes wrong, and it just causes spew.

commit 47bf19763b030ad00764efb31451714843d06b3a
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 12 12:00:29 2014 -0400

    in test-multi, exit every million syscalls.
    
    We sometimes find teardown bugs on process exit.

commit ac963b59e442758acd8d615e00daac6723cbe1a8
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 12 11:59:24 2014 -0400

    This isn't an error, so should use output()

commit 7289ff90f79de4045febea6b08f4cc9ad06eb287
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 12 11:59:00 2014 -0400

    silence some spew

commit 969bcb05d3062a6d89aa80a3a001d0f83fc77f3d
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 9 12:33:26 2014 -0400

    fix leak of namelist ptr

commit aa533bbd88e421aa3a5ad58c8a7176480c82ccba
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 8 14:06:23 2014 -0400

    add a missed rand_range conversion in rand32()

commit 84f90a9f347b4da6658576ad80862919b2b0f353
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 8 13:59:57 2014 -0400

    use rand_range in get_len

commit 1b83da7b23d49a08272239b8afce7eed7bb5d2c2
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 8 13:58:26 2014 -0400

    use rand_range in rand8x8

commit a659cf60a6550bd2734a61f823b811f11877fd37
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 8 13:55:52 2014 -0400

    use rand_range in random-address routines

commit 61c766b76cffcf6b68d965a15914600c1d41f019
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 8 13:53:39 2014 -0400

    use rand_range in generic sanitize routines

commit 0015109ff656334e786a3c2bf59e1675ffd800b6
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 8 13:52:33 2014 -0400

    use rand_range in fds.c

commit b8619dcbe3d42ba213583f3ea6e1b444a3c1ba36
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 8 13:52:15 2014 -0400

    use rand_range for listen backlog parameter.
    
    Also bump up upper bound to 128.

commit 4fd84b8869068ac941336a2fa532c29a5ead9421
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 8 10:48:27 2014 -0400

    add a rand_range function

commit 838d5e6d7b9d99c172236b7197943f0b6f161c4e
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 8 10:32:55 2014 -0400

    cpu numbers start at zero.
    
    Also, don't bother with two separate "out of range" cases.

commit c9f335c1f5bdfb3fb15010cf7ebfc129fa5b1c18
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 7 00:37:47 2014 -0400

    CFLAG ordering

commit 2b8ba829cb0f34358f418a6ce0b3763d8d4641b4
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 7 00:27:02 2014 -0400

    destroy the linked-list of filenames when we build the index.
    
    We don't free the ->name entry of each node, because it will
    now be pointed at by the index instead.

commit 51c3742b8ac22a94c06990f4e44081bc186e1468
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 7 00:25:16 2014 -0400

    use a temporary buffer for debug output
    
    Passing va_list's doesn't work, duh.

commit 7a5044acd8a506610a4f7ef77f1ce0a29d17ab39
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 6 22:11:44 2014 -0400

    update syscall list for superh

commit 712824b6c57f524b4ba6297b0761bc2892ce45de
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 6 22:10:22 2014 -0400

    update syscall list for parisc

commit 66ecba33967554477629f6c9a6d77b729d347ad1
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 6 22:09:01 2014 -0400

    update syscall list for ia64

commit a1506b327c7cd983f94ea569503c266b28e5623c
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 6 22:04:02 2014 -0400

    update mips32 syscalls

commit 719f75d3d99e06aee61ec17107893598c0693c25
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 6 22:01:49 2014 -0400

    update syscall list for aarch64

commit c38324026b7b5589810d4c34e25b6efe0a37f4e3
Author: David Miller <davem@davemloft.net>
Date:   Tue May 6 20:24:40 2014 -0400

    sparc: Update syscall table.
    
    Signed-off-by: David S. Miller <davem@davemloft.net>

commit 65df833cd7d5ef6aba82099dc5a5adbc2272371d
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 6 18:37:29 2014 -0400

    Move eventfd2 defines to compat.h
    
    Reported by Dave Miller on Sparc.

commit 2d67b0aab48b2d28cf5fd18409651daa9500be0f
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 6 13:15:42 2014 -0400

    use zmalloc on log startup in lieu of error checking

commit 1fdd4f2e6ea542d5e50d09b846b30534e1fe52e9
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 6 12:28:09 2014 -0400

    take syscall lock when writing new registers.

commit d1deea94d4a36f37e01f01d2f0907f7539659213
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 6 12:27:47 2014 -0400

    watchdog: reduce debugging spew using debugf

commit b368a26fef7771b09d095c1fb55ea23f22bf7ca3
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 6 12:27:26 2014 -0400

    invert test in debugf to reduce indentation.
    
    add commentary

commit bacaf7d52fbd7b567c0b0d67a7242dcbd810fc02
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 6 12:16:22 2014 -0400

    remove unused validate_address function

commit 5179820a920360a9a6a3d3b33f3c5386ba8c92ea
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 6 12:15:56 2014 -0400

    remove unused within_page function

commit 342f2f2646b2b9d34c2158f6752f3feba359bfcc
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 6 12:13:02 2014 -0400

    introduce get_writable_address

commit 0338515089597b8c0e6a2b38eabbabf35e0fda32
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 6 01:17:49 2014 -0400

    fix up dead lock detection.
    
    pid_alive return value needs to be checked.

commit 6e04c8ae4690056e0927441200912029d9c15524
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 6 12:10:45 2014 -0400

    remove address munging.
    
    This causes more problems than it solves, and will be done in a better
    way in follow-on patches.

commit 7b8669132a62eb03bda36c02b3284cbcc7aba23c
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 6 00:40:33 2014 -0400

    limit the shifting in mremap a little.

commit 569cc143f93a68cce1e580507bda5ca9607012d7
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 6 00:10:51 2014 -0400

    introduce a nr_pages function to make the dirty mapping code clearer
    
    Also, change the 'every other page' routines to work sometimes
    on the odd pages, sometimes on the even.

commit d4533124eb0df7993350f71b190b9b732ceeb39f
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 5 19:43:52 2014 -0400

    bring back the previous table in mremap, but use them as alignments instead.
    
    This seems to be a lot more useful than randomly corrupting glibc.

commit 21feb66d5e5e693a482e27101a691bbff1f3e053
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 5 17:39:15 2014 -0400

    mremap 5th arg is an address, not a length.
    
    Make something up for now. Later we'll do something smarter.

commit a6a54cd8ecb71e27abcb7693b648cd31b3ca7e01
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 5 15:43:29 2014 -0400

    remove mystery +1's from page counts in page dirtying code

commit e2c765d7ab585c66fdc217efd1605d8a20e05e1d
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 5 15:39:01 2014 -0400

    fix two signedness issues in mapping dirtying code

commit 9d35588bd4b54db395e74814df68318d1a18a33e
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 5 15:38:31 2014 -0400

    fix up a bunch of off by ones in the dirty mapping code.

commit 2e9314b8ce474b85f5e7924ecda9b9b9ddf22f4f
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 5 15:28:22 2014 -0400

    mremap sanitize fixes
    
    - do the mapping dirtying in the ->post not in the ->sanitize
    - only do the mapping dirtying on success.

commit a567d9debe12b45bbfeb4ed9ea3b8996e4f1f400
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 5 15:11:03 2014 -0400

    Copy correct number of bytes in unicode fuzzer

commit fd6b46fe9ac421e41cae7836a7d17060a1ad196e
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 5 13:11:55 2014 -0400

    Make the syscall lock be per syscall instead of global.

commit 13cfc90e1f17bd474c5fbe0574255e7e2845545f
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 5 13:10:34 2014 -0400

    add missing include for pid_t

commit dff7e7a53beb0394b72fca7cb88d3b8cf92baaa9
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 2 11:58:57 2014 -0400

    change formatting of stuck syscall msg

commit c7db86f099f82b0dd419ca413cb6fa40dcf8ace0
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 2 11:51:19 2014 -0400

    reduce some noise in the watchdog kill spew

commit db50bd6b46179607c1813c452120eb8b29ca7bd7
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 2 11:03:56 2014 -0400

    watchdog: remove some debugging leftovers

commit 4aaed2509b4dd901c83d22fd08d824030ba5de8a
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 2 11:02:39 2014 -0400

    watchdog: simply the 30s/40s codepaths

commit 59910c69778b657b95fb712a8dc8465ac37e3915
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 2 10:44:52 2014 -0400

    watchdog: remove unnecessary sleep from child checking loop
    
    We don't need to wait for a child to exit here, because we don't
    interact with it any further in this function.

commit 8132b7599a1a6866101146265e3012eb746eabd5
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 2 10:43:22 2014 -0400

    Increase time for follow-up SIGKILLs
    
    Sending one at 30s and then at 32s (next time round the loop)
    might not be a long enough delta if we're swapping like crazy
    and taking a while to page stuff in/out. Bump it up to 40s.

commit e852c267200c1c6825f383ee3339eea2759fcfaa
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 1 18:17:37 2014 -0400

    if we tainted, tell the children to exit as the first thing.

commit 8a24a432ff9d887e243cadbd1e4639352d37f4d9
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 1 18:13:42 2014 -0400

    Add some comments to the watchdog.

commit bb35d76614ed19a6703dbef8c77a4dffbd013da2
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 1 18:10:07 2014 -0400

    just duplicate the unlock instead of the goto.

commit 070821322c5ec89e71dba4ac3c017e0b758008f2
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 1 18:07:27 2014 -0400

    add missing SIGKILL to check_lock
    
    and factor out a common kill_pid routine.

commit 0cb2b9c48a3f67908ab3fb6b5aa2aadbf8f7f538
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 1 18:02:13 2014 -0400

    move the lock stealing code to the watchdog.
    
    Rather than have each child try to do this, which can't really work well.
    This does require that the watchdog needs to know about every lock we
    define, but that might not be a bad thing.

commit ed57247c3bddcade15a397ecf288a3ce0c945a95
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 1 16:47:13 2014 -0400

    move multi-arch syscall disabling to tables.c
    
    and delete all open coded variants

commit 875584c815be8d088a6745a05dcb2e5594fedb99
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 1 16:38:58 2014 -0400

    shorten previous syscallrec assignments

commit 6cb1d590e591063581300f2d323bacb765250901
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 1 16:37:17 2014 -0400

    use syscallrecord struct for shorter code

commit 69bacaa296232c90666627b72ff6c25198187063
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 1 16:18:39 2014 -0400

    replace a gcc all with $(CC) in dependancy generation.
    
    Broke building with CC=clang

commit 43e5322c693ca58d73285417a8484a9f2f048fc6
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 1 16:17:08 2014 -0400

    remove unused macros

commit e8fa9d6533378d50a99fa4120f09299b7074d89a
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 1 00:24:48 2014 -0400

    update min/max to match Linux kernels variants.

commit abe2605a3ff1e5193374ef1fd7d68c0dc18607f7
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 30 19:01:32 2014 -0400

    make all test scripts understand and use $DROPPRIVS

commit b2ccced06dfaa17303bc2b108e46f6a04c750991
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 30 18:58:05 2014 -0400

    unify taint handling in scripts

commit 96a3e2e371f5a2c09009423c03d876ba55f95972
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 30 18:54:52 2014 -0400

    consolidate path munging in test scripts

commit 119addbee392bd7340a3b12b2749e775240c0689
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 29 17:00:42 2014 -0400

    update find.sh

commit 60a94d8164dab96f3d7f6a7f89c24493a7284df7
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 29 16:58:55 2014 -0400

    Make -c followed by -x do the right thing.

commit 8d713f1509ec14fa8f6273e85a15bdffde2a9813
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 29 16:42:45 2014 -0400

    non-biarch arches aren't necessarily 32-bit.

commit aa74dfc73f5e9639c4647ca5b491cb3de657e070
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 29 15:11:07 2014 -0400

    disable printing the operation name for now.
    
    For 1.4, we'll only have one op still, so it's pointless to print it.

commit dba35e5828a1f85d03730455ca10e1e31a78b5f9
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 29 15:09:49 2014 -0400

    increase lastop variable scope
    
    Otherwise the setjmp resets it.

commit 24a2df41ca18c30891c2ba2f047157895629a924
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 29 14:05:31 2014 -0400

    Clean up coverity files

commit b9bf52b481b02703d6ef048501d16968fcbac29e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 29 13:29:17 2014 -0400

    if zmalloc fails, do an munlockall.
    
    We may have done an mlockall() prior to this call, which would
    screw up even tiny allocations, like a single page.

commit 71c2b4fe76a609236410f0f197eb38fb90b34b00
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 29 11:30:01 2014 -0400

    locks: make change lock state the last operation.

commit b52a6e6796422d4ac9d4557f9be142084bbf584d
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 28 18:06:57 2014 -0400

    make main process clear its pid before exit.
    
    Suppresses an extraneous "main pid has disappeared" message
    on successful exit.

commit cb6d60ad2ecbceebbb8de5e32f4f687a8c2f2990
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 28 17:57:13 2014 -0400

    remove misplaced \n

commit b0764811b3b7c5f20e47a40c45f3fa9b06c34fa9
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 28 17:53:02 2014 -0400

    hush the children at exit time a little.
    
    This is too noisy when there are lots of children.
    Rather than adding a mutex or something, just remove the output

commit 8f14f080f891f7677cbc988d54a5eb6702363dec
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 28 11:20:42 2014 -0400

    rename file-private locks to "open file description locks"

commit 43a46693b7803b6a4290d97a53474f215229f244
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 25 00:15:14 2014 -0400

    remove unnecessary headers

commit 1f469f22451d785602d892730d9535dadfeaef8a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 25 00:13:22 2014 -0400

    remove dead prototype

commit 72bc7deed6998c5c3523727e946e779b41c110cd
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 25 00:12:54 2014 -0400

    kill off constants.h
    
    move leftovers into trinity.h

commit 01f1aab34f7bbf5bc7b064be770363b5823e80e0
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 25 00:09:48 2014 -0400

    move syscallrecord struct to syscall.h

commit dc42692a22fe591b32b236d50363fd1f6a1f9cdf
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 25 00:07:55 2014 -0400

    move MAX_NR_SYSCALLS to syscall.h

commit e863f0cdcc9945e1cd31e4a35f296cb770c9a0e4
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 24 17:05:10 2014 -0400

    remove duplicate spawn_no_more check.
    
    This is done in the fork_children() loop

commit 55eb1a867fa4ae88597743b91a57fe0d2257f1a8
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 24 17:02:53 2014 -0400

    remove the regenerate code
    
    it's never been particularly useful, and actually has a bunch of bugs.

commit 8627c579f299b5e27f4dbbf944cff56ae89c28c7
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 23 11:58:59 2014 -0400

    bump up the regeneration time.

commit f25973c127b8189dab47684fdd8c814f6415a900
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 23 11:29:33 2014 -0400

    remove mirror target

commit 490c7686d13bd1e404c2c3651ae9c083de796450
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 23 10:51:00 2014 -0400

    handle some ftw errors properly.

commit 1348d254a03f9740eb4c832a65ba990296907544
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 22 17:03:51 2014 -0400

    split up constants.h some

commit 3bed99d23fb1df13a6bbb04570d1d99e8e80eddb
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 22 12:37:37 2014 -0400

    add infrastructure to generically free arguments post syscall
    
    For now, just used by the iovec alloc.

commit 3c6c47c05408abb3b3b4e8e597b29a69ce84c435
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 22 12:00:48 2014 -0400

    note a leak that needs fixing eventually

commit db3af2b86506b80484beac76bf21804301e10b98
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 22 11:59:18 2014 -0400

    make iovec creation just use mappings, so we don't leak mallocs.

commit 1712e46326f3b64935635373bf41f7efe3dbc26a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 22 11:58:49 2014 -0400

    don't allow 0 as a valid iovec count

commit 6cf81719516e404f9aabacd4e3552d46115adfa1
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 22 11:57:13 2014 -0400

    limit the number of iovec entries to 1-256 for now.

commit 0446b1f12138ff00b70d701dbea711f579559794
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 22 11:56:37 2014 -0400

    bump child retry limit
    
    3 is a bit too easy to hit in some cases.

commit 6da753fbf31aeae8490b3c6eafafdb4bb9d3fff8
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 21 22:29:49 2014 -0400

    move_pages: just use calloc

commit c975ad4e3b5bffa4b7376950dfb6aa801f96ca9e
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 21 22:21:55 2014 -0400

    print where the zmalloc failures are coming from.

commit 828617878ee37f8313f5c205855bf3e8c4abdaf0
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 21 18:00:38 2014 -0400

    clean up another syscalls -> operations usage

commit a60f9ada79eb07ed3642deeb2a504afd5beacd4f
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 21 16:05:14 2014 -0400

    move away from syscall counts towards operation counts
    
    mostly just shm renaming:
    previous_count -> previous_op_count
    child_syscall_count -> child_op_count

commit 8540ef8b90fef9a13ad99e597f3f396b107f3375
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 21 13:57:15 2014 -0400

    rename the sockets code to align with all the other fd generators

commit a708a00c0391cb9e1a9b7c6335c2754542b784b4
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 18 14:05:28 2014 -0400

    plan to cleanup the fd generation code

commit 5477d13a3c2574988623f91c85fc541a4b3837ab
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 17 14:44:31 2014 -0400

    rename

commit fc3e0e7fa9b35944a1231919cd9347d54b4ce80a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 17 14:43:35 2014 -0400

    rename

commit 172420f8b3f22d8e3336740294c6333211d282d1
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 17 14:43:16 2014 -0400

    move eventfd stuff out to own file

commit 02aaa54b1a7b45bab220b36766006eb71f6717d7
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 17 14:41:44 2014 -0400

    move epoll stuff out into own file

commit 04930cbb0e5e5e2bacfd1355cda803fd012d3698
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 17 14:40:09 2014 -0400

    move rand_pipe_fd into perf.c

commit eb48e3d1ac76457b52aa8286ce19d7ef5a66b6fd
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 17 14:38:46 2014 -0400

    move pipe related stuff out of fds.c into own file

commit 7922ab27eb524a21c1fa06b7b457d4bd4ce7e432
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 17 14:36:28 2014 -0400

    move file related stuff out of fds.c into files.c

commit e1c1dd65dd51adb0a9c17366af48c44a331bd8da
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 17 13:40:04 2014 -0400

    make generate_filelist private to files.c

commit f0b4b05814b27c61331610e92919da87c2da0010
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 17 13:20:07 2014 -0400

    create a set_rand_bitmask function and use instead of open-coding.

commit 67c18ff7c954f68ba9c135667d8a016377280b4a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 17 13:19:57 2014 -0400

    make syscall list values longs.

commit 87caed837a3b8f9ec76ff2982e802994113d1442
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 16 22:58:46 2014 -0400

    s390/s390x: update syscall list

commit d427f438841f593cfc480c356c9454e32a683476
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 16 22:52:48 2014 -0400

    i386 supports renameat2 now

commit ade8a78b7d20c677d8e2341a36404d897ef2e18e
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 16 10:55:56 2014 -0400

    use random open flags during fd initialization on startup.

commit 176621e936f0f8d494ce663bd552a540c15c89aa
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 15 17:23:23 2014 -0400

    reordering

commit 7f332ebab6f1b9f99e3fe3d8066d38ee97039b38
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 15 17:20:38 2014 -0400

    merge open_by_handle_at into open

commit c1a65fdf397d9b4d84bf8fe9c608da234bf3854d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 15 17:15:42 2014 -0400

    divide open flags into ARG_OP stuff and ARG_LIST stuff.
    
    O_RDONLY and O_WRONLY etc don't make any sense together.
    By splitting out the not-exclusive stuff we could later change
    things up so that check_stat_file() can OR in random flags
    during the initial fd opening.
    
    (and any other places we end up doing fd opening)

commit 0c67637c68f72a395dbdd46eb52d7f94e77a317a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 15 16:51:51 2014 -0400

    missed a listxattr variant

commit 5593976aa6961328b218b8abf3077c2dfe00bce3
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 15 16:50:36 2014 -0400

    merge listxattr variants

commit a252dcfd3a7ae8a31e5331bb4364e42b46c9ac10
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 15 16:50:03 2014 -0400

    merge recvmsg into recv

commit 0ba6aaadcc00bcda12df414cf5eb2b44bd9a1cc0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 15 16:48:29 2014 -0400

    merge send* syscall variants

commit 79ab0e51c7693f6093079858eea7dab104cc4997
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 15 16:45:36 2014 -0400

    merge recv* syscalls

commit 3718ede3f9d392e6daa2a8d00528bab2a828b89d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 15 14:39:38 2014 -0400

    locking: include a memory barrier before unlocking.

commit ef5a238606462ef7b0ec9165616bafa594b9795c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 15 10:55:01 2014 -0400

    lastop can be clobbered by the setjmp
    
    gcc 4.9 fix.

commit 0123adad5547743ac898b0cdbe19002167cceafb
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 8 15:04:12 2014 -0400

    clean up enable_random_syscalls_biarch()
    
    It's still pretty ugly, but a fair bit simpler than it was.
    The actual number of syscalls enabled with -r will differ.
    Think of -r as "enable at most this many" rather than "exactly this many"

commit da1e1065ab5599cf334ab80a4c01f152b4cd0abe
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 7 23:09:53 2014 -0400

    merge signalfd/signalfd4 syscall entries

commit 7a6b90616518a70255e50e5d6e3f0c07840e2538
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 7 22:54:26 2014 -0400

    merge accept/accept4 syscall entries

commit 182bddd605f982c8e5683ea74721d055f8370b82
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 7 13:26:27 2014 -0400

    add the new fd private lock flags to fcntl

commit 8746fce8699da946f46a2e5cc744cf0cc2ae703a
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 7 13:23:36 2014 -0400

    add missing fallocate flags

commit 970f5c38753fb7a71d4c1eeb92f0b225cdf2f8e1
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 7 11:14:38 2014 -0400

    add new renameat2 syscall

commit 2f68e5a3118f9ab0008a9a2c80ba7b886dd534d1
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 7 11:03:19 2014 -0400

    make-it-fail fixes
    
    - if it doesn't exist, don't try again ever.
    - don't retry if we get -EPERM
    - rename variable

commit a46cd318ac4a136832444a5c10509d2de32b2c1a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 4 20:37:14 2014 -0400

    reorganize to group similar function/tables.

commit 471c5639963df12c7a0edef34b7ca7c413e20067
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 4 20:36:21 2014 -0400

    add routine to fault in last page in a mapping

commit fbd146efe302de5a27b00aec4d35cbaf447cab31
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 4 20:34:42 2014 -0400

    add routine to read in a bunch of random pages from a mapping

commit a2ff88c7f8f6449ecf8ce37186fe3fea9bea0c84
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 4 20:32:08 2014 -0400

    add routine to fault in whole mapping in reverse

commit f193ca276d6c801ea33e51e6e35c2f619d71653a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 4 20:30:54 2014 -0400

    add routine to fault in every other page

commit 2c1b912197de685727c74d6621a85e511137bb91
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 4 20:29:30 2014 -0400

    add a 'read whole mapping' operation

commit 6fd19e89923f693c320c9d16395a14c21a915ae9
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 4 20:26:48 2014 -0400

    start some fault-in paths too.

commit 3e10fd4b729cbfd61f26466e97fc2130242ab7ad
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 4 20:14:13 2014 -0400

    change the page dirtying code to use function pointers.

commit 696332bc067a454919267f08ab5794a3cf076745
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 4 20:03:46 2014 -0400

    include cleanup

commit 9e86e2c5f5e02eec1f7f93624a725691c8bcef77
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 4 20:00:17 2014 -0400

    split out the page dirtying to its own file.

commit a5a08438d8a831c78bb678ece6ab26b35d9892ba
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 4 17:19:00 2014 -0400

    shorten perf fd setup using syscall record struct

commit 65cbcb99b8d6fa1a324e889691d7c470ccc6386a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 4 16:57:14 2014 -0400

    add to, and prioritize the TODO a little.
    
    I'm not going to get back to networking stuff for a while.

commit 318ebfdfec70e106859f539347c25cc9bcbbe17e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 3 17:41:23 2014 -0400

    syscall32 can be static

commit e6cbe3c724d4ff4097e32049f5eec064becdd894
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 3 17:38:21 2014 -0400

    reindent flags and renumber to group active/tobedeactivated flags

commit 6a59c20fcbaea414834f233eefc5fc8ccab7479c
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 3 17:37:29 2014 -0400

    remove unused syscall flag

commit 5aec4e8080524733566044f8dc9b97153166713c
Author: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Date:   Tue Apr 1 13:57:28 2014 +0300

    uid: initialize nobody_uid/nobody/gid only if --dropprivs requested
    
    I don't have nobody user in my dummy rootfs for virtual machine and
    trinity stops on getpwnam("nobody").
    
    Let's request uid/gid for nobody only if --dropprivs requested.
    
    Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>

commit cdd474d4f3be5d30bc0ef7cf0fff7f79bf64dd0e
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 2 14:50:20 2014 -0400

    change child op return type to indicate if we can make progress within this process.

commit 03148b18180333be45b89a9caa72eb88d7c15521
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 2 14:45:20 2014 -0400

    disable the recent execve special casing.
    
    This needs more thought, as it leaves a bunch of unparented processes
    and zombies.

commit 2b74898f2e3577f964e776cf239b5f97676ed0be
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 27 17:48:44 2014 -0700

    don't propagate up a syscall return, it's meaningless.
    
    instead all we care about is whether the iteration of the child op succeeded or not.

commit 022c38b5a1f2f13a516a90634889efe0dfdf11c9
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 26 14:39:48 2014 -0700

    more TODO updates.

commit 5d40c6d9af1a036907273f5793d2b8f79fdda69d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 26 14:06:49 2014 -0700

    do execve in a separate pid.
    
    Because on success, we won't be coming back, and trinity will wonder
    what happened to the pid.

commit 364ee8423b43298f790152b73e971cf732d8f1a6
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 26 14:06:42 2014 -0700

    var scope reduction

commit 641cec928ff7cdd5df564a5b4234ab01ca88c8e2
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 26 13:04:59 2014 -0700

    check_parent_pid can be static

commit 2656c9b63582b341d4e64bf1f389716ee8521387
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 26 13:02:57 2014 -0700

    remove the ->type field in the child ops

commit 72c006a60934c4c1264f12bb33885130a1a533ff
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 26 12:22:15 2014 -0700

    update comment describing child

commit e16fdf8dd2b3dec1bf2f930fe2f41c416d6436d8
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 26 12:20:29 2014 -0700

    we don't care about child_process's return value.

commit cb8a33f7ae1144ad079cb74b6302b80503cf7c14
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 26 12:12:26 2014 -0700

    repurpose the child_functions array to be child_operations instead.
    
    Instead of doing different things per child, we want to be able to have
    a child do things other than just random syscalls.
    
    This involves turning the child_random_syscalls() & child_process() functions
    inside out, but should allow us to do a lot more interesting things in
    children/ now.
    
    For now the child_read_all_files() child is disabled (as it was previously).
    
    I might reintroduce multiple child types later, but right now things in children/
    should be 'one shot' and not involve lengthy loops, because we're going to be
    moving the "we are making progress" indicators up into child_process too (See FIXME).

commit 4d9e1e32968c0ba54e8d2e4ea5f8eedc7edcf67a
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 26 12:01:03 2014 -0700

    Fix up a typo, and a no longer true statement in README

commit f4c9bffa9f15a45da4cbcb81c8029b9512a593ac
Author: Vinson Lee <vlee@twitter.com>
Date:   Mon Mar 24 14:46:56 2014 -0700

    Add SWAP_FLAG_DISCARD symbol for older distros.
    
    This patch fixes this build error on CentOS 6.
    
      CC    syscalls/swap.o
    syscalls/swap.c:16: error: ‘SWAP_FLAG_DISCARD’ undeclared here (not in a function)
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 9f259162c1a24b31d8cc88c92271ac028e901d10
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 20:20:35 2014 -0400

    remove periodic reseeding entirely.
    
    Given we never run for that long before a pid dies, and we reseed anyway,
    chances of us needing this code are slim.

commit 340f8882241458cbd85ed67c569bd7e6aaf391a7
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 20:06:42 2014 -0400

    fork: something to experiment with later.
    
    For now, leave marked as AVOID until I've played with it a little.

commit e44f2bba2931d1901d655435d3c1c9d9c57948c1
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 19:36:56 2014 -0400

    store the previous syscalls using struct syscallrecord too
    
    bonus: we now store do32bit correctly

commit a82fde3ddbcb4b92421b8083150269dc6c124492
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 19:33:23 2014 -0400

    shm syscall entries reorg
    
    create a syscallrecord structure, and move all related fields into it.

commit 9b97024d0decd042e52ce42378786c4dd4725ca0
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 19:04:59 2014 -0400

    watchdog: assign potential fd to local var

commit b1216a3e4fcf5957b3cc501e31452aba3932e52f
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 18:55:41 2014 -0400

    watchdog: remove redundant recalculation of callno

commit e6bcdc2b5724bcf73e97bec2b00a36ad6e3089f9
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 18:50:28 2014 -0400

    remove broken sanity check from syscall path
    
    It was missing the [childno], but is pretty pointless these days anyway.

commit 72095db3e6efb34ebc4502d5c94c1bb5c70a06c5
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 18:09:28 2014 -0400

    remove scaling TODO.
    
    we dynmically allocate these arrays now

commit d4bb3d0583e1c0770c0791a7f387c093810b811b
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 18:08:48 2014 -0400

    remove [64] cleanup TODO

commit cc8fd3cf85966678dc03c61de8a7a20c3cd4f6fc
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 18:04:09 2014 -0400

    swapon: add swapflags definitions

commit 46bed392f07fe032dd993c52324a595cacfa2c18
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 18:02:05 2014 -0400

    move swap on/off variants into same file

commit 73e22375677a78573071ae2c5446f3805342c70d
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:57:36 2014 -0400

    merge pipe variants

commit 0964b1dce356f99807735584a659b987e019bf14
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:56:19 2014 -0400

    merge clone variants

commit 2d43659e5424c174e2e3fff89750eee5a3163e05
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:47:53 2014 -0400

    merge unlink variants

commit 5199f177bf5b06a8517cc525a2ff9a806c0f1127
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:47:20 2014 -0400

    merge symlink variants

commit ecbdee368d0e3040d3fabf6133de8a5e9a2c6aa4
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:46:52 2014 -0400

    merge rename variants

commit 7334c73eb304f9ac69e7c0b4164593f1dd713fe4
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:46:34 2014 -0400

    merge readlink variants

commit b4b6fc746c38dcae7e8d095fa9d3be2b7325f887
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:46:01 2014 -0400

    merge open variants

commit a5b396cf082dc7b2356d0d048d97702874e32c62
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:45:10 2014 -0400

    merge newfstat variants

commit a19fdfc65cd1e0fca9524daa50dbdb2a0ac8e5c8
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:44:09 2014 -0400

    merge mknod variants

commit 7f7b1e62d33eacd08611c5479486260232fb1f4e
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:43:29 2014 -0400

    merge mkdir variants

commit 856dc3012491573a1a56d66053081d616a5c4547
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:40:08 2014 -0400

    wire up setuid16 to sanitise routine.

commit 7a165985b36c79c1e6f36e11dca274deeeaa617d
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:39:56 2014 -0400

    merge setuid variants

commit 10ff5c9ad9954bf204e4cc80df8bd094a1fa2de5
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:39:17 2014 -0400

    merge setreuid variants

commit f640266a08c07dcb51801590fbb8c45a2a5db02b
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:38:54 2014 -0400

    merge setresuid variants

commit 65e7af7376933ee9280abb0e3cf28fb638e1fd61
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:38:29 2014 -0400

    merge setresgid variants

commit 86b31a498e5c562726b6d0a9f5b03fd494299752
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:38:02 2014 -0400

    merge setregid variants

commit 94fbad3d73487fe7f79e5a3c07a30d41ce41aeb5
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:37:35 2014 -0400

    merge setgroups variants

commit 8a47c0c3a2b16c4b30850673e2afcbc87b98e59a
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:37:07 2014 -0400

    merge setgid variants

commit c9660f6f927cd46983fc84df7b3dea6c2c244e0f
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:36:35 2014 -0400

    merge setfsuid variants

commit 29da19bcbe4af25541be7f8d4ed9525aec98b5ed
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:35:40 2014 -0400

    merge setfsgit variants

commit 8d9d0b4b70bb78f8ad98c6e1a9f17f4a2a49ee7d
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:35:04 2014 -0400

    merge lchown variants

commit 820f56ae86d20bd62ee5b7447b948b2e95b9f33d
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:34:32 2014 -0400

    merge getuid variants

commit b6257e2ac325ebe9edb06b252b800a58101786e0
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:34:07 2014 -0400

    merge getresuid variants

commit dc27a5fbb38f1c8f888b9990abde061b360720db
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:33:44 2014 -0400

    merge getresgid variants

commit 3a873e2a0bedc5f3d68fb960c20b64ccad4c9147
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:33:13 2014 -0400

    merge getgroups variants

commit a0815fdb94b883b2fea73f16e4c1e1d5889030fd
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:32:49 2014 -0400

    merge getgid variants

commit f998f8ad3501e4499c490773e158cb07ecbbb1da
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:32:21 2014 -0400

    merge geteuid variants

commit 66d242f569ca99bd1d274671fcbe1cac41922681
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:31:37 2014 -0400

    merge getegid variants

commit 62b5a5d93cf9d80b2eddd4e48e8ffc0810c4d35b
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:30:59 2014 -0400

    merge chown variants

commit a07e6d09548e8532b5217043f14030f9eb94710a
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:30:04 2014 -0400

    merge fchown variants

commit 9e409b237ad17799bb6218b49c298ed08a0af6df
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:29:18 2014 -0400

    merge fchmod variants

commit bed4e53382e49292992f56f56a5bed23f3fa3824
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:27:34 2014 -0400

    merge truncate variants

commit b8da7e6396714054b5fd0b587edd9a392141f59a
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:27:09 2014 -0400

    merge statfs variants

commit 41a2d9c858433b9c5a484f09f024f625b26d126e
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:26:40 2014 -0400

    merge stat variants

commit 1c1751e6d1f946de60f9f508198c0cfa572c9966
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:25:43 2014 -0400

    merge sendfile variants

commit cb47c094bdd7d683965b88d38401fdfd09659049
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:24:19 2014 -0400

    merge lstat variants

commit 8ec59d2cae4f9cef79ad2351b8ddc3839da6e77f
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:23:39 2014 -0400

    merge getdents variants

commit 135c59d652b915dde0439c8177c398c03d1e82c5
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:22:50 2014 -0400

    merge ftruncate variants

commit 13bfefc3828c76a56a7a031a91d12d945a45236e
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:21:37 2014 -0400

    fstat64: add missing NEED_ALARM

commit 1cb7af69ca1e94f24ecfa57eddc6ae2ca24c236f
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:21:12 2014 -0400

    merge fstatfs variants

commit bc4f38244832784bdbcf96b14fa963955b6cce73
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:20:23 2014 -0400

    merge fstat variants

commit db303e3ba06669560f75be0e75b16e06e7f67fac
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:06:32 2014 -0400

    add missing NEED_ALARMS to fadvise variants.

commit 5c2fd4b08260f347107584ac8dbecf3deca8326d
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:04:36 2014 -0400

    make ARM specific syscall dependant on __arm__

commit 482f66c2d1e40dd76ed2535b0dda2613132cd364
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:02:17 2014 -0400

    move all fadvise variants into same file

commit f55bd249ace90213973968ebfeeef6f776aef459
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 17:01:41 2014 -0400

    mmap: for non-anon maps, set size to one page for now

commit 3ff71f33e0f40a06aea3b32a2700d4f82f0d293b
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 23 16:51:38 2014 -0400

    newfstatat: fix typo in arg name

commit 93e9998249fdf5390f60d61743ce46f83215506c
Author: Dave Jones <davej@redhat.com>
Date:   Sat Mar 22 18:05:11 2014 -0400

    move kernel_taint_initial from params to taint

commit d56242b9814f7faeefbe4aa799b55a566787e4cf
Author: Dave Jones <davej@redhat.com>
Date:   Sat Mar 22 18:01:42 2014 -0400

    move the taint stuff out to its own file

commit d4589e41f2f789d368c60ec4fa99538e6e27f304
Author: Dave Jones <davej@redhat.com>
Date:   Sat Mar 22 17:54:20 2014 -0400

    When we detect a taint, record the time it happened.
    
    This will be useful for comparing child timestamps when debugging.

commit 22180547dac2e2c4d7cbb7fe28f0dbb60dd7202d
Author: Dave Jones <davej@redhat.com>
Date:   Sat Mar 22 17:33:08 2014 -0400

    remove dead prototype

commit 59ba0013feb009539b0565391723cd9d9e9c5684
Author: Dave Jones <davej@redhat.com>
Date:   Sat Mar 22 17:26:15 2014 -0400

    handle mmap/mmap2 difference accordingly.

commit a650957a9d2d91564e150c8cf3a1c6613dc51cbe
Author: Dave Jones <davej@redhat.com>
Date:   Sat Mar 22 17:21:19 2014 -0400

    factor out syscall name checking in sync_file_range
    
    we'll use it in other syscalls too.

commit 531f3fa7798b323bdf1235beec7d08cb5c8eda68
Author: Dave Jones <davej@redhat.com>
Date:   Sat Mar 22 17:09:40 2014 -0400

    sanitise_mmap can be static now

commit 76b44097355844146f1ad486c748da0c2d970ad4
Author: Dave Jones <davej@redhat.com>
Date:   Sat Mar 22 17:08:08 2014 -0400

    unify mmap syscalls into a single file.

commit da7b4745752de213c5e843fd435793577293a47f
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 20 16:47:14 2014 -0400

    make send() use the same flags as sendto()
    
    The kernel treats them both as the same function.

commit 5d0340482091e3b54f3de1b51d367748cbbfc7ac
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 20 16:47:04 2014 -0400

    remove unneeded include

commit 666b512b019ea13e5d31fb4d58c9ec078b98cfeb
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 19 10:55:59 2014 -0400

    implement lock stealing.
    
    This is kind of gross, but we have a rare situation where we end up
    with a process holding a lock and then crashing.  If that happens,
    the other processes all end up stuck waiting for a lock that never becomes free.
    
    This isn't perfect. If we have multiple waiters, we can end up stealing
    from a live process. I'll revisit this later, for now I just want to
    fix the obvious deadlock.

commit 185c96e59664bf1b756194a2c33419d7490bd36e
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 17 22:10:19 2014 -0400

    disable the setsockopt from regeneration.
    
    This is causing too many deadlocks.

commit 43bdf2b44626933bc08ef42c83632519ff58a427
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 14 00:38:44 2014 -0400

    reduce scope

commit 7bda94e997c69a21ebbf7cbdefbc7a4a6228ee25
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 14 00:37:40 2014 -0400

    print out child no on exit.

commit 25387378d66457c9322ae81b5c651c7462299460
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 14 00:36:21 2014 -0400

    simplify forking code by inverting test

commit fb747fda4b2a791dfea4c47bb56878cf0bcba5fc
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 14 00:34:02 2014 -0400

    move more child init into child.c

commit 43d1bf6f3875084283cba23102e9d281084f907f
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 13 22:35:06 2014 -0400

    only allow one -V for now

commit ce91aa43572f0e728c34e331348c56881e868113
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 13 22:32:08 2014 -0400

    clear uninitialized variables.

commit 445dac47841d39a49f64fa78341d921e3b9b2c0b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 13 20:25:38 2014 -0400

    Add an owner pid to a lock so we can figure out who owns it.

commit 312aac2811d05dc7330ac2c19a67e75388d0e22c
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 13 17:04:15 2014 -0400

    use lock_t for lock definitions

commit 24941b7501afa7687c75f0a2fa59ae5ce64e5962
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 11 14:40:38 2014 -0400

    add a script to do a random subset of vm syscalls

commit 69fcae317fe0924001094fbe0a8ab3e5c29260cc
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 11 13:54:43 2014 -0400

    update path setup in scripts/test-all-syscalls-sequentially.sh

commit b20ef2eb1b835b41d56f0892ab3edf2ad033262e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 11 13:54:03 2014 -0400

    make test-vm cycle through mmap + 1 other VM syscall.

commit 4e4d7e8c56b5d28e91c2f9a3a71f278454bc71ec
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 11 01:29:11 2014 -0400

    move_pages commentary, cleanup

commit 67dd1ca4f0bacadff90750d44769bb84e1b86d1e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 11 01:23:13 2014 -0400

    more commentary for the mappings code

commit a8e397666b3b9c8fe05817d7e2570ca5adc94f8f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 11 01:13:00 2014 -0400

    enhance commentary in get_map and clean up logic.

commit 97111ef39bc5c3be0abc93f3318cf13aae9f1fcc
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 10 20:03:09 2014 -0400

    Make -Werror happen with > 4.8

commit 026f4ee5a26ff2fbea2bfa83a1c523bb8d5b2807
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 10 19:18:59 2014 -0400

    make -Werror only apply on gcc 4.8 and above.
    
    I'm tired of fighting older compilers.
    I'll take patches fixing warnings as long as they aren't crazy,
    but I'm not going to waste time chasing down older issues more than necessary.

commit 0c8550724bdb31c5f46e4aa3289675081b86f2aa
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 10 19:16:38 2014 -0400

    automate the -Werror setting / devel mode check.

commit 5967a4a41870a98781f3cd3cace463274720cfa7
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 7 17:41:53 2014 -0500

    debugging for the watchdog segv.
    
    This is still there, and I still don't understand it.

commit 487d652cc244f81384bdbf4b5f200140823a327c
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 7 13:14:39 2014 -0500

    write: don't use page_rand directly
    
    using get_non_null_address gives more variation.

commit abcc9262c0121e8af048122176462976293ccb0a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 7 13:14:28 2014 -0500

    setsockopt: don't use page_rand directly
    
    using get_non_null_address gives more variation.

commit adab24e07666f548e666d6094d74bd947193435e
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 7 13:14:13 2014 -0500

    TODO: come back to sendmsg later.

commit 7398ccdf078404c298703fac0cc470c40850aa86
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 7 13:13:34 2014 -0500

    read: don't use page_rand directly
    
    using get_non_null_address gives more variation.

commit 62ba4d6b4e382e720eb77d70c7c7a5788a6a3cbf
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 7 13:13:14 2014 -0500

    perf_event_open: don't use page_rand directly
    
    using get_non_null_address gives more variation.

commit 9db3e0d52b2b6c5d55e3daa90f8a3bbcf0c71c5f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 7 13:12:24 2014 -0500

    ioctl: don't use page_rand directly
    
    using get_non_null_address gives more variation.

commit 29fda84940c0a3b41e3bfeb41dd367b92c09365d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 7 13:11:31 2014 -0500

    special case calls to get_address from the main pid.
    
    This is prep work for moving syscalls/* away from directly using
    page_rand and friends.

commit 16713c4e6434df78f979c7c4e7db5a5d1007ef0f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 7 00:33:17 2014 -0500

    move check_uid to uid.c

commit 3389ba4de410216b672c9951f73c88a10c5d4f29
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 7 00:30:51 2014 -0500

    hint about --dropprivs

commit 0c433dbdc0f6e64c9a6a01cde473bdb6f4d98abf
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 6 21:01:52 2014 -0500

    reorder, and comment some things.

commit 928da3eea9dff5bd06b1069266015039cca67f36
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 6 20:58:35 2014 -0500

    make choose_syscall_table do one thing only.

commit 7c2263f3d5e8da5a729e5231ecb58c2acd5c16be
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 6 20:57:22 2014 -0500

    remove unnecessary nr_active_syscalls

commit 86562ff17a1f2233db9fae6f06963efe59388dab
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 6 20:55:43 2014 -0500

    when picking a syscall number, use the correct range.
    
    This was likely what was causing the watchdog to segv, not a race after all.

commit b9706ae3232a9a439f1bfe96c59056d77e17c7c9
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 6 21:03:23 2014 -0500

    commentary on the active_* vars in the shm

commit f83c4238f7d8789c2740fe35eb1dbc5783c18dd3
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 6 13:18:28 2014 -0500

    fix compile on gcc4.7
    
    -Wshadow is broken in older gcc, so needs unique func/var names.
    
    reported by Vinson Lee <vlee@twitter.com>

commit cfd2f89cec8f4b6bc9b9eab142c62d5bc9bc48b3
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 5 22:20:10 2014 -0500

    use TRUE/FALSE in params.c

commit c698fa766d401554e5ed89e1f9ee3b493deca9ef
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 5 21:45:59 2014 -0500

    rename acquire/release to lock/unlock
    
    I never liked those names from the start.

commit f73f0fa8794cd16ee030496dc3a7cd8f923be92d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 5 21:43:49 2014 -0500

    add locking around shm->do32bit / syscallno accesses

commit 21b85b036c00088be9d4e11891fcb81e957b9fca
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 5 11:16:00 2014 -0500

    taint mask is unsigned

commit 4b948f9706d4047fc22109203db50dd1fca4ac81
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 5 11:09:51 2014 -0500

    remove unreachable code

commit 39d661957838491bf580412ba3d93110ec7df0cb
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 5 11:08:47 2014 -0500

    use usleep for <1s sleeps

commit 6e7d90846e1cdb1ab46bfd440e7b35eeb1e38d88
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 4 22:42:42 2014 -0500

    limit shm->do32bit accesses in child.
    
    Do a single write at the end.
    This really needs some locking so it doesn't race with the watchdog
    reading this and shm->syscallno. (It segv's when it tries to read
    past the end of the wrong syscall table if the race occurs).

commit 9c0ddb63d778cca18bc39c05005c3b934194ae7a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 4 10:41:51 2014 -0500

    typo

commit 3e7691cf0afb34a03bf43ce124fef086b19cb9df
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 4 00:46:25 2014 -0500

    don't bother printing errors about re-enabling coredump failures.
    
    It's usually EPERM, and there's nothing we can do about it in that case.
    (and we're about to exit, so hopefully we'll exit without dumping anyway)

commit e6d20e8dfd3e68d854933075e425f86567bde0b0
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 3 23:28:05 2014 -0500

    coredump enable/disable cleanups

commit 160a713457318c65946d9c686e4a10ecff0bf6cc
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 3 22:09:20 2014 -0500

    make disabling the coredumps happen after we've finished setting up.

commit 5bb7809c2f0819a5b33a891936170671a005087c
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 3 22:08:44 2014 -0500

    setup shm->mappings[child] list before we use it in setup_page_maps

commit 93da5b7c3f73de5043d5fe4f0fa9078bbaf93de6
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 3 21:29:13 2014 -0500

    fix inverted logic

commit f10247a10cba431d9e042bf44b2ee3786a5ddc6c
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 3 21:28:24 2014 -0500

    move locking out to separate file

commit d575a18e49dd7c381758a2fc148c3c7cd1fc2afe
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 3 20:15:33 2014 -0500

    use page offsets instead of dirtying the same page.

commit 11fe8fc8a1da885323840cbe899d47f442f3c936
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 3 20:14:39 2014 -0500

    remove erroneous ; in maps dirtying

commit c82b64a53207309e050869a7635f31a1d85e8e69
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 26 16:49:37 2014 -0500

    page_0xff wasn't getting memset'd.

commit e8791eb9e774a3deba92a3ae2312e0cee77d13b0
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 24 22:34:36 2014 -0500

    make a bunch of things static

commit 482fa13b96ba467f1476cdf7086e0c083d929ee8
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 24 22:34:22 2014 -0500

    remove unnecessary include

commit 82d5f7072ffb4400924cd0b0bc93b612db82599f
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 24 22:31:25 2014 -0500

    fix sparse warnings in sync_file_range about long constants.

commit fb2b1525213f4c7a0f72dd89c7853d587b4d884c
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 24 22:29:52 2014 -0500

    make a bunch of sparse missing prototype warnings go away
    
    really every syscalls/*.c should be including this, but as they all include
    sanitise.h, we can just put it here once.

commit d287f410a5366ac863a396657d87ae789a341e5d
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 24 22:29:22 2014 -0500

    use NULL as pointer in shmget, not 0

commit e5b0848619c733eca05f8dffe53b668ae99e77a1
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 24 22:18:04 2014 -0500

    use NULL instead of 0 for gettimeofday 2nd param

commit 7f456f82bf3d03165d570c9da99008a9c6e9adfc
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 24 22:17:30 2014 -0500

    add -D__linux__ just for sparse use.

commit 991b2cc253be7a22c913ac6b3e330c9466be62db
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 24 22:17:10 2014 -0500

    move sigwas extern to header

commit 787c70b248c4191acb107bc06652bc142da720a8
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 16:46:36 2014 -0500

    update paths in analyse-dumps.sh
    
    also mask out childno to make hashing work better.

commit 22201624ac3e877e4a14605a330258a5d20405fc
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 16:44:19 2014 -0500

    remove more no longer needed cruft

commit 8739410ceec01b839d1d40b59cc7d6ae1f549a07
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 16:44:01 2014 -0500

    remove now pointless check

commit 0183571cb4d3dcb9e84e97c3152af763fc301bf8
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 16:43:03 2014 -0500

    now that -C can scale better, we don't need the extra shell loops

commit e6e10e81434ccceb4d5f35e26d0213315c54961a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 16:40:50 2014 -0500

    if we have a tmp dir in the trinity dir, use that instead of /tmp

commit 4805a83d5ec5e74a11085d3924706850bba00e41
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 16:39:11 2014 -0500

    add --dropprivs if we were run as root

commit 69b4cbb5ee710969c15805eab0ed8e064eed685b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 13:59:31 2014 -0500

    move_pages: clear old pointers after freeing

commit 49be05191949927d21e2858bab2ce6d10f4b52aa
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 13:34:35 2014 -0500

    move_pages: if pagetypes alloc failed, don't deref it in ->post

commit 93f4fad3de65e2161d0c1a373f578671fb5bc059
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 13:26:10 2014 -0500

    dirty_mapping: fix write past end of mapping bug.

commit 0ba5c543cfb702d0d3a711622456c3bfbd8160f3
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 13:25:42 2014 -0500

    mprotect: clear scratch after use

commit 610215169fed9f4eeb49316403b30ee0b3135877
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 13:25:07 2014 -0500

    Revert "mprotect: clear scratch after use"
    
    This reverts commit 1726cb0ea3ca5fef8dd0990e512cf21c578c18d7.
    
    accidental commit of a bunch of local-only changes

commit e584af944fa63a760214ab91bdbad0f28f8484b1
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 12:45:09 2014 -0500

    sendmsg: clear scratch after use

commit 3347bb5014f05b0211462e41382f2ac2eee48a05
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 12:44:38 2014 -0500

    munmap: clear scratch after use

commit 46a4485c55ddcf8af38638bced41f4a18bc7bbad
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 12:44:11 2014 -0500

    mremap: clear scratch after use

commit ef088af51350b2ddeb1eab094825bd62f2cd64a8
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 12:43:28 2014 -0500

    mprotect: clear scratch after use

commit 93184898f7035a4435322682e90a0deb3b78b4f1
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 12:42:53 2014 -0500

    modify_ldt: clear scratch after use

commit d0c9987c439bd852aab98825d5a6d8708361f939
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 12:42:23 2014 -0500

    mincore: remove unnecessary use of scratch

commit e3266589ea69f673c3e398f9d65127f3c86938d4
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 12:41:13 2014 -0500

    move_pages: clear out scratch register after use

commit 0deeca66c80f6219f33d10780855292263b4b85b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 12:39:42 2014 -0500

    handle malloc failure better in move_pages

commit 8650870ce4b566b4a62f98442a098257ff4789a6
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 21 12:23:34 2014 -0500

    make get_address aware of map lengths

commit fa358dc291c9d88a2fa616a50e97f98ef877779b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 20 22:30:08 2014 -0500

    watchdog: scoping reductions

commit 6683e249dfaba367b2a35ab74117e3124b4b5392
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 20 22:15:26 2014 -0500

    start of sockaddr generation for sendmsg

commit 2e12dc3a75e739e886f49ab826a00da4bd2d8b37
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 19 21:06:45 2014 -0500

    make all the shm array allocations shared.
    
    I should have just done this from the start.

commit bc5d2914acb7b0a3386444ec2a8b97b147ae79ae
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 19 20:59:46 2014 -0500

    make alloc_shared exit if we fail an allocation.

commit a69f00dfd1d86bd1ea05e2763851a68abb37169c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 19 20:32:39 2014 -0500

    shm timevals need to be shared.
    
    Otherwise watchdog just sees 0 all the time.

commit e1b3697749348a7b9dca26a4a0d506df2f6d62a6
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 19 20:31:38 2014 -0500

    watchdog: print child numbers, and timevals as unsigned

commit fdc27b1b284a1b56f515370eccfc66e694003c31
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 19 16:05:23 2014 -0500

    use better function for looping over pids

commit 3302d46a402ee16a67bc91a59f5c8f0edf0f0c2d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 19 16:01:02 2014 -0500

    split up shm create again.
    
    Now I remember why we had that _postargs function.
    The params parsing toggles shm->syscall* so it needs the shm set up
    before we do the param parsing.
    And we need to have done the param parsing before we do the array
    allocation in case we've passed -C.
    
    The answer is to do this in two phases.

commit c9563f4337800650c88852d2319aa2d9fd30879f
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Tue Feb 18 12:44:07 2014 -0500

    update perf_event_open support to 3.14-rc3
    
    Update perf_event_open support to 3.14-rc3.
    The only real change is the addition of the
    PERF_FLAG_FD_CLOEXEC flag.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 934e427a94a72ea08801052300b9141e2882c465
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 19 13:08:10 2014 -0500

    remove FIXME. tried it, looked ugly.

commit ff9ffd53bc032ef1465cd074a953cdf4d8cf6264
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 19 13:00:29 2014 -0500

    function reordering

commit d9646baa54ed0e75a936f4b856b19bb40a22a48e
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 19 12:59:44 2014 -0500

    move alloc_shared to utils

commit dfc36a1bcbc7b22050609075013362e78d24ba83
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 19 12:14:03 2014 -0500

    split up create/init shm so we can initiate logging in between

commit 2038f5949237e0fe188f355b9bff820f5a82459f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 18 22:33:29 2014 -0500

    move max_children to child.h

commit 27ff91ff902c8c2cd18a6ff50ffb0f3e8bf983bf
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 18 22:31:55 2014 -0500

    kill off MAX_NR_CHILDREN
    
    We are now totally dynamic, and -C has no limit other than memory.

commit efcff4d7913f66055b426eb56802f11f04cf2206
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 18 22:29:01 2014 -0500

    create_shm either succeeds, or exit()'s

commit 8efd43f1583c414932278713f60fa11c36528f31
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 18 22:27:25 2014 -0500

    parse args before setting up the shm.
    
    this allows us to kill off the setup_shm_postargs function

commit fbe4c001a7e32656224e10ac57f20404bb4ddecb
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 18 22:25:13 2014 -0500

    remove a MAX_NR_CHILDREN loop

commit e755fb9fbc2078444a01be5d67933d230f845cd1
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 18 22:24:44 2014 -0500

    do runtime allocation of arrays in the shm

commit b26ebd9ffdf732ba22203aa40506bca7a2ab0bf3
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 18 21:05:36 2014 -0500

    shm: split up create/init

commit 0a5b653c0f2233a9bdfaebe0a7b102ccba50db33
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 14 12:57:10 2014 -0500

    bump up the max size of iovecs that we create.

commit 4977ef046954805312993ca3e73d9b5833a370d6
Author: Daniel Borkmann <dborkman@redhat.com>
Date:   Fri Feb 14 17:25:11 2014 +0100

    bpf bug.
    
    Patch attached. If you have any questions feel free to ask.
    
    Best,
    
    Daniel
    
    >From ac47891915fc128350f5e904bb78fd5eb745bad8 Mon Sep 17 00:00:00 2001
    Message-Id: <ac47891915fc128350f5e904bb78fd5eb745bad8.1392394906.git.dborkman@redhat.com>
    In-Reply-To: <cover.1392394906.git.dborkman@redhat.com>
    References: <cover.1392394906.git.dborkman@redhat.com>
    From: Daniel Borkmann <dborkman@redhat.com>
    Date: Fri, 14 Feb 2014 12:35:35 +0100
    Subject: [PATCH trinity] bpf: improvements in code generation and fixes
    
    Dave reported that commit 3a83e2c2f893d6 seemed to have broken BPF
    generation. This patch fixes callsites and improves BPF code
    generation in general. Also, for debugging purposes, it adds
    a BPF code disassembler. At some point in time, code generation
    needs to be further improved to be more intelligent, for now,
    we try to be less random and slightly adapt parameters as well.
    
    Reported-by: Dave Jones <davej@redhat.com>
    Signed-off-by: Daniel Borkmann <dborkman@redhat.com>

commit f15bdfdb8ffe2514fc85ea76ddf01a24575ff869
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 13 17:44:12 2014 -0500

    annotate option arg for sys_sysfs

commit 47de52e96e1c6b547a53b218300409995861c547
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 11 23:27:12 2014 -0500

    Don't set arg1 type for mmap.
    
    We explicitly set it to 0 right now in the sanitise routine, so passing in
    a mapping address is pointless.

commit 8c77e060919605b71c2d624f72ec5c67d4a7f56c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 11 23:16:28 2014 -0500

    fix operation when starting already tainted.

commit eadf424de08b73ac0c7f0a5d39911b5a5c182295
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 11 16:23:06 2014 -0500

    Make mincore operate only on map structs
    
    And also alloc/free a struct for the vec argument instead of randomly
    scribbling over something.

commit 2a83f99ea65140c61983e4c8350f65c52d66d0b2
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 11 16:22:19 2014 -0500

    don't print ARG_MMAP as a struct map
    
    By the time we get here, we're not a map struct, but a map->ptr.

commit 445507f54735069f5a5db612868315412a4f546d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 11 11:17:39 2014 -0500

    don't deref potential NULL mappings

commit 65aa7c52cfdc8bf985e023c626621b484b55ffb5
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 11 11:15:54 2014 -0500

    __get_map returns a pointer, so return NULL, not 0

commit 70276c96b4bd86777201b7d731cd497e4260e6ff
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 11 11:04:04 2014 -0500

    watchdog: check_main_alive cleanups

commit e377020c7f75c1f584ff61259a3d2ac33cd4cc64
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 11 11:02:21 2014 -0500

    watchdog: fix incorrect format string

commit 2bc0fc10ebe1f52dedadbb14c1835e22bda8af51
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 10 18:46:59 2014 -0500

    move assignment to only case that uses var.

commit 585707feaf333e4dad4caa3ced66a034cffda305
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 10 18:43:08 2014 -0500

    Check for fork() failure and just exit if it happens.

commit b1b06568dff71f22f43e76893aa25bae9dbadc69
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 10 18:41:53 2014 -0500

    if all children exit, and the pidmap is empty, be sure that nr_running childs is 0

commit 98b74f8a54df5a86c2f8428a249fbe268791bf6a
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 10 18:38:44 2014 -0500

    watchdog: Add additional check for main pid alive or not while spinning

commit c2a1d3c59357d9a60bbefbf7bc27160c287ab1cc
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 10 18:37:42 2014 -0500

    remove unnecessary futzing with shm->mainpid
    
    We can just use the return from fork here.

commit e537c49dd679b3442bde5199e4b235a7d88b89fb
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 10 18:36:48 2014 -0500

    reduce var scoping

commit d17b6b558a4d85339ac0e520ba9ffa9b1aee357f
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 10 18:36:25 2014 -0500

    size is unsigned

commit 8a0fa8cc36819e63deb6908afbd0ffde5d56a576
Author: Dan Horák <dan@danny.cz>
Date:   Sat Feb 1 11:12:04 2014 +0100

    load correct syscall table for s390 and s390x
    
    The actual syscall table for s390/s390x must be loaded depending on
    the __s390x__ symbol.
    
    building on s390 (32=bit) ends with
    ...
    In file included from /builddir/build/BUILD/trinity-1.3/include/arch.h:27:0,
                     from tables.c:10:
    tables.c: In function 'select_syscall_tables':
    /builddir/build/BUILD/trinity-1.3/include/arch-s390.h:24:18: error: 'syscalls_s390x' undeclared (first use in this function)
     #define SYSCALLS syscalls_s390x
                      ^
    tables.c:347:32: note: in expansion of macro 'SYSCALLS'
      syscalls = copy_syscall_table(SYSCALLS, ARRAY_SIZE(SYSCALLS));
                                    ^
    /builddir/build/BUILD/trinity-1.3/include/arch-s390.h:24:18: note: each undeclared identifier is reported only once for each function it appears in
     #define SYSCALLS syscalls_s390x
                      ^
    tables.c:347:32: note: in expansion of macro 'SYSCALLS'
      syscalls = copy_syscall_table(SYSCALLS, ARRAY_SIZE(SYSCALLS));
                                    ^
    make: *** [tables.o] Error 1

commit 2f1b2ba73cae2628aa5bc734013f6ac04957de7a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 31 01:34:05 2014 -0500

    use enum to keep list of strings in sync

commit 4f8c97e01f52d5ae03285a5fbc229e533cd87bea
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 31 01:25:42 2014 -0500

    pretty printing for allocation sizes.

commit 2f048300ccac8a53b361e61af49e97172103a4c1
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 30 01:36:15 2014 -0500

    add missing break.

commit 8efd34eb11340761ed7f5b96711f42c506498e4c
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 30 00:23:01 2014 -0500

    Make format logging use enum.
    
    Also add missing ARG_MMAP entry which -Wswitch-enum picked up on
    now that -Wdefault-switch isn't hiding it.

commit 82303aec9b440f0acf0ef6d2b24dcb67be437133
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 29 23:57:32 2014 -0500

    remove -Wswitch-default
    
    This didn't really find any bugs, and just uglies up the code a lot.
    (in at least one case, it caused a bug to be hidden)

commit c5102da9c5031492587e74a1e98804cf22d07423
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 29 23:50:03 2014 -0500

    fix off by one

commit 5e6168305aa209e099ae97ef3f4de5b1f65c8ae2
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 28 22:23:12 2014 -0500

    use mappings in iovecs

commit cefba51013f744e93960eac69c8dac932431cd65
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 28 19:12:05 2014 -0500

    some tweaks to the 32bit interesting numbers routine

commit f195fb56bf001a239ce87379c1bc09d4266b40f8
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 28 15:56:39 2014 -0500

    move some parameter parsing into params.c

commit e9445c9d0685d73bd03222a943b763e714696a3c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 28 15:49:15 2014 -0500

    move uid0 checking into uid.c

commit ec928c14818e150efff318ec7c175ab3cda942e5
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 28 15:45:00 2014 -0500

    remove dumb waiting for pid code
    
    we know it from the fork above, duh

commit 6cfba1989dfc3d089736c795f39731db8863e87a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 28 15:37:03 2014 -0500

    move the getopt string nearer to the longopts to keep in sync easier.

commit a071a5eb6e041d90b298360c3d40f56391f26cd6
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 28 15:34:35 2014 -0500

    initial pass at writing support for running as root and dropping privs.
    
    This is very work in progress, and may do entirely the wrong things right now,
    use at own risk.
    
    Side effect of this work:
    - Inability to write socket cache is no longer fatal.

commit af4623c77d0c84a51aa3288d38e501b0d295e235
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jan 26 20:56:15 2014 -0500

    add support for SO_BPF_EXTENSIONS

commit 48b5678ac0c935935d1a6070b98e4b1d8a4813fc
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 24 15:16:49 2014 -0500

    use common exit path.

commit c455fdcfa385b459433f7a3b5c9056c718c3f674
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 24 15:16:35 2014 -0500

    fix memory leak in error path

commit f702718b0110ffea12f96674bc5f4d86f8d6e650
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 24 14:53:32 2014 -0500

    search_syscall_table needs to have its return checked.

commit 58a7fa000ee6271d8953c148c1d70816fb70a69a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 24 14:34:48 2014 -0500

    remove one of the ctrl-c checks in socket creation

commit d0851a08665226456f462a36089ce228d3378711
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 24 14:29:10 2014 -0500

    unlock and close cachefile if we bail out of socket creation.

commit bd351e8edab40dceea2fe698f1785a6ffedc05d8
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 23 11:31:51 2014 -0500

    fill final page in a mapping with ascii

commit 2b33ff5ca945eebcf1926429eaa1592396ae9ead
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 22 22:58:04 2014 -0500

    improve printing of unspecified longs.
    
    - Widen the range of "shall we print hex or dec" from 8k to 16k
    - Make decimal default to signed
    - print down to negative -16k as decimal too.

commit 8f7b6d1f56a6e0599f7367624369a53cf359b33c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 22 22:56:01 2014 -0500

    always print addresses as just %lx
    
    Also make ARG_OP/ARG_LIST always be hex.

commit 59c45e1f550987286ac114346d4b598ec518c7c3
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 22 22:45:19 2014 -0500

    always print out fd's as signed.
    
    So that '-1' works correctly in mmap for eg.

commit 75e6bf4e528f4145a8f3726cef600a2c3eb94c95
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 21 16:27:48 2014 -0500

    update madvise() flags

commit e0f79e0ce7a88694029ce71a402c3caf3b1b7b44
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Tue Jan 21 14:47:57 2014 -0500

    fix divide by zero in random/taviso()
    
    On Tue, 21 Jan 2014, Dave Jones wrote:
    
    > random.c:57:16: warning: Division by zero
    >                 if (!temp) r %= temp;
    >                            ~~^~~~~~~
    > random.c:62:26: warning: Division by zero
    >                 if (!temp) r |= rand() % temp;
    >                                 ~~~~~~~^~~~~~
    >
    > Vince, want to send a follow-up fix ?
    
    Yes, brown-paper bag time.  Never trust "it compiles and runs" as
    testing for a bug that only shows up 1 in 4 billion times anyway.
    
    I've copied the code into a test harness and actually ran it with code
    that returns zero more often to verify it works (after hacking things
    so gcc didn't "helpfully" optimize the zero generating code away).

commit 4e591fd0eff4e9344a1a344b785caf996f1b9573
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 21 14:26:03 2014 -0500

    fix cut-and-paste error.

commit 8a426a35d3246f6749b6c6fb7315b2ed9ab703b3
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 21 14:05:06 2014 -0500

    do some more interesting things with munmap.

commit f71351dcd4225279e925949d847a824b189a0634
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 21 12:45:53 2014 -0500

    fill out policy field for sched_setscheduler,
    
    and add new SCHED_DEADLINE

commit 3f4773e366ee26212fe47448c3447948bdd6f0db
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 21 11:14:56 2014 -0500

    add syscall def for sched_getattr

commit 06da91c42a53fbf11af88028e3e91cafbe852202
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 21 11:08:50 2014 -0500

    add sched_setattr for x86-32 too

commit c5d385f20792d05b56facee0fdb87395e60f84e3
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 21 11:07:37 2014 -0500

    add syscall def for sched_setattr

commit 4e41247abdd013001de96344c6fd27f149cf3871
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jan 20 16:57:53 2014 -0500

    Switch %as usage to %ms
    
    As well as appeasing cppcheck, this also keeps newer toolchains happy,
    which for some reason started warning.

commit 23099ef02b1a4de635a02dba0fee1c7e5557489d
Author: Jiri Slaby <jslaby@suse.cz>
Date:   Mon Jan 20 22:45:51 2014 +0100

    handle SIGABRT similar to SIGSEGV
    
    Sometimes, glibc aborts when some assertion is not met. In that case,
    we want to have a coredump, so do not ignore that signal when
    debugging is enabled.

commit 25be9c010fe64b3688dc81d5583551bf681baeb0
Author: Jiri Slaby <jirislaby@gmail.com>
Date:   Mon Jan 20 21:15:35 2014 +0100

    modify_ldt: initialize scratch
    
    We store a pointer in the scratch area, but only conditionally. Hence
    there can be an uninitialized value which causes problems in the
    tear-down path:
    *** Error in `trinity': double free or corruption (!prev):
    0x0000000002484530 ***
    ======= Backtrace: =========
    /lib64/libc.so.6(+0x740af)[0x7f2a66eb60af]
    /lib64/libc.so.6(+0x798de)[0x7f2a66ebb8de]
    /lib64/libc.so.6(+0x7a5e6)[0x7f2a66ebc5e6]
    trinity[0x40abc8]
    trinity[0x40dabc]
    trinity[0x40338a]
    trinity[0x40704c]
    trinity[0x402c47]
    /lib64/libc.so.6(__libc_start_main+0xf5)[0x7f2a66e63b15]
    trinity[0x402f13]

commit 75eb4d2e8f89a159dadfd22d7394a0fc5b08e278
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jan 19 20:00:45 2014 -0500

    link with LDFLAGS if set.
    
    (another patch from debian)

commit 12e15ef29272807d8ad2969ddf3640d20f6c4ac3
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jan 19 19:59:12 2014 -0500

    add to CFLAGS instead of overwriting them.
    
    One less patch for debian to carry.

commit 9d09c6480d2e89c134d943bd84065899be3170de
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Fri Jan 17 15:41:58 2014 -0500

    fix divide by zero in random/taviso()
    
    In random.c the taviso() routine uses "rand() % rand()" which
    can cause a floating point exception if rand() returns 0.
    Add some tests to make sure that doesn't happen.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit eab8022959136abda87ceb3ad83661c2e0a10c07
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 16 16:45:32 2014 -0500

    put the masking on the variable that changes, not the original alloc.

commit daff5dc8c8a4b94a3df5a802d50085ebc7a9b8a0
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 16 11:00:57 2014 -0500

    reduce variable scoping.

commit 728f09b51a8a6cc4350738aaab519f86c0a7e6d5
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 15 19:30:13 2014 -0500

    guarantee page alignment for page allocs.

commit 85fdd1cfaa5a8893e9163e4dea54b1f5ef275abf
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 15 19:28:31 2014 -0500

    factor out the move_pages page freeing to a common routine.

commit 63536981bc217da5e26f35c11d198cb29f1e44e6
Author: Jiri Slaby <jslaby@suse.cz>
Date:   Tue Jan 14 16:57:06 2014 +0100

    compat: define more ETH_P_ constants
    
    Otherwise for 3.0, we see:
    net/ether.c:23: error: 'ETH_P_TDLS' undeclared here (not in a function)
    net/ether.c:24: error: 'ETH_P_AF_IUCV' undeclared here (not in a function)
    
    Signed-off-by: Jiri Slaby <jslaby@suse.cz>

commit c10cef8958fee14167ee8b1b5438860cdae94c32
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jan 13 17:39:45 2014 -0500

    (entry->num_args >= argnum) is always true.

commit 96105054746eb19f09b17815bfceae5dc25077e7
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jan 13 17:38:25 2014 -0500

    we always have name set.

commit 7cfb325b7a6f70029cc53d1505ab12c1806309f2
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 10 17:31:35 2014 -0500

    only process the number of args a syscall entry has.

commit f5654048e527aa2230db9ed05822999b702f1cfc
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 10 16:15:36 2014 -0500

    rename the syscall struct to syscallentry
    
    and rename all instances of variables using it.
    
    (all because an older ver of gcc doesn't like shadowing syscall()
     with a struct of the same name. Garg.)

commit e3cc257f659e5f67144cbff1ff6b47d648fb18b0
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 10 13:43:36 2014 -0500

    don't check for uid changes when we run as root

commit a1699bd7f2399c9a7ba18e908122616204c6a48b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 10 13:37:02 2014 -0500

    more table munging syscall entry caching

commit 9930a7e3f63d53cdb8566e68bfc46f74c0b5a8de
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 10 13:30:33 2014 -0500

    syscall entry caching for uniarch table munging

commit 5c707919678e1d662bb0b422b4caf09bc0578de1
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 10 13:23:11 2014 -0500

    syscall table entry caching for biarch table munging

commit b40616174614805602558ecc870a118a822847a7
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 10 13:10:18 2014 -0500

    syscall table caching for find_previous_arg_address()

commit 3aa13787ba43d2a3b8e26205d73b3ef4d05898de
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 10 13:09:05 2014 -0500

    more syscall table caching

commit cfb93f71b9b71b28d9f140d70a5e3ef563ec659c
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 10 13:06:40 2014 -0500

    shut up llvm false positive.
    
    The default case won't ever be hit, but we might as well just set it to NULL.

commit d2c98956076bab28198b1f942b61a314034cf377
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 10 13:05:16 2014 -0500

    use syscall entry ptrs to clean up a lot of code.

commit 69f65b7eb84b56cf121b72d93b8eae23d8a7e375
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 10 12:54:32 2014 -0500

    cache the ptr to the syscall entry.

commit 0d7e7c7450b0e4753e489df1bb4e32bc74915c31
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 10 12:08:49 2014 -0500

    when faulting in a page, don't just do the first page in the mapping.

commit afcace5e4cc89d5fc62a588abffe1497ebaca345
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 10 11:57:06 2014 -0500

    We don't need to pass syscall no, now that we're passing the right syscall struct.

commit 315480142865c8871209b1e2b5e9eefaec3ca473
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 10 11:55:32 2014 -0500

    we can also figure out the reg arg for output_reg now that we pass childno.

commit a010d683fb4bf9a6dde85e6731130f73dd7bdc70
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 10 11:53:00 2014 -0500

    remove some parameters from output_arg

commit e78fb6eb64cc4409e0756a15bb3369fa772465e1
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 10 11:43:47 2014 -0500

    introduce a ->decode function for syscalls to output decoded parameters.
    
    decode mmap PROT args as an example, but it's going to be more useful
    for decoding structs.

commit 117ae39d5cc13d367ffc62d0bb992cf43bd83128
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 10 11:09:40 2014 -0500

    skip over some other protocols too if passed with -P

commit acd1482f8ae8d259a73ab185beb1972999ed32ae
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 10 00:06:10 2014 -0500

    ugly workaround for bug introduced by the format string cleanups
    
    Better would be for ret to be passed down as a signed long, but
    this at least fixes things for now.

commit b3f237fe091c9dcebfc7149db2029418dcc3c4c2
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 9 16:02:17 2014 -0500

    cppcheck format string corrections.

commit 4670c111a9970c6e54c6c7151803ee61131b7616
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 9 15:54:49 2014 -0500

    cppcheck variable scoping reductions.

commit f7d6051077f05720f5ad491dae9d715ae98de86e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 9 14:42:27 2014 -0500

    remove unneeded includes

commit 87f61566afea2053cb99d95c1117f3509e8b2b40
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 9 13:19:15 2014 -0500

    rename global -> shared

commit 0a9cddd469d261e8bdf9c292dbe57ab7ab2bd89b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 9 13:17:04 2014 -0500

    move printing shm ptr to shm.c

commit 0949843b85ca227a8d11dd3a84ed1f35661358fb
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 9 13:15:55 2014 -0500

    split up maps.c some.

commit f75ac8f8b1e171562af78a09c20e9252d0e2f73d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 9 12:56:48 2014 -0500

    rename variable

commit 0ad13ac0de29d43da2be390af2e57e4ec775f0fb
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 9 12:56:16 2014 -0500

    remove some other getuid calls.

commit 15166538b2f16569d30921db5300a84cbd2e38d3
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 19:35:15 2014 -0500

    missing proto change.

commit 849cb18ce1711c672e478c3e13bdcb21fb47b164
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 19:34:22 2014 -0500

    validate protocol when passing -P and handle accordingly.
    
    graceful exit means we have to pass failure upwards.

commit a879858207e96e95a2febe86f09de1ba8f10a55f
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 15:32:24 2014 -0500

    fix off-by-one that could lead to div by 0

commit e4a2721e33b015ec2632998ed6dcbd17afd21650
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 14:47:54 2014 -0500

    hand out page_maps as an address

commit 7eb991c78aa62e150e31bf21ddbb74ae192523db
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 13:03:05 2014 -0500

    save a getuid() call each loop

commit 5f718c23844b0867a5f57e950d834b93e3b51ea3
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 12:41:49 2014 -0500

    reuse the same IP five times.

commit e0f35f382e20da746aca346f931ba9ae71374356
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 12:41:33 2014 -0500

    fix output ordering

commit fe1469d0de3a63629627b6253a8c489ac124492a
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 12:29:07 2014 -0500

    some debug info.
    
    I should really add a ->dump operation for such things post ->sanitise

commit ef009eecfefb05d882ede6d8ae6b39974563ecfa
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 12:28:05 2014 -0500

    do address munging all the time.

commit c54e6d879567ecdb0153793a199266ca5e3ca5f0
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 12:17:06 2014 -0500

    use network order for port number in ipv6 too

commit e9d1678a93ea56ef0806b6550adc8f2f98e3d4d4
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 12:16:34 2014 -0500

    use network order for port number

commit ca58ca5df4673a9419f378b8605552cac0bfa5bf
Author: Toralf Förster <toralf.foerster@gmx.de>
Date:   Tue Jan 7 17:27:51 2014 +0100

    be more quiet in [main] Setsockopt(...
    
    Signed-off-by: Toralf Förster <toralf.foerster@gmx.de>

commit c911b0da841eaaa1cf7bdea3641cec44819a01c8
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 11:36:50 2014 -0500

    rework checking for network syscalls.
    
    Now -n will run syscalls with ARG_SOCKADDR and ARG_FD as args.

commit 93db46f534f6baaa82c2e94795728d568a30b477
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 11:23:00 2014 -0500

    factor out the argtype checking

commit 4d34a66b3f453ab372d5804b7843405c1c7c3ba4
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 11:08:22 2014 -0500

    only allocate page_types array once

commit 916d23f81ba59457211f47eb604ad06e44894a32
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 11:00:50 2014 -0500

    fault in pages in random order.

commit 562916ee922c1f6640deb8a8efbdf91222989baf
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 10:56:59 2014 -0500

    fault in mapping in reverse order

commit c7a1c4482539d1284bc27e59e9864f591f821479
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 10:55:38 2014 -0500

    fault in every other page of a mapping

commit 10b2ac14f6635916f1f5181463e4b60ce3aaf13b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 10:44:56 2014 -0500

    remove unnecessary assignment

commit 73aaa2fe44accfe108226c8ea6b330fe516a17b7
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 10:43:30 2014 -0500

    move table munging to tables.c

commit 52198a49092f13440293c72e3bc6e29438d31334
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 10:41:58 2014 -0500

    move shared page stuff to maps.c
    
    still not the perfect place for it, but better.

commit 994894f6003b9c3a896f608a88dc48895001f5a9
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 10:37:28 2014 -0500

    split out the page allocation routine.

commit 024fb1aca4d1bc1369a6b15172fcd809111f3c21
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 8 10:29:20 2014 -0500

    pass around a page of ptrs to the shared mmaps

commit def0454f9527af795ba54bcc915563aa11e62156
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 7 19:42:10 2014 -0500

    some initial documentation on adding ->sanitise routines

commit 753c9d5f6c05bb49fbbe46bc685a21de6ac68b42
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 7 12:11:11 2014 -0500

    teach move_pages about the maps list

commit dab22745b1675fe04e6e86d93f116907684b1417
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 7 12:03:57 2014 -0500

    set scratch later

commit 61133e0c06a655383b1eef255b85ecfd6ddc431a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 7 12:01:49 2014 -0500

    small move_pages improvements

commit f3dcb04fafa9e8541437123bf625f6455c02bc92
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jan 6 22:47:45 2014 -0500

    max_nr_children doesn't need to be in the shm

commit e2b6b482675e6afa6b588d0c89aa7936aa7bdac6
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jan 6 22:13:06 2014 -0500

    use pragma once instead of ifndef guards.
    
    also adds some missing guards to some files.

commit 4a5d28007573e1e48a0920d31bfda5d9c1baae38
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jan 6 22:01:10 2014 -0500

    make 'make devel/release' also commit the configure.sh change

commit 4d7f8b4ffba38dafe1566bf07da6236670ad5074
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jan 6 22:00:24 2014 -0500

    enable devel mode

commit 893c6e00ec0e9de804eb87aee0c5d596ef06c0da
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 20 15:08:59 2013 -0500

    Enable -Werror

commit db9aae17e4f44fceab801bb2118c472a244eadf9
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 20 15:08:51 2013 -0500

    begin 1.4

commit fcfd191f6185d9f76e1b6dd33dca0ed127418933
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 20 14:29:37 2013 -0500

    1.3 release

commit 108de0d0fb27782987046f1b89ead6399a676601
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 20 14:28:50 2013 -0500

    Disable -Werror

commit e4bbda4674089e57b22235f7d65c8bda1b8d7f4e
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 20 11:55:57 2013 -0500

    rename some of the network files
    
    Making it clearer which files are protocol agnostic

commit 81d40adc000e1cab21d3490f7c6a136255f678b6
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 20 02:10:53 2013 -0500

    don't free(page_rand) when we don't have a sockaddr generator.

commit 1f9633ddf9eaec1b5471a4000b035d35d8bf73a3
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 20 01:33:14 2013 -0500

    TODO updates

commit 52f8d8dd2429a0d5969cadcb51ae83ee60c2d6ad
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 19 21:14:39 2013 -0500

    change sockaddr generation to pass around correct types.
    
    These are pointers to sockaddr structs, not unsigned longs.

commit 05b96ee3b1b9901f98b48be0e9b3bcc9f613c331
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 19 16:52:35 2013 -0500

    fix centos6 build failures
    
    reported & tested by Vinson Lee.

commit ca5b4135e26473d9d5f58e19ea8ac29baa1e58d3
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 19 16:09:21 2013 -0500

    extra debug

commit a12d99e6e3ff5fe0f455d61b7f5ff660bfd67b95
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 19 02:04:27 2013 -0500

    collapse the various sockaddr/setsockopt files into per-proto files.
    
    I got tired of bouncing between the two sets.

commit e60aefa4670118d92a4557650cadb03c0f6a3138
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 19 01:42:55 2013 -0500

    beginnings of rds sockaddr gen

commit 4afac475a411f0c929e780c31f84e76ede76c921
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 19 01:27:13 2013 -0500

    generate random sockaddr contents if we don't know the protocol yet.

commit a372d6d75ab183140920aab3bc73de81f5323091
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 19 01:26:28 2013 -0500

    Use correct type.

commit 56a76ff8abcf8c21ad2eb694aae3fd2b6ead4aaa
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 18 21:20:12 2013 -0500

    store mapping type in the struct.
    
    right now we could figure this out by looking what list it's on, but
    actually storing it is simpler.

commit 66b09f46cea7b4c0ba79d663608f9b043a835f3c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 18 21:03:51 2013 -0500

    bump up maximum children to 256.

commit 77412490200b44bc835b8a8859e9fb9a580f3ee9
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 18 20:23:00 2013 -0500

    remove misleading message.

commit 56e386b71ee0bbdef47fcdd95ca299d6f0cb31ae
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 18 19:41:56 2013 -0500

    update 'now' every pidslot walk instead of once per loop.

commit 6c7cd9918e415896bd8b85cad1094c49708499c9
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 18 18:43:53 2013 -0500

    remove duplicate varaible.
    
    pidslot == childno

commit ad41079671378e55321d1f0656b3b83f8d55a3a3
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 18 11:27:56 2013 -0500

    throw away result, we don't care.

commit 5db2da27a14b6efc6b37e569ce73b6d66c507c85
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 18 11:26:51 2013 -0500

    fix memory leak

commit de732d832b418f41bc11d312bd8393b990463a70
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 18 11:26:45 2013 -0500

    fix memory leak

commit b1efdb89c9df58b4f744eba15742bff80d9d21ba
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 18 11:14:54 2013 -0500

    remove unnecessary init

commit 21b2e92ce0cd09557b539c2ee9eeb44fa1d48c19
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 18 11:12:35 2013 -0500

    only set CC if it's unset.
    
    This makes 'make scan' do something useful again.

commit 8773d5e22519f511cf9ab0faeb2424f85460e494
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 18 10:46:55 2013 -0500

    remove pointless duplicate break

commit 8e51acb7bad48d11beb4cf4dca4de409ae2dccc4
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Fri Dec 13 15:40:43 2013 -0500

    perf_event_open update to 3.13-rc3
    
    Update perf_event_open support to Linux 3.13-rc3
    
    This mostly just enables Haswell PERF_SAMPLE_TRANSACTION
    support.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 4cf3a15d2553c32343812532d796d1151ac68f03
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 12 20:42:39 2013 -0500

    if this fails we're SOL anyway.
    
    But it shuts coverity up

commit 8f6b3d17d3685276c0bac133da870d652ca7f7ae
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 12 18:34:42 2013 -0500

    actually, forget that idea.

commit b324e5144fcbb37ee6db6f7bf6fed40c2e0bd15e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 12 18:27:49 2013 -0500

    backtrace needs -rdynamic

commit 20e3caa935a692d9482da8c87fbb1f614f20b186
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 12 18:27:02 2013 -0500

    add a show_backtrace function for debugging.

commit 33a5f453811ca80c5cbc8e158b09f1cdd8b25a2a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 12 18:00:37 2013 -0500

    make -g followed by a -x work.

commit 327e9082eb766ae620b2c0029e5170d2b1d09567
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 12 17:30:58 2013 -0500

    clean up the remap_file_pages sanitize routine

commit a8b291736e2a4e66c7b7e3f81c0c97677533fff7
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 12 16:24:40 2013 -0500

    make 0.0.0.0 be a /8

commit 87b2976fb49b634ff0cf6da029aa1dd2e256bb3e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 12 13:12:53 2013 -0500

    msync can't be ASYNC and SYNC

commit 88e11a3532198a1d0385815733be207202f665ee
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 11 18:37:50 2013 -0500

    centralise the 'dirty a mapping' code.

commit cc5acbf0b8c39b78ca687675e3661cf5ff1bf49f
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 11 11:45:50 2013 -0500

    remove outdated comment

commit 36d85c271a5262854772c92bee1d21688652a152
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 9 21:29:13 2013 -0500

    explicitly clear the other arch if not specified in -a

commit c651cd4c72c8ccf77527bde8474faa1d1690bef0
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 9 21:29:02 2013 -0500

    add more missing flushes

commit c7bb14e032795465f62290f2594566b2953232d1
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 9 21:25:19 2013 -0500

    output message when parsing failure for -a

commit b61d7f0713e45ddc2c99763675190285a7269319
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 9 21:14:12 2013 -0500

    avoid dereferencing null maps after we munmap everything

commit 810119f384e525bce23bc2bafff6303f1d4046b2
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 9 21:14:00 2013 -0500

    flush partial out to stdout.

commit fb05e6d3d34ad6c6889a61edf0b7e531e957f200
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 9 16:15:22 2013 -0500

    increment nr_sockets after we've done using it.

commit e90f96ae972504b31dc599dc4ea3607d7cebdfae
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 9 16:15:06 2013 -0500

    explicitly list l_linger as 0

commit dc640fbd6715ad521dace53865e2886827734559
Author: Vinson Lee <vlee@twitter.com>
Date:   Mon Dec 9 11:18:11 2013 -0800

    syscalls/prctl.c: Fix build with USE_SECCOMP disabled.
    
    This patch fixes this build error on CentOS 6.
    
      CC    syscalls/prctl.o
    syscalls/prctl.c:48: error: expected ‘)’ before ‘int’
    cc1: warnings being treated as errors
    syscalls/prctl.c: In function ‘sanitise_prctl’:
    syscalls/prctl.c:63: error: implicit declaration of function ‘do_set_seccomp’
    syscalls/prctl.c:63: error: nested extern declaration of ‘do_set_seccomp’
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 0143dd9ec00d2b75b48ecbf79896529380a27ed7
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 9 13:35:36 2013 -0500

    make remap_file_pages map list aware

commit f10193c9d08911c0dcb439991fd15b07f7293ddf
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 9 13:27:52 2013 -0500

    make msync aware of maps list

commit 77818cc12f4099417e25ff76f1e52bfe2a0bd953
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 9 13:26:20 2013 -0500

    make munlock aware of maps list

commit 32234b0eac364cc0c4393ea40d0d0b5a4d197a77
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 9 13:24:55 2013 -0500

    make mlock aware of maps list

commit b36beaf8d2fc68738b250975900bcce462c7c96f
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 9 13:23:41 2013 -0500

    add common routine for setting a1/a2 for start/len of mmap.

commit fd1a80fa89d6ebdb356e209b5ade1565fac43f54
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 9 13:13:21 2013 -0500

    make mbind map-list aware

commit 237261e71e3219e3b550493a59e21abd4805c4f7
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 9 13:09:36 2013 -0500

    make madvise map-list aware

commit 4c9192ba56b0540168a1771699a6feba0221c769
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 9 13:07:04 2013 -0500

    add a routine to delete local mappings
    
    (so I don't forget to decrement the mapping count)

commit 221a93cc8af21c778640b78501b22a6fd1f6db5d
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 9 10:31:33 2013 -0500

    make munmap aware of the mapping lists

commit 73ffa319b263b7b9fe62be4f0c8699426db15512
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 6 16:18:21 2013 -0500

    move some macros out of trinity.h into utils.h

commit 3c0af08949da9c160a7874547a1255758aab8543
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 6 15:41:11 2013 -0500

    set mmap sizes to something more interesting

commit 99e8177b371ef821aeb9827a235fde665b416a34
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 6 15:28:33 2013 -0500

    make get_map choose from both local and global mapping lists.

commit c0edb87de9cf2f77e7e9a7dc1796b8291e07b5bb
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 6 14:54:34 2013 -0500

    add results from mmap to child-local list.

commit a27ffc18ae201377a4ac99a8a687560b7663e4b5
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 6 14:51:31 2013 -0500

    rename variable

commit 45d2597415202e08f531724a16063025cc69e376
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 6 14:00:12 2013 -0500

    add per-child mmap list.

commit 50e37c071611f6ee264144d43cefe2564aae9a73
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 6 13:56:02 2013 -0500

    use common definition for MB/GB

commit e1a97d518b8e69a976d00a2b099032737a91e733
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 6 13:53:04 2013 -0500

    move shm code out of trinity.c into own file

commit c4b97b09366b48839c61aa10fc81644f5e1aba0b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 6 13:42:54 2013 -0500

    move some non mmap stuff out of maps.c

commit 8681279c2b4f04d35f18f76ffea66fca4ac41f3a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 6 12:59:48 2013 -0500

    rename maps variables

commit 96222a943e486c1436fabc22f8af3395873e2569
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 6 11:50:27 2013 -0500

    add a fixme for later

commit f55bd3cff87d89a317c1dc809e4257eefc76892f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 6 11:48:33 2013 -0500

    use correct error check in mmap->post

commit e24960c22d5e9a8014d9555225722dbd1c0708be
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 6 11:43:50 2013 -0500

    make mremap use ARG_MMAP properly

commit e8e4b73ae9623617a3e3fd898eaf7ede662cca82
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 6 11:42:03 2013 -0500

    fix up mprotect to work with new map struct

commit ad4d77e7726fb8c49214417e58f25dbb6b47469e
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 6 11:34:20 2013 -0500

    add ARG_MMAP for routines that operate on mappings we're going to track.

commit cf2812bc955e9ea06392e0134437fa0bf4d1ec59
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 6 11:28:05 2013 -0500

    don't write to read-only mappings.

commit bea9e287695378b84cc04735aecbf4cdb3d29ef4
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 6 11:21:39 2013 -0500

    update the mapping pointers on successful mremaps

commit 0164d3ecbc8ede86955b884b01f6ebfd3ffb7730
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 5 19:30:45 2013 -0500

    only do the guard page on single pages.

commit e91a95f48c0daa297cf17126fb3aafc03e1f00ca
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 5 17:37:30 2013 -0500

    don't do 3GB worth of mmaps on startup, it ooms way too quick.

commit 65ea62752a47d22cdecf954eb2782b0b841d5094
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 5 17:25:52 2013 -0500

    add a TODO for tomorrow.

commit 498fb8bc2824325645749ead7345b319d40617e4
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 5 17:08:27 2013 -0500

    mremap sanitise improvements.

commit c6a60e0b052c8df12d3139502d8da389d049f5b6
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 5 17:07:27 2013 -0500

    do more mappings in the main thread.

commit 2fa8f4a3429836691bd4a832939263c720475288
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 5 17:01:58 2013 -0500

    change get_map to return the map struct, not the ptr

commit 2a6c7ae42c75d2fd001a1847ca13d47eaed3b9a3
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 5 16:59:19 2013 -0500

    add missing hunk of previous commit. oops.

commit ea87188a2c92d7748e6e23383f1d6c1208cc7195
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 5 01:39:39 2013 -0500

    Convert maps.c over to use list.h

commit 3febc40a8c757446cd7f6a050d002c3625256d29
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 4 17:54:40 2013 -0500

    add F_GETOWNER_UIDS fcntl flag

commit ca377105b4222a2e77e5d75a346e16a175896408
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 4 17:38:04 2013 -0500

    add a ->post for mmap

commit 3f02a37a9147ee6746a5917787b2c67d55511757
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 4 17:35:38 2013 -0500

    move the anon mmap code out to its own subroutine.

commit cb4aba8f03a8f1c63e128d102a4f3a796a9fc44b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 4 17:33:21 2013 -0500

    [watchdog] if we're exiting don't bother checking main.
    
    Fairly big messy diff, but it cleans up some occasional spew at process exit time.

commit 9de1e42ac22de2b17d62ffe031eb01408afa3cfb
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 4 17:25:57 2013 -0500

    change ->post to have the childno as the parameter, instead of the retval.
    
    Stick the retval in the shm.
    
    This will allow ->post ops to figure out what our params were.

commit fa868e38a3fd5784b3efcd7cf4459b289ebac392
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 4 13:57:02 2013 -0500

    Fix compile warning when USE_SECCOMP disabled.
    
    Reported-and-Tested-by: Vinson Lee <vlee@twitter.com>

commit cc610cbb1b90a9aba5a5396e57e8f07cd288fb18
Author: Ildar Muslukhov <ildarm@google.com>
Date:   Tue Dec 3 16:44:46 2013 -0800

    Fixes incorrect syscall indices for shmat and shmctl in X86_64 arch.
    
    This patch fixes incorrect indices for shmat (should be #30) and shmctl
    (should be #31).
    
    Signed-off-by: Ildar Muslukhov <ildarm@google.com>

commit b21a91e92ed13c9b52b4bfa77ebb8dcb84bf2bd5
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 3 16:05:46 2013 -0500

    missing address-of op.

commit b7b6d219b112bf7955e4cf2140c0d17e8595c42e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 3 14:59:21 2013 -0500

    fix up a missed pointer conversion after Ildar's last change.

commit 23402b71653f20cfa42a71b0886b2c57cbf33bc9
Author: Ildar Muslukhov <ildarm@google.com>
Date:   Wed Nov 27 16:30:16 2013 -0800

    Fixes bug with incorrect casting of on stackvariable to a pointer.
    
    This patch fixes a bug in open_socket function with sa variable of type
    struct sockaddr. In particular, this variable is alloctaed on stack, and
    then address to it is passed to specific functions that allocate memory
    and overwrite the content of sa with the address of the newly allocated
    object, but not with the values that were assigned to the object. Thus, 4
    or 8 first bytes of the sa struct are always initialized with an
    address of newly allocated object.
    
    Signed-off-by: Ildar Muslukhov <ildarm@google.com>

commit 960100fa71e96c78be7d944021c85848b0c745e9
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 3 12:24:13 2013 -0500

    fix build error on 32bit

commit be71ae591eccaf8a7e8ef85cd4a4346a709cc008
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 2 18:03:37 2013 -0500

    add and use zmalloc

commit 790c134494c870bf3ac92eae08838e365d004a10
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 26 12:42:46 2013 -0500

    centralise the setsockopt code in one place instead of duplicating workarounds.

commit 01449221a577fa5febb118f175d19be934f90408
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 26 11:50:34 2013 -0500

    fix bad cut-n-paste causing inverted state when using -r on uniarch

commit 6b9d15df44ae475fcc7e80d11a62b0fc1e874a57
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 26 11:14:01 2013 -0500

    work around bluetooth linger bug

commit 831f62bf90849d83ee58bd7fd8e3184b7a62f961
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 26 11:13:05 2013 -0500

    save/restore r11 in 32bit syscalls.

commit e77bc5780303b04296be390cb73a3e77599590a9
Author: Vinson Lee <vlee@twitter.com>
Date:   Fri Nov 15 17:20:59 2013 -0800

    net/bpf.c: Remove AUDIT_ARCH_H8300.
    
    AUDIT_ARCH_H8300 was removed from include/uapi/linux/audit.h in Linux
    3.13 commit 6f6f467eaaa0ac4fb77714d0172d65c781dabb8c.
    
    commit 6f6f467eaaa0ac4fb77714d0172d65c781dabb8c
    Author: Guenter Roeck <linux@roeck-us.net>
    Date:   Fri Aug 30 06:12:40 2013 -0700
    
        Drop remaining references to H8/300 architecture
    
        With the architecture gone, any references to it are no longer needed.
    
        Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
        Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
        Signed-off-by: Guenter Roeck <linux@roeck-us.net>
    
    This patch fixes this build error.
    
      CC    net/bpf.o
    net/bpf.c:135:18: error: ‘AUDIT_ARCH_H8300’ undeclared here (not in a function)
      AUDIT_ARCH_FRV, AUDIT_ARCH_H8300, AUDIT_ARCH_I386, AUDIT_ARCH_IA64,
                      ^
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>
    Acked-by: Daniel Borkmann <dborkman@redhat.com>

commit 229d8911424b97ffb30da52b3a941023a2fad9b4
Author: Vinson Lee <vlee@twitter.com>
Date:   Fri Nov 15 17:01:29 2013 -0800

    net/rds_setsockopt.c: Always define SOL_RDS.
    
    This patch fixes this build error on CentOS 6.
    
      CC    net/rds_setsockopt.o
    net/rds_setsockopt.c: In function ‘rds_setsockopt’:
    net/rds_setsockopt.c:33: error: ‘SOL_RDS’ undeclared (first use in this function)
    net/rds_setsockopt.c:33: error: (Each undeclared identifier is reported only once
    net/rds_setsockopt.c:33: error: for each function it appears in.)
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 5aeeb0e435f3983d15d0cbe80f896e125145a36a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Nov 15 19:37:38 2013 -0500

    clear out pidslots after we kill all children

commit 4501a885a1d7f81f8c2ff7cd31de33df37de2e91
Author: Dave Jones <davej@redhat.com>
Date:   Fri Nov 15 12:39:50 2013 -0500

    reduce ifdeffery in caif

commit a1654d5bddbb55d3dfd625bf376c6deb88703905
Author: Dave Jones <davej@redhat.com>
Date:   Fri Nov 15 12:38:29 2013 -0500

    reduce ifdeffery in rds

commit d071a67bd8843a743d471b192da5e966805aacd5
Author: Dave Jones <davej@redhat.com>
Date:   Fri Nov 15 11:23:05 2013 -0500

    don't fill out args that don't exist.
    
    Not sure how I never noticed this before.

commit 0cf5a77c4fb160c099f63cf24c05f6c19121dd72
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 14 18:49:47 2013 -0500

    remove double assign of variable.

commit feafc056870fc97b5042fda93531b29d771101db
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 14 18:49:12 2013 -0500

    workaround a lockup with bluetooth sockets

commit e96e672ca3cca927c25320432422450717f103b4
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 14 18:46:38 2013 -0500

    stub out syscall32 entirely on non-biarch builds.

commit 102b3414e7dda799744222f2b46a3ebe664fad32
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 14 18:22:09 2013 -0500

    ugly hack for 32bit warnings.
    
    This needs to just go away entirely on 32bit builds. Will do later.

commit 9f44280b17ad1c277fb539cbbba3187c3dc50861
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 14 18:21:28 2013 -0500

    fix 32bit warnings

commit a6445a230832c4bf239685a4b8d01af32d93999c
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Thu Nov 14 12:35:27 2013 -0500

    perf_event_open more likely tracepoint values
    
    Tracepoint ids tend to be relatively small (less than 1024 or so).
    This change makes it more likely to hit valid ones.
    
    Once the fix to avoid the troublesome 0x18 (irq_work_exit)
    tracepoint gets into the kernel maybe we can start stressing
    things with trinity again.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 2342edca85cfd050eab5c8da251d4ace63f09c01
Author: Sergey Bronnikov <estetus@gmail.com>
Date:   Thu Nov 14 15:19:32 2013 +0400

    Add missing command line option to help.

commit bfbc9a5c3ec3fd19e7d2b50e3f62e3a7487f45a5
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 14 10:55:04 2013 -0500

    add ETH_P_PRP to ether types

commit 6b0d67124d1654b53898f7511700098a292c096f
Author: Vinson Lee <vlee@twitter.com>
Date:   Wed Nov 13 11:17:27 2013 -0800

    Fix unused variable build errors.
    
    This patch fixes these build errors on CentOS 6.
    
      CC    net/caif_setsockopt.o
    cc1: warnings being treated as errors
    net/caif_setsockopt.c: In function ‘caif_setsockopt’:
    net/caif_setsockopt.c:20: error: unused variable ‘val’
    
      CC    net/rds_setsockopt.o
    cc1: warnings being treated as errors
    net/rds_setsockopt.c: In function ‘rds_setsockopt’:
    net/rds_setsockopt.c:23: error: unused variable ‘val’
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit cc6f77eb72ccc775f8f680115ffd909b15ff4959
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 13 12:51:58 2013 -0500

    Add SO_MAX_PACING_RATE to setsockopt options

commit 1ea0ce763074d06b7c9d9dc718a3d34235db340a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 12 17:34:44 2013 -0500

    Enable -Werror

commit 87b5f553b8dcbb05cfd531beacd9a0f975f9a9e9
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 12 17:33:49 2013 -0500

    silence -Wshadow warning

commit 753b2076210dd5f6297a504c8e5a9520f36f095f
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Wed Nov 6 15:49:06 2013 -0500

    perf_event_open: vary pid and cpu values more
    
    While digging through long execution traces working on the ftrace
    bug I noticed we weren't really varying the "cpu" or "pid" values
    very much.  This attempts to add a bit more variety.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 244f3431e4e11f1ab379899b4493e0f82de12c50
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 5 20:39:40 2013 -0500

    informational use of socket triplet on closure failure

commit 60de13ef94be6ca5197ff8017835e5d4fb0ba967
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 5 20:38:00 2013 -0500

    make socket shutdown at exit use the code in sockets.c
    
    Also improve the sockets.c variant to do the SO_LINGER tweaking.

commit a9daf334a86c848bd2db04c61cd5842f36ae2f85
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 5 20:33:03 2013 -0500

    store the family/type/proto on socket creation.

commit 6a9fef06c2756d73c0c98a7718d3abac9b9ca1e6
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 5 20:30:37 2013 -0500

    replace shm->socket_fds with a struct to map back to family/proto/type

commit 50840aca75bb60484489ee046c648a11ac5005d2
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 5 19:58:31 2013 -0500

    make ctags rule dependant on source files.

commit 5a19920ee552d9552c983428f10f0323b5a2b522
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 5 19:54:05 2013 -0500

    sockets code shouldn't really know about things like victim path
    
    Move that to the caller.

commit 13d6573b9c15538b8c022913091cf417d35ded0b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 5 11:45:08 2013 -0500

    reenable prctl, but just for seccomp for now.

commit c7d38d957a0ae338f021d0ad1360560d3be3f5ef
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 4 22:33:03 2013 -0500

    hush some more.

commit 333f2eb761a4175a2d37a63977566c7dd1c6e9d0
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 4 22:28:32 2013 -0500

    add a new rand-page generator
    
    spew an ascii variant of a random number into the buffer

commit 4ce56deb91ca4209be632013ce7bdaaeab3a7e6d
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 4 22:26:09 2013 -0500

    remove logging usage of previous_a*
    
    This is bogus. The old code compared an unaltered version of the register
    to the post-sanitised version.  This 'oldreg' bears no relation to
    shm->previous_a* which is the _previous syscall_ that was made.
    The latter is a debug thing, not something useful in this way.
    
    We only did this to colorize the output some, so it's simpler just
    to remove that whole thing.

commit f4e3553b608f6692971f591a3f28f6ec905805f3
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 4 22:22:56 2013 -0500

    remove i386 int80 code.
    
    This is unnecessary. x86-32 uses syscall(), not syscall32().

commit c77098a16863e2a821e84263616b66916b6c70bb
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 4 22:21:48 2013 -0500

    cleanup syscall32 some more.
    
    remove some unnecessary ifdeffery

commit ec14561f9b5d763cb2e3d01d127f47e364ec3237
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 4 22:06:09 2013 -0500

    remove the num_args checking in syscall32.

commit 40e163664229b3264bd40fde20bf5d9a0f467182
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 4 21:46:05 2013 -0500

    save/restore r10 around 32bit x86-64 syscalls.
    
    This fixes the problem where we segfault as soon as we hit __syscall_return(res)
    res was being held in r10, and went out of scope once we hit the syscall,
    so we were scribbling through random ptrs ->segv.
    
    We should probably be saving more registers here to be completely safe,
    but for now this seems to do the trick.

commit b344761bde490d0442976b1f3453af2f69dae237
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 4 17:11:00 2013 -0500

    restore the unconditional <timed out> message.

commit f07fd31c9c7c3a05300a84680f0ea2927b43db40
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 4 17:08:42 2013 -0500

    quit child if we keep stalling without making progress.

commit b7ef58a959446c55be277f16948a1066050eeff4
Author: Cyrill Gorcunov <gorcunov@openvz.org>
Date:   Wed Oct 30 16:15:46 2013 +0400

    net: sockets -- Don't generate sockets with disabled protocol families
    
    Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>

commit 69dda3f4f19e70a0cd5090e5790f577dabc725f4
Author: Cyrill Gorcunov <gorcunov@openvz.org>
Date:   Wed Oct 30 16:15:45 2013 +0400

    params: Add --no_proto|-E option
    
    This option will track the socket families which user
    prefer to exclude from testing, for example PF_PACKET
    sockets and such.
    
    Note I've had to include net.h in two files only because
    TRINITY_PF_MAX is declared there. Maybe it worth to move
    this constant to constants.h.
    
    Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>

commit f47ca6719452851afe1607d2d37def53e66fe6f8
Author: Cyrill Gorcunov <gorcunov@openvz.org>
Date:   Wed Oct 30 16:15:44 2013 +0400

    net: protocols -- Move protocol function declarations to protocols.h
    
    This allows to isolate functions and include the header only where needed.
    
    Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>

commit 65bdaff64906d299afa189a82b00eb2b8ceae5d2
Author: Cyrill Gorcunov <gorcunov@openvz.org>
Date:   Wed Oct 30 16:15:43 2013 +0400

    net: protocols -- Add lookup_proto helper
    
    We will use it for parsing disabled sockets families as well.
    
    Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>

commit aa69ee2e8e0e258549ac5e8b8f84094b8c5d2d22
Author: Cyrill Gorcunov <gorcunov@openvz.org>
Date:   Wed Oct 30 16:15:42 2013 +0400

    net: protocols -- Use ARRAY_SIZE helper
    
    We're looking up over @protocols array so use appropriate
    helper here instead of TRINITY_PF_MAX constant.
    
    Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>

commit d763b037ed040787b539055b7d15bb4489119a0b
Author: Cyrill Gorcunov <gorcunov@openvz.org>
Date:   Wed Oct 30 16:15:41 2013 +0400

    log: Add BUG_ON helper
    
    We will use it in sockets creating routine.
    
    Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>

commit e27360bdec4713dacea67a45449af63b4d888ab0
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Mon Oct 28 12:48:01 2013 +0200

    configure.sh: support cross compilation with sysroot
    
    SYSROOT points to the path where libs and headers are to be searched
    from.
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit 56d9636602f4a359a664c3e889b9d7c1b061dc42
Author: Rolf Eike Beer <eike-kernel@sf-tec.de>
Date:   Tue Oct 22 18:42:32 2013 +0200

    add more headers to fix compile on PA-RISC

commit 7873b549a3f0ac4ceb2e3d47fe032fe7a0249bf7
Author: Rolf Eike Beer <eike-kernel@sf-tec.de>
Date:   Tue Oct 22 18:39:48 2013 +0200

    add basic information for PA-RISC architecture

commit 2eac69c33d7110a34969fa197c3b6ce633ba1e7a
Author: Cyrill Gorcunov <gorcunov@gmail.com>
Date:   Thu Oct 24 11:53:34 2013 +0400

    params: Add missing \n into usage output
    
    Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>

commit 904c9c208dbe8d4b139bd08ccbff67b5d00e73ab
Author: Toralf Förster <toralf.foerster@gmx.de>
Date:   Thu Oct 31 20:30:30 2013 +0100

    set syscalls_todo to N, not N+1
    
    I really do not see any use case to increment users choice.
    
    Signed-off-by: Toralf Förster <toralf.foerster@gmx.de>

commit 82750a7970c99a8b81593fe56d6988bbe1d1eb81
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 25 09:06:59 2013 -0400

    hush some noise if we're running with -q

commit 7ccf18ee785035ac306cedae76f48284cecdddc9
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 23 11:57:39 2013 -0400

    fix accounting so it counts syscalls before we do the syscall.
    
    Otherwise when we jump back after sigalrm, we don't increment the counters.

commit f488f90c0a1e98a01bd85277cfaa623e2256db8b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 23 11:57:18 2013 -0400

    flush previous syscall output when we get SIGALRM

commit f06f9478c528549de7bea4e1ff0788aa5372a396
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 23 11:56:39 2013 -0400

    jump back to child after getting SIGALRM
    
    This should work better now, so we should be able to jump directly back into
    the same child and just do a different syscall.

commit dce794a2c778a12bffda51c634cfa2323243bd6f
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 23 11:37:41 2013 -0400

    use rand32 for netlink groups.

commit f106c780b2b9a722e0927bbebc88a14e54339057
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 23 11:05:05 2013 -0400

    use a trinity child pid for the netlink pid

commit 39edfd6d34f1137cb2d01700897f6daa081a8193
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 23 11:00:40 2013 -0400

    Add missing NETLINK_RX_RING, NETLINK_TX_RING definitions.

commit d176ec0c8cbc1a80857e867eb3ac3ecec21a5a16
Author: Vinson Lee <vlee@twitter.com>
Date:   Fri Oct 18 19:35:34 2013 -0700

    compat.h: Add if_ether.h symbols for older kernels.
    
    This patch fixes these build errors on CentOS 6.
    
      CC    net/ether.o
    net/ether.c:12: error: ‘ETH_P_BATMAN’ undeclared here (not in a function)
    net/ether.c:19: error: ‘ETH_P_LINK_CTL’ undeclared here (not in a function)
    net/ether.c:21: error: ‘ETH_P_8021AD’ undeclared here (not in a function)
    net/ether.c:21: error: ‘ETH_P_802_EX1’ undeclared here (not in a function)
    net/ether.c:22: error: ‘ETH_P_8021AH’ undeclared here (not in a function)
    net/ether.c:22: error: ‘ETH_P_MVRP’ undeclared here (not in a function)
    net/ether.c:23: error: ‘ETH_P_QINQ1’ undeclared here (not in a function)
    net/ether.c:24: error: ‘ETH_P_QINQ2’ undeclared here (not in a function)
    net/ether.c:24: error: ‘ETH_P_QINQ3’ undeclared here (not in a function)
    net/ether.c:31: error: ‘ETH_P_CANFD’ undeclared here (not in a function)
    net/ether.c:34: error: ‘ETH_P_CAIF’ undeclared here (not in a function)
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 81063fb12e7e79299e131cb32a5a3223844730fe
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Wed Oct 16 13:43:30 2013 +0300

    Makefile: install depends on trinity target
    
    After this 'make install' just works.
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit 8c12db27144fa298e0bd5b2238758e5265bef95a
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Wed Oct 16 13:43:29 2013 +0300

    net/netlink_setsockopt.c: workaround for old headers
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit f237586eaad062a6f1c7f489a7359d190788caf1
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Wed Oct 16 13:43:28 2013 +0300

    net/netlink.c: support older kernels and headers
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit 0cc8dffceddcf1eb11d8c00b19f3aba193485633
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Wed Oct 16 13:43:27 2013 +0300

    net/irda_setsockopt.c: workaround for old irda headers
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit 17f2a50b8efb516f5291316dd64615ee75644fed
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Wed Oct 16 13:43:26 2013 +0300

    Define IP_MULTICAST_ALL if header files don't have it
    
    Older systems have it defined in <linux/in.h> which conflicts with
    <netinet/in.h>.
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit 5c1edb2fcb0346c2c1311fef26cbc290a6328e2f
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Wed Oct 16 13:43:25 2013 +0300

    Makefile: support sysroot for libs and includes
    
    Helps with cross compilation when correct lib and include files
    are used.
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit bf4ac8aa5a56fe143e305cc40e5337c526bd7727
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Wed Oct 16 13:43:24 2013 +0300

    Makefile: use gcc as compiler
    
    Cross compilers might not have cc symlink to gcc.
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit 4e3bf7e2ab735c090f3f276540a8756f8c3e62a2
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Wed Oct 16 13:43:21 2013 +0300

    configure.sh: expand tilde from CROSS_COMPILE variable
    
    Users can define cross compiler paths like
    CROSS_COMPILE="ccache ~/bin/arm/bin/..." and would like this to work.
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit e997ca403c8cbff2615ff9d9641506c7ac65a6cd
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Wed Oct 16 13:43:20 2013 +0300

    configure.sh: use gcc as default compiler
    
    Cross compilers might not have cc symlink but do have gcc.
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit d0c7ece98e4cf1b6b1aae4c146471abce71ce402
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Wed Oct 16 13:43:19 2013 +0300

    configure.sh: use colors only with terminal
    
    The color codes mess up less and log files.
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit d1a88b8fdef6b33f9fbee69162a97c962723b38e
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 18 13:49:35 2013 -0400

    ETH_P_802_3_MIN back compat.

commit 59a95dc27439f5f55c48e996eebd899feaa457c0
Author: Daniel Borkmann <dborkman@redhat.com>
Date:   Fri Oct 18 16:56:09 2013 +0200

    net: bpf: fix off by one in seccomp_choose
    
    We were off by one when iterating through states, so simply fix
    that by choosing the correct max limit.
    
    Signed-off-by: Daniel Borkmann <dborkman@redhat.com>

commit 3a5737c8ad870e5f67d6e2012a04bf6323050df1
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 17 23:07:46 2013 -0400

    Generate random ethernet types for packet sockets.

commit d472a9e737eb2e687e97a0ce429c24c798da2dd8
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 17 22:00:24 2013 -0400

    add a hack to shut up clang.
    
    It thinks page_size can be zero here, which is crap.

commit 7a66731bce2ba412009f6a52c7e8369631720d94
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 17 21:51:32 2013 -0400

    remove dead code.

commit 183973ce1657ec4241d3a782490a3c4412a02ec1
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 17 21:05:45 2013 -0400

    fill in proper multicast structs for IP multicast setsockopts

commit 02e5018042e41bbc75819e0617179c561f497b0d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 17 20:04:39 2013 -0400

    multicast address block is a /24, not a /4

commit 313b1944dd33336444bedd1bd014932b2315bd03
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 17 16:54:58 2013 -0400

    make rand32 do the length dividing thing too
    
    on the off-chance we're using it to unannotated length args

commit 655e88db349f997e874a1a7e7c8a46a61ff5cb90
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 17 16:54:39 2013 -0400

    shorten the length dividing code.

commit 5342489474e557454465657a06395c1aa902e395
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 17 16:03:19 2013 -0400

    don't try truncating logfiles if logging is turned off.

commit a0b5732e442ffcfede765686ab8aab1dcc338b85
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 17 15:47:43 2013 -0400

    output the pid when we BUG()

commit 49e103306ef9a8aa29fdb563655c8a0d91d5f24d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 17 15:21:01 2013 -0400

    if we bug(), handle ctrl-c in a cleaner manner.

commit d78bf4136bcd1b1284ac546a8bd29287723efe81
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 17 15:12:46 2013 -0400

    add missing \n

commit 9126ab3e7f0e6fc80caeeb6333001b9636ac7c70
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 17 14:00:47 2013 -0400

    truncate log files to 0 when starting a new child.

commit 7199726e9b5b739bd60674fd5183c572d23ffdca
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 17 12:26:03 2013 -0400

    do setsockopt on sockets at creation time.

commit 5e0be4f098cd5d05c029df49f47ede63ead4aa88
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 17 12:12:35 2013 -0400

    remove unneeded include

commit 320b9abd9b42443b5ff14a24848a624d23f8d5da
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 17 12:12:19 2013 -0400

    add missing proto

commit c874dd87ee3ae913135418f263ea6ab9cf00d030
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 17 12:10:51 2013 -0400

    split the regeneration code out to its own file.

commit cc04123b3b7cae40e3096f09bb77885b7b7f97a9
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 17 12:06:55 2013 -0400

    do perioidic setsockopt on opened sockets.

commit 6ece28705f1b1d0644fae17de87fb380797530e8
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 17 11:49:18 2013 -0400

    split up setsockopt struct generation to ease reuse.
    
    I want to use this from the main process, which means not screwing
    with child shm entries.

commit 682b2488c076d7c41c13ae06d588d3f39603a8f9
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 16 16:02:08 2013 -0400

    maintain a 'hiscore' of nr of syscalls a child did.
    
    Not especially useful, but I was curious

commit a9b41773057df0075afb4506b9fbe346ddf0ae60
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 16 15:58:58 2013 -0400

    move mainpid into the shm
    
    Watchdog needs to read it, and we forked off the watchdog before we
    set mainpid.  This meant we were doing kill(0, 0) in check_main_alive()

commit 4a6ddd73b8cc3f64c89dc5025df6fe7e540c50ff
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 16 15:57:29 2013 -0400

    Wait until children have been started before starting watchdog checking.

commit 63725cd355961fb5c5c20f69bd9a32416c181394
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 15 15:57:07 2013 -0400

    Add missing format string

commit f2e6014d11ecc454de20ae0b2fda8c675927f1ec
Author: Will Deacon <will.deacon@arm.com>
Date:   Sun Oct 13 14:17:52 2013 +0100

    alpha: add Linux syscall table numbers
    
    Crikey, Alpha's syscall table is a right mess. Anyway, with enough sed
    and perl-fu, here's something that looks about right for trinity. I had
    to hand-hack the result to remove a whole bunch of obsolete syscalls
    that trinity doesn't care about.
    
    I also fixed up some missing Alpha ptrace requests with dummy values.
    
    Signed-off-by: Will Deacon <will.deacon@arm.com>

commit e2e62fcd9176c54bdef36a8b981f06c85eea1af1
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 11 11:53:50 2013 -0400

    terminate the page of rand format strings at a random part of the page.

commit 43699d80654a7afc4804a7d23ba2f011df16439d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 20:02:35 2013 -0400

    TODO updates

commit 68e596c0ec82152434f09c2d863fd152cc61fed5
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 19:47:25 2013 -0400

    sort

commit 89e57be9a9f5ab8508704c66e961b556145a7b4a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 19:25:51 2013 -0400

    more sorting

commit 0c0e7d1a0b2a551c4948aecc19f365e6b9cab687
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 19:21:08 2013 -0400

    clean up arch syscall setup

commit 9a1152d647461e42b08b2fa756fd670b28b343cc
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 19:11:14 2013 -0400

    remove excess whitespace

commit 5060ba22e9b5f9a811fa93db14784444844f8097
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 19:07:59 2013 -0400

    remove leftovers

commit 663d6967f88dd2b5b141b980da896db494fab36b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 19:06:46 2013 -0400

    split the table related stuff out of syscall.h
    
    this means we can change tables headers without rebuilding every single syscall.

commit e5014d609381e03e6d06ef1a30f06f5f11d10e9e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 18:57:26 2013 -0400

    split display_enabled_syscalls into uni/biarch

commit 78b47fb84e469cc45637f0f48f79cb279349482c
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 18:54:36 2013 -0400

    remove unnecessary variable

commit 3b4937f36ee46276a5c9f71b16ae7ae3e43f360b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 18:53:48 2013 -0400

    split dump_syscall_tables into uni/biarch

commit 8dca299faf23cb9a1cffd266e02412856be5e6ac
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 18:49:51 2013 -0400

    add back a missing space that got inadvertantly dropped.

commit f806befda6c219c6fa8e2054840a80be48bc52d9
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 18:46:29 2013 -0400

    split deactivate_disabled_syscalls into uni/biarch

commit 4bf5ff36faca8a7db5aa817039ffdc13bedbf9e1
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 18:42:12 2013 -0400

    split init_syscalls into uni/biarch

commit ed8aa13cba513703fbee5f33fe7b7e5d7bb1bd09
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 18:29:19 2013 -0400

    split mark_all_syscalls_active into uniarch/biarch

commit ddae190e6a41afa9f7d0de55412bbc91697c21d4
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 18:25:27 2013 -0400

    split setup_syscall_group into uniarch/biarch

commit b14608eb78d86038a0be42e0e8922bcad636e227
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 18:19:20 2013 -0400

    split disable_non_net_syscalls into uni/biarch variants

commit 04c47b9fffdec7de374bf848a727ce0c4096afc9
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 18:13:05 2013 -0400

    split tables.c into uniarch and biarch variants

commit 3ead8f1b2c3d63ca6e5f7dc9396ec7356e3d7805
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 16:02:14 2013 -0400

    more test-multi tweaks
    
    silence the pushd/popd
    chmod the /tmp.... , not ../tmp...

commit cfcd2be96990020a53e37c75e67b91692d19a47e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 15:58:49 2013 -0400

    Store old path in test-multi.sh
    
    we need to know where to copy the trinity executable from

commit d293785631c4c456d95a142e15fec3022a350bed
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 15:00:19 2013 -0400

    split enable_random_syscalls in two
    
    this function is a mess, and breaking it apart into pieces is the only way
    it's going to get better.

commit 18d50bf9fa31bd0793021cdd2737d9001af97dbd
Author: Ildar Muslukhov <ildarm@google.com>
Date:   Tue Oct 8 14:32:45 2013 -0700

    converted debugf from macro to a function
    
    This patch replaces debugf macro with a function.
    
    Signed-off-by: Ildar Muslukhov <ildarm@google.com>

commit b941dd2a2f4d58fd944af0f05d779bc6f2136b72
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 10 14:21:09 2013 -0400

    remove redundant check.

commit 31df1ae2f4a9b455e2f9a2df4f365b8e38a7faa5
Author: Anders Roxell <anders.roxell@linaro.org>
Date:   Thu Oct 10 10:41:15 2013 +0200

    tables.c: change the output format of -L
    
    So the scripts can works for both biarch and !biarch
    
    Signed-off-by: Anders Roxell <anders.roxell@linaro.org>

commit a16554a045149af795557660353c7b07ee7c5dff
Author: Anders Roxell <anders.roxell@linaro.org>
Date:   Wed Oct 9 12:56:25 2013 +0200

    scripts: change mkdir tmp to mktemp
    
    Signed-off-by: Anders Roxell <anders.roxell@linaro.org>

commit e224771c6303e2f420902390c47596d63a9dcd2c
Author: Anders Roxell <anders.roxell@linaro.org>
Date:   Wed Oct 9 12:56:24 2013 +0200

    scripts: TRINITY_PATH defaults to "."
    
    TRINITY_PATH may be changed in the enviroment
    
    Signed-off-by: Anders Roxell <anders.roxell@linaro.org>

commit b5d853dc37f214b3ba55de434e82454d8dd5c6cb
Author: Ildar Muslukhov <ildarm@google.com>
Date:   Wed Oct 9 10:32:17 2013 -0700

    Added and wired bufferless loging function for syscall parameters
    
    Reworked color_arg function call and output into a bufferless approach in order to fix stack smash detected bug. Moved the whole logging logic to log.c file (seems to be more appropriate place for it)
    
    Signed-off-by: Ildar Muslukhov <ildarm@google.com>

commit 2e9fdedbc54582a94e25d28e2d8fe8b33f3fd384
Author: Ildar Muslukhov <ildarm@google.com>
Date:   Wed Oct 9 10:32:16 2013 -0700

    wired in output function instead of printf (and some missing outputstd)
    
    This patch replaces the remaining printf with output function call. Some of the printf are replaced with outputstd.
    
    Signed-off-by: Ildar Muslukhov <ildarm@google.com>

commit 37ff9885576d1f93594a91296582c52ff7f8b3ff
Author: Ildar Muslukhov <ildarm@google.com>
Date:   Wed Oct 9 10:32:15 2013 -0700

    refactored output function
    
    This patch refactors output function and adds boundary checking to avoid overflow.
    
    Signed-off-by: Ildar Muslukhov <ildarm@google.com>

commit ca939c8f7346b2474b28a4500749af584df51780
Author: Ildar Muslukhov <ildarm@google.com>
Date:   Wed Oct 9 10:32:14 2013 -0700

    wired outputstd/err functions
    
    This patch replaces printf call with outputstd or outputerr functions where appropriate.
    
    Signed-off-by: Ildar Muslukhov <ildarm@google.com>

commit 04f12d2d4c2fdce52bd1d3527bb91f326066b49f
Author: Ildar Muslukhov <ildarm@google.com>
Date:   Wed Oct 9 10:32:13 2013 -0700

    added outputerr/outputstd log functions
    
    This patch adds two function that is going to be used for output to stdout and stderr. This is done in order to have printf called from one file.
    
    Signed-off-by: Ildar Muslukhov <ildarm@google.com>

commit fcbde1c14e7257a51a8edb00490dc4f89d7fcb0e
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 7 20:03:03 2013 -0400

    add missing reason string.

commit 4dcc6cde3c195796bc9e15319eedd99d3f756938
Author: Ildar Muslukhov <ildarm@google.com>
Date:   Fri Oct 4 11:29:41 2013 -0700

    Fix to broken random selection of syscalls.
    
    This patch fixes random selection of system calls, introduced in the
    earlier commit.
    
    Signed-off-by: Ildar Muslukhov <ildarm@google.com>

commit 0115860dd6d51eec2bcc493454a6a6ed7fadeee0
Author: Jiri Slaby <jslaby@suse.cz>
Date:   Sat Oct 5 09:33:48 2013 +0200

    fix broken includes
    
    In 3.0, we get errors like:
    /usr/include/linux/netlink.h:34: error: expected specifier-qualifier-list before 'sa_family_t'

commit 16afb681a470e632b7d52b5879c2e6a556efd302
Author: Jiri Slaby <jslaby@suse.cz>
Date:   Sat Oct 5 09:33:47 2013 +0200

    net/netlink: add missing compat.h include
    
    On 3.0, we get:
    net/netlink.c:32: error: 'NETLINK_CRYPTO' undeclared (first use in this function)

commit 4cdd41bce7591f7384cdd81491334d775b9fa067
Author: Jiri Slaby <jslaby@suse.cz>
Date:   Sat Oct 5 09:33:46 2013 +0200

    compat: define some net constants
    
    To be able to build for 3.0.

commit e7fc5569209caf35802811cf989f15376422dcb9
Author: Dave Jones <davej@redhat.com>
Date:   Sat Oct 5 11:38:54 2013 -0400

    remove unnecessary assignment
    
    This actually prevents the watchdog exiting in some cases.

commit 50dadfd4f48e82d2c713e116bc1240caa0eef7a6
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 4 18:06:38 2013 -0400

    on nftw failure, print the errno string, instead of just '-1'

commit 4cc09d9e4ccdd0db51361e843d7db1ec1cfbfac6
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 4 17:29:31 2013 -0400

    clarify comment

commit 5521490b4b338f801031c5f1f5de4309f82ba465
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 4 12:37:22 2013 -0400

    fix indentation.

commit 2c8514347247c5d38a7b3b167d3f52c284458fe8
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 4 11:50:54 2013 -0400

    check if uid changes across syscalls.

commit da50e86e649f60cc819301d52c582720f80e3d51
Author: Ildar Muslukhov <ildarm@google.com>
Date:   Thu Oct 3 16:09:55 2013 -0700

    fix arch selection parameter
    
    This patch fixes arch (-a) selection parameter when applied to all
    syscall.
    
    Signed-off-by: Ildar Muslukhov <ildarm@google.com>

commit 1f54a078b7ed87d547f0cf1c7e1f6d79006e0b7c
Author: Ildar Muslukhov <ildarm@google.com>
Date:   Thu Oct 3 15:45:21 2013 -0700

    Re-add checks for AVOID and correct syscall activation.
    
    This patch fixes commit 6e296f3a7cab6d946479e76398535f107e80ac4d, that is
    it adds back checks for AVOID types of syscalls and corrects syscalls
    deactivation.
    
    Signed-off-by: Ildar Muslukhov <ildarm@google.com>

commit 8c6e68d78008b1870f93816a5b010823e4a1e44a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 3 14:18:33 2013 -0400

    build in sorted order
    
    Kinda OCD, but gives an idea how far along a build is on slow build machines.

commit d4363ee8908b7ce84f2cd00921d8ecfde71c9017
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 3 13:51:30 2013 -0400

    fix incorrect syscall table usage on non-biarch
    
    Spotted by Josh Scheid <jscheid@velocetechnologies.com>

commit 31b4121c196aaadce8b4b585b86294fa5afb94d2
Author: Ildar Muslukhov <ildarm@google.com>
Date:   Wed Oct 2 16:46:41 2013 -0700

    Fixes weird looping in child_random_syscalls function.
    
    This patch removes makes the loop in child_random_syscalls aware of which
    system calls are enabled, thus, avoiding unnecessary looping.
    
    Signed-off-by: Ildar Muslukhov <ildarm@google.com>

commit a57043bdb8f0cda7962bb0db7bf49a1fe9ddc1ea
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 3 00:51:04 2013 -0400

    limit setuid params to sizeof(uid_t)

commit 5f76c4eb193e2f44c03df3eb28c0f19d2820c2cc
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 3 00:46:22 2013 -0400

    argument annotation for setns

commit 05a1a0a21cdbf64efdbaed06a24247d0f2daaca1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 2 15:46:58 2013 -0400

    remove duplicate strings
    
    These are unnecessary now that output does it for us

commit 4fd99d83311ae7b7c5a28176d7935bd58e02a19e
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 2 13:37:31 2013 -0400

    rand_page: return page of format strings.

commit 3672c4d619f4ec4d6406039922788ef1e9d27180
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 2 11:50:08 2013 -0400

    move the prefix printing into output itself.
    
    small bug right now: Split calls to output() without a newline get
    two prefixes logged. Will fix later.

commit 369dfa3c0fabdc49c7eda6eb145bdbb143bf6911
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 2 10:44:55 2013 -0400

    Fix logging when -q is used.
    
    The COLOR_ARGS conversion broke this combination, leading to log files containing
    things like..
    
    [24966] [10] linkat) = -1 (No such file or directory)
    [24966] [11] rt_sigsuspend) = -1 (Invalid argument)
    
    Add a comment explaining the intent, and further optimise by skipping over
    the output call too.

commit 856e0b5acb2f830d1f9380b3092770311fe00c20
Author: Ildar Muslukhov <ildarm@google.com>
Date:   Tue Oct 1 10:54:39 2013 -0700

    convert COLOR_ARG macro to a function
    
    Signed-off-by: Ildar Muslukhov <ildarm@google.com>

commit 9d949ebe0d681aa12602046d03a04f5ecef1d0ec
Author: Dave Jones <davej@redhat.com>
Date:   Mon Sep 30 13:33:02 2013 -0400

    reword parameter text for specifying whether to do 32/64 bit syscalls

commit 04ddbabd1cb270a3f677cc20367dab8ee72d9343
Author: Dave Jones <davej@redhat.com>
Date:   Fri Sep 27 17:39:07 2013 -0400

    change the watchdog sigkill usage.
    
    kill off the 60s counter, and instead use a 'kill count'.
    After 30s, we send 1 kill per second, which seems more succesful.
    
    (With -C64, we still see high kill counts before success, but we do
     eventually seem to make forward progress.)

commit 336e1330ca5051fa13ae2c28672aca5352b98862
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 26 17:42:39 2013 -0400

    add missing brackets to disambiguate

commit 0c5c830b9cf53fc0a2003c1435eaf4e6f4859373
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 26 12:15:19 2013 -0400

    fix up corner case where trinity binary got accidentally chmod'd.

commit 8936dfdf541744dbc82d6131192ce59f7429fce3
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 26 12:14:16 2013 -0400

    when we get back from the signal handler, print what happened in the child.
    
    Ignore SIGALRM, it'll happen too often.

commit 46f6ce9818dd0a7f9a1a3ba1ee8280c893d43039
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 26 12:12:35 2013 -0400

    print out pid in ENOSYS output

commit 8b16ff9e6c3fd2d3a89d7faa6e59721a5144a38b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 26 12:11:31 2013 -0400

    signal handler improvements.
    
    When the alarm goes off, check if we're stuck in an fd.

commit 00ccd6c964219463dc6d7684ff43ca223c2448df
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 26 12:09:15 2013 -0400

    check return from kill, just in case

commit d0dc7b12a97256a67d1e27528dc8873491008b4c
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 26 12:08:28 2013 -0400

    factor out the stuck syscall info to own function

commit ee98797ca94b2952a3eadb36d828d9574e3c6b42
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 26 12:06:38 2013 -0400

    print out the fd as an int.

commit fd60190a184c1b12753072cec788496f9e7a5718
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 26 12:04:38 2013 -0400

    remove pointless local var

commit 689073fb779fe7cdd95a8c91f23447dc7366a812
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 26 12:00:56 2013 -0400

    remove callno param from check_if_fd

commit 2a588d6b684c29949fc34f34526c38c29ed9766e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 26 11:56:20 2013 -0400

    We should be safe to do 32-bit syscalls from 64bit again.

commit f0e36b0533739ef2221027bc3454921170296fc3
Author: Ildar Muslukhov <ildarm@google.com>
Date:   Mon Sep 23 19:41:14 2013 -0700

    Added parameter that controls which taint flag causes trinity to stop.
    
    This is reworked patch for the parameter that allows to control which
    kernel taint flags causes trinity to stop. In particular, now the short
    parameter is -T (instead of -t) and flags has to be specified as literals
    (e.g., DIE, WARN, USER, etc.). All 13 supported flags are specified in
    README.
    
    Signed-off-by: Ildar Muslukhov <ildarm@google.com>

commit ca5f169a2d72a4a2d3cfbe9e5fbdbcd0763105c3
Author: Ildar Muslukhov <ildarm@google.com>
Date:   Thu Sep 19 15:52:11 2013 -0700

    Reduce the number of pages allocated in copy_syscall_table function.
    
    This patch reduces the number of pages allocated in the copy_syscall_table
    function.
    
    Signed-off-by: Ildar Muslukhov <ildarm@google.com>

commit 5984d37b47583ed343933edb3c0dfee1e94cf4c9
Author: Ildar Muslukhov <ildarm@google.com>
Date:   Thu Sep 19 14:24:24 2013 -0700

    Fix syscall32 to supply 6th arg to syscal
    
    This patch fixes missing 6th param for 32bit syscalls in i386 and x86_64
    archs.
    
    Note: I have to remove static from syscall32, because gcc messes up with
    inline asm becuase of that.
    
    Signed-off-by: Ildar Muslukhov <ildarm@google.com>

commit 284d0ca1a9a750d38188db9478c885a163c06d32
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 17 15:32:14 2013 -0400

    make handle_arg_randpage copy the page_allocs page instead of passing it.
    
    also fix broken indentation.

commit a595f3ea00aa770aac3f7229b53303123ee99f68
Author: Ildar Muslukhov <ildarm@google.com>
Date:   Mon Sep 16 15:05:56 2013 -0700

    Selection of 32 and 64 bit variants of syscalls
    
    This patch adds parameter that enables explicit selection of either 32 or 64 bit version of syscall(s).
    
    Signed-off-by: Ildar Muslukhov <ildarm@google.com>

commit f4017be9058c919c8d87eb4513ddcff47c22e3eb
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Tue Sep 17 15:11:57 2013 -0400

    update perf_event support for 3.12-rc1
    
    This patch wires up PERF_COUNT_SW_DUMMY and PERF_SAMPLE_IDENTIFIER
    which appear in 3.12-rc1
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit de10f80281c994969f3a0fe4f4cf6a92693e2543
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 17 11:54:44 2013 -0400

    more thoughts on binary logging

commit bad589667f56a55dbf2b78da05c1d3136e8b0714
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 17 11:01:20 2013 -0400

    TODO: --fd param

commit 573e039c357036275a47a1d6e52cdb842b63b216
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 17 10:58:06 2013 -0400

    idea for binary logging at some point.

commit e812f24e5ba61f7b77a1e5c2ffa4da13ada71d25
Author: Dave Jones <davej@redhat.com>
Date:   Mon Sep 16 14:21:47 2013 -0400

    change how we use the lower 32bits when picking 64bit interesting nums

commit 12fa238aff9c24aaaa2432ac15d85f8c52420c5d
Author: Dave Jones <davej@redhat.com>
Date:   Mon Sep 16 14:19:14 2013 -0400

    some new 32bit 'interesting' numbers

commit ba56072a09bcf6fa91f867e0383350638717972e
Author: Will Deacon <will.deacon@arm.com>
Date:   Fri Sep 13 13:56:31 2013 +0100

    add support for the AArch64 architecture
    
    This patch wires up support for the AArch64 architecture, which has been
    supported in the mainline kernel (as arm64) since 3.7.
    
    Signed-off-by: Will Deacon <will.deacon@arm.com>

commit 8c09dd46261381d34b57a159f69e30d7b04289b1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 11 10:31:48 2013 -0400

    clean up coverity stuff before running

commit bc38ec5312aa7fd7d5707a757ab626fbe90801b6
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 11 00:50:17 2013 -0400

    add some comments

commit aa6c50867dfb873ba1fd86c61dacc639b81cdb51
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 10 21:21:23 2013 -0400

    rand64: bits are numbered 0-63, so we need to subtract one.

commit 73ffeb0aa79a1219dccef3edb19016bcfc132ec7
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 10 21:19:09 2013 -0400

    rand64: we want a bit number here, not a value.

commit 3ca53cd95abd09aaa1873a22eb18e3a99e68ae1a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 10 19:24:33 2013 -0400

    bugs update

commit 0ca49066e432fefba1887a47810e9d47db337d8c
Author: Will Deacon <will.deacon@arm.com>
Date:   Tue Sep 10 13:43:04 2013 -0400

    wire up support for the Alpha architecture
    
    Will Deacon has more exotic hardware than I do: he sent this patch
    against my perf_fuzzer to add Alpha support (he already used it to find a
    kernel bug).
    
    Anyway since I use the trinity headers, the patch applies equally well
    against trinity.
    
    Signed-off-by: Will Deacon <will.deacon@arm.com>

commit 98c413417a6ce264d570a4d9766e8981f514ac2e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 5 13:51:41 2013 -0400

    attribution

commit 989e5d8746a69676416be338b160f0471ee7b3ca
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 5 12:02:31 2013 -0400

    Add MAP_32BIT for x86-64 to mmap flags

commit d61fc5993a674716cea089c166d34f43385ae94e
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Tue Sep 3 15:08:00 2013 -0400

    update perf_event_open() for Linux 3.11 release
    
    Update the perf_event_open syscall support for new ABIs added
    in the 3.11 release.
    
    Thankfully this is pretty minor.  Lots of big changes to the ABI
    seem to be afoot for 3.12 though.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 5775f864d812313b445f5305149d55a5a16caa63
Author: Dave Jones <davej@redhat.com>
Date:   Fri Aug 30 13:39:25 2013 -0400

    perf may not be available
    
    We had an infinite loop here if we got an ENOSYS back. oops.

commit 9091fccbacc503b4dc47f58b02c1e0da92505a85
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 29 17:37:05 2013 -0400

    generate eventfd fd's on startup

commit 5b9e7228d165a08e86c3afaca10a99dcdd1608c1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 28 21:27:56 2013 -0400

    add epoll fd's on startup

commit 5913b86cd8c1176f8eed172c075f9cb98c6b75d6
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 28 21:04:46 2013 -0400

    crazy idea: actually use those shared perf fds.
    
    derp.

commit d74a6ff0e9dc2d581ad9b597a47cdbc1d8e268b7
Author: Vinson Lee <vlee@twitter.com>
Date:   Fri Aug 23 20:04:52 2013 -0700

    can.c: Include compat.h.
    
    This patch fixes these build errors with glibc before 2.10.
    
    net/can.c: In function ‘can_gen_sockaddr’:
    net/can.c:17: error: ‘AF_CAN’ undeclared (first use in this function)
    net/can.c:17: error: (Each undeclared identifier is reported only once
    net/can.c:17: error: for each function it appears in.)
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit ba9ba17ae1dfa656343a16fe5af433c5b8bbd445
Author: Vinson Lee <vlee@twitter.com>
Date:   Fri Aug 23 19:10:32 2013 -0700

    Add symbols from sys/mount.h for older glibc.
    
    MNT_DETACH and MNT_EXPIRE are not available with glibc before 2.10.
    
    UMOUNT_NOFOLLOW is not available with glibc before 2.12.
    
    This patches fixes these build errors with older glibc.
    
    syscalls/umount.c:16: error: ‘MNT_DETACH’ undeclared here (not in a function)
    syscalls/umount.c:16: error: ‘MNT_EXPIRE’ undeclared here (not in a function)
    syscalls/umount.c:16: error: ‘UMOUNT_NOFOLLOW’ undeclared here (not in a function)
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 44e3a632d4a429a7d2a735c26ab7e277ef28b463
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 28 13:46:39 2013 -0400

    remove dead prototype

commit 4ed32cf9663b68ca7a4990346a1f0d24cfc168ef
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 28 13:26:11 2013 -0400

    generate some perf fds in the parent for sharing.

commit 36b69e2db113087e536310b3b666a6eaeeebaa8b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 28 12:51:27 2013 -0400

    reduce the amount of casting in check_if_fd

commit 249ec0834e8acce2a8e916f17f2a700efcbbd4b5
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 28 12:48:14 2013 -0400

    disable that 'Chose' output by default until we have more children types

commit 5737f5d473aa624ff51b99ed3d1c6f5ade1acab7
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Fri Aug 23 13:56:51 2013 -0400

    perf_event_open() branch sample support
    
    The following patch exercises some of the recent additions to the
    perf_event_open() syscall.
    
    The PERF_SAMPLE_WEIGHT and PERF_SAMPLE_DATA_SRC sample types were
    added in Linux 3.10.
    
    It also adds more complete support for PERF_SAMPLE_REGS_USER which was
    added in Linux 3.7
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit a9e2fa00f320be159e7380942a077731661b5fb9
Author: Dave Jones <davej@redhat.com>
Date:   Fri Aug 23 12:12:33 2013 -0400

    Fix -P usage, broken by recent reworking of socket generation.

commit bde1d56e99424d452e4a6be8a256999b07d3ffb9
Author: Vinson Lee <vlee@twitter.com>
Date:   Thu Aug 22 18:07:27 2013 -0700

    timerfd_create: Include time.h instead of sys/timerfd.h.
    
    This patch fixes these build errors with glibc before 2.8.
    
    syscalls/timerfd_create.c:4:25: error: sys/timerfd.h: No such file or directory
    syscalls/timerfd_create.c:15: error: ‘CLOCK_REALTIME’ undeclared here (not in a function)
    syscalls/timerfd_create.c:15: error: ‘CLOCK_MONOTONIC’ undeclared here (not in a function)
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 8361603cecc4de0eb18a83f8a04b743ff68e2573
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 22 18:03:46 2013 -0400

    rework the socket generator a little to give more varied output.
    
    cycling through all available address families should give us more assortment.

commit 323fee3803d3b4003f07a7e6ba8a14a715da9b3d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 22 15:07:55 2013 -0400

    clean up the cachefile locking some.

commit 0f7aa1c1624be55e25ede463dbb76baae6e4f45b
Author: Vinson Lee <vlee@twitter.com>
Date:   Wed Aug 21 21:22:57 2013 -0700

    compat.h: Add SOCK_CLOEXEC and SOCK_NONBLOCK for older glibc.
    
    This patch fixes these build errors with glibc before 2.9.
    
    syscalls/socket.c: In function ‘gen_socket_args’:
    syscalls/socket.c:82: error: ‘SOCK_CLOEXEC’ undeclared (first use in this function)
    syscalls/socket.c:82: error: (Each undeclared identifier is reported only once
    syscalls/socket.c:82: error: for each function it appears in.)
    syscalls/socket.c:84: error: ‘SOCK_NONBLOCK’ undeclared (first use in this function)
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit c1dcc4a2e4e345390299e6cad47b6b39e01ef98f
Author: Will Deacon <will.deacon@arm.com>
Date:   Wed Aug 21 11:36:08 2013 +0100

    ARM: tidy up arch-specific constants
    
    arch-arm.h is basically a copy of arch-i386.h, but there are differences
    between the two, so tidy up the ARM header to match exactly how the
    kernel is structured.
    
    Cc: Russell King <linux@arm.linux.org.uk>
    Signed-off-by: Will Deacon <will.deacon@arm.com>

commit b4d35a6d8206c6d3dc1b671932a5416c4212dfd9
Author: Vinson Lee <vlee@twitter.com>
Date:   Fri Aug 16 19:26:30 2013 -0700

    compat.h: Add more address family symbols for older glibc.
    
    This patch fixes these build errors with glibc before 2.10.
    
    syscalls/socket.c:27: error: ‘AF_CAN’ undeclared here (not in a function)
    syscalls/socket.c:33: error: ‘AF_LLC’ undeclared here (not in a function)
    syscalls/socket.c:38: error: ‘AF_PHONET’ undeclared here (not in a function)
    syscalls/socket.c:39: error: ‘AF_RDS’ undeclared here (not in a function)
    syscalls/socket.c:40: error: ‘AF_TIPC’ undeclared here (not in a function)
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit b7453d650f9cee38d330f304d6f162d12e787898
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 14 21:06:48 2013 -0400

    convert pathname code from switch to an if, as there are only two cases.

commit 06dc212849b831c5f4956cb39ae4a5cbc69b2093
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 14 21:03:27 2013 -0400

    minor fixes to random pathname code.

commit 3a18674abc36834ab3d64e668fa11699ffc4a44f
Author: Dave Jones <davej@redhat.com>
Date:   Sat Aug 10 16:33:53 2013 -0400

    add a case where random pathname can be ""
    
    brought on because of this kernel commit:
    "fs: Allow unprivileged linkat(..., AT_EMPTY_PATH) aka flink"

commit 2ee48f098f571545fdb0ff1f492008989052612f
Author: Dave Jones <davej@redhat.com>
Date:   Sat Aug 10 16:28:19 2013 -0400

    move random pathname code out to own file

commit 28c446dadaf3a6a2e358a8847d40e5ce7b80690e
Author: Dave Jones <davej@redhat.com>
Date:   Sat Aug 10 16:24:04 2013 -0400

    add AT_FDCWD special case to linkat

commit 744e34a229c067d3998fa3261be34f52b63e9890
Author: Dave Jones <davej@redhat.com>
Date:   Sat Aug 10 16:18:10 2013 -0400

    these are pathnames, so we can annotate them as such.

commit c526d0b37e0a6289eff7d4f90fbfa5ac8b47d11d
Author: Vinson Lee <vlee@twitter.com>
Date:   Thu Aug 8 15:01:14 2013 -0700

    compat.h: Add MSG_CMSG_CLOEXEC for older glibc.
    
    This patch fixes this build error with glibc before 2.7.
    
      CC    syscalls/recvfrom.o
    syscalls/recvfrom.c:28: error: ‘MSG_CMSG_CLOEXEC’ undeclared here (not in a function)
    make: *** [syscalls/recvfrom.o] Error 1
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 23eb7370c6bfc1e52c26b61f4b8a4043a7baa903
Author: Vinson Lee <vlee@twitter.com>
Date:   Wed Aug 7 17:26:14 2013 -0700

    net/caif.c: Include compat.h.
    
    This patch fixes this build error with glibc before 2.13.
    
      CC    net/caif.o
    net/caif.c: In function ‘caif_gen_sockaddr’:
    net/caif.c:22: error: ‘PF_CAIF’ undeclared (first use in this function)
    net/caif.c:22: error: (Each undeclared identifier is reported only once
    net/caif.c:22: error: for each function it appears in.)
    make: *** [net/caif.o] Error 1
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit f4cb80e2981dee1b7d690a5c242d73029f454b97
Author: Vinson Lee <vlee@twitter.com>
Date:   Wed Aug 7 17:17:14 2013 -0700

    net/alg.c: Include compat.h.
    
    This patch fixes this build error with glibc before 2.13.
    
      CC    net/alg.o
    net/alg.c: In function ‘alg_gen_sockaddr’:
    net/alg.c:21: error: ‘PF_ALG’ undeclared (first use in this function)
    net/alg.c:21: error: (Each undeclared identifier is reported only once
    net/alg.c:21: error: for each function it appears in.)
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 904b4ed45475c8c774905b2ce471b03fbed3903f
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 1 16:58:24 2013 -0400

    add a second child type. (read all files)
    
    This is disabled for now, because it's pretty dull, and there is
    some more groundwork to do before we can have multiple child types running.

commit 9992ad1b8a4bdaad545373b75ffbf3603f9c6972
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 1 00:25:53 2013 -0400

    Create a directory for all the child processes

commit 00962606aab65dfba77a8d778da430d55a0cdf0d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 1 00:21:54 2013 -0400

    Introduce child process types.
    
    'do random syscalls' is just one type. (And right now, the only one).
    
    More to follow.

commit 95bfd8dd939b65a74bec774cd33e52bb0caac86b
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Sat Jul 27 17:35:16 2013 +0300

    Do not busy loop child processes on startup
    
    Stop busy looping the child processes on trinity startup, so that they
    do not compete with the main process for CPU time.

commit d27fb192f85d3aa9405e655e55352c06a04b074e
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Sat Jul 27 17:35:15 2013 +0300

    Fix child crash when disabling some syscalls on biarch
    
    Check if search_syscall_table() fails and returns -1. This avoids
    crashing the child processes, that I was seeing on x86-64 with the
    msgrcv() syscall.

commit 5cf997201ce2c634c11bb80cbcdb870735357861
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 31 22:14:17 2013 -0400

    annotate flags for sys_timer_settime

commit a1312d382650c28b42e32ccea1cf5b82124d7d11
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 31 22:10:43 2013 -0400

    annotate flags for umount

commit 4e89292e8a64af4d16112b8fffabd9ee55e849a0
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 31 22:04:58 2013 -0400

    fix off by one in ARG_LIST number of bits.

commit 26b31b854854cfd1e3e0390c301268400a67b388
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 31 21:59:04 2013 -0400

    annotate flags for utimensat

commit cbff25be80ba1dd61801cf9b59fc9ed4347ac19c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 31 21:55:35 2013 -0400

    add CLONE_NEWUSER/CLONE_NEWPID to unshare flags

commit 1fa30227a0a757483c352e2b50120d760bfac855
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 31 20:39:12 2013 -0400

    Add FALLOC_FL_NO_HIDE_STALE fallocate flag.

commit 4913d4a2c2fe7321fb4fcee9e76346e4ad9b9b35
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 30 12:24:01 2013 -0400

    fix another use-after-free in devices.c
    
    I missed one variant in 0d6921aa64dafa990cdc6fcbdf1589a8ace82a69

commit 22760c09cb76fd77060b2e765f3ea4d31dc4f9b6
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 25 23:59:47 2013 -0400

    Fallback to random proto/type if we haven't got a socket->func yet.

commit ac9590828baca531b35080f0752067f848857bd6
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 24 22:17:36 2013 -0400

    remove unnecessary allocation

commit c9acfc3111a3cce6de7abe09b8173226ff2a018c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 24 21:01:21 2013 -0400

    move the 'change to tmp/' code out to its own function.
    
    Coverity #1042334

commit 8f74b2cf0190824a91b5cbda5316c52a506c601b
Author: Andi Shyti <andi@etezian.org>
Date:   Wed Jul 24 12:31:05 2013 +0200

    ioctls: btrfs-control: fix header inclusion order
    
    This patch enables newer linux/btrfs.h in Kernel headers 3.11,
    where NULL is not recognized as a define. This is done by
    including  stdlib.h before linux/btrfs.h. It gets rid of the
    following compiling error:
    
    In file included from ioctls/btrfs-control.c:5:0:
    /usr/include/linux/btrfs.h: In function ‘btrfs_err_str’:
    /usr/include/linux/btrfs.h:486:11: error: ‘NULL’ undeclared (first use in
        return NULL;
               ^
    
    Signed-off-by: Andi Shyti <andi@etezian.org>

commit 560093ff84e7f67ab9473b0478e22347f47a84c9
Author: Andi Shyti <andi@etezian.org>
Date:   Wed Jul 24 00:42:07 2013 +0200

    ioctls: videodev2: check if defines are obsolete
    
    In kernel headers 3.11 the VIDIOC_DBG_G_CHIP_IDENT define is
    considered obsolete, therefore removed, generating the following
    compiling error:
    
    ioctls/videodev2.c:72:15: error: ‘VIDIOC_DBG_G_CHIP_IDENT’ undeclared here (not in a function)
      IOCTL(VIDIOC_DBG_G_CHIP_IDENT),
                   ^
    
    Check the definition of it before using it
    
    Signed-off-by: Andi Shyti <andi@etezian.org>

commit 186bb446290bed94e5297d517993267a5571227c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 24 20:46:50 2013 -0400

    fix potential use after free if memory alloc fails.
    
    Coverity #1042351

commit b02de1236e747dd0992038ef3b460eb7be113ab6
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 22 19:37:25 2013 -0400

    ..and begin 1.3

commit 124fa6f380e55b7f58d171da92875cdb23288b06
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 22 19:35:31 2013 -0400

    1.2 release.

commit cda821ac2e53f12d7bc289ab83fe71f6a3bf14b5
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 22 19:35:08 2013 -0400

    Disable -Werror

commit 95acd49851fde0e3e302092ab531bb5a127e7f99
Author: Andy Honig <ahonig@google.com>
Date:   Fri Jul 19 14:39:15 2013 -0700

    improve randomness in rand64()
    
    Commit 6112813df53ce6762f5c8de1d4382d67f9f2f769 introduced
    code to occasionally invert the valuerand64().  However the code
    mistakenly zeroed the value instead by xoring it with itself.
    
    Signed-off-by: Andy Honig <ahonig@google.com>

commit 6f8f150c356d89e418d17fb578a9f7e0c7c093ab
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 19 14:35:21 2013 -0400

    add a coverity helper target

commit e0f83bc5a9462349aa0ac51186a0ab16b9144a6f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 19 14:32:33 2013 -0400

    more 'should never happen' bulletproofing.

commit 1c90d4bf5ca59b9471e257b672114a988b70e496
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 19 14:28:21 2013 -0400

    ARG_OP has the same issue.

commit 851ad2f1a4b0c5614565a41b5bd8b1ff50b9fa90
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 19 14:26:45 2013 -0400

    Fix potential div0 in ARG_LIST handling.
    
    appease coverity.
    
    In theory there's a divide by zero here if we ever had syscall arg
    annotated as ARG_LIST, with 0 elements.  Hopefully I'm not dumb enough
    to create such an entry, but just in case..

commit fe8a9d2fa00a6be64372927ea8bbb9c1c8717999
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 19 13:56:17 2013 -0400

    Add PTRACE_GETSIGMASK and PTRACE_SETSIGMASK
    
    ptrace still disabled, but adding for completeness.

commit d4f1139ccf4da72fb50e7015f532355e83d6b098
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 18 18:33:11 2013 -0400

    Move socketcall to use a table of function ptrs instead of a big switch.

commit b13f73cbb2282c15b8328dd7eeef8b311b6c477a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 18 17:53:53 2013 -0400

    kill off struct proto_type
    
    Seems pointless given we have the triplet

commit fa141a7e2d9088dfc772e8bb530bd7d65386480c
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 18 17:43:49 2013 -0400

    Fix up i386 socketcall segfault.
    
    This was an ugly bug.
    Occasionally the watchdog would segv after trying to print info
    about syscall number -1.  It happened because socketcall was trying
    to lookup "socket", which doesn't exist on i386.
    
    Instead, we repurpose the sanitise function from socket into a
    function that generates family/type/protocol triplets.

commit 977603b0ee74c742d1a5619d042a52a04fb9d3d4
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 18 17:15:36 2013 -0400

    Sanity check to catch a rare segv

commit 947f5f56cc6b1dbe463a02a0b34ea93ebb087c9e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 18 14:19:04 2013 -0400

    add special case for 'main dead' to bail the watchdog loop.

commit 8ac8e538edcbc5aa636e0cda9ff48e190f3ee43b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 18 14:15:58 2013 -0400

    If the 'main' process crashes during regeneration, avoid hanging the watchdog

commit b33983228ac3536c8dc7bd55aed7c594af695071
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 18 14:11:25 2013 -0400

    move exit related stuff out of the watchdog function.

commit 746a63f632a900232d914e359da6078db2953128
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 18 14:09:30 2013 -0400

    move function to avoid need for prototype

commit 19ed6917382a49f1712cbc1c5af02a45598ce68e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 18 14:08:35 2013 -0400

    move watchdog_pid out of shm, take 2.
    
    This way should work better.

commit cf5f6ea034f15871466624c3a309b90bf3a7e1d5
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 18 13:52:30 2013 -0400

    Revert "Move the watchdog pid out of the shm too."
    
    This reverts commit b1a2fa9391cd2e8bf40601ee2df3b374ab846d2a.
    
    This is nonsense, of course it needs to be in the shm.

commit 37e4344e0ccc202d275a0323991e9c7f3351799c
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 18 13:33:22 2013 -0400

    watchdog should check shm snity & main alive even if we're regenerating.

commit 4db192bfe5a19efac702d1e402b170b8b479790f
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 18 13:30:23 2013 -0400

    Move the watchdog pid out of the shm too.

commit fbdc6d721e3ae020cc324e520dde6ac617daaafe
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 18 13:07:41 2013 -0400

    Split shm->parentpid into two.
    
    In some cases this was 'initial process', and the rest of the time 'forking thread'.
    Split into two so watchdog monitors only the forking process.
    
    This way, if for whatever reason the forking thread dies (usually a segv)
    then we don't hang, because the watchdog will see it's no longer around
    instead of it checking on the init process.
    
    Additionally, these pids don't need to be in the shm.

commit 21f5d34cca8254a2c75500dc5f75d757b79bd6f4
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 17 02:25:34 2013 -0400

    Rewrite the tree-walk permissions checking
    
    It looks like there were some cases where we'd miss a file we could read/write.

commit a6daf66bef05ed36db9e2c60592050cba7fc32c5
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 21:02:43 2013 -0400

    rename ARG_RANDOM_INT to ARG_RANDOM_LONG
    
    This has been bugging me for a while.

commit d5fdc8b84681e9f8f5d64d5eda3e52e8beb479c3
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 20:48:22 2013 -0400

    move ARG_MODE_T handling out to own function

commit 8dfdae220f2840e1662c0d92522a056e10a7ce37
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 20:46:07 2013 -0400

    move ARG_SOCKADDR handling out to own function

commit e5ecb483f5f1427653e21a7e3cce5849a84f5f6a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 20:44:17 2013 -0400

    move ARG_IOVECLEN/SOCKADDRLEN out to own function

commit a5ab93c72bec6624b42eb2f58f9fe064308b34fd
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 20:38:41 2013 -0400

    move ARG_IOVEC handling out to own function

commit 6b05042b0f8e276dcae52d54377260b8641742e9
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 20:36:39 2013 -0400

    move ARG_RANDOM_PAGE out to own function

commit a92eb4f7f1d3a17df491a1a443aa23d484348d5f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 20:35:42 2013 -0400

    move ARG_LIST handling out to own function

commit 0e6d472ee38e0c4124bccfb6660b46c53ea29f43
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 20:34:00 2013 -0400

    move ARG_OP handling out to own function

commit 39990e6efe47f10f8d4b642e33d269b13e6b1fcb
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 20:32:05 2013 -0400

    move ARG_RANGE handling out to own function

commit 03775516c60d160824b027ffa997d3d3c5b20a99
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 20:29:20 2013 -0400

    move ARG_ADDRESS handling to separate function

commit c1ae6bb157bd3c2eea72b64b674003a28dff9d61
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 20:21:46 2013 -0400

    device names can be const.

commit aa968cb1136e8137c81446ae1617fd05e476c012
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 20:19:48 2013 -0400

    use rand_bool everywhere instead of rand() % 2

commit b323968b328c96ac439eeb26dbbdedfac3fa06d6
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 20:06:04 2013 -0400

    add an extra case to the taviso rand function

commit 436837c1e7ac9c9fe74e5b583fe9eabd1f63d1bd
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 20:04:58 2013 -0400

    split random.c into seed.c

commit 6b86f619d8cfdba778d12ba5efd0b9de38e65135
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 19:56:00 2013 -0400

    constify file/path names

commit 37f626f09c6227fcd25b8a786a64d5bec6c2b700
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 19:44:41 2013 -0400

    make get_cpus return something in the range of num online cpus most the time.

commit d30509b8c7174abaf61c6ef747c1683354eab398
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 19:37:53 2013 -0400

    The remainder is more interesting here.

commit 7d38b16d57fe1270ae6f7255342501f143e59a17
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 19:32:38 2013 -0400

    fix inverted logic

commit a7770fe9a36b2b5911c71518c9445a513e0586dc
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 18:57:52 2013 -0400

    socket options are unsigned

commit f8faecfd702a151f3ade06e04804f3d8d5d04d90
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 18:56:27 2013 -0400

    constify socket options

commit 708178727b49f8c26240b1028c5b77c9d1723b6a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 18:51:55 2013 -0400

    constify

commit ba4054a09256e3bd9573174c37f85fc01cd0d5a8
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 18:46:16 2013 -0400

    change sockaddr generation to use a table of func ptrs instead of a huge switch.

commit 5de7b003917620ff90c666106c563588ffc1443c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 18:28:48 2013 -0400

    change socket to use a table of function pointers instead of a huge switch

commit 809ebbd0a98c704a3087203c54f2854cdb0df722
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 18:13:09 2013 -0400

    change setsockopt to use function pointers instead of a huge switch

commit 0461015ac1dd46d83d9199e505285e0b27ff32e6
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 16:12:56 2013 -0400

    rename function to match others from same proto

commit b4988c0b8dc27bd5ccabb0cd746b1aa0eaa0bd7b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 16:10:39 2013 -0400

    reorganise

commit 7daba573b1ec5e3dad3fa192e684118e761afd51
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 16:06:19 2013 -0400

    rename sockaddr generation functions

commit ae89dd22828ff20ad960b2788f7a12d5d83aa929
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 15:56:12 2013 -0400

    silence some llvm warnings

commit 25f5f5466c8c353a807e862e116da31f626a4a88
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 15:43:01 2013 -0400

    remove duplicated code now that all cases are the same.

commit 019c7a25f8a2cb93451f37df2abfd0283be1239e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 15:38:27 2013 -0400

    move nfc setsockopt out to own file

commit 5c43d8fc06dfd6d10dd1305ce30392f642984550
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 15:36:59 2013 -0400

    move alg setsockopt out to own file

commit 55677a07fbde22129bb2010c4fac80bb6c01e7a3
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 15:35:09 2013 -0400

    move caif setsockopt out to own file

commit 931e156df341b2c48c7b47858bae7b19341500ac
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 15:31:54 2013 -0400

    move iucv setsockopt out to own file

commit 22e1e0371320f3304e972ade962f07cb5d536591
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 15:29:43 2013 -0400

    move rds setsockopt out to own file

commit 5d93ced0721c6ca945b9ce00a65d06501e8d2a09
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 15:26:08 2013 -0400

    move pnpipe setsockopt out to own file

commit 7a27c7212a6bfa0693f6fe6601c0da02d714e757
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 15:24:06 2013 -0400

    move bluetooth setsockopt out to own file

commit fbe5b83a5bd340ae673a483792df7518eed119c4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 15:17:20 2013 -0400

    move pppol2tp out to own file

commit 10875ce1ace9bbe0e12933df15aeba16e110d9a6
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 15:12:35 2013 -0400

    move rxrpc setsockopt out to own file

commit 5aab2d714829a378069786ff251a7235c970cfa0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 15:10:27 2013 -0400

    move tipc setsockopt out to own file

commit e571956d5f10d6ec54366181b89d7d2e383888fa
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 15:07:46 2013 -0400

    move netlink setsockopt out to own file

commit 697fb534eea7bacc90c0ff4e2b14a3a5ab7e699a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 15:05:06 2013 -0400

    move dccp setsockopt out to own file

commit dd6eff724b18d5da828bb811154a09290601c5c3
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 15:02:58 2013 -0400

    move llc setsockopt out to own file

commit 1c023b97c30ae571b44e93c9dc0892711180d061
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:59:15 2013 -0400

    move netbeui setsockopt out to own file

commit 1b821fc421f14a96511ac70ecb3568368ea331bd
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:57:17 2013 -0400

    move irda setsockopt out to own file

commit f36101704c89358a9711f734adf0b19c22585713
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:54:59 2013 -0400

    move aal setsockopt out to own file

commit 2d16994d65b9d8ac949308437dd4428e089d78b7
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:53:41 2013 -0400

    move atm setsockopt out to own file

commit d097cf658dc14551863233678e0f772b6e43b7cc
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:50:39 2013 -0400

    move packet setsockopt out to own file

commit 3c2950d1c4d98d3856acc1610b6715acf2da340b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:46:33 2013 -0400

    move x25 setsockopt out to own file

commit 1dc9e9bd201cf9a09e4aadd8dd28a00de87bc2ce
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:43:58 2013 -0400

    move decnet setsockopt out to own file

commit 6d69ee7baee2b12cf03e46a38fe35186f8a319e1
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:41:10 2013 -0400

    move ROSE setsockopt out to own file

commit ab7d8d5f2a111cc05f72871898211b193d7e6744
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:36:37 2013 -0400

    move netrom setsockopt out to own file

commit 54769d59b3e2614af0c30e5dcaa1df355b251e6f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:33:21 2013 -0400

    move appletalk setsockopt out to own file

commit 0d2ff85061574b8b255aebeb8aaca5f3c1d42d95
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:30:59 2013 -0400

    move ax25 setsockopt out to own file

commit 482e75fa321f528703aa8223607135683c28cb87
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:27:53 2013 -0400

    move ipx setsockopt out to own file

commit e9985cde5c3058cf921b673c5cb9dd52d05653b4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:25:36 2013 -0400

    move raw setsockopt out to own file

commit e89e029de0aab8259fc9ee575bd17d6c5a3cbd67
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:23:04 2013 -0400

    move udplite setsockopt out to own file

commit 6acc90bd0980be1734836426110b72767cc500e9
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:19:24 2013 -0400

    fix cut-n-paste typo

commit bee8f464212c1fc4e6666193819f1092769f2d05
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:18:34 2013 -0400

    move sctp setsockopt out to own file

commit 8a3e53dd8c3c9b9b793d8d1d019cdf90c7cbc40c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:16:20 2013 -0400

    move icmpv6 setsockopt out to own file

commit 97fc3a9872c1fbdb2e3464be2817ad41251d3795
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:12:46 2013 -0400

    move ipv6 setsockopt out to own file

commit a1871409ebb2c94b8957cfb220cd4db218192688
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:08:42 2013 -0400

    move UDP setsockopt out to own file

commit 8f31fa6573e00baca4209a8e8da39fa6e641f1c4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 14:04:30 2013 -0400

    move tcp setsockopt out to own file.

commit 952b58de3d7f7d432c74ab3684ea83b15dc3084b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 13:59:35 2013 -0400

    move socket setsockopt stuff out to own file

commit 4994f967520b1ff1d215b188fec271267e380c8d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 13:48:03 2013 -0400

    pass a struct of socketopts around instead of requiring protocols to know shm/syscall internals.

commit 7df4836e4d78d79a2521b82ab38eaa0e0f5c50fa
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 13:31:12 2013 -0400

    remove unnecessary includes

commit 4746eeb14974780328ea28aaaf80b584aaa0de3a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 13:26:34 2013 -0400

    in the default case (unknown family) we still need to set the prototype

commit 9aa040bb26e7f59bd9e09aa13bcd34a863e64da8
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 13:25:15 2013 -0400

    now that all cases are converted, remove the local type/proto args.

commit 5517f0b14a630c04ad5d9c2afa8947ae87c5a74a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 13:22:38 2013 -0400

    x25: move random socket generation into per protocol files.

commit 9eb150c0f1fa600279823ec32d71844b39e64efb
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 13:21:07 2013 -0400

    unix: move random socket generation into per protocol files.

commit 6ba6a45185ea98fb62d11c14ba1f36a8fba8ad29
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 13:19:35 2013 -0400

    tipc: move random socket generation into per protocol files.

commit 5197fab635e469d379063d323bbc1a2f47666b48
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 13:18:00 2013 -0400

    rds: move random socket generation into per protocol files.

commit 5c545ea145146a7902c1ef8810fa179e2c2af4e3
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 13:12:48 2013 -0400

    phonet: move random socket generation into per protocol files.

commit ab968bfe362d006d71479041c6bd379424fab14e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 13:08:26 2013 -0400

    packet: move random socket generation into per protocol files.

commit 8c269d5c046bf38c1a076d1eff9a452fca6e039a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 13:05:50 2013 -0400

    nfc: move random socket generation into per protocol files.

commit 4a981712fffef4996d71653b5b4993deabc4cb3e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 13:03:15 2013 -0400

    netlink: move random socket generation into per protocol files.

commit c36a0893e7b7f7f6ce83b24a44f73c07c1239677
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 13:01:16 2013 -0400

    llc: move random socket generation into per protocol files.

commit 5604d30f1880390d492c281e85293aed7e286f5a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 12:59:01 2013 -0400

    irda: move random socket generation into per protocol files.

commit f680547535e1c230ae0be93381befc69277109f8
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 12:56:24 2013 -0400

    ipx: move random socket generation into per protocol files.

commit aab3046027f6d3059bfde3961726e9fc76bd556d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 12:54:52 2013 -0400

    inet6: move random socket generation into per protocol files.

commit 0c418d0a4547529fdf8845271b9f0acd7210969f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 12:51:59 2013 -0400

    inet: move random socket generation into per protocol files.

commit a3a460d5a6cfcea2888cd87e8a56f8ba50b3fbed
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 12:48:40 2013 -0400

    decnet: move random socket generation into per protocol files.

commit e89bcd3716ca4788896c4f87c6605ae83b1b709f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 12:46:30 2013 -0400

    can: move random socket generation into per protocol files.

commit 9dc57dd8b363ae9da9679234162d637c207b2ec9
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 12:44:24 2013 -0400

    caif: move random socket generation into per protocol files.

commit d26e9d24298e43e3f1c3e24d7533f70720df4785
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 12:41:15 2013 -0400

    ax25: move random socket generation into per protocol files.

commit f4dcf900e724714562c3ff6f7cca7696b4014599
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 12:35:56 2013 -0400

    appletalk: move random socket generation into per protocol files.

commit a5eb588fa6f0bfa2ed03e46f0cb6ec07cb85ef14
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 12:27:33 2013 -0400

    Add AF_IB address family

commit cbac524e59541c722d1a5d02dfcc743d2c98c5b0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 16 12:22:27 2013 -0400

    Add SO_BUSY_POLL setsockopt

commit 9ec7e3c2bf981689c1daa0db69829496cbda4dd6
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Tue Jul 16 00:27:48 2013 -0400

    improved perf_event_open() sysfs parser
    
    The following updates the perf_event_open() sysfs parser to fully
    implement the ABI as described in the kernel ABI documentation.
    
    This includes non-contiguous format field values, as well as support
    for the config2 attribute.
    
    This should fix the warning about commas in the format field.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 8730b126e5d7491e51ce6c6d889e65a6ccd4fc33
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 15 22:56:17 2013 -0400

    Use child offset into array.

commit a912df8699a21535687faf501bc68584ed6a7834
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 15 22:27:43 2013 -0400

    fix compile on 32bit

commit 461f492d00dac9e4dcd5da8e6f140ba6bf0837d6
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 15 22:24:49 2013 -0400

    don't panic about lost pid slots if we reaped a pid.

commit 0ddeb020616bcebffc39afa5dfb583800cdef13d
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 15 22:20:00 2013 -0400

    The rand() functions do the single bit case.

commit 85f2d96a637c1534ad8631fe029dd86e07f1c970
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 15 22:17:15 2013 -0400

    Don't directly call the 'interesting number' functions.
    
    rand32/rand64 will sometimes call these, so just use those.
    That way, we'll also get some randomness used.

commit 57943ad7de4588d0fa4414ed7907f7d9b606b0ed
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 15 22:10:14 2013 -0400

    This sanitise routine appears to be nonsense.

commit ebe1de9374a1a31499f92a7d8685fbfb8a41cb33
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 15 22:07:32 2013 -0400

    This probably works better if it works on the right register.

commit 68a8ad30f2ff3e0f680b451eec2cd2c60162ad52
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 12 17:48:58 2013 -0400

    Make rand64 return a u64 always, even on 32-bit.

commit 409cbc4c012625fa1d079cef1532b0b08777d9e0
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 12 16:09:54 2013 -0400

    limit rand64 to 32bits when building on 32bit.
    
    I might later change this to make this function return u64 if necessary,
    but for now, the rand64() consumers on 32bit don't care because
    they work on ulongs.

commit 233a68cda23fee61712e2321ba45ba8f0f5faf6f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 12 15:58:03 2013 -0400

    add u32/u64 typedefs

commit e9eaf12a68b13cf2ea6496c1042a491735b6f7ba
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 12 15:03:50 2013 -0400

    be a bit more anal about missing prototypes.
    
    This highlighted a bunch of functions that could be made static,
    as well as some missing includes.

commit cafb703c206a2b5d35716718c6d67320baf735d6
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 12 13:35:32 2013 -0400

    do the pause from -p /after/ the syscall.
    
    Keep the pre-syscall sync where it is though.

commit 06d66b56904836602f68b0d9aa751f8c3bb01eeb
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 12 00:35:42 2013 -0400

    sendmsg is fixed now

commit 8054957078a6eb5753565c552420d22b40a60353
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 12 00:29:55 2013 -0400

    fix up perms on parent dir first

commit 540660287faa9527d710c6416395e8f1ecb231cf
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 12 00:29:14 2013 -0400

    make a copy of trinity in the harness
    
    This should be useful in case we unlink(/proc/self/exe)

commit e5e836c83b6dd4f3bfd05de4442ac7419dd423f1
Author: Nico Golde <nico@ngolde.de>
Date:   Mon Jun 24 18:03:57 2013 +0200

    parse_proc_devices: cut off device string on newline
    
    I didn't see that the device name includes the newline so that map_dev() will fail.
    This was kind of hard to see unless you have a driver and you know how the
    command ids should look like.

commit 64e62af1d226c839f7f759fef62c3843cc24b3e8
Author: Michael Ellerman <michael@ellerman.id.au>
Date:   Tue Jul 9 18:44:51 2013 +1000

    Fallback to the git SHA1 if there are no tags
    
    We use 'git describe' to generate the GITVERSION. However if the
    repository has no tags for some reason that fails, and generates no
    output. We should use 'git describe --always' which will fall back to
    just printing the abbreviated SHA1 if there are no tags available.

commit 24e724112ecf454bf7b4e674fba2c7543796e494
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jul 7 21:48:25 2013 -0400

    wait until all children have started.
    
    Prevents the reseed storm on startup.

commit cd96c619c63b638b8ca43d1be4a980c872782103
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jul 7 19:57:06 2013 -0400

    fix up 'dir' leak
    
    caught by coverity.

commit 8dd2396b8e1c16c056e460e7b01edd2c2b55848c
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 21:50:08 2013 -0400

    convert strncpy->memcpy.
    
    this time for real. Not sure what happened in the earlier commit.

commit 134b48e5da21c971db5640ce19a95536ea4a1cf2
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 21:43:06 2013 -0400

    Revert "bail watchdog pid walking immediately if something weird has happened."
    
    This reverts commit e2eda88807c9ad98b4d8f9a050bef674eebfa9ba.
    
    We can't do this because we call the reap_dead_kids function on exit
    when things go bad, and this change would leave the watchdog
    thinking there are still child processes around.

commit 8d59bd523e8bd0502d40ad0dbb66f675bbf9f9df
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 16:16:26 2013 -0400

    output pid in error message when failing to restore limit

commit ec6873616a31c6312e0144f30b626b44b262c81a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 16:15:20 2013 -0400

    bail watchdog pid walking immediately if something weird has happened.

commit fb43e028083f528a5bb88c0ffc42135f1277d781
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 16:07:28 2013 -0400

    get pidslot offset correct
    
    also don't colorize the empty pidslots

commit 665d62f5f2cc360f54c98f5b5d650382b1dcdb07
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 16:00:08 2013 -0400

    Fix looping in pid dump.
    
    I'm not sure how many more ways I can get this wrong.

commit f203fec4997a07c54cfc909334a56ec1b5e9cfe5
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 15:10:36 2013 -0400

    don't mix up printf/sprintf

commit 82577e7053808154c483794ff0282e4043d11d68
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 14:18:36 2013 -0400

    print out the pid in both cases, duh.

commit 226b171546c91847a0c2fff3164bd9ace57683ad
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 14:17:07 2013 -0400

    use correct iterator.

commit 1b36d95db41e66288e27ef5b5f85c67aff407d65
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 14:12:36 2013 -0400

    O_TMPFILE support

commit 973c33273c4003917e0731405dd632a14a409959
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 13:57:00 2013 -0400

    null terminate at end of string, not eol

commit 2446b91627aeb47ea7777822eb576a45d8857b41
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 12:48:30 2013 -0400

    colorize pids in the pidmap that are dead

commit f179caf25c9ef5dc8f83d4cb29eed0baa367ea2a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 12:41:39 2013 -0400

    stagger the output of the pidmap

commit fc4a4a2925c4841388579cfc0196031d006d1da0
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Fri Jul 5 19:15:54 2013 +0300

    trinity.c: log errors if socket calls fail
    
    Maybe that's all that needs to be done at this point.
    
    Coverity CID 1042335 (#1 of 1): Unchecked return value from library
    (CHECKED_RETURN)
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit 7523f6f3df9b9b870f105e8784cd57f25ad48336
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Fri Jul 5 19:15:53 2013 +0300

    trinity.c: fix uninitialized variable
    
    Coverity says:
    
    CID 1042350 (#1 of 1): Uninitialized scalar variable (UNINIT)
    23. uninit_use_in_call: Using uninitialized value "ling": field "ling"."l_linger" is uninitialized when calling "setsockopt(int, int, int, void const *, socklen_t)".
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit 7d96eb9b3a4f6de309ad843bf75b35b980f08a58
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 11:47:53 2013 -0400

    use 64bit constants

commit 277ce723ad01884224555b57436f94b6242067a9
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 11:53:35 2013 -0400

    use memcpy instead of strncpy on non-string data

commit 601981453abe64a7ddc693e71998888ae0f66bc1
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 11:52:24 2013 -0400

    use memcpy instead of strcpy, as this is not a string
    
    (side effect of silencing coverity)

commit a8305397084500d4d5d76f02981007e20e35bc19
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 11:45:28 2013 -0400

    use 64bit rand for sync_file_range

commit 8ac597597d6ed11a0f78b8112f1228cca8ad6dae
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 11:39:59 2013 -0400

    pick the right sync_file_range syscall table entry.
    
    Because we copy all these structs, the comparison is never going to
    compare to the correct version. Check the name instead.

commit a4ed25a040573b5cf300c3bdfb3400fe61284a6a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 11:15:15 2013 -0400

    actually use the zero page in the mmap

commit 4837e46b26595066bb39e817b305c37e1990bfd4
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 11:07:33 2013 -0400

    fix error checking for open

commit bc905242537378a1b3ee48365703374cf5c4cf99
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 5 10:50:47 2013 -0400

    Add IP_UNICAST_IF back-compat for old distros

commit 652c247deccccab95c4e964d7d0c74681d4895d7
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Thu Jul 4 22:55:44 2013 +0300

    maps.c: only close() if fd is valid
    
    Fixes Coverity CID 1042340.
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit 5559c7e756f95b7919160e9a466ba235e4b120aa
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Thu Jul 4 22:55:43 2013 +0300

    sockets.c: don't leak cachefile on return paths
    
    Fixes Coverity CID's 1042341 and 1042342.
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit 9b62af2c8d7f53d6039444c3ce2107f6d3ed0756
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Thu Jul 4 22:55:42 2013 +0300

    perf_event_open.c: close dir's on exit paths
    
    Don't leaky fd's so much. Fixes Coverity CID's 1042345, 1042346 and 1042347.
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit 910e1798b363a999713950d47723dba603990dc7
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Thu Jul 4 22:55:41 2013 +0300

    perf_event_open: initialize chars
    
    Just in case if they get used like in Coverity CID 1042349 and 1042348.
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit f8257a067a0857aa685ae0ab49615c0bac3def15
Author: Vinson Lee <vlee@twitter.com>
Date:   Wed Jul 3 16:03:50 2013 -0700

    Add MRT_ADD_MFC_PROXY and MRT_DEL_MFC_PROXY for older kernels.
    
    This patch fixes these build errors on kernels 3.8 and older.
    
      CC    net/ip_setsockopt.o
    net/ip_setsockopt.c:32:22: error: ‘MRT_ADD_MFC_PROXY’ undeclared here (not in a function)
    net/ip_setsockopt.c:32:41: error: ‘MRT_DEL_MFC_PROXY’ undeclared here (not in a function)
    make: *** [net/ip_setsockopt.o] Error 1
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit b2df339a4e82ccf056adb2075e25bcbc2ea2abf6
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Thu Jul 4 00:28:02 2013 -0400

    Fix up clang warnings in perf_event_open.c
    
    I was slightly off, the problem turned out to be if we fail at fopen()
    we never got to the fscanf(), but we'd then still call parse_format()
    with an uninitialized format_value.  So good catch by llvm.
    
    The following should fix things.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 638ab468e392e18127df0bb4072cb94550109301
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Wed Jul 3 17:07:38 2013 -0400

    Handle decimal entries in /sys/bus/event_source/devices/cpu/events/
    
    The following patch should properly handle decimal values in
       /sys/bus/event_source/devices/cpu/events/
    files.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 784880d5a47e7c55d6cac4706b1229ff4036962d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 3 15:57:49 2013 -0400

    remove a clang warning, and clarify indentation

commit f7e3d23c51f77534cbd2e4261ef5ab7e3b43d021
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 3 15:52:57 2013 -0400

    check for ctrl-c before we check for 0 files in filelist.
    
    Saves printing out an unsurprising message.

commit 2482bbf4d759df1abc84505588311c140f9969f6
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 3 15:52:16 2013 -0400

    remove unnecessary assignment

commit 1e26aa377ebff61452236e5fb116ea8fd478daa2
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 3 15:36:37 2013 -0400

    collapse the setsockopt switches into a single switch statement

commit df5b70fb04455b836e5b2050f3d6ff9b3752f620
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 3 15:12:13 2013 -0400

    That's pretty much it for ip_setsockopt
    
    All other options will do fine with just the default sizeof(int)

commit fe6a005bf494a5500b7e53659d5d692b38434e71
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 3 15:06:19 2013 -0400

    move the IP setsockopt stuff out to its own file

commit 9208a749a1580f74480a1e74c634ed9323d37e67
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 3 14:42:21 2013 -0400

    try some more random optlens for some of the as-yet unannotated protocols

commit be97f4ffc952494057d6b662df3c283ce3e2e207
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 3 14:40:27 2013 -0400

    add the MROUTE setsockopt optlens

commit 7ff620b2e2e7a9189cfbe5e2385d5a9c209a92c7
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 3 14:33:35 2013 -0400

    Add the MROUTE socket options

commit 5a032ca6009740ac1828d42287e6a4e644e417aa
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 3 14:12:24 2013 -0400

    start fleshing out some more of the IP socket option optlen cases.

commit 6bf82780cd483582e87158c5e5035cf9394b800b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 3 13:35:23 2013 -0400

    Add some missing IP socket options.

commit f1b3a9bd2c402fd7f5c26ac940ed4707cd576375
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 3 12:50:29 2013 -0400

    TODO updates
    
    Remove some done things, sort/reorganise

commit 9c35c598267f8f88c3512c2a3d140cdd7d5b9bbd
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 3 11:06:26 2013 -0400

    more get_new_random_fd cleanup
    
    Remove the goto targets, and split out the code that gets used in
    multiple places.

commit 747a6d65e7f6d12d6a9677b6dddbd742740d0b31
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 3 11:05:33 2013 -0400

    remove the fd 'retry' loop
    
    This is all crap. We're dealing with array indexes here, not absolute fd numbers.

commit bef44a2c14cafe2d1269a0f11d393fdc9602e3c6
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 3 10:41:11 2013 -0400

    watchdog should be avoiding anything that looks like a logfile fd too

commit cf5647853b089dea6aafa9060bfe3ae4cb35b945
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 3 10:40:05 2013 -0400

    factor out the routine to calculate the fileno of the last logfile

commit 4e12a4ff255c98afe97886f844e3e9c6415c981d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 3 10:29:09 2013 -0400

    additional sanity check on the fd before we print it in the watchdog.
    
    If it was out of range for whatever reason, it would overflow the buffer.
    Also move it out to a separate function.

commit cff4638daafc3704bb80f64901e895f36625e76e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 2 12:39:21 2013 -0400

    Convert filelist over to use the list.h primitives instead of our own.

commit 0a18eaae3fd8102cc7f437f73ccd4c4e6a45571e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 2 12:38:41 2013 -0400

    add a copy of the kernels linked-list routines.

commit fadb8ea910ec0c201cbf2bfd81b69481512e2b0c
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Sun Jun 30 22:33:12 2013 +0300

    ioctls/vhost: VHOST_SCSI_* ioctls

commit 87a62fadabe90f76b6264ef341086e99fa8010d9
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Sun Jun 30 22:33:11 2013 +0300

    ioctls/drm: add deprecated nouveau ioctls
    
    nouveau has some deprecated ioctls that the kernel implements, but the
    ioctl definitions are no longer available in the kernel headers.

commit 658ae4e60874a3fa99bd8d9a6c13f0e482db2825
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Sun Jun 30 22:33:10 2013 +0300

    msgrcv() and msgsnd() IPC_NOWAIT msgflg

commit a8496b21de1268fdd5c619b9720411f0c7f4c3a7
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Sun Jun 30 22:33:09 2013 +0300

    move perf_event.h to include/
    
    The perf header will be used also in a few other places beyond syscalls,
    so move it to the include/ directory so that it is easily available.

commit 059f038ddcd85137ea6560a823a7f2d311e7053f
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Sun Jun 30 22:33:08 2013 +0300

    Fix double rand() in generate_pathname()
    
    We take 90% chance two times in the "unmangled" path in
    generate_pathname(); looks accidental to me.

commit cee2984bf420cb68ed16623b5c47b4528c216479
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Sun Jun 30 22:33:07 2013 +0300

    Avoid needless get_filename() calls in generate_pathname()

commit 733bc695f5102e8bc89094eb171eec8fbe52179e
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Sun Jun 30 22:33:06 2013 +0300

    Fix open() result check in check_tainted()

commit 8509d5cf12330d1a08d00c75aa02484eda4d7402
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Sun Jun 30 22:33:05 2013 +0300

    Use TRINITY_PF_MAX instead of PF_MAX in syscalls/socket.c

commit 16d6cb1d4562c3d3ccda5dad5bd7a8d146ba0b63
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Fri Jun 21 16:39:54 2013 -0400

    Remove noisy 'Unable to opendir /sys/bus/event_source/devices' messages
    
    It's not fatal if the directory doesn't exist.
    
    A patch that removes the message is included below.

commit ed6369b072c57f5866b798059841e11a6f869fdb
Author: Nico Golde <nion@debian.org>
Date:   Fri Jun 21 15:18:58 2013 +0200

    devices.c: dont use %as scanf format string, but instead allocate and parse the device name manually for compatibility with non-glibc libcs

commit e68fc0220813a09e14ddfc37932237fb230e40f5
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 26 20:46:59 2013 -0400

    document some reseeding bugs

commit f8b57006a6f9148587996780e736d3cb95359d11
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 21:04:51 2013 -0400

    Increase random distribution in MSB of rand64

commit e58f75246a2d7761e0fd19a4fec99f9a9ee27e99
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 21:01:30 2013 -0400

    Occasionally invert rand64

commit 69e0478d7ab45f7bb3305629ffda9bb9cae6375e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 21:00:59 2013 -0400

    add some 64-bit truncation

commit e6260a5b821ab6ca68625c26f03f54e906e387e2
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 20:59:44 2013 -0400

    add a crappy rand|rand<<32 64bit case

commit a462695b486717e1f843823254a2a735ab1fdefe
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 20:58:00 2013 -0400

    remove duplication in rand64.
    
    Just call rand32 instead of reimplementing it.

commit 742c4da1f0bf13e681c5a0db3c5154c56528d375
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 20:56:57 2013 -0400

    re-use __rand32 in the mangler
    
    This was the reasoning behind splitting it out.

commit 57fbcd526f1fdeb05ccedf7987f43ddfab61284a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 20:55:48 2013 -0400

    don't always mangle, sometimes let rand32 just pass through

commit 10ae02c0a5aaf75e5cb2940b56354f158c4831a4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 20:54:41 2013 -0400

    don't take bits away in rand32 mangler

commit 322e97a2aee1bc9655db257f7d29bcaf927716b9
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 20:53:44 2013 -0400

    split up rand32 into generation/mangling

commit 222930294dac6d25ad984f58b3fd1aeb3d3ab66f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 20:48:34 2013 -0400

    add rept8 random function
    
    generate 8 bit, and copy it throughout a long.
    Like rand8x8, but same for every byte.

commit 6a18dc5470386ce8ebb7ebe29f9db166640c835d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 20:44:01 2013 -0400

    introduce rand8x8
    
    Construct a long made out of 8 bit chunks.

commit d99a6547b431b20f583af6d0896d49c86483ddaf
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 20:39:34 2013 -0400

    use randbits() in rand32

commit 188f9f0913d428cfab6a4c40b3f1e34d79eaa76a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 20:36:39 2013 -0400

    Move randbits out of the way of forthcoming work.

commit 0687aa50544913617bf634042a8eeb03aa872097
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 20:34:54 2013 -0400

    give rand64() a 50/50 chance of returning a 32bit number.

commit d5b1e6e112ee8d48544f57f393747cd59101fb44
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 20:28:41 2013 -0400

    factor out Tavis' random munger into its own function.
    
    And also call it in the 32bit path

commit 13c22e42a3efd4f7c8a357ebc5c53fa818c43562
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 20:22:09 2013 -0400

    collapse single bit setting functions into one

commit f4e9b23aa88d80dde15f845525fb4c2ee3b088d1
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 13:50:22 2013 -0400

    fix race at shutdown where we could spawn a new child as the watchdog exited.

commit 8ab9aaab4ea6da9915a82beed9446ae3419be23e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 13:48:03 2013 -0400

    sometimes don't mangle

commit 084e56d21f769c904d9e2bd40f7dd1b6ec8a521e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 25 13:47:44 2013 -0400

    set multiple random bits

commit b459cfcac3aa9a559aafa4536fc3da7a6fb2de0d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 21 14:35:52 2013 -0400

    move X86 definition into arch files

commit 605b2fba91c799beb2b4b494e0f83dfaaa83e51c
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 21 14:30:18 2013 -0400

    move page_size to arch.h

commit d50d0e5afa36bf05c9321ddff220b3925d55d265
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 21 14:21:19 2013 -0400

    introduce some more randomness to lengths.

commit c64c3ec63671286b40fa25631d3e2ed16d7c0325
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 21 14:20:49 2013 -0400

    add -1 - page size to interesting numbers

commit cdc9098acbd6c12b18d777a6d89367d50ee9d36b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 21 11:37:13 2013 -0400

    extra debug when nftw fails

commit 51cae47fe693d5401eea6a81db5c7db85e1547a2
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 21 11:42:00 2013 -0400

    fdstr needs to be bigger to handle larger pids

commit 2360fabd0ecfdcdbac7735b1b2612d07f1a1f0b2
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 20 14:17:17 2013 -0400

    dump the fd (if present) when we're stuck.

commit 2131e8fa03cf34a1c5fc11cfd86153bf70281494
Author: Michael Ellerman <michael@ellerman.id.au>
Date:   Thu Jun 20 14:15:15 2013 +1000

    Rework dependancy generation to work with subdirs.
    
    > Changes to the syscall struct should cause everything in syscalls/*
    > to be rebuilt. But somehow the dependency magic in the Makefile doesn't pick it up.
    
    I'm not a make expert, but I think it's because it's looking for .deps/syscalls/foo.d
    whereas we generate .deps/foo.d
    
    This patch works for me, touching include/syscall.h rebuilds a few
    things and then everything under syscalls.

commit a3ddc471d76f3e4a8b723c9cf245fdf75f165c2b
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Wed Jun 19 22:10:54 2013 -0400

    make perf_event_open() use init routine
    
    This makes the perf_event_open() syscall use the new init functionality.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 16d8b2d8a55c29b8f72c7a039d4820f3186ff953
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Wed Jun 19 22:09:09 2013 -0400

    add init() function to struct syscall
    
    While working on the perf_event_open() sysfs init problem, I was wondering
    if it might be easier if we added the possibility of an init() routine
    to each syscall structure.  That way trinity can support doing setup
    before fuzzing begins.
    
    Below is a quick patch implementing this, it seems to work but I still
    don't have the best grasp of trinity internals.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 7db0ea96781981e178cb4f0a56eb8df606c76dfb
Author: Sravan V Dodla <sravan@linux.vnet.ibm.com>
Date:   Wed Jun 19 18:45:14 2013 +0530

    Removing arguments of KVM_S390_STORE_STATUS
    
    Removing unused arguments of KVM_S390_STORE_STATUS.

commit 1cac391f2a17a66a64c11a842482f756a74e72e9
Author: Michael Ellerman <michael@ellerman.id.au>
Date:   Wed Jun 19 19:02:58 2013 +1000

    Sanitise F_SETSIG argument to avoid SIGINT
    
    It's possible to send ourselves a SIGINT using F_SETSIG. This is then
    interpreted as a Ctrl-c from the user and trinity quits.
    
    Avoid SIGINT, instead use 0, which switches back to the default
    behaviour of sending SIGIO.

commit d52f473fc7f61cc020058ca1c9a841fb355f01f2
Author: Michael Ellerman <michael@ellerman.id.au>
Date:   Wed Jun 19 18:59:34 2013 +1000

    Add compat definitions for recent powerpc KVM features

commit 035aac83b3343c65793f3696e06b11cdc914469a
Author: Michael Ellerman <michael@ellerman.id.au>
Date:   Wed Jun 19 18:59:33 2013 +1000

    Move x86 KVM ioctls under #ifdef X86

commit b192f49f0e1ab919ac160267403b63ac90e78ac2
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 19 00:00:53 2013 -0400

    print the taint reason in the watchdog.

commit 679c9537a24a1a1debc40a06c8c99242ff7603a0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 18 23:55:46 2013 -0400

    fix up s390 compile problem.

commit 2960893e3e96beafc09e89182449c814100bab2a
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 17 14:06:00 2013 -0400

    idea to avoid problems once reported

commit c40e7123c7d0d489d3663a15a86848c5dd173186
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 14 11:27:08 2013 -0400

    we need to do the SIGKILL 'helping' sooner.

commit bb0ec6163a80af8aca0b0f3ba6af8190ce0c73c0
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 14 10:02:36 2013 -0400

    add missing arg

commit 0ccc6ad2ae253ccc8a0afbceb64c9ddb44d61f68
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Thu Jun 13 15:58:37 2013 -0400

    Fix divide by zero in random_sysfs_config
    
    > [221186.493638] traps: trinity-child30[375] trap divide error
    > ip:40dbae sp:7fff3c1d7ae0 error:0 in trinity[400000+28000]
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 22135323dfb99dc2223a5c9f56930f99034391ce
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Wed Jun 12 14:34:26 2013 -0400

    more perf_event_open cleanups
    
    This has some cleanups for the perf_event_open call
    + Makes the switch statements all have explicit case statements
      for all the cases we care about (instead of sometimes relying
      on default)
    + Uses rand32() or rand64() over rand() when appropriate
    + Adds a lot more coverage to the create_random_event() function
    + Some other fine tuning
    
    In the end I didn't run lindent on it.  I agree the file could use it,
    but I didn't want to mix indentation changes in with functionality
    changes.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 13bf6f7ed9bfb3a1cb5f1348821f88c06aaf1a02
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 13 19:41:39 2013 -0400

    don't quit the watchdog process while there are still children running.
    
    If we have a child stuck in a sleep, we need the watchdog around to kill it.

commit 0152344abf32bda2ee8051d04a93d2b2ef87df8e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 13 19:41:15 2013 -0400

    comments

commit 51a6eca3a88d277950f08c4fbfe59e6c741c2587
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 13 00:18:18 2013 -0400

    don't check ppid every syscall.
    
    this dominates profiles, and isn't really necessary in the common case.

commit a9ff74db3cb335dfc7db3417feffe7b62e7979c5
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 13 00:17:18 2013 -0400

    try some variations in rand32

commit fd93c2de9519439305189faca27a944857e41430
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Tue Jun 11 00:33:23 2013 -0400

    perf_event_open enable sysfs exported events
    
    perf_event exports a bunch of event info via /sys under
    /sys/bus/event_source/devices
    
    By parsing these files it's possible to generate more likely valid
    (and almost-valid) events.  It also makes exercising expanded PMUs
    (such as uncore events on recent Intel machines) possible.
    
    The included somewhat ugly patch enables generating events based
    on the sysfs files (1 time out of 9).  It sets up a table of all
    possible event fields and generalized events for all detected event
    types the first time such an event is requested, then uses this table
    to generate events subsequent times.
    
    The parser is possibly fragile.  The perf tool actually has a full
    lexer to handle the parsing of these files.  I've tested on a core2
    and sandybridge machine but it might possibly fail on more obscure
    machines, especially if any have non-contiguous event fields.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit df7f86cd9a83010bd0b56aa6432abbbe64445c70
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 11 17:28:07 2013 -0400

    closer to what I want, but still far from perfect.

commit 8d4d72050a2e789b28be2bb4de544b72f67d6475
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 11 17:11:34 2013 -0400

    be explicit with braces.

commit 17480df40ba963d80d95e7a0bfeb0aca1b79ec92
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 11 17:07:43 2013 -0400

    steal a copy of Lindent from the kernel

commit 794b2e5aed4c077fffac369801e7dbf642e101dc
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 11 15:08:20 2013 -0400

    pacify sparse/smatch

commit af502a09fea94caa89b701d65d7de554895c0d59
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 11 15:00:43 2013 -0400

    document leak

commit c489e17e8f0349310a96b3aec30428966f40c95d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 11 14:58:22 2013 -0400

    missing proto arg

commit 2d1843aa3500fed8ca152be52874699f56e80348
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 11 11:48:07 2013 -0400

    long, not long long

commit 648f70f8fb588aaefc381c2619c486c3b16eaf95
Author: Hangbin Liu <liuhangbin@gmail.com>
Date:   Sat Jun 8 11:03:39 2013 +0800

    Specify domain in sanitise_socket
    
    We should specify the domain in sanitise_socket().
    If we pass a specific domain, we may otherwise get unrelated type and protocol.
    
    Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>

commit c6f028f4e883e84e91c1aea8467074e2a74f2b0b
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Sat Jun 8 11:51:36 2013 -0400

    Use local copy of perf_event.h
    
    Attached is an updated patch that adds perf_event.h from linux-3.10-rc4,
    updates syscalls/perf_event_open.c to remove all ifdefs, and removes
    the perf_event detection code from configure.sh
    
    This will allow the perf_event_open code to compile on my machine
    (and not depend on a possibly old version of linux/perf_event.h)
    plus give some extra coverage by testing the added attr fields
    on older kernels.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit fb5c38fde5045aa450b821145cf6e9d2b1211956
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 10 19:16:59 2013 -0400

    fix up dupe chmod
    
    meant to be working on parent dir.

commit 7bd0f7cb7a6c48736e3ff0c05ecdd84e3ffb1119
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 10 16:21:54 2013 -0400

    delete the old tmp dirs on each run

commit 9f4b6a8174a6439204713492d1c4d20ece8bd83a
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 10 16:20:36 2013 -0400

    make the watchdog keep check on the parent too, in case it crashes.
    
    having the children be reparented makes things go crazy otherwise.

commit ce100e29701cebc841198129325af8b602bd2116
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 7 17:26:34 2013 -0400

    todo based on Tommi's hack.

commit efcdd4f5da7ab09fe5d86447ae84e5701a6a2454
Author: Michael Ellerman <michael@ellerman.id.au>
Date:   Sat Jun 1 21:46:18 2013 +1000

    Sanity check the syscall number in print_syscall_name()
    
    In case it's out of bounds, causing us to segfault.

commit 8fcdab63ae1e753622b7083c523d7825f21d530e
Author: Michael Ellerman <michael@ellerman.id.au>
Date:   Sat Jun 1 21:46:17 2013 +1000

    Fix compile error in net/bpf.c for non x86
    
    In bpf.c there is an architecture check with cases for i386, x86-64 and
    "other". But the other case typos TRUE_ARCH, leading to a build error.

commit 9a672ac4e357b78c2fcd651d726b35eaac0605a2
Author: Michael Ellerman <michael@ellerman.id.au>
Date:   Sat Jun 1 21:46:16 2013 +1000

    Ignore vim swap files

commit 1bdfde4f5e32b061341a2966692db63b9383d021
Author: Michael Ellerman <michael@ellerman.id.au>
Date:   Sat Jun 1 21:46:15 2013 +1000

    Add support for cross building using CROSS_COMPILE
    
    Similarly to the kernel.  CROSS_COMPILE is optional, when specified it's
    a prefix, eg. myarch-linux-gnu-
    
    We still allow overriding CC, separately, or in addition to setting
    CROSS_COMPILE.
    
    There is also support for cross configuring, because at the moment all
    the configure tests only need to build, they're not run.

commit 6b41a6e72a9c011b5fadd0f0fa1aac6b5e741c89
Author: Hangbin Liu <liuhangbin@gmail.com>
Date:   Fri Jun 7 00:28:37 2013 +0800

    syscalls: use ARRAY_SIZE(x) instead of hard code define
    
    Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>

commit 17bf5d4175824afd71e86009990bcf365d58cd28
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 6 19:07:51 2013 -0400

    treat return value as a long
    
    can't believe it's taken this long to notice.

commit 02c6cbf62edbece93ed7c02b5a2a6e40a4d71da3
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 6 18:54:20 2013 -0400

    Cleanup usage of -L in scripts

commit f7342b2bd957554934b8a5edd74e78a2ab20d604
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 6 18:51:23 2013 -0400

    Handle error return from syscalls correctly.
    
    Successful returns from mmap were being flagged as errors for eg.
    
    Also if successful return > 10000, print it in hex

commit f70f1f269437b4160dc44d8b3b4733afa5314524
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 6 18:45:23 2013 -0400

    simplify output string in syscall

commit 04975f83a45ec8849cf1e83dd482d1b7d5b9c400
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 6 18:43:51 2013 -0400

    skip ANSI_RESET correctly in log files.

commit 141d5e365aa017cfbbc80a5d5a994cdb4e641cd7
Author: Hangbin Liu <liuhangbin@gmail.com>
Date:   Wed Jun 5 17:41:24 2013 +0800

    gitignore: ignore patch files
    
    Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>

commit 4e77d557077b480ae3028ad34c05c26041e58da5
Author: Hangbin Liu <liuhangbin@gmail.com>
Date:   Wed Jun 5 17:41:23 2013 +0800

    gitignore: ignore cscope files
    
    Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>

commit ddbf1dc74d012cd160c7c5db520df9fdf00c7bd4
Author: Jiri Slaby <jslaby@suse.cz>
Date:   Tue Jun 4 11:56:10 2013 +0200

    compat: define some more
    
    syscalls/prctl.c:27: error: 'PR_MCE_KILL_GET' undeclared here (not in a function)
    syscalls/socket.c:178: error: 'NETLINK_CRYPTO' undeclared (first use in this function)
    
    [v2]
     - include compat.h at the end, otherwise we fail to build on some
       configs

commit 1213046620f1d1a4a97f3ea84fc8e13c894e79e4
Author: Jiri Slaby <jslaby@suse.cz>
Date:   Tue Jun 4 11:48:23 2013 +0200

    detect seccomp
    
    It is not defined on some older systems:
    syscalls/prctl.c:7:27: error: linux/seccomp.h: No such file or directory
    syscalls/prctl.c:37: error: 'SECCOMP_MODE_FILTER' undeclared (first use in this function)

commit 0f75e96cc24c3cdb0ce364fd96b5e13a1fbbb773
Author: Jiri Slaby <jslaby@suse.cz>
Date:   Tue Jun 4 11:48:22 2013 +0200

    use config.h in USE_* users
    
    Sometimes we use USE_* macros but do not include config.h. Fix that.

commit 198eb0afc9777b1c43b378d8af2fe8a806b3640a
Author: Michael Ellerman <michael@ellerman.id.au>
Date:   Mon Jun 3 23:32:13 2013 +1000

    Read pid_max from /proc
    
    I was hitting this on my system, where it's 65536. So fix the todo and
    read it from /proc.
    
    I left a fallback to the hardcoded values in case /proc is not mounted
    or something else goes wrong.

commit a9461f0747f32b684db38eccd875f6b90be9fb7d
Author: Daniel Borkmann <dborkman@redhat.com>
Date:   Mon Jun 3 21:07:43 2013 +0200

    prctl: enable seccomp generator invocation
    
    Allow trinity to use the recently added seccomp bpf generator
    that was introduced in 3d0d78 ("bpf: add randomized seccomp
    filter generated by markov chain"). The last two arguments for
    that call can be left as random, the kernel does not seem to
    care about it.
    
    Signed-off-by: Daniel Borkmann <dborkman@redhat.com>

commit a7f8c5fcaa2eb2185b2867b176856b0461754880
Author: Daniel Borkmann <dborkman@redhat.com>
Date:   Mon Jun 3 21:07:42 2013 +0200

    minor: move offsetof into trinity.h
    
    Move it there since it's of generic use somewhere in future
    else as well.
    
    Signed-off-by: Daniel Borkmann <dborkman@redhat.com>

commit cccfe8313549596e7271906e8a0e1130167b6695
Author: Daniel Borkmann <dborkman@redhat.com>
Date:   Mon Jun 3 21:07:41 2013 +0200

    net: bpf: let clang shut up
    
    Let clang warning shut up and initialize the allocated area
    with zeroes.
    
      CC    net/bpf.o
    net/bpf.c:297:16: warning: The left expression of the compound assignment
           is an uninitialized value. The computed value will also be garbage
                    curr[1].code |= (uint16_t) rand();
                    ~~~~~~~~~~~~ ^
    net/bpf.c:299:16: warning: The left expression of the compound assignment
           is an uninitialized value. The computed value will also be garbage
                    curr[2].code |= (uint16_t) rand();
                    ~~~~~~~~~~~~ ^
    
    Reported-by: Dave Jones <davej@redhat.com>
    Signed-off-by: Daniel Borkmann <dborkman@redhat.com>

commit 6d6892257d2ec168183d70441578d2d904feb70d
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 30 21:41:28 2013 -0400

    TODO updates
    
    The socket generation code could be improved in a number of ways

commit 02ae7c684f9557ed83f15219b07a30f2901fa23e
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 29 14:05:43 2013 -0400

    use rand_bool in boolean decisions

commit 19195a1af9127a7ec260a553879a2db2a18a917e
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 29 13:21:22 2013 -0400

    fix highest netlink socket proto

commit d34b016c87000671e359f903c8cd52a3b3d49de8
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 28 21:20:07 2013 -0400

    avoid restart_syscall

commit 828eef44d12648d1bc77d5b4c0edd575bea2c932
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 28 15:59:27 2013 -0400

    handle tmp perms better

commit 223b5ad9c3be2de496661de6469dea9b36b44971
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 28 15:58:57 2013 -0400

    check for syscalls still enabled in loop

commit 2ab26e38e7de917630b480962266d08e6f99e813
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 28 15:56:37 2013 -0400

    when disabling a syscall for ENOSYS, handle biarch too
    
    This isn't strictly the right thing to do here, but it works around
    an ugly infinite loop bug that I don't have a good fix for yet.

commit 1bed3675851ceb9a47b1bd726d5385a94e9084c8
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 28 15:56:16 2013 -0400

    cruft

commit 1a4563c26fcc1f0709ae3f3cb1d3e2eba9424d9b
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 28 14:44:53 2013 -0400

    don't ignore no-input syscalls.

commit e7603699f57e2f68ce0166158787b2e73690731c
Author: Kees Cook <keescook@chromium.org>
Date:   Mon May 20 11:01:28 2013 -0700

    random: use /dev/urandom for reseeding if available
    
    Use a real source of randomness for reseeding, if available. Otherwise,
    reseeding is only permuted by time.
    
    Signed-off-by: Kees Cook <keescook@chromium.org>

commit 30ace5e0b7f2827cd7b2780dbc08a0c7daa424ae
Author: Vinson Lee <vlee@twitter.com>
Date:   Thu May 23 21:08:47 2013 -0700

    Fix build on kernels without perf_event_attr exclude_callchain_kernel.
    
    This patch fixes these build errors on 3.6 and older.
    
      CC    syscalls/perf_event_open.o
    syscalls/perf_event_open.c: In function ‘create_mostly_valid_counting_event’:
    syscalls/perf_event_open.c:382:6: error: ‘struct perf_event_attr’ has no member named ‘exclude_callchain_kernel’
    syscalls/perf_event_open.c:383:6: error: ‘struct perf_event_attr’ has no member named ‘exclude_callchain_user’
    syscalls/perf_event_open.c: In function ‘create_mostly_valid_sampling_event’:
    syscalls/perf_event_open.c:431:6: error: ‘struct perf_event_attr’ has no member named ‘exclude_callchain_kernel’
    syscalls/perf_event_open.c:432:6: error: ‘struct perf_event_attr’ has no member named ‘exclude_callchain_user’
    make: *** [syscalls/perf_event_open.o] Error 1
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 7b6c91ed7c1807e0dea77fccdd9c920ecf62d60a
Author: Daniel Borkmann <dborkman@redhat.com>
Date:   Thu May 23 16:41:28 2013 +0200

    bpf: add randomized seccomp filter generated by markov chain
    
    This generates semi-random seccomp BPF filters for trinity.
    Seccomp BPF filters are generated slightly different from
    networking BPF filters, since their main purpose is to
    generate a whitelist for syscalls that are allowed.
    
    Signed-off-by: Daniel Borkmann <dborkman@redhat.com>

commit 876e523743dad4631352e278d2838e59d7521b16
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 23 11:09:51 2013 -0400

    Lindent

commit 9ff3caff9047aa4ebcf49c65d5a968dfbb0be817
Author: Daniel Borkmann <dborkman@redhat.com>
Date:   Thu May 23 14:15:16 2013 +0200

    socket: packet: mostly use ETH_P_ALL protocol, but also randomize it
    
    Probably that's jut minor, but mostly use ETH_P_ALL as a protocol
    and in some minor cases allow to randomize it.
    
    Signed-off-by: Daniel Borkmann <dborkman@redhat.com>

commit 2847069924d5ffa4316f9c1d744b3a769f238e2e
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 23 10:13:53 2013 -0400

    Some older distros don't have hw_breakpoint.h
    
    Just dupe it in compat.h, seeing as it's so small.

commit d9d1b6a99be6af6a382f5b17f3f9c25273b74e6a
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 22 13:18:26 2013 -0400

    32bit fix

commit 89978a0682e5e035a562258d400947f0e293c895
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 21 22:28:49 2013 -0400

    demux the fcntl op parameter.

commit 3510e05e4f7f73acf1a002f6b36741773e6f2673
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 21 18:26:44 2013 -0400

    use rand_bool

commit 13a592051ca51d37a6458a4cab6b57ec94ca37a8
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 21 17:42:12 2013 -0400

    flip high bit randomly.

commit 6d73dfa13a2a9f8ace7a9a4e6eecd22523590800
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 21 17:39:04 2013 -0400

    introduce rand32
    
    to be used in places instead of rand(), to stress the single-bit cases some.
    (later, I'll bias it so that rand() is what gets called most the time)

commit 5c6f822ef79214b8cc32e148f371058a4524351d
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 21 17:30:24 2013 -0400

    add an extra short circuit if 0

commit 2a7b98382cd7cf4a6a03bf4ec15c808a577b6d5e
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 21 17:21:24 2013 -0400

    introduce some routines to return single bits.

commit 1de2ce9a9083d368332f278ccd73ae2c44dbc509
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 21 17:14:31 2013 -0400

    add a random bool helper

commit 30f561945927f4d389009889584831d6367aae83
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 21 17:12:03 2013 -0400

    move rand64 to random.h

commit 02efd8092c60326ba0b51f9b207715af6c5658fc
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 21 16:54:52 2013 -0400

    Make ARG_RANDOM_INT a bit more interesting than rand()

commit 1e03b0d670cba0a82ed7224923d1e77bb95b1881
Author: Jiri Slaby <jslaby@suse.cz>
Date:   Mon May 20 16:05:21 2013 +0200

    fix stupid header requirements
    
    pppox needs if_ether.h for ETH_ALEN in if_pppox.h otherwise on old
    systems we get:
    In file included from net/pppox.c:6:
    /usr/include/linux/if_pppox.h:39: error: 'ETH_ALEN' undeclared here (not in a function)
    
    rose needs ax25.h for ax25_address in rose.h otherwise on old
    systems we get:
    In file included from net/rose.c:5:
    /usr/include/linux/rose.h:49: error: expected specifier-qualifier-list before 'ax25_address'
    /usr/include/linux/rose.h:57: error: expected specifier-qualifier-list before 'ax25_address'
    /usr/include/linux/rose.h:65: error: expected specifier-qualifier-list before 'ax25_address'
    /usr/include/linux/rose.h:78: error: expected specifier-qualifier-list before 'ax25_address'
    net/rose.c: In function 'gen_rose':
    net/rose.c:24: error: 'struct sockaddr_rose' has no member named 'srose_call'
    net/rose.c:26: error: 'struct sockaddr_rose' has no member named 'srose_ndigis'
    net/rose.c:27: error: 'struct sockaddr_rose' has no member named 'srose_digi'

commit ef739c124b6f586cf4813bc04bf4b41847bbe89e
Author: Jiri Slaby <jslaby@suse.cz>
Date:   Mon May 20 16:05:20 2013 +0200

    compat: define some KVM_* constants

commit bd2e72355053e9e47f8cdbf5fd77620c2d12f5e2
Author: Jiri Slaby <jslaby@suse.cz>
Date:   Mon May 20 16:05:19 2013 +0200

    compat: define MS_BORN constant

commit fe63ee5b41432f6590d9f7c858935a6f7ecaf639
Author: Jiri Slaby <jslaby@suse.cz>
Date:   Mon May 20 16:05:18 2013 +0200

    fix tpacket3_hdr build failure
    
    It is declared iff TPACKET3_HDRLEN is...

commit 098ea4c9d5ef24e2623611cff265becd877714bc
Author: Vince Weaver <vincent.weaver@maine.edu>
Date:   Sun May 19 23:34:53 2013 -0400

    perf_event_open() coverage improvements
    
    Here's an updated patch that more or less exercises all of the
    perf_event_open() bits, up to around the 3.6 kernel or so.
    
    The patch is standalone against current trinity git.
    
    It could definitely use some review and tuning.
    
    I haven't turned up any issues with it, but that's partly because in my
    experience (the recent exploit being an exception) perf_event bugs rarely
    manifest with just an open call; they usually involve
    reads/writes/ioctls/mmaps/signals on the fd that is opened, as well as
    problems after fork/exec or else on complex events made of multiple
    perf_event_open() calls chained together (the call can take a previously
    opened fd as an argument).  Testing things like this will be more
    difficult.
    
    Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>

commit 5ecf6eed1df7857f38c1137e466826cdf2d74588
Author: Kees Cook <keescook@chromium.org>
Date:   Sun May 19 18:57:01 2013 -0700

    fix random switches and page construction
    
    Fix truncated page construction, as well as several unreachable switch
    statements. Make interesting 32bit number actually 32 bits. Add additional
    64 bit construction with high 32bits from interesting-32.
    
    Signed-off-by: Kees Cook <keescook@chromium.org>

commit bc33f19e9db3462cbd289299f9eddb699391d110
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 17 20:57:24 2013 -0400

    bring back rand64

commit 5cbe1c2d78e5dc6f5b6b0c80eba60660769ddf4d
Author: Vinson Lee <vlee@twitter.com>
Date:   Thu May 16 18:35:40 2013 -0700

    Add KVM symbols for older kernels.
    
    This patch fixes these build errors on 3.4 and older.
    
      CC    ioctls/kvm.o
    ioctls/kvm.c:21:2: error: ‘KVM_S390_UCAS_MAP’ undeclared here (not in a function)
    ioctls/kvm.c:22:2: error: ‘KVM_S390_UCAS_UNMAP’ undeclared here (not in a function)
    ioctls/kvm.c:23:2: error: ‘KVM_S390_VCPU_FAULT’ undeclared here (not in a function)
    ioctls/kvm.c:55:2: error: ‘KVM_ASSIGN_SET_INTX_MASK’ undeclared here (not in a function)
    ioctls/kvm.c:56:2: error: ‘KVM_SIGNAL_MSI’ undeclared here (not in a function)
    ioctls/kvm.c:114:2: error: ‘KVM_DIRTY_TLB’ undeclared here (not in a function)
    ioctls/kvm.c:115:2: error: ‘KVM_GET_ONE_REG’ undeclared here (not in a function)
    ioctls/kvm.c:116:2: error: ‘KVM_SET_ONE_REG’ undeclared here (not in a function)
    ioctls/kvm.c:117:2: error: ‘KVM_KVMCLOCK_CTRL’ undeclared here (not in a function)
    make: *** [ioctls/kvm.o] Error 1
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 86a511edcf6460fdcf8e4d514bd2ce9b90d4ca66
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 16 19:38:05 2013 -0400

    fix off by 1  rand usage

commit 7737311307910f852ca3b1cb9ad59c4231383c50
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 16 15:07:30 2013 -0400

    Fix ARM compile when lacking kvm.h
    
    Reported by Folkert van Heusden <folkert.mobiel@gmail.com>

commit 724f52df90925131baae9dfea4a2b5a030860d87
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 16 14:56:58 2013 -0400

    add some improvements from test-multi to the test-sequential script

commit d9c892871d9cadb38ebcfdd3cf0152f20d15abd6
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 16 13:02:01 2013 -0400

    fix off by one.

commit 3f9366d5f7427c502be77950a7de3e1682ff5fe0
Author: Daniel Borkmann <dborkman@redhat.com>
Date:   Thu May 16 16:10:25 2013 +0200

    net: bpf: increase the chance that the semi-random bpf is run
    
    The kernel filter code expects us to have a return instruction and
    that won't be changed in the kernel code. So increase the chance that
    we can also get past the filter precheck and run this fuzzed filter
    code.
    
    Signed-off-by: Daniel Borkmann <dborkman@redhat.com>

commit 3f99a2ce19d3852a414d9cb7f26e1ccd5b644cee
Author: Daniel Borkmann <dborkman@redhat.com>
Date:   Thu May 16 11:54:16 2013 +0200

    net: add generation of semi-random BPF filters
    
    BPF filters are used in networking and in application sand-boxing
    and even have a arch-dependant JIT compiler in the kernel, so add
    a more detailed semi-random BPF generator.
    
    Signed-off-by: Daniel Borkmann <dborkman@redhat.com>

commit fb7f7bb631643fe5a6b38583e89f679958de56c6
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 14 18:19:41 2013 -0400

    fill out some more fields

commit 6c56fd583659ddceb1b5d8098de0dc40acbba834
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 14 18:15:31 2013 -0400

    rename hw to attr

commit b25a5464d70eb5877ee956c156752f3b3907f5ad
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 14 18:14:31 2013 -0400

    make sure we clear reserved bits in perf struct

commit f45f3ff38014be120c2c111ea0bffa5e991a8dac
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 14 18:05:01 2013 -0400

    perf_event_open updates
    
    - add missing PERF_COUNT_HW_REF_CPU_CYCLES
    - use get_len for more interesting lengths than rand()

commit f6a5eef0f5a3c2fa1b2a8b00c64c06e3d3d23d97
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 14 16:34:23 2013 -0400

    change permissions back on tmp dir if they have changed.
    
    also, exclude sendmsg for now until oops w/connect is fixed

commit cc9f32712830b56f42018482e616d34309252260
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 14 16:33:29 2013 -0400

    50% of the time, write a single byte

commit 97cbe77cab34e11593819d39b07513458e95a13e
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:42:01 2013 -0400

    split the protocol decoding code out of sockets.c to own file in net/

commit c44d0d707bd1c7eea19702de1c3945cbb01dfdd8
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:34:00 2013 -0400

    Move sockaddr main file to net/

commit e5c373238aa276161968451225dce9b799cd1b09
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:32:41 2013 -0400

    move phonet sockaddr code out to own file

commit 3222debb0f048517a2777f8f5bc97a506bdb9bbb
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:28:59 2013 -0400

    move tipc sockaddr code out to own file

commit 51d33ec17b079476cd8af9be71e1872b1914bc3b
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:27:42 2013 -0400

    move CAN sockaddr code out to own file

commit cf1fdf333a7d58a9b7f148df7df061c8b3e61d5a
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:26:16 2013 -0400

    move irda sockaddr code out to own file

commit 032f2628aba8cd688486b3688c16c6d2a3f75790
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:24:25 2013 -0400

    move atmsvc sockaddr code into own file too, along with the other atm code.
    
    (and rename)

commit b7cfa0ff18503ab7271eb116c281b6b8fb23d47e
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:22:12 2013 -0400

    move econet sockaddr code out to own file

commit af776bc0954b14e96a8aed8dcb24c476c2ebd347
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:20:39 2013 -0400

    move packet sockaddr code out to own file

commit c4981220114b350d37d346fd2d9dfa16749b5a08
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:18:36 2013 -0400

    move netlink sockaddr code out to own file

commit 4d1d3b9ce939595ce019a18a512a3dc38e1c9f62
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:16:49 2013 -0400

    move llc sockaddr code out to own file

commit 9c124144cac540a864d5184ef42422ce3d45965a
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:14:27 2013 -0400

    move decnet sockaddr code out to own file

commit e752cdeb44eafdb8d570ba7cf97ad6f71145f866
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:12:54 2013 -0400

    move rose sockaddr code out to own file

commit f6b10f731f4919133b2f42f766969c51c15f01ad
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:11:11 2013 -0400

    move x25 sockaddr code out to own file

commit 95b9600ace019602cc21c3d1c81c2415a4f2eb3d
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:09:20 2013 -0400

    move atmpvc sockaddr code out to separate file

commit d9a2ff9363b50b68e250b82a9f06e393b3975475
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:06:51 2013 -0400

    move appletalk sockaddr code out to own file

commit 9799c1807847453459f2f540f9cb962784b24ec1
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:05:00 2013 -0400

    move ipx sockaddr code out to own file

commit 0940d86a6abc50c604a26cd3a48eb82bec1fcb8c
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:03:27 2013 -0400

    move ax25 sockaddr code out to its own file

commit 12b589f35f31012571ff2b90ce1c090a47555f9d
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 19:00:43 2013 -0400

    move NFC support out of sockaddr.c into its own file

commit d9adcb149473b98cc084d44e92f27df4acb1eff1
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 13 13:38:24 2013 -0400

    new rand-page mode: page of 0's and 1's

commit a7ac465c8a498a696ea3d52a51b271c8f557e8e1
Author: Daniel Borkmann <dborkman@redhat.com>
Date:   Fri May 3 13:38:26 2013 +0200

    setsockopt: add missing SCTP socket layer options
    
    There are a couple of missing socket options for SCTP, so simply
    add them to the list.
    
    Signed-off-by: Daniel Borkmann <dborkman@redhat.com>

commit 9492df9c26d468b2da88880dff05f40a327abe9a
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 2 12:08:51 2013 -0400

    skeleton for SOL_NFC setsockopt support later

commit cb3fda02e8154153e1ba60c449d0e0e2f32e3f7a
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 2 12:04:07 2013 -0400

    add SO_SELECT_ERR_QUEUE flag to setsockopt

commit 7632369e3d082d99e5cfd8c2cb0442551eb69a56
Author: Daniel Borkmann <dborkman@redhat.com>
Date:   Wed May 1 17:10:15 2013 +0200

    setsockopt: sanitize PF_PACKET SOL_PACKET options for zero-copy rings
    
    In SOL_PACKET, we can sanitize the setsockopt() syscall a bit in the
    following ways:
    
    i) PACKET_VERSION is always checked in the kernel and it is quite
       likely to return -EINVAL here, very unlikely to crash this option.
       However, if we pass the correct values to it (TPACKET_V1, TPACKET_V2,
       TPACKET_V3), we can jump into this version specific code on other
       syscalls on that socket.
    
    ii) PACKET_{R,T}X_RING never gets a structure of size int, so it might
        always return -EINVAL here. Depending on the TPACKET version, it
        can either be tpacket_req or tpacket_req3. Make it more likely to
        have size tpacket_req though.
    
    Signed-off-by: Daniel Borkmann <dborkman@redhat.com>

commit 1bd60e46538f6cd60d3a56e8a9fa0a86cfdb606c
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 29 16:37:32 2013 -0400

    compile fix for older headers

commit ac80a1cdcd9bc92676284ce9fe158d3f645c23ad
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 29 14:34:21 2013 -0400

    run each child only for a finite time
    
    (until all the leaks have been plugged, otherwise we invoke
     the oom-killer eventually)

commit f5cfd3c1d8f5a1ccdce28c8b15889c5caf3ee72f
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 29 14:33:33 2013 -0400

    rework the pathname generator for more varied mangling

commit 3580faf996ded86ecce48cdaffe5f066db12e0ca
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 29 14:33:00 2013 -0400

    symlink takes pathnames

commit 6d19fc2832845203a10dd7bf4a5668538d030d60
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 26 15:23:22 2013 -0400

    hush llvm warning

commit a6f52daad613cb4dc6bc0a272831db2eb5770b3b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 26 12:10:41 2013 -0400

    add kvm ioctls

commit 38fc4cd25e90aee20d9a1de205725bcd6c1ada33
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 26 11:04:57 2013 -0400

    use correct type for mode

commit 6ca1de4c1681692676224a246cee1ff58465fd43
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 26 09:14:47 2013 -0400

    ARG_MODE_T

commit 11b94416e7a83f041aa9728f330186ced6e6aa2e
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 26 08:20:11 2013 -0400

    actually, it's the pathnames that are the problem, not the fds

commit 951d3e438cd381e0fba90436cf6d651b47aee0b0
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 26 08:18:09 2013 -0400

    improve comment

commit 6afd823bb46595433b67a57fab10fc1c30589088
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 25 17:02:58 2013 -0400

    duh, missing breaks.

commit 286beff1ac152a96a5473509380c190180ee7b66
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 25 17:00:57 2013 -0400

    if we pass -n, don't do syscalls that need pathnames.

commit 504163376e7fbb0ae67da19bbb8094e239638342
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 23 17:42:59 2013 -0400

    Add an example of how to search for an interaction between two syscalls.

commit b258e149331f1c773ed15cf10881e613faeb0e59
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 23 13:44:06 2013 -0400

    lower priority of debug message

commit a463edbcdc356dc0e332d5239cbc0ee0b64d8649
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 18 11:27:45 2013 -0400

    print out syscall number for non-biarch -L

commit 8beb858b1a0244757a444aa0a9688a9dfb00d2af
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 17 17:48:30 2013 -0400

    update clean rule to delete all objects
    
    (Was missing net/ but this seems cleaner)

commit 74811b6e7b7ed225de192c17b64ef72dcecf7e82
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 17 16:31:12 2013 -0400

    better variable name

commit b957c2644c2c9da32be1c50ab4358d6e6f003778
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 19:43:12 2013 -0400

    disable VM/VFS syscalls if we -n

commit 20dee0cf8cd0a534a31b355722865c7f307a339a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 19:09:10 2013 -0400

    fix mistake where we used the same syscall number for both 32 & 64 bit tables

commit 66c5f0daf74b071021509cfe99cfbf48c8de45c7
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 18:14:37 2013 -0400

    split out routine for checking if a syscall is in a net-related group

commit d4cd3c802b7a7e5b9d81b63077e4976f866044a4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 18:12:44 2013 -0400

    do validation at the end, after we've done with -r etc

commit 30a29094add2436d659b0b96c6a76c621894f381
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 17:49:37 2013 -0400

    revert debug code accidentally committed.

commit 5f9258c775490f899c25ac0cc334bca8a88e6ef8
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 17:47:20 2013 -0400

    move scripts to scripts dir

commit 43170d7da19d8c0996d66c5518d03aa5f90983ef
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 17:43:18 2013 -0400

    when verbose, display all enabled syscalls.

commit 325a0bb7615050befb23d89d36c443d04d88991d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 17:33:57 2013 -0400

    Remove duplicate information.

commit 7694ead104609683be5802398c26b6f810c574d5
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 17:18:35 2013 -0400

    don't ignore enabled syscalls

commit d63a18dce98828e3543b0dfdf18a8e2265b866d0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 17:18:05 2013 -0400

    display count of dis/enabled syscalls.

commit 4bb6dc04e1701138d677dfd7d2364735196dd601
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 16:54:03 2013 -0400

    fix -x behaviour when no -r

commit b65f0abfd17c81303e0d51540a446b6357027aec
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 16:30:42 2013 -0400

    move ustat into the vfs group

commit d34b72955a14338bf3224a73a5bbcdf8f66643db
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 13:48:34 2013 -0400

    move alg out to separate file

commit 42fe618e6579a38ef703f04e04e54ade420de42f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 13:46:14 2013 -0400

    move caif out to separate file

commit fd644055c617756ee62210407df23f1a12c3c416
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 13:45:46 2013 -0400

    add net/ to source list

commit c3203f619843b6f78d0c6787d197ac8cb79e8640
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 13:07:42 2013 -0400

    rename function

commit 6238e8ab59d603b90955e2d3e6283b332ef70d4f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 13:05:31 2013 -0400

    make -r take the number of syscalls to choose from

commit 4bfe05b2c072b7b13dc4abd56884e721f36a02cb
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 12:37:32 2013 -0400

    pick a better name for DISABLED flag

commit b7e4c0b26c23b70aff0b9c0dc82f194bab08806d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 12:34:09 2013 -0400

    fix 32/64 typo

commit 9d3e6badbb9838bbf96a1aefb10e39f1a9f5bec4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 12:32:08 2013 -0400

    move the -x processing out until after we're done processing args.
    
    we need to collect things and process them afterwards rather trying
    to process as we go.  This should make a few things work better, such
    as having '-x foo -r' and then having -r re-enabling the x'd option.

commit f39c47938647c8458cb07ea0015878dbc5d3f6e7
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 00:59:31 2013 -0400

    move unix sockets out to seperate file

commit 68059b74b17049332b92ca65bf3d5404771e7961
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 00:52:00 2013 -0400

    move pppox code out to separate file

commit 7ec70bc5a44791fa7ba1aa3b9acf69a90484f3cf
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 00:48:16 2013 -0400

    move out ipv6 code

commit 6aa22030a6615e9c75dbaedadfc7ab4d1898dd97
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 00:45:16 2013 -0400

    compile fix

commit ae206d39bdae25e631837cb6d505c5c89631fae9
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 00:39:56 2013 -0400

    add moved out ipv4 code.
    
    (from previous broken commit)

commit d69b6052d145e7080bc3c86e7bc5680a37cb588e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 16 00:38:50 2013 -0400

    Start splitting out network protocols into per-proto files.

commit 4119f7097b9e6876457499f9f94ed58deb5c3236
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 15 21:26:43 2013 -0400

    and more VFS syscalls

commit 99c72bdf5e940b218f3ece871de738ea2396a194
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 15 21:24:08 2013 -0400

    more VFS grouping

commit 0cf3df3b36c46035d5272a35b7d116996b94b12c
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 15 21:16:59 2013 -0400

    add VFS group

commit da3745b4135718580b5684fd1949789e6a3ab3dc
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 15 19:56:56 2013 -0400

    if we pass -r and -n, might as well skip the VM syscalls.

commit 636e57197c26134860a29cc48d86ef093812e1eb
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 15 14:05:17 2013 -0400

    exit, not return

commit c2ba0f26bbd9e2ee93f07335a63d9b59febfe450
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 15 14:03:00 2013 -0400

    -x options need to be after -r

commit c0fa5a3259eba0b4e3f676a57ca8d7fa8df33c82
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 15 10:37:45 2013 -0400

    don't enable already enabled syscalls

commit 86d95293150e55048aa33365d7c0ce9bc4d6c4f2
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 14 20:08:38 2013 -0400

    split up shm init
    
    we need randomness in parse_args if we pass -r
    but we also need to handle -s, so post-setup the shm too if necessary

commit 99c1979433c1ccb43c94e5b16f6ff85b2df4bdf5
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 14 19:42:34 2013 -0400

    make !biarch variant of toggle_syscall work the same way as biarch

commit dc18b9e23aa904fdfef777b2d7c118ffe79e3517
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 14 19:40:32 2013 -0400

    remove duplicate message
    
    toggle_* prints this already

commit 7b427393b818fd49a3dea565b2a8e99bad7ea314
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 14 19:06:01 2013 -0400

    condense 'enabled' output a little by printing both arches for biarch on same line

commit affbc2e9637f631d28eb462b42a596b1418e0fef
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 14 18:50:00 2013 -0400

    make sure -x doesn't enable all syscalls if we've already done -r

commit bdda801dc316136bbf7823a79766f0d92a7dd4ea
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 14 18:43:26 2013 -0400

    clarify

commit 5ee5e12af33b6ce4191b386c6696f5c53621e578
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 14 18:19:33 2013 -0400

    remove level of nesting

commit 462c1d4e76ae71cb963ce8c7141a74ac37045a7a
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 14 18:08:21 2013 -0400

    remove hand-rolled validate function

commit d8306ac3b16deb72410890cd6597cd62639e8f3c
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 14 18:03:26 2013 -0400

    remove duplicate test
    
    already done (better) in the validate routine

commit 04c1035123155e33896bbd8b795a96b9ceb9333c
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 14 17:49:33 2013 -0400

    validate the random syscalls we enable, to make sure we do 10 worthwhile calls.

commit 626703057d5dfb1d57f1c98cf9e7a6d172307e15
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 14 12:36:56 2013 -0400

    remove debug code

commit 475fe5d9d58f03d8fefb616a030939b90bfdc41a
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 14 12:36:23 2013 -0400

    evaluate -r earlier
    
    this way we can use -r and -x together

commit fe2ab7ac0562979173b4de3436181d0954f12ac1
Author: Dave Jones <davej@redhat.com>
Date:   Sat Apr 13 22:58:47 2013 -0400

    only output progress every 10k

commit 26b3dbd0b484e953c016bacfdaea1062b31d204c
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 12 18:17:44 2013 -0400

    more -n fixes

commit ba7a8b37b596f2cc6dcafeb48d5a650aae9e8ffe
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 12 18:00:31 2013 -0400

    avoid potential div by 0 when running with -n

commit 66a879ab22d9dae86d99a6ed00b372507806f3e2
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 12 17:51:01 2013 -0400

    add 'random 10 syscalls' mode.

commit 180f963e9290094dc700a415769f3ad48c106f60
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 12 17:50:05 2013 -0400

    factor out the syscall munging to a separate func

commit c04e01b37c3af07fa11309f6204e3ace2bdbec5a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 12 16:35:47 2013 -0400

    add a mode to only use network sockets

commit 9663bf2373c9d1d7917e10b05b692b7387601217
Author: Nikola Pajkovsky <npajkovs@redhat.com>
Date:   Thu Apr 11 23:50:04 2013 +0200

    mount: define MS_NOSEC for older kernel
    
      CC    syscalls/mount.o
    syscalls/mount.c:32: error: ‘MS_NOSEC’ undeclared here (not in a function)
    make: *** [syscalls/mount.o] Error 1
    
    Signed-off-by: Nikola Pajkovsky <npajkovs@redhat.com>

commit 7c5546f8c95974d6fd5fb59698cbaf4978a7fff1
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 12 16:27:43 2013 -0400

    idea to shorten runtime for hard to reproduce bugs

commit 0d39194f0cb9817d695f40205c12a43339797972
Author: Vinson Lee <vlee@twitter.com>
Date:   Fri Apr 5 20:03:14 2013 -0700

    Include compat.h in syscalls/msgsnd.c for MSG_COPY for older kernels.
    
    This patch fixes this build error on Linux 3.7 and older.
    
      CC    syscalls/msgsnd.o
    syscalls/msgsnd.c:19:40: error: ‘MSG_COPY’ undeclared here (not in a function)
    make: *** [syscalls/msgsnd.o] Error 1
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 31487cc0a50693a4f70dd4ac653f8290661a60ff
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 10 21:28:52 2013 -0400

    decode protocol number correctly.

commit 6938a6260a273fc6a6c5fb74e0bfc26b970529f8
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 10 21:14:11 2013 -0400

    decode errno when a syscall fails.
    
    Not sure where this negative check idea came from, but it was misguided.

commit 65098c6d4d6194cd0fe1e39440b466aef9ddbc6c
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 4 18:18:22 2013 -0400

    occasionally don't add in the usecs

commit 5add7bbe5fca5751f5bee9546640397361680163
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 4 18:14:50 2013 -0400

    simplify seed generation.
    
    this seems good enough

commit cdfc8bdd311cf27e78662b5eb5d220176e762bae
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 4 18:03:14 2013 -0400

    ugly hack to support 32bit pids.
    
    clean up later.

commit c146ee1e1bcf409226e5e503ab85bffd6a8eb705
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 4 12:20:17 2013 -0400

    backwards compat for MS_SNAP_STABLE

commit 765b4968989bec7f1ba72a5ca7c8ae8cd0a54b0b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 3 23:09:32 2013 -0400

    annotate flags for sys_mount

commit 6b6fcbe2f273da4981f62b107a0003de89b2e2fd
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 2 21:02:35 2013 -0400

    break up do_random_syscalls some more

commit ced955e3aae99876c6e7e8989f7bf68ad77fd300
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 2 18:33:56 2013 -0400

    move the random syscall loop out to its own file

commit 98d7848a88fd6f9c37133534db8a17ff1a79eb70
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 2 17:44:54 2013 -0400

    TODO updates

commit 9adb89bf6a2c86da8cfcc75d02221adab52e1f0f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 2 17:35:56 2013 -0400

    child process factoring
    
    child_process will eventually pick a random 'thing' to do,
    and random syscalls will be just one possibility.

commit 95b106d2e182abace4f9f75c0016b2d9593eeb59
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 2 16:11:39 2013 -0400

    make syscall table validation non-fatal
    
    Toralf Förster noticed that trinity would exit immediately when
    if you ran with '-x munmap'.
    The reason for this is that munmap is currently marked AVOID,
    and we exit()'d if validate_specific_syscall() returned FALSE.
    
    Just leaving it be informational should be good enough.

commit 3743b97dbc535a798800782ca6dc177fbe48f621
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 2 14:07:43 2013 -0400

    Tommi pointed out that I've no had enough caffiene today.

commit a2515c26bdba305d56536c7bb130567df3425d23
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 2 13:59:14 2013 -0400

    old distros don't have MSG_COPY yet
    
    (pointed out by Tommi)

commit 07fdb8be37e193833203452d3236014cb165a4af
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 2 12:19:46 2013 -0400

    flags arg for msgsnd

commit 3e5905fd223c68ebd8577eec85252d5330acad2a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 2 12:19:06 2013 -0400

    Forgot to add the type

commit 2aecd7bea72d59b88346f4f8dcedb55e0e8ff2d1
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 2 12:16:29 2013 -0400

    flags arg for msgrcv

commit c65cf75f01c0d7fc37fb156a0cc85953a155b912
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 2 10:24:55 2013 -0400

    factor out the parent pid checking from child process

commit 4dc29ad34404cebbf07b8158be040e4f6c095673
Author: Nicolas Kaiser <nikai@nikai.net>
Date:   Tue Apr 2 13:40:03 2013 +0200

    overwritten assignment in sanitise_setsockopt
    
    I noticed an assignment that gets immediately overwritten, contradicting
    a nearby comment.
    According to the comment, 'optval' should get disabled half of the time.
    
    As at the time of the conditional assignment optval was already set to
    different nonzero values, it looks to me like the trailing assignment,
    which was introduced in commit 737e1a71, may be unintended.
    
    Signed-off-by: Nicolas Kaiser <nikai@nikai.net>

commit caefd22e7f0f8ac2202fa5498ad1c12113c39ecf
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 1 21:01:48 2013 -0400

    todo: globbing in ioctl devs.

commit c26783fa99e6de2ee39c0979990a7507b22526d5
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 1 18:23:36 2013 -0400

    Let people know configure 'failures' aren't fatal.
    
    Apparently this is confusing people.

commit ebf7a69ff58f6b31b5888288de0f40e5761dece9
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Apr 1 21:58:39 2013 +0300

    Build syscalls/ia64/*.c

commit 9d066a3799057f45f8cc1781e0d06d5764f7b7eb
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Apr 1 21:58:38 2013 +0300

    Fix SYSCALLS_ARCH
    
    SuperH compilation was broken, since we are adding SYSCALLS_ARCH to
    SRCS, and expect SRCS to only contain *.c files.

commit 9a6bfcdfb5e648622869f86b6f3e03780b5484f6
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Apr 1 21:58:37 2013 +0300

    Check headers with the compiler in configure.sh
    
    To allow for proper cross compilation support, pass the headers to the
    compiler to see if they are available. Checking the return value from
    the pre-processing (-E) seems to do the trick.

commit 739ce5f7eadb4096a41767d4ff7a6eb773a2e8a4
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Apr 1 21:58:36 2013 +0300

    Dummy PTRACE_GETFPREGS and PTRACE_SETFPREGS for IA64

commit 094e0c9d65db30371e2de6a8237ba6b1393585f0
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Apr 1 21:58:35 2013 +0300

    Fix double .arg3name assignment in syscalls/ia64/perfmonctl.c

commit e33bcb5620f96c5895e8067906a3f068ebb6d7d7
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Apr 1 21:58:34 2013 +0300

    Include asm/perfmon.h for the PFM_* constants in syscalls/ia64/perfmonctl.c

commit 1a1a5e8b01d7adc5aebed4b32c8ad28514a559f8
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Apr 1 21:58:33 2013 +0300

    Fix bitrot in include/syscalls-ia64.h

commit c7cd76f2f6f9f7a675175ac6afcae000b617eeca
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Apr 1 21:58:32 2013 +0300

    Check 'drm.h' existence
    
    None of the DRM headers were available when I tried IA64 cross
    compilation with some old kernel headers.

commit d08cb6b103528750587c34ad2a15d838b0fcb7fa
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Apr 1 21:58:31 2013 +0300

    Use color clearing also in Makefile

commit fd37506cc21f06036ecde5e72cf518ed350981e8
Author: Vinson Lee <vlee@twitter.com>
Date:   Mon Apr 1 11:40:45 2013 -0700

    Add #ifdefs around missing exynos ioctls on earlier kernels.
    
    This patch fixes these build errors with Linux 3.4 and earlier.
    
      CC    ioctls/drm.o
    ioctls/drm.c:179: error: ‘DRM_IOCTL_EXYNOS_GEM_GET’ undeclared here (not in a function)
    ioctls/drm.c:181: error: ‘DRM_IOCTL_EXYNOS_G2D_GET_VER’ undeclared here (not in a function)
    ioctls/drm.c:182: error: ‘DRM_IOCTL_EXYNOS_G2D_SET_CMDLIST’ undeclared here (not in a function)
    ioctls/drm.c:183: error: ‘DRM_IOCTL_EXYNOS_G2D_EXEC’ undeclared here (not in a function)
    make: *** [ioctls/drm.o] Error 1
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 5e2cab3e04d5d7a4ef817afc92de78d9d891e2e8
Author: Brian Haley <brian.haley@hp.com>
Date:   Thu Mar 28 22:41:35 2013 -0400

    Fix 2.6.32 compiler errors
    
    Fix compiler errors found when building on ancient 2.6.32 system.
    
    Changes since v2:
    
    Fix inotify error using compat.h
    
    Changes since v1:
    
    Added missing #include of config.h in ioctls/vhost.c
    
    Signed-off-by: Brian Haley <brian.haley@hp.com>

commit 749d01edd7c27a6fa63287491d4f9338bb942c8f
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 27 14:40:02 2013 -0400

    output ANSI reset codes, instead of defaulting back to white

commit 31fc3582e08a04b61322cfc1546b1fc380d43321
Author: Denis Kirjanov <kda@linux-powerpc.org>
Date:   Wed Mar 27 10:06:20 2013 +0400

    Fix BITS_PER_LONG for ppc32
    
    Signed-off-by: Denis Kirjanov <kda@linux-powerpc.org>

commit d262dca98c451ae02a2bb0595f22cbf064c8527d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 26 10:32:38 2013 -0400

    rename test-random
    
    this made sense in days of old when we had --mode.
    Now it's the generic multi-threaded test.

commit a4d8fc64f43f8556043075cc1f5964ab132555e4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 26 10:31:35 2013 -0400

    print out the git version when using a devel tree.

commit 14ab731f64f11640d4b494cdf47654adf6a7c1b6
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 22 17:57:37 2013 -0400

    hush nftw walk error if we ctrl-c'd

commit 7a3e20940b3b1994a0d754e55bc2fc0a6ea13969
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 22 16:40:59 2013 -0400

    add missing biarch check
    
    watchdog was segfaulting on i386

commit 41f472019db5e7d20a9b3c50d97b2ae48fed2026
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 22 12:05:02 2013 -0400

    some useful advice.

commit c99fe208c5206141cc3926ebe104298c5fa6430f
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Wed Mar 20 21:59:32 2013 +0200

    prefer to OOM kill the children
    
    Tell the kernel to prefer to kill the children first when running out of
    memory. We do not currently handle the watchdog or the main process death very
    well.

commit a4df50d4742828a85b9d0b87f7b15db80f31c689
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Wed Mar 20 21:40:58 2013 +0200

    uinput ioctl support

commit a0d85093258c86ce845ed8fac4502ca6347b78b1
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Wed Mar 20 21:40:57 2013 +0200

    vsock ioctl support

commit 02a46b0ebabae1ee8fcc9d20156922d5baca3a17
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 20 10:19:40 2013 -0400

    blacklist some dangerous files in /proc

commit 427cf9a51e6cc55466527f321fa31e41527e9bff
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 19 17:34:08 2013 -0400

    make the ignore files code match full paths
    
    This is pretty crappy and should be fixed up later to use actual globbing

commit 2024d22c11fb1c509465b47f89474c3410b331a4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 19 15:29:52 2013 -0400

    for whatever reason, nodes is signed.

commit 8a1c2ea0a18b365a3384d968c8b3ed249e919f27
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 19 14:13:45 2013 -0400

    hush llvm

commit 2ecf72a3dd71fc0ecc737088eebeb737356577e7
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 19 14:08:07 2013 -0400

    fix clang to work with latest llvm from f19

commit 6715356fc457b422e877a410f59e10afe1a2ba70
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Mar 18 21:48:31 2013 +0200

    syscalls/inotify_add_watch: enumerate known mask bits

commit a558ff25903abb1558a1c050f36c47fd6da6492c
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Mar 18 21:48:30 2013 +0200

    ioctls/vt: add TIOCGPKT, TIOCGPTLCK, TIOCGEXCL
    
    Add a few more ioctls introduced in upstream linux commit c6298038 ("tty,
    ioctls -- Add new ioctl definitions for tty flags fetching").

commit 02142481ef204fc574e7c2a746c80d888f6c56f0
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Mar 18 21:48:29 2013 +0200

    ioctls/videodev2: add VIDIOC_EXPBUF

commit 451db6c41993785f03c5161a870b11c992461897
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Mar 18 21:48:28 2013 +0200

    ioctls/snd: include ioctls from <sound/compress_offload.h>

commit 88011d6a457bad4cc8a48c14bfec0af91c763e20
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 17 19:28:29 2013 -0400

    only decode errno for negative errno's

commit 339812c9c388e48a7851c7d9ff0beaf8b1832ba6
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 17 19:25:41 2013 -0400

    todo: multiple -V's

commit d977e1e64d850b7806b1a8aceafade1927482aa3
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 17 17:42:45 2013 -0400

    vger list.

commit c63f4be3dbb91443b4ed95308807c42cc255b200
Author: Dave Jones <davej@redhat.com>
Date:   Sun Mar 17 11:02:19 2013 -0400

    depends fail

commit 82cfb851a7faef0d45d44d39fcae4625c26270e5
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 15 16:55:14 2013 -0400

    make fabricate_onepage_struct an internal implementation detail of generate_random_page

commit d9da33b9623ae1612e66acf5aa86298a5a1669f3
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 15 16:54:59 2013 -0400

    make execve generate pages of junk instead of structs as args

commit 6321f623fbdf7fba32444989e87aae1dde281d4d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 15 16:46:05 2013 -0400

    use more random page contents for ioctls

commit b8eabc707addd19ede330ca84c3e7df4ee1a3926
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 15 12:44:28 2013 -0400

    don't fuzz writing MSRs, it's a terrible idea.
    
    Pretty much guaranteed to crash if you're unlucky enough to hit
    the wrong MSR.

commit de50264289d46afeb780623136e9aa0d73922b4f
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 14 16:42:08 2013 -0400

    update setsockopt options

commit 172d55181ed21add159c1cbd07a3c153d5f76b39
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 14 21:58:14 2013 +0200

    guard against empty ioctl groups
    
    We assume elsewhere that all groups contain at least one ioctl operation, so
    filter out any empty group at registration time to avoid crashing.

commit f13a0d064e0ed39bec3011c2229bc46dc9307d4a
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 14 21:58:13 2013 +0200

    autofs ioctl support

commit fdf20ce152a5d7e0e1250142600c451dfa38dd79
Author: Kees Cook <keescook@chromium.org>
Date:   Wed Mar 13 23:47:32 2013 -0700

    socket: close all the sockets
    
    As in close_files(), close all the sockets, not half.
    
    Signed-off-by: Kees Cook <keescook@chromium.org>
    Reported-by: Brad Spengler <spender@grsecurity.net>

commit 4f99af21dc7a8cfa6e7a3aeb19a607a3d5925d32
Author: Mathias Krause <minipli@googlemail.com>
Date:   Wed Mar 13 22:20:16 2013 +0100

    Fix the colouring in configure.sh
    
    Don't switch to white, just return to the normal mode. Not everybody
    uses bold white as foreground text colour.
    
    Signed-off-by: Mathias Krause <minipli@googlemail.com>

commit 2bab22c01aafac207b1fe6b8cdf30dec0d0cb41b
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Wed Mar 13 22:19:07 2013 +0200

    btrfs-control ioctl support

commit 4811fe8acac57a509dcbe776b50a933473437ea4
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Wed Mar 13 22:19:06 2013 +0200

    bump IOCTL_GROUPS_MAX
    
    We hit the limit, so allocate some more.

commit 8eeb004753ed8f033ebab663d749a359c7d22667
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Wed Mar 13 22:19:05 2013 +0200

    include 'config.h' for USE_DRM_EXYNOS

commit 13aec0fc2981d49cea013a5f26c0c5edd686a0d3
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Wed Mar 13 22:19:04 2013 +0200

    include 'config.h' for USE_VFIO

commit e680a2493b886cf72e17b33d5a677b9536f6f851
Author: Jan Stancek <jstancek@redhat.com>
Date:   Wed Mar 13 16:32:01 2013 +0100

    update arch-s390.h, MODULE_ADDR
    
    KERNEL_ADDR
      # grep _text /proc/kallsyms | head -n1
      0000000000000000 T _text
    
    MODULE_ADDR
      arch/s390/include/asm/pgtable.h:#define MODULES_LEN   (1UL << 31)
      arch/s390/kernel/setup.c
            vmax = 1UL << 53;       /* 4-level kernel page table */
            MODULES_END = vmax;
            MODULES_VADDR = MODULES_END - MODULES_LEN;
    
    This patch updates MODULE_ADDR to 0x1FFFFF80000000L for 64bit systems.
    
    Signed-off-by: Jan Stancek <jstancek@redhat.com>

commit ef2c37213dca7d07ff5733a529f37c3e68b019a1
Author: Jan Stancek <jstancek@redhat.com>
Date:   Wed Mar 13 16:32:00 2013 +0100

    update arch-ppc.h, add defines for 32/64 bit
    
    This patch updates KERNEL_ADDR and MODULE_ADDR for
    32 and 64 bit powerpc systems.
    
    KERNEL_ADDR
      # grep _text /proc/kallsyms | head -n1
      c000000000000000 T _text
    
    MODULE_ADDR
    As far as I can tell powerpc is using generic module_alloc(),
    which allocates space via vmalloc. VMALLOC_START for 64bit
    is 0x8000000000000000 (CONFIG_PPC_BOOK3E=y) or 0xD000000000000000.
    This patch sets it to latter.
    
    Signed-off-by: Jan Stancek <jstancek@redhat.com>

commit cdfa7367edbca33820672ad7bbae86a34190cbd7
Author: Jan Stancek <jstancek@redhat.com>
Date:   Wed Mar 13 15:16:38 2013 +0100

    ioctl/vt: fix compilation on ppc (RHEL7, 3.8)
    
    on x86_64:
    asm/termios.h includes
      asm-generic/termios.h
        asm/termbits.h
          asm-generic/termbits.h
            defines struct termios and termios2
    
    on ppc64:
    asm/termios.h includes
      asm/termbits.h
        defines struct termios
    
    Including asm/termbits.h creates a conflict on ppc, because struct
    termios is defined in /usr/include/bits/termios.h and also in
    /usr/include/asm/termbits.h.
    
      # make
        CC  ioctls/vt.o
      In file included from ioctls/vt.c:4:0:
      /usr/include/asm/termbits.h:22:8: error: redefinition of ‘struct
      termios’
      In file included from /usr/include/termios.h:40:0,
                       from /usr/include/bits/ioctl-types.h:5,
                       from /usr/include/sys/ioctl.h:29,
                       from ioctls/vt.c:2:
      /usr/include/bits/termios.h:33:8: note: originally defined here
      ioctls/vt.c:120:2: error: ‘TCGETS2’ undeclared here (not in a function)
      ioctls/vt.c:121:2: error: ‘TCSETS2’ undeclared here (not in a function)
      ioctls/vt.c:122:2: error: ‘TCSETSW2’ undeclared here (not in a function)
      ioctls/vt.c:123:2: error: ‘TCSETSF2’ undeclared here (not in a function)
      make: *** [ioctls/vt.o] Error 1
    
    Also asm-generic/ioctls.h which defines TCGETS2 and others is not
    included by any other header file on ppc.
    
    This patch adds check if struct termios2 is provided by glibc headers,
    if not it is defined in place instead of including asm/termbits.h.
    
    Signed-off-by: Jan Stancek <jstancek@redhat.com>

commit 703c280c7a9f8ab10fff00b533d8316adc1f1a14
Author: Jan Stancek <jstancek@redhat.com>
Date:   Wed Mar 13 15:16:37 2013 +0100

    add define for s390 PAGE_OFFSET
    
    Signed-off-by: Jan Stancek <jstancek@redhat.com>

commit b0bdb4869499c774fab2da131dba781db87ad158
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 12 13:28:50 2013 -0400

    some groovy thai gibberish

commit ae5edfc9d4bbb39eeff4156d1c1674b94c77bb2c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 12 13:24:31 2013 -0400

    fix some warnings

commit 98f9615de6c6d0e02265d198a9676f20b7f35cac
Author: Vladimir Davydov <vdavydov@parallels.com>
Date:   Tue Mar 12 18:19:20 2013 +0400

    fix ioctls/vt compile problems on older kernels

commit 374d8cc008ca9b45092e6543c0a4fa789c3da3a7
Author: Vladimir Davydov <vdavydov@parallels.com>
Date:   Tue Mar 12 18:19:19 2013 +0400

    fix ioctls/videodev2 compile problems on older distros

commit ab0bbfbe039f3ef34ecb01df1c373f867ac83f76
Author: Vladimir Davydov <vdavydov@parallels.com>
Date:   Tue Mar 12 18:19:18 2013 +0400

    fix ioctls/vfio compile problems on older distros

commit cb69da8402d36fa3d2274a0f28d8ceefa9a3c648
Author: Vladimir Davydov <vdavydov@parallels.com>
Date:   Tue Mar 12 18:19:17 2013 +0400

    fix ioctls/rtc compile problems on older distros

commit 1097daea3094410455fa41abbe2c42397fbe602f
Author: Vladimir Davydov <vdavydov@parallels.com>
Date:   Tue Mar 12 18:19:16 2013 +0400

    fix ioctls/mtd compile problems on older distros

commit 22f3d9f612c914f7885c2444956d4c8be49dbb58
Author: Vladimir Davydov <vdavydov@parallels.com>
Date:   Tue Mar 12 18:19:15 2013 +0400

    fix ioctls/loop compile problems on older distros

commit 51ccb01fd2989538671ea61dc401dec1a34bf3a9
Author: Vladimir Davydov <vdavydov@parallels.com>
Date:   Tue Mar 12 18:19:14 2013 +0400

    fix ioctls/input compile problems on older distros

commit 17e4a6bb207f4e710f1e24914059ef45ef4630f8
Author: Vladimir Davydov <vdavydov@parallels.com>
Date:   Tue Mar 12 18:19:13 2013 +0400

    fix ioctls/firewire compile problems on older distros

commit 1113ee15fe125f22bd8452244be684ee39e0933e
Author: Vladimir Davydov <vdavydov@parallels.com>
Date:   Tue Mar 12 18:19:12 2013 +0400

    fix ioctls/drm compile problems on older distros

commit dc151fda043343b600508032c3a157b3db6a90d4
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 11 21:01:24 2013 -0400

    fix ignore_files()

commit c57e90e54ddc85b31cb1b59967bd97e85948d12e
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 11 20:34:34 2013 -0400

    TODO: unicode tty corruption

commit fcb5dcc15d4d363d1ec14111fa5bd7ff985d5590
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 8 22:42:55 2013 -0500

    idea to parallelize the nftw's

commit 088c7faa88388687395195dfa1ce2d25515153c4
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 8 22:09:28 2013 -0500

    some ideas from todays fun.

commit d29ef2d96c8e80be21e4519dacc009c58970f75f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 8 17:44:44 2013 -0500

    colourise the output of configure a little.

commit d6c406db421cc60dad277e94ed2c2ef2999b56cb
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 8 16:57:57 2013 -0500

    I missed the ending.

commit 3e1e8c82f8559604cdc85e1ff8d7dd0e2a561576
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 8 16:53:46 2013 -0500

    HE COMES.

commit 8434f7cd3e6d544071075490bc1c4f47ca8f3fa8
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 8 16:41:07 2013 -0500

    constify

commit 8f827634bf8d4ceb3b681d29c7c898c9bfd39ea7
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Fri Mar 8 22:06:40 2013 +0200

    Take compiler name from $CC in configure.sh

commit e87f5d2f4e67e98056987ab680df6c54b9debdb0
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Fri Mar 8 22:06:39 2013 +0200

    Wrap more ioctl commands in ifdefs in ioctls/vt.c
    
    These were not available when cross-compiling to MIPS.

commit bd54b6be3c2fe01c947f96e12a27be53d8eb82da
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Fri Mar 8 22:06:38 2013 +0200

    Use KDGKBMUTE and KDSKBMUTE ioctl commands conditionally
    
    KDGKBMUTE and KDSKBMUTE do not appear in upstream linux kernel, but are
    present e.g. in fedora 18 kernel headers package.

commit 4259d53c7543fa3c49c8a3420ed1a54263d24c9e
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 8 11:22:19 2013 -0500

    README cleanup

commit c4739e45863480db6d2cc42f10d2725814bc8d99
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 23:53:50 2013 -0500

    Move architecture specific addresses out to own function.

commit e4ba32e98535eaa042d4749aff53b80becf8f65b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 23:27:52 2013 -0500

    some TODO updates

commit 1102d410cf0fa921b56ee18076cef6c36f27efd4
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 23:08:07 2013 -0500

    language cleanup

commit 99b0b2092ecb15103f0a06274caaa8c54b22dbc4
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 23:02:04 2013 -0500

    comment cleanup

commit 78e63f4a0bfb50210606811c9caf2a5d3df41e82
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 22:50:18 2013 -0500

    move the device scanning code out into its own file

commit 20a481dadd97a92992890dc70d981174f47ff418
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 22:42:16 2013 -0500

    fix path mangling to not scribble over memory

commit 675698280979c64bcdcac3c4ec65515491f6854e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 22:31:33 2013 -0500

    source formatting

commit f72f4b14fb6a5c4f1b66bc7d52139c78955e40cc
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 22:29:44 2013 -0500

    move the ARG_PATHNAME logic out into files.c

commit 8ec1541ad59127e15e6b02881e295c371109fd9e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 22:20:28 2013 -0500

    move ARG_LEN stuff out to own file

commit e7bc57bcef169b574018464303d5fdd1bedd0c07
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 22:16:29 2013 -0500

    kill off global 'filebuffer'
    
    re-use page_rand instead.
    It's properly shared between threads.
    
    The only downside is that we're only going to do at most 1 page of I/O,
    but we can revisit this one day maybe.

commit 7782fd34a6afabc98e2c64848ead73804c07f585
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 22:07:31 2013 -0500

    move ARG_ADDRESS stuff out to separate file

commit 90b133bd6d8c274761a3110bcef48d094ab2662b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 21:59:49 2013 -0500

    some comments

commit f2924eacbd22151eb75f262f003b01949430dc10
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 21:53:36 2013 -0500

    move the random page generation out to separate file

commit fef90edf83caedfc684dd34e3d21871e236ccfd7
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 21:48:03 2013 -0500

    move get_filename to files.c

commit 7ac73def7ff6dee84d5f27db79267d73ac2b329d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 21:46:21 2013 -0500

    move the interesting numbers out to their own file

commit 30d216e846d3014c9ea862bc1ab9b1e180d63909
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 21:43:09 2013 -0500

    remove unnecessary trinity.h includes from syscall table definitions.

commit 225b264e3ef7c59e5eb0be647e83cb45ec24a3fe
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 21:40:29 2013 -0500

    syscall.h only needs types.h, not the whole trinity.h
    
    Fix up fallout from places which were missing explicit trinity.h includes.

commit 393a420b8ea8f68c7ef2c765cbe01ee990795592
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 21:32:35 2013 -0500

    shm.h cleanup. Split exit stuff from trinity.h out to separate file

commit 688ed9b0a792a25b86ba33fcbae24529f181991a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 21:29:15 2013 -0500

    move the unicode stuff out to its own file

commit 40b842205e3abe16bc334c2c5e22566c5a269577
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:50 2013 +0200

    binder ioctl support

commit 0a4ab7a73b1789e69c9b67cdef186f4ed5f6b938
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:49 2013 +0200

    logger ioctl support

commit b7a2e1cee02c56b97d7129ee0fe969be83c57d1e
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:48 2013 +0200

    msr ioctl support

commit 0f0013cb39b7bd1ededa511e743dc7030872655c
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:47 2013 +0200

    video4linux ioctl support

commit d400afed923d4b89079ddbeac490fc5cadd43ab6
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:46 2013 +0200

    firewire ioctl support

commit 7f5b9f0a4d4dfcbcb7cfcfe980a83d6234e71558
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:45 2013 +0200

    input ioctl support

commit b9450941e8d2418eda0c73158aefb69389c3aa9d
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:44 2013 +0200

    ozwpan ioctl support

commit a344f52c5a486784a8505acd16a6594e21d567c0
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:43 2013 +0200

    usbmon ioctl support

commit 85d027493312279c5a9bcad0c2eea035e8efd3e0
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:42 2013 +0200

    loop ioctl support

commit 59ffa694a2edf4b4d86fda548637b2a14e5df0c2
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:41 2013 +0200

    hpet ioctl support

commit dd490aa0ad2f57d54587b789b2a2de5bacb245bf
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:40 2013 +0200

    vmci ioctl support

commit 5d56f6a23b8def0bbd55f4e017c429bc15ade522
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:39 2013 +0200

    device mapper ioctl support

commit 9548df24ff40d0a8adb1eccbc33df8b33b067ed9
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:38 2013 +0200

    drm ioctl support

commit 2869e7a677a5d2129980caff800f381c6848c791
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:37 2013 +0200

    vhost ioctl support

commit dac6b7af412dccadfa9cf69f6a4da481cbc368fa
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:36 2013 +0200

    vfio ioctl support

commit c3dc4359d59e3b7e12990e3852dc8613268d3200
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:35 2013 +0200

    rfkill ioctl support

commit ffd0db0d767334ac9d317c47da44930539bf8e05
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:34 2013 +0200

    rtc ioctl support

commit 2a80e6a1d63c21095563e2e19d2db9f05c4260c1
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:33 2013 +0200

    mce ioctl support

commit e629b9230ea645a06ff203dd99a26339126cc87b
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:32 2013 +0200

    watchdog ioctl support

commit 08a60a02d1b899e9761d3924416ec5ecb4e68a4b
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:31 2013 +0200

    /dev/random and /dev/urandom ioctl support

commit 81d10db2f429cebf81fde81a995ea197c08b0af6
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:30 2013 +0200

    new style console/vt/tty support

commit c84a2edf554a59853bed1678e54aebbd35b0b20a
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:29 2013 +0200

    new style scsi ioctl support

commit 71232fc71508242a5c039e03ecf88bbaee8a657e
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:28 2013 +0200

    new style mtd ioctl support

commit 7084fcfa74fd08c2378623d510947164558d775f
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:27 2013 +0200

    new style socket ioctl support

commit 29f51c0db73657fd416c4d89b421232352d87d43
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:26 2013 +0200

    new style sisfb ioctl support

commit b59450d7ff73a1e89ff127b45c0e069dccd674b3
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:25 2013 +0200

    new style sound ioctl support

commit ff37a51a86076b18c8b7f50b5ab08864736c412d
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:24 2013 +0200

    new style framebuffer ioctl support

commit 646e64ea1aea15ca9bfe450fe1874232c3bcb7b5
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:23 2013 +0200

    new style cdrom ioctl support

commit 757ec27a8c35e857b2db7bc0296bc4d552d6a5a7
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:22 2013 +0200

    Add --ioctls/-I for dumping known ioctl operations

commit 2cab2fbb11bb6e35b25e76394a6ba62a125e8b8e
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:21 2013 +0200

    new ioctl() sanitise
    
    This changes the ioctl() sanitise to prefer valid ioctl operations given
    the file descriptor. The old version would simply pick an ioctl
    operation at random, but this is not very interesting in most cases,
    since unrecognized ioctl() operations do not tend to have any effect.

commit 95499ff5a38ccc33b0db0cb4ea14c7521597f25d
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:20 2013 +0200

    Parse /proc/devices and /proc/misc for new ioctl support
    
    Parse /proc/devices and /proc/misc. These will be used in the new ioctl()
    fuzzing code to match a file descriptor to a device name.

commit f9cd919107b8721e320d715b8a3e9a5b647435dc
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Mar 7 22:34:19 2013 +0200

    longopts[] constness

commit ddc3e6a1f75d140cc3c75cf722a231fd91290639
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 10:37:22 2013 -0500

    change the filename suffix generation to use more varied data
    
    including unicode. Bring the noise.

commit 6101f3ab2c9efa37d20aef96aef0ef8aa4e59b23
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 09:58:08 2013 -0500

    Use LLC_OPT_UNKNOWN definition instead of 0
    
    From Vlad's variant of the LLC_OPT_PKTINFO patch

commit 47457952ed5a7c35d794e8a10235b239521835aa
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 00:56:05 2013 -0500

    terminate the unicode page somewhere randomly

commit e8e0e928e5fa1d1e1f2fd3c353a7d13c995f09b0
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 7 00:03:03 2013 -0500

    Add check for LLC_OPT_PKTINFO for older distributions.
    
    This patch fixes this build error on distributions, such as CentOS 6 and
    Ubuntu 10.04, that have older kernels.
    
      CC    syscalls/setsockopt.o
    In file included from syscalls/setsockopt.c:13:
    ./syscalls/setsockopt.h:167: error: ‘LLC_OPT_PKTINFO’ undeclared here (not in a function)
    make: *** [syscalls/setsockopt.o] Error 1
    
    Tested-by: Vinson Lee <vlee@twitter.com>

commit 535bb87480007004cd2cdaea4e78986e0cef74a9
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 6 21:31:50 2013 -0500

    typo

commit 847b3618c014d2535dccf62d518c8f72a198c587
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 6 19:59:07 2013 -0500

    fix up comments

commit d0dc34aa6533de22909fddcb8e6a3572331685b1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 6 19:50:00 2013 -0500

    generate a page of unicode garbage

commit 9c18d041aac2e953cda959bd9122f68de84f93fa
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 6 19:43:57 2013 -0500

    clean up the page_rand randomizer some

commit 186cbe217025be36d2ec91f7bbbc00f58d6e09e1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 6 17:32:18 2013 -0500

    path mangling.

commit a2ad581b022d4ac98e9afd41c12964149b849a84
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 6 17:19:14 2013 -0500

    execve sanitise improvements

commit 985d066e2ce3a2621cf9324b69afc647a411c2b0
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 6 16:23:48 2013 -0500

    remove debug code

commit 52db96326331e3b12edf8860db47817cbdf56a33
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 6 16:22:30 2013 -0500

    Check files are readable before adding them to the file list.
    
    - Avoid -EISDIR on directories
    - consolidate stat'ing code to one function.

commit 969db374f354137ca1d5799e5a4a1e6164c64a62
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 6 15:56:38 2013 -0500

    Fix handling of forced 32bit syscalls.
    
    if we ran trinity -c lstat on a 64bit machine, we would never
    do any syscalls, because lstat only exists on 32bit.

commit a6ffc4c505206099eddbac724235cc9c4fe0d47a
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 6 15:48:40 2013 -0500

    Reorder syscall tests in child loop
    
    This way, we retry faster on disabled syscalls.

commit d0ac978f8d7cb4ad44b4ce9265b891bcefb092e3
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 6 15:46:54 2013 -0500

    Use nftw instead of our own dir scanning code.
    
    Among other things, this handles symlinks better.

commit dfb389e171e5d87393af64c4dc5f4dd95d288177
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 4 18:49:17 2013 -0500

    Allow -V to take a single file as an argument
    
    Suggested by Tommi.

commit fae73d45e8ebc0798e72799ff5195211d1ba73aa
Author: Vinson Lee <vlee@twitter.com>
Date:   Fri Mar 1 18:05:14 2013 -0800

    compat.h: Add MADV_MERGEABLE and MADV_UNMERGEABLE.
    
    This patch fixes these build errors with glibc 2.11 and older.
    
    syscalls/madvise.c:28: error: ‘MADV_MERGEABLE’ undeclared here (not in a function)
    syscalls/madvise.c:28: error: ‘MADV_UNMERGEABLE’ undeclared here (not in a function)
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 11bbbb274758eda01b9d676bda3ed7744eb77b76
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 1 16:46:04 2013 -0500

    forgot to bump the rev post release.

commit 7b22e8259ee0f5d1ef51cd6d6c09580025b09d12
Author: Heiko Carstens <heiko.carstens@de.ibm.com>
Date:   Mon Oct 22 07:55:40 2012 +0200

    add s390 backend
    
    Add s390 31 and 64 bit system call lists and defines.
    Since s390 doesn't support all ptrace requests that other
    architectures support, define those request to zero, so we
    don't get a compile error.
    All those test calls will then be mapped to a "zero" ptrace
    request, but that doesn't hurt.
    
    Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>

commit b6e0a8208747110c54fb386bc2e07a85a98da1db
Author: Jiri Slaby <jslaby@suse.cz>
Date:   Wed Feb 27 00:03:48 2013 +0100

    compat.h: handle new KEYCTL seprately
    
    Some KEYCTLs were added later, so handle them by separate ifdef/endif
    pair.
    
    Signed-off-by: Jiri Slaby <jslaby@suse.cz>

commit 28d1e6874672565b936f385905ad40ee7f3b22f6
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 28 15:46:41 2013 -0500

    use mktemp

commit 9495e0e452e8221f63ba70bbc560f5bd614d764a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 28 15:44:19 2013 -0500

    Revert "add stub for LLC_OPT_PKTINFO"
    
    This reverts commit c23a262f165cc51fa73f1881b7e26f6744b9ad1d.
    
    This is an enum, and as such won't ever be #define'd

commit d9fe3c63b6c517bed4b3227aa66d700a241f4f78
Author: Jiri Slaby <jslaby@suse.cz>
Date:   Wed Feb 27 00:52:40 2013 +0100

    compat.h: add RDS constants
    
    Signed-off-by: Jiri Slaby <jslaby@suse.cz>

commit 156cbc52319511bac177ca938c5df45c2bd32350
Author: Vinson Lee <vlee@twitter.com>
Date:   Thu Feb 28 11:26:06 2013 -0800

    Add PF_CAIF and PF_ALG for compatibility with glibc 2.12 and older.
    
    This patch fixes these build errors with glibc-2.12 or older.
    
      CC    sockaddr.o
    sockaddr.c: In function ‘gen_caif’:
    sockaddr.c:585: error: ‘PF_CAIF’ undeclared (first use in this function)
    sockaddr.c:585: error: (Each undeclared identifier is reported only once
    sockaddr.c:585: error: for each function it appears in.)
    sockaddr.c: In function ‘gen_alg’:
    sockaddr.c:609: error: ‘PF_ALG’ undeclared (first use in this function)
    sockaddr.c: In function ‘generate_sockaddr’:
    sockaddr.c:788: error: ‘PF_CAIF’ undeclared (first use in this function)
    sockaddr.c:792: error: ‘PF_ALG’ undeclared (first use in this function)
    
    Signed-off-by: Vinson Lee <vlee@twitter.com>

commit 776dd5dce4008308b6f55038b3d34e84f363330d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 28 14:34:00 2013 -0500

    remove signal.h check
    
    this was there purely as an example

commit a7f8cb38da0f1301faa757bbc394c7230af651f0
Author: Vladimir Davydov <vdavydov@parallels.com>
Date:   Thu Feb 28 14:36:53 2013 +0400

    add stub for LLC_OPT_PKTINFO
    
    The definition of LLC_OPT_PKTINFO is absent on some older stable enterprise
    kernels like rhel6. The patch adds stub for it to fix compile problems there.

commit fd0e6a1da838581d6970c62f47dfe8243c413533
Author: Vladimir Davydov <vdavydov@parallels.com>
Date:   Thu Feb 28 14:36:52 2013 +0400

    fix rds compile problems on older kernels

commit da854b694054770209b1f4baf0ed5d38c3780f49
Author: Vladimir Davydov <vdavydov@parallels.com>
Date:   Thu Feb 28 14:36:51 2013 +0400

    fix if_alg compile problems on older kernels

commit 1069dedec6f58ade69a5e3e8be7140a68f8441ac
Author: Vladimir Davydov <vdavydov@parallels.com>
Date:   Thu Feb 28 14:36:50 2013 +0400

    fix caif compile problems on older kernels

commit bcdcc86938642f870bd53aa02f7933408be53b55
Author: Vladimir Davydov <vdavydov@parallels.com>
Date:   Thu Feb 28 14:36:49 2013 +0400

    fix pppox compile problems on older kernels

commit 909e436ce71cb3688e70bde524ffc4aa0a013770
Author: Vladimir Davydov <vdavydov@parallels.com>
Date:   Thu Feb 28 14:36:48 2013 +0400

    configure: overwrite "$TMP.c" when checking features
    
    Just in case it already exists.

commit f3f3086fc155fed502e577434599385585c4549a
Author: Dave Jones <davej@redhat.com>
Date:   Sat Feb 23 22:51:46 2013 -0500

    remove old commented out code

commit 1343de404ae5ccad1228fdade86c41829e83e9da
Author: Dave Jones <davej@redhat.com>
Date:   Sat Feb 23 18:37:06 2013 -0500

    another potential div 0

commit 18261b26a962073788a8919fa8721eebffb5c09d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 22 15:46:53 2013 -0500

    llvm shortcut

commit 25b67176a5839296892cc891308ab006ceeb4329
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 22 15:45:42 2013 -0500

    Fix another potential divide by zero.

commit d2eb9840488687ee5c190889f518dad88746dc49
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 22 00:16:14 2013 -0500

    move child init code into init_child

commit 6d5a262fb4b0ae54c90a394bffbfeead7be1e1fd
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 23:50:51 2013 -0500

    clean up the TODO some, braindump some ideas about syscall chains

commit f08c6fff9112fbf6fe24a8d670dff257fdc31383
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 23:28:44 2013 -0500

    FIXME for dependancy tracking

commit 5f229f05951b9fdd302ccdf2e55b9534f4dc0d06
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 23:24:39 2013 -0500

    remove unnecessary includes of trinity.h

commit d1b72933fc2efaff17bfe6e5f6006e724f5baf5b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 22:58:44 2013 -0500

    move some networking bits to net.h

commit e6f9280f7981583371f305e2881fcbc401cdf4c0
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 22:55:45 2013 -0500

    move random seed stuff to random.h

commit b7bb092cdb270bb72317775d85fe1abd56ce1c2e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 22:51:28 2013 -0500

    move child related bits out to child.h

commit 41c0724a01dd5411ba3e4a0e0773e4bcccd7155c
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 22:42:07 2013 -0500

    reduce includes in trinity.h

commit 9c0d87b82ddb6e37b0d8ba25650060612b1bab43
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 22:35:47 2013 -0500

    move signals related bits out to signal.h

commit 3dba14de463322c96a44334c783cce4949bcaf71
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 22:31:44 2013 -0500

    remove unnecessary include

commit fe3468e291f2d1258b4cb0621c19ee7cdf500ff2
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 22:31:06 2013 -0500

    move networking bits out to net.h

commit 1667527d33b51296c3d98afb7cee6de594f06f7d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 22:27:26 2013 -0500

    don't care about splint any more

commit 72aa201ecefa9a55ffffcd161fe798a5c3e22de4
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 22:26:13 2013 -0500

    move parameters stuff to params.h
    
    also create types.h (so far, just bool)

commit 446d81cedf7bc0fd7b2ddc29b647e2860a2a479a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 22:17:10 2013 -0500

    remove old unneeded externs

commit acdf453b7a2f1da183b30609262ca3b4710e6396
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 22:16:03 2013 -0500

    move mappings stuff out to its own header file

commit 8b9e4543df6ea769873fe77bbe8add328e4e3847
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 22:10:38 2013 -0500

    Enable -Werror

commit 49744cdf0f414e1f025ef2ebd0abae7713e75a22
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 22:10:26 2013 -0500

    move logging stuff from trinity.h to its own header

commit ee3ed4c872109b7a9e75007251159b7fd4ee2385
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 22:03:53 2013 -0500

    move pid related stuff to its own header

commit e4d1ecc16b3010c1c03b3ae0c7c2cb2b763af74d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 21:55:50 2013 -0500

    Move headers to include/

commit a6a73f5ea1fe1e2a75e1fc3ebea55f7808e8cb65
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 12:00:02 2013 -0500

    hush the build process a little.
    
    (stolen from sparse)

commit ae11223b44f894bcd073f562d16a9a0cde6b9298
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 21 11:27:08 2013 -0500

    beginnings of vsock fuzzing.
    
    Just an empty template for now.

commit 9619740bced0034dc727176d9ef911ad95b8af50
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 20 15:41:47 2013 -0500

    When using -V, we can divide by zero in get_new_random_fd
    
    Spotted by Tommi.
    
    This whole function needs rewriting really, it's getting a bit crufty.

commit 3bde656ca301df6307708a851fbb1d9b3cc6b69f
Author: Dave Jones <davej@redhat.com>
Date:   Sun Feb 17 19:38:22 2013 -0500

    make 64bit -1 > -255 be potentially 'interesting' numbers.

commit b1bd9198bf4d0cd444d6279cfbb6dcf03a76a0f8
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 15 16:54:28 2013 -0500

    shortcut

commit ed428932181749f6a8aa6cca8a0af9e60b7623b9
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Feb 11 21:42:20 2013 +0200

    Do not disable sendfile() on ENOSYS
    
    Noticed that sendfile() was getting inactivated due to returning ENOSYS when
    passed some suitable arguments:
    
    [pid  3615] sendfile(304<socket:[73572]>, 622</sys/kernel/slab/:t-0000096/alloc_calls>, [5], 4096) = -1 ENOSYS (Function not implemented)

commit 751523c2c775b9bdb5cf5fffd0884bbb3d3755c7
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 15 13:31:31 2013 -0500

    unnecessary assignment

commit 1fe44da37e0e56344a9e50d06a3c336b0fb334da
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 15 13:30:37 2013 -0500

    remove unnecessary assignment.

commit e37c9d60e09dfb6d1eb9d651674c33816150ec4d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 15 13:27:58 2013 -0500

    check read() return value

commit 7cae89dde55740cf5fe22f973a22127278d6a633
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 15 13:25:23 2013 -0500

    when child logging fails, set all logfiles to the main logfile.

commit 2cb0aec14ec3f23ec06f87f134888e2cda66e7e3
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 15 13:22:48 2013 -0500

    fix leak

commit 2cda23a7434bd77a0d2ddf64da6ffd5883600d9f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 15 13:17:21 2013 -0500

    Avoid potential malloc(0) when no syscalls match group.
    
    Right now this shouldn't be possible, but this sanity check
    will be useful when I start adding additional groups.

commit e6276ffbd179f6e43e738234e3a085c439d10593
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 15 13:11:49 2013 -0500

    Avoid division by zero

commit b649346ec9617c509ad60c15e9c05e833713a4cf
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 15 13:05:57 2013 -0500

    count needs to be at least 1, to prevent malloc(0)

commit 81e8e444c0d6e4431cc39dd9c7e0d144a7d106b9
Author: Dave Jones <davej@redhat.com>
Date:   Fri Feb 15 13:03:40 2013 -0500

    some currently unused sanity test routines for later.

commit 944e911d50e0a0ba8709adbe0b42222358445791
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Wed Feb 6 22:19:02 2013 +0200

    Always allow --list/-L as root
    
    When running as root, we insist on passing --dangerous before proceeding.
    Allow --list/-L to work as root even without --dangerous.

commit 71617a34363e32db4af893b904dbbe324cacbf01
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 5 00:12:30 2013 -0500

    fcntl todo

commit be772a970c0819bbce0ff7aae55e5d70ccf3f5d0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 5 00:11:51 2013 -0500

    Correct the flags arg count for pipe2
    
    Noticed by Michael Kerrisk <mtk@lwn.net>

commit c9faaaf06d24fa6639382430937959e17178a94d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 29 17:01:04 2013 -0500

    Don't bother generating parameter output if we're running silent.

commit 8c5f7de2cbe255893d4abeca0b6c9dc0409c70ac
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 29 01:37:58 2013 -0500

    decode the PF type on socket construction

commit 83775df5cfaf3f5b61339cc6cf3f8111cc059935
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 29 01:25:06 2013 -0500

    Add verbose mode
    
    Move the locking info to be verbose only

commit 5d1f460b64543b4738907e08a678a647170c707e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 29 01:14:04 2013 -0500

    Use fewer pipes.
    
    10 is a little excessive, and doesn't really gain us anything.

commit 063b98d2eb1ef562fa83cb776cc4084a86970536
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jan 28 19:33:57 2013 -0500

    close stdin before calling execve

commit 966ea1372d29d98f5dc0ff04510f4846b3792c25
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 24 13:55:57 2013 -0500

    use some groovy colours to highlight the absense of config.h

commit e3e83635891b1be12ce23a5a73ae602533781fe6
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 22 21:50:11 2013 -0500

    Remove incorrect (and pointless) date from startup message.

commit b9df4a7a9f2f25540305a54ec67c60da726bab7d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 22 11:43:08 2013 -0500

    tag shortcut

commit bf99fe48a15cd7b402d8d5c3e40088feaadec908
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 22 11:40:25 2013 -0500

    v1.1

commit afec2284aae6a074e9d3d494cbd091f8f2b33fe3
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 22 11:40:10 2013 -0500

    Disable -Werror

commit 6c055544ae242572995b585ed19df4c985d882d7
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 22 11:39:18 2013 -0500

    fix regexps.

commit 444c8a67f5bad595e72cee4a468aadfb59bf9b24
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 22 11:33:08 2013 -0500

    split up release/tarball generation

commit 1a5471826efd4a2daa6faec0f4a1efa57fc8fec6
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 22 11:31:41 2013 -0500

    Append, not overwrite.

commit baf36cc1be74ebefa1aa17f97d92f5777e162851
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 22 11:05:01 2013 -0500

    Move VERSION definition into config.h
    
    Otherwise it doesn't trigger rebuilds when it changes.

commit 6d48b1bf664bb641a60e1edd01bcc01eed777805
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 17 12:06:57 2013 -0500

    check taint

commit c7c19c2fde3689cc530bd12ca5f65dadc408dfdc
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 17 10:55:17 2013 -0500

    if we pass a specific net proto, make sure the sockaddrs are specific too

commit c76a9c82c5fd25e3cfbceb38db63da6090f003ed
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 17 10:54:44 2013 -0500

    ignore SIGPIPE in the child

commit c77927f6342fb642f648fa97f53bb5012731ce1b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 17 10:49:05 2013 -0500

    just remove the -N

commit 4d4086b29fca73a8585134335b6f057a811d0075
Author: Nico Golde <nico@ngolde.de>
Date:   Sun Dec 23 19:26:47 2012 +0100

    prevent generate_filelist() from failing if only one file was found due to the current linked-list logic
    
    This comes from the way I use trinity currently, only opening one specific
    character device. The linked-list implementation by default points ->next to
    the first element so that if only one element is in the list, the loop logic
    fails to add that file.

commit 1cf136ec10f5a669af8906e2945ebbdd429f0566
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 11 16:31:10 2013 -0500

    template for file_exists tests

commit 9cd230814ef503c700d8831769b510c2700f9e28
Author: Jiri Slaby <jirislaby@gmail.com>
Date:   Fri Jan 11 21:08:01 2013 +0100

    compat.h: add newer net constants
    
    They are not on older kernels.

commit 65dde7b821f082d79ce7981580db65fb8d0d0008
Author: Jiri Slaby <jirislaby@gmail.com>
Date:   Fri Jan 11 11:51:05 2013 +0100

    compat.h: add RLIMIT_RTTIME

commit 87041dd53931db8922f543511792231024feade7
Author: Jiri Slaby <jirislaby@gmail.com>
Date:   Fri Jan 11 11:51:04 2013 +0100

    compat.h: add SO_BSDCOMPAT
    
    We do not have that in suse at all.

commit 68389b0a204f0f6f47ea5c54b2c6c2ddf318324b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 11 15:02:25 2013 -0500

    Change from return in switch to drop through
    
    No functional change, just debugging cleanup

commit a6e2851d84070c7ab1d771066b3315a9ef82a7cf
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 11 15:00:39 2013 -0500

    initialise sockaddr stack vars.
    
    Occasionally we got sockaddr init trying to happen on shm addresses.
    I think this is just stale stack contents, and this seems to be
    where it's coming from.

commit 7ee0eacbe51d3298babed6799ead6faa82800ebd
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 11 14:58:33 2013 -0500

    sanity check the arguments for addresses before we do the syscall.
    
    This will be expanded over time, but for now, I'm just concerned
    about shm corruption, so check if the address we pass is that.

commit bb90342134c0dc6cc2d220ffd06f1c98cf72c36d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 11 14:57:43 2013 -0500

    When we BUG, lockup afterwards.

commit 25f7f01c9d325c62faaf0876a460e879a3f70e44
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 11 13:28:41 2013 -0500

    todo: redirect stdin/out when we execve

commit 7f56f3d888818a76fb4804386d8746ad244dc42d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 11 12:28:19 2013 -0500

    ignore SIGIO

commit c60e869e756b25ec3a7abbccc733bc5a7d2835db
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 11 11:47:31 2013 -0500

    If there are no pids running, checking the pidslots is a waste of time.

commit 011393b7fd2c99a2ffdc50500c867a66da4292ce
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 10 11:09:23 2013 -0500

    clarify compat.h/configure.sh differences

commit 2ee78ccd4ced8c499fbdc188c841b4d719bd5a02
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 10 11:02:04 2013 -0500

    more backwards compat for older distros

commit b4ea96a5f3d3ef687758f36c654e896423375314
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 9 15:27:39 2013 -0500

    consolidate struct fabrication routines

commit 9f74d81766b6caab92fb40610f528b01f06dd1b8
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 9 13:29:00 2013 -0500

    Use the .deps dir, instead of sprinkling new .d's everywhere

commit c5b254da0ac44edf70ce159da9fb83d74a4d7e7f
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 9 13:03:13 2013 -0500

    smaller deps.

commit ce83ecd37ca667e2b047681844a378afe2242f74
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 9 12:04:20 2013 -0500

    Fix compile.
    
    uglier, but works.

commit d1904911d368e3e9acb41209d4295cbe240bf693
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 9 11:52:27 2013 -0500

    Check that config.h exists before trying to build

commit 84bc42708f7f6dca8159ca92d4486c1aa801421b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 9 11:35:07 2013 -0500

    Fix dependency generation
    
    Now changing header files should trigger a rebuild correctly.

commit 408476393ee10d91fbb3d75165d591e04e97a412
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 9 11:13:22 2013 -0500

    reenable the known ioctls
    
    might as well at least pass down some sensible ioctl numbers.

commit 5015c58395b7249ce136fc917d169f3247bafa38
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 9 11:02:22 2013 -0500

    remove some used-once variables.

commit 8ac260e58198ce911ed7a640f98fcb8376fd315b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 9 10:57:25 2013 -0500

    Use wildcards for source files

commit 0b4acd80d66c9a91bc16c884d45d54478de889c3
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 9 10:56:15 2013 -0500

    remove redundant variable

commit cc0ffd76cb852ed871da731e55fe42e33afdea91
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 9 10:19:42 2013 -0500

    Fix the pppox compile problems. Hopefully for real this time.

commit 5808b927dc9f316f0752f0197e6682c48e2a9f9f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 8 18:03:12 2013 -0500

    don't use sockets when we pass victim files

commit 36e5062cddb7ee1d25df81d37705f8fee1615a35
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 8 18:02:56 2013 -0500

    Make victim files be exclusive.
    
    Don't worry about gathering fd's from other places.

commit 47de2dc70583e9e24be435f95abccb43252930e6
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 8 16:36:10 2013 -0500

    actually, 8.

commit 593306ac7e8461200db402cb17581f43d8f1e783
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 8 15:16:35 2013 -0500

    print syscall names
    
    - split out into separate function
    - document new bug

commit 54387f6bde6d94da99047cab2c45b61af28b2f25
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jan 7 13:50:10 2013 -0500

    Just deal with seeds in decimal everywhere.

commit 193b1f43edf5d101a08b3afd22a0dae217845d2b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 4 17:30:37 2013 -0500

    spend slightly less time regenerating, if we can squeeze out another 50k calls.

commit 90ddf0faaf766d340cf931f5b712157a9d3e119f
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 2 11:00:17 2013 -0500

    Revert "Fix -c to only call the syscall specified."
    
    This reverts commit 819c566ed0c277b8ce80267da12a8ced3e2b8b00.
    
    Turns out that wasn't quite so dumb after all.
    Sasha points out that I did it this way so that multiple -c parameters
    can be passed.  With the commit above, we could only have a single -c.
    
    To do this properly we'd have to construct a special syscall table,
    and then change the rand() to be nr_enabled_syscalls.
    
    For now, just revert. I'll revisit this post 1.1 sometime.

commit c37e1603f7c1b6b4c7afee608bd316f10c9472f2
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 28 22:11:08 2012 -0500

    postfix the test-random tmp dirs with cpu# instead of pid.
    
    We don't need quite so many subdirs.

commit 778d85301eb6f255736c5ba73160101dbb67d327
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 28 22:09:42 2012 -0500

    Fix -c to only call the syscall specified.
    
    We were picking a random syscall, and retrying until it was the right one,
    which was uh, really dumb. Like really really dumb.

commit 66c32f5e4ca18d0b8ee8532a85c763ccf7750338
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 28 17:05:28 2012 -0500

    syscall table walking helpers

commit a218ddec21f2294c2eca5ec5253d42298554eefa
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 28 11:47:00 2012 -0500

    wait for main to be done with regeneration before exit.

commit 400c7a575dd490f700d4f4aacade51ca2dd2a1d3
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 28 11:42:54 2012 -0500

    better name for variable.

commit fd08f260771b503e668d9fa4a3a87c8a6e495f61
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 28 11:38:15 2012 -0500

    english++

commit 17c1ff05dea59badf6ba02b3dacab77fa858994d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 28 11:36:12 2012 -0500

    introduce for_each_pidslot helper macro

commit 78b5133e4d0173b25c06a990a86b64517420e4ec
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 28 11:28:01 2012 -0500

    reap dead children when we wait for the watchdog to exit.
    
    As we're now doing this from two places, factor it out to a separate function.

commit 9db569cfb09b2082210f42ea2f869f9f0e05283b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 21 09:49:00 2012 -0500

    mlock with a len=0 causes us to hang a lot.

commit 9c7192c5badb53a381b9827e3e2ead48b3b47b72
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 21 09:37:24 2012 -0500

    decode the syscall name when we get stuck

commit 0eb61bef44e853d459e1ba388a6537a43c3e240f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 21 00:20:58 2012 -0500

    move the config.h include to the file that actually needs it
    
    instead of needlessly including it dozens of times.

commit 850ba54201ba9250fe2eb47dea23556551f641f7
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 20 12:42:31 2012 -0500

    idea about unions in the syscall struct

commit 266c1ec2c712780fce7cd3500877c9b5e69ce192
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 20 12:34:55 2012 -0500

    Add finit_module definition

commit aff53133f67ad6f434e33b80e5f0a667448fa4c2
Author: Nico Golde <nico@ngolde.de>
Date:   Thu Dec 20 12:22:23 2012 -0500

    add missing header for syscall()

commit 699f1ac3b863d83bae99ed1186defd93c096db63
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 18 10:48:26 2012 -0500

    introduce a configure script
    
    This is just the beginning of a bunch of backwards compat changes
    that allow trinity to be built on crusty old enterprise distros.

commit 0a6bca492d12b40b8ee140af7a91641075a37aa7
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 13 15:35:08 2012 -0500

    use some different sizes on the mmaps

commit 45c2410bf55ac5f2d0babd26a889871459ced722
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 13 15:33:41 2012 -0500

    add a fixme about the shared mmaps

commit 826bc4034c7beec4f5c5a380698cba06af2ad354
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 13 15:16:31 2012 -0500

    Document why we copy the syscall tables

commit 3b795db31aa774dd5c4eb194a8d2ac5a04e1d1eb
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 12 16:59:35 2012 -0500

    Add some hacking roadmaps for the various processes

commit 049e5b2f421ab646a501ab14b0b2bcb79b0f5257
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 12 11:04:56 2012 -0500

    limit the number of syscalls each child does in the harness
    
    Otherwise we can get 1 single long-running process

commit 3a87e2b199e430074b8737a9327a7a15d19fbb73
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 12 11:03:28 2012 -0500

    remove some unnecessary looping in socket generation

commit b1a64a216d6c6833e292a18038e7581e0dd2e55c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 12 10:58:12 2012 -0500

    ctrl-c handling improvements.

commit d7251f58c649829fc9e491a4577bea45052c3ba4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 22:24:38 2012 -0500

    fix the taint check

commit 20cbb22bf39d8bce82d2f18c8c7b509d1f47af3f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 18:13:02 2012 -0500

    skip boring entries when dumping pidmap

commit 2a96abe4f40a71395c8121f17f961ad48baef605
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 18:04:28 2012 -0500

    IPX is kinda dull.

commit 3dfef8d61bfeaa0f988c7efc11e28f85e242ac72
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 18:01:44 2012 -0500

    PF_AX25 support for sanitise_socket

commit c5a36d1a6d350b1f9b6c162c49d734278cf1b97d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 17:51:13 2012 -0500

    AF_CAIF support for sanitise_socket

commit 0f87c1b307f8d4a192f0500ce425c4a7a20ea798
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 17:48:59 2012 -0500

    AF_CAN support for sanitise_socket

commit 8bb5694e3fe83ce711c9528838c1438ce678c0db
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 17:45:57 2012 -0500

    DECnet support for sanitise_socket

commit 0e2a45851fc3d2f6c42860c6ab25204aa66227ba
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 17:43:03 2012 -0500

    AF_IRDA support for sanitise_socket

commit 29abc767a95a818e750a549fe3768650aebfe6e4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 17:38:57 2012 -0500

    AF_LLC support for sanitise_socket

commit 84c287bb087e34220b284fcc75fb3e47e53d72d5
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 17:37:07 2012 -0500

    fix netlink max protocol

commit d7e2e15974a9ef9b5133f29ef302377aa75103de
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 17:34:21 2012 -0500

    PF_PACKET support for sanitise_socket()

commit 64a13b7a6161ef941be7c6431988784f79d714d0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 17:31:45 2012 -0500

    PHONET support in sanitise_socket()

commit 5b49d336a55cad8e441e9bba4a93ae070352002b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 17:28:22 2012 -0500

    RDS support for sanitise_socket()

commit 676227719d9951169fc4d2a2b88db127340eac8c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 17:26:03 2012 -0500

    sort cases alphabetically

commit b24ecc69596bce94cc8506e5092e75db63c1d68a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 17:24:45 2012 -0500

    TIPC support for sanitise_socket()

commit 86e46013006758d6fb12cfafb2bacbbb46ae32fb
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 17:21:05 2012 -0500

    unix sockets need protocol==PF_UNIX

commit 8c028acb7e1c2eaea141c7b96fc72379fbb6e023
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 17:18:15 2012 -0500

    x25 sockets must have protocol==0

commit 58a00d8ed128b352650973ffcc0f53aa29b70d9c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 17:08:41 2012 -0500

    rework the test harness to use different socketcache per instance.

commit a8019441099dd225a6c093321047b2433f77906d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 17:08:09 2012 -0500

    use the domain from the sanitise_socket call instead of the loop index

commit 324ac7d0f6201720da2725112633cd5b6ddfa12d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 00:20:55 2012 -0500

    factor out the signal masking code.

commit 1c9dcb38aadc00aad2578be41e3a920039d0694a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 11 00:19:31 2012 -0500

    Ugly special case that triggered when we generated pppox sockaddrs.
    
    This was a pain to track down..
    
            Program received signal SIGFPE, Arithmetic exception.
            get_new_random_fd () at fds.c:47
            47                      fd_index = rand() % nr_file_fds;
            (gdb) print nr_file_fds
            $1 = 0
            (gdb) bt
            #0  get_new_random_fd () at fds.c:47
            #1  get_new_random_fd () at fds.c:31
            #2  0x000000000040382d in get_random_fd () at fds.c:102
            #3  0x000000000040772c in gen_pppox (addrlen=0x7fff2a202dec, addr=0x7fff2a202df0) at sockaddr.c:417
            #4  generate_sockaddr (addr=addr@entry=0x7fff2a202df0, addrlen=addrlen@entry=0x7fff2a202dec, pf=<optimized out>, pf@entry=24) at sockaddr.c:726
            #5  0x00000000004063d1 in open_socket (domain=24, type=2, protocol=0) at sockets.c:47
            #6  0x000000000040671c in open_sockets () at sockets.c:223
            #7  0x00000000004038eb in setup_fds () at fds.c:119
            #8  0x00000000004032fe in do_main_loop () at main.c:341
            #9  0x000000000040249a in main (argc=<optimized out>, argv=<optimized out>) at trinity.c:179
    
    Later I'll do something a bit less.. dumb, but for 1.1, the quick band-aid.

commit 4c3587bcb4305a4cf18b8c34a50ca497c6435191
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 7 13:10:22 2012 -0500

    explicitly size variable

commit c2ed64de3a9d7aa1d7ba1d961ac3d36a9232c07f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 7 13:09:30 2012 -0500

    remove unnecessary cast by declaring variable as signed

commit 14e48e7914336abd990a2242fb32e54edf6ade3a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 7 13:08:27 2012 -0500

    seeds are unsigned

commit 7592b51dfb3d61f0030339a801fb0dab699f077c
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 7 13:05:17 2012 -0500

    Add -Werror when we're in development

commit ad1b1c9a7fcca0f2afec193e6feb67d5d95de669
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 7 12:58:33 2012 -0500

    Reduce clutter on warning cmdline

commit ee0794ff8ddbc6284e8688183f171fd1dc39a4df
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 7 09:18:51 2012 -0500

    reduce the alarm timeout.
    
    We don't need to wait a whole 3 seconds, 1 should be sufficient.

commit cf6c6fa00d3d359d9b580961572f90110ec5eaa7
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 7 09:17:20 2012 -0500

    fix race when we become tainted.
    
    I saw a case where trinity tainted the kernel, and processes started exiting,
    but one started up in "already tainted" mode.
    
    I'm not sure this will fix it, but at this point, I'm out of ideas.

commit f067c0a7e5d04924b1ee8810cd400a4a0a26f55e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 6 16:10:28 2012 -0500

    duh, missing case statement.

commit 6fe4546288defdc7d25a9f3d0e385be6c75bb3e7
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 6 10:00:19 2012 -0500

    Prevent hang in watchdog on exit.
    
    The children should have exited at this point, but there seems
    to be a race where sometimes the pid map is still populated with
    dead pids.

commit 7386c9171301a11a802256173496c47996c26e47
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 5 20:08:40 2012 -0500

    Move the periodic reseed counter to the shm.
    
    This allows us to reset it when we're doing reseeding anyway
    (Like if we spawn a new child).

commit 8bf813e10207bbcfd5803afbf9d4114c04057e2b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 5 18:10:27 2012 -0500

    pass the domain arg down to generate_sockaddr as a hint.

commit b74b43bc76df23200e515e500d1c2b6543289b1e
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 5 17:53:20 2012 -0500

    some work in progress stuff for the socket creation.
    
    This will fail most of the time right now, as the PF of the sockaddr
    will most likely not match what we passed to socket()
    
    Will need a new variant of generate_sockaddr that takes a hint.

commit bdc8323fa75e464722d82f438e0eb288fc78f5cc
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 5 17:00:37 2012 -0500

    remove unnecessary sleep
    
    we go around the loop again, and will sleep when we hit the waitpid(-1,..)
    in handle_children

commit e1bcc7aee3abb5a019ebf76210e60fbd2da6abb3
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 5 16:50:49 2012 -0500

    add some commentary on the child handling.

commit c03d42f48d005e438b8181282a0b218f48a242a1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 5 16:47:48 2012 -0500

    output exit reason after the children have exited

commit 5b3c5c0033e8b77c423939f79afcd1ea6be0e38a
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 5 16:40:25 2012 -0500

    explicitly wait for the main loop process to exit instead of 'any child'

commit e77fb12ddd878be0bbc4fdea12ae6d2fb658a739
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 5 16:39:54 2012 -0500

    move some non-loop stuff out of main_loop()

commit 26871e824a7e75efbdfbadb4f40a5ff653d63659
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 5 16:38:45 2012 -0500

    wait for watchdog to exit before we exit main

commit 48427fd5b37cb5f4e7d722efbc0876580c5878be
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 5 11:26:22 2012 -0500

    another potential warning

commit 256dfc16c234a4afb8a32b2c6379ab7bff51b9ba
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 5 11:01:50 2012 -0500

    factor out the actual socket creation

commit 56441502b2d2927782047e5c535921b8e5d4fa3a
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 5 10:41:06 2012 -0500

    ignore tmp dir in tag creation

commit 3a8c25502cae0e5d8bd5831564aacdb7dd5ec0ac
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 29 18:29:55 2012 -0500

    ctags helper

commit 3bdb77841f71bb8e9709a8a6c67692fe3d624a36
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 27 22:50:29 2012 -0500

    TODO: per-arch address munging.

commit 2847357026baebd7fabe84855b6228b15001b95e
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:51 2012 +0200

    Nuke the dummy "reserved-223" slot filler struct
    
    Just use syscall_ni_syscall instead of defining a separate struct. The
    dummy support just bloats Trinity.

commit f15419e888f7cbf651085babfd222766499b07ee
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:50 2012 +0200

    Nuke the dummy "vserver" syscall support
    
    There is no "vserver" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit 0975fa4c612b07511e4135803e5e649b51f01389
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:49 2012 +0200

    Nuke the dummy "tux" syscall support
    
    There is no "tux" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit 87151656934139c19a1230ec58b18b8df827fa1b
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:48 2012 +0200

    Nuke the dummy "stty" syscall support
    
    There is no "stty" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit 6527fb898e7317194d4dffccbf5a54d0cd734894
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:47 2012 +0200

    Nuke the dummy "streams2" syscall support
    
    There is no "streams2" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit 5c0a508fc6ede7126620ef155740daee465efde7
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:46 2012 +0200

    Nuke the dummy "streams1" syscall support
    
    There is no "streams1" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit 61d645564dd091f17861adbd10b3a494076e8206
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:45 2012 +0200

    Nuke the dummy "setaltroot" syscall support
    
    There is no "setaltroot" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit fc0f91005407f05a6fd70c098eaa6e512610f11e
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:44 2012 +0200

    Nuke the dummy "set_zone_reclaim" syscall support
    
    There is no "set_zone_reclaim" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit aeb454dc3ab6b9654f6ae5d3b19cd1b99f811bd0
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:43 2012 +0200

    Nuke the dummy "set_thread_area" syscall support
    
    There is no "set_thread_area" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit 9f17fbbce077dc5b5fd1310b4fbd227f87d80803
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:42 2012 +0200

    Nuke the dummy "security" syscall support
    
    There is no "security" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit e463ab50aa8722e918e88ea909427faba0908e25
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:41 2012 +0200

    Nuke the dummy "query_module" syscall support
    
    There is no "query_module" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit 11d00e067edc98d6bbd794ad1408d3e2b3d276bb
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:40 2012 +0200

    Nuke the dummy "putpmsg" syscall support
    
    There is no "putpmsg" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit 26bf5f8ff72b0936a64ff86fb000d0330c679998
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:39 2012 +0200

    Nuke the dummy "ulimit" syscall support
    
    There is no "ulimit" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit 362a09a0a90dbcc399cf51c2150882a2aea825d7
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:38 2012 +0200

    Nuke the dummy "profil" syscall support
    
    There is no "profil" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit 8af7f7f9531eb7bfae589d1ef3305bbe9f82d912
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:37 2012 +0200

    Nuke the dummy "prof" syscall support
    
    There is no "prof" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit d474fc96d634fef10adc1bf09623d37877db1991
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:36 2012 +0200

    Nuke the dummy "mpx" syscall support
    
    There is no "mpx" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit ed529bce25377aac0fa70a7d01cc240928f38ea9
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:35 2012 +0200

    Nuke the dummy "lock" syscall support
    
    There is no "lock" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit 1efb2eecf438204d113f8f356bb3c805243f227f
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:34 2012 +0200

    Nuke the dummy "idle" syscall support
    
    There is no "idle" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit 12fe09e185d3da55aba15afb45f026eaac72a76d
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:33 2012 +0200

    Nuke the dummy "ftime" syscall support
    
    There is no "ftime" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit f32ef5e17e7682428d97d8d1d2b8509bd5380ca5
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:32 2012 +0200

    Nuke the dummy "break" syscall support
    
    There is no "break" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit 2ce2db977eb6b3970043e032ad00a82712ef8d42
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:31 2012 +0200

    Nuke the dummy "gtty" syscall support
    
    There is no "gtty" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit d4a4a018379cade481b7b29bce48ab1e28066913
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:30 2012 +0200

    Nuke the dummy "getpmsg" syscall support
    
    There is no "getpmsg" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit 5e7c6fcce068279a3baab2ef9793b416f0689cb8
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:29 2012 +0200

    Nuke the dummy "get_thread_area" syscall support
    
    There is no "get_thread_area" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit e8c088722686f3fe1eea57c393eb036a11b6c692
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:28 2012 +0200

    Nuke the dummy "get_kernel_syms" syscall support
    
    There is no "get_kernel_syms" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit 9fd8f6994edaaed43b1ef3c5d2d86ed5d58ae8ca
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:27 2012 +0200

    Nuke the dummy "epoll_wait_old" syscall support
    
    There is no "epoll_wait_old" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit 7e166c9c05090dc95b489be3ee3ff84ba9556278
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:26 2012 +0200

    Nuke the dummy "epoll_ctl_old" syscall support
    
    There is no "epoll_ctl_old" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit 3e014c8c31b2e943541264441d6e4f20c24f2a54
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:25 2012 +0200

    Nuke the dummy "create_module" syscall support
    
    There is no "create_module" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit d5d9e3e709e8d4decb116d6c3e666084859dd057
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Tue Nov 27 14:36:24 2012 +0200

    Nuke the dummy "afs" syscall support
    
    There is no "afs" syscall in current kernels,
    so do not bother with it.
    
    The dummy support just bloats Trinity.

commit a6fd4cf9a1409e75e590e26aa65db9cc0993e1c8
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Nov 26 11:20:59 2012 +0200

    resurrect uselib syscall
    
    As far as I can tell, the uselib syscall is still present on some
    architectures. It's easy to fuzz, so bring it back.

commit a49572b870edb4d693bb7a57e99d4221fbede87d
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Nov 26 11:20:58 2012 +0200

    Prefix the watchdog messages with [watchdog]

commit 19ef2acf8487fcc86d13dbd0861ff34d5480fd34
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Nov 26 11:20:57 2012 +0200

    Stop calling the watchdog a thread
    
    The watchdog is a process, not a thread.

commit c9d25a2ec316c690656fd67c9c46ec1cb9eb8ddb
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Nov 26 11:20:56 2012 +0200

    Hide the watchdog() function

commit a53d4f35fb0d7c54efc6e81414ff8b8f470a1156
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Nov 26 11:20:55 2012 +0200

    Really sleep 100ms when spawning watchdog

commit b80550dc6513c57915febaca4e563c1c1b96d1e1
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Nov 26 11:20:54 2012 +0200

    Fix nr_file_fds getting out-of-sync after regeneration
    
    Fix nr_file_fds getting out-of-sync after regeneration. This was causing
    the main process to SIGSEGV itself when accessing the file_fds array
    out-of-bounds in close_files().

commit 884559170d66797fb184941357f085828b0213ee
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Nov 26 11:20:53 2012 +0200

    Fix bogus '&syscall_ni_syscall.h' in syscalls-ia64.h

commit d9db09faad671a19ba33b63e10f6eeba1b03462b
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Nov 26 11:20:52 2012 +0200

    Include syscall number in inactivation message

commit af157b9334bda9884038304fd61957932c05b7fc
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Nov 26 11:20:51 2012 +0200

    Remove useless return statement from end of find_specific_proto()

commit 50133c55d2e0e447d31b6dd264b39f907dcd8da6
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Nov 26 11:20:50 2012 +0200

    Access array directly in find_specific_proto()

commit 2ea340ad8e4970aefffb6962979a2d5b156ec517
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Nov 26 11:20:49 2012 +0200

    Apply the ARRAY_SIZE macro

commit d330d88e402491d410357038e636ec58bfd5325d
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Nov 26 11:20:48 2012 +0200

    One-time protect the shm
    
    Commit bc8bb18a ("move the zeropage mappings further away from the
    shm.") added some protection to the "shm" object we are sharing across
    the processes, by mapping /dev/zero a few times. However, we call
    setup_maps() whenever regenerating, so the /dev/zero mappings will pile
    up.
    
    Provide similar protection by allocating bunch of pages around the "shm"
    object, and mprotect() them to PROT_NONE.

commit 02c34e062d14af9d112d141327784df773fffeaa
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Nov 26 11:20:47 2012 +0200

    Guard against strerror(0) in syscall.c
    
    Printing the error string when there was no error is not very
    interesting.

commit f084003a49738f41a6739c11f1e9188cb0fb3162
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Nov 26 11:20:46 2012 +0200

    Grab errno carefully after syscall
    
    Handle errno carefully: clear it just before making a syscall, and grab
    a copy before we make a system or library call that could modify it.

commit 8360c169e084297d6edfc45491408224cc81736b
Author: Dave Jones <davej@redhat.com>
Date:   Sun Nov 25 15:57:57 2012 -0500

    work on the 'child gets reparented' bug some.
    
    Still not sure why this occasionally happens.

commit 03793ba166a163a38a832fdc8e76d8b31e45d160
Author: Dave Jones <davej@redhat.com>
Date:   Sun Nov 25 15:57:15 2012 -0500

    mbind sanitise improvements

commit 344b70b712a75d030689a63878ee86a2ae251254
Author: Dave Jones <davej@redhat.com>
Date:   Sun Nov 25 15:55:31 2012 -0500

    test-random improvements
    
    - more sanity checks to make sure we're in the right dir
    - bail early when we become tainted instead of starting new instances

commit 441419f811961b15c9f65f777107f0c7f6622907
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Nov 22 09:39:22 2012 +0200

    Avoid unneeded string formatting in output()

commit befef645efdc2c129dcf45833f5c7cdef80206c2
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Nov 22 09:39:21 2012 +0200

    Scale down struct arglist size
    
    Scale down the struct arglist values array size. The most we are using
    at the moment is 23 entries from the array in a few syscalls, so 1024 is
    way excessive. It also bloats the resulting binary size: after this
    patch the stripped trinity binary size went from ~10 megabytes to 500
    kilobytes on my system.
    
    GCC also gives a nice warning in case the array size limit is exceeded.

commit 5535d1f455796be8210d46c47c31f39c40ee749d
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Nov 22 09:39:20 2012 +0200

    show /dev/zero mapping size in debug message in maps.c

commit c81bc9703510126db56391a4c58db112af86ff20
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Nov 22 09:39:19 2012 +0200

    unmap /dev/zero mappings in destroy_maps()
    
    We call destroy_maps() repeatedly during fuzzing, so let's also actually
    unmap the /dev/zero mappings to avoid them piling up.

commit d7279d0b6f2fdd32e106ae565777880198e93abd
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Nov 22 09:39:18 2012 +0200

    hide alloc_zero_map() in maps.c

commit d652556ebadeebed7b660d4ea168d47dcd718cfe
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Nov 22 09:39:17 2012 +0200

    Check malloc() failure in alloc_zero_map()

commit 10e91ea62accf067a2fa6dae8f0cfb6816cca52f
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Nov 22 09:39:16 2012 +0200

    Check mmap() return value against MAP_FAILED, not NULL in maps.c

commit 48dd9cd6e1d96892f4e89365bdddbe8a700f9057
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 20 15:32:47 2012 -0500

    more seeding improvements
    
    - when picking a new seed, make it a bit less sequential
    - return the right variable. we were returning 0 when no -s was specified.

commit bc5eda424abb706297b6421ac54948d8ceeeefd0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 20 11:42:44 2012 -0500

    mapping improvements
    
    - sometimes do huge page mappings
    - make them shared between children

commit 04fd0496b7815bcf78571a9a35d32ef158f33d1e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 20 11:41:55 2012 -0500

    set the seed the forked 'main loop' too.
    
    so we get properly randomised sockets/fd's

commit 21ff624adcb46daeaff97bdc35a42208ac7c2af0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 20 11:41:10 2012 -0500

    move init stuff out of the main_loop function
    
    Leaving it actually be just 'main loop'

commit b15c665f22aa1e2a5dd3e8686edd9524e49d709a
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 19 19:47:33 2012 -0500

    some more thoughts.

commit 414b5548bc8fdbc617429ccf2fe704ef321a56c3
Author: Dave Jones <davej@redhat.com>
Date:   Fri Nov 9 15:13:21 2012 -0500

    socket flags cleanup
    
    - remove duplicate definitions
    - use system headers & compat.h for newer defs
    - use FASTOPEN

commit b971154d5486191a86ac9da9f101c35db1ccd6bf
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 6 16:58:09 2012 -0500

    Seed the parent too.
    
    Otherwise we end up with the same sockets/fd's every time.

commit 60817bd9fc13bb55e4a0c1001d24307f193f80ae
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Oct 25 13:36:25 2012 +0300

    Comment out unimplemented syscall references from `syscalls-sparc.h'

commit 0db5b5938fb50377e202c2ed15386a16c9892261
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Oct 25 13:36:24 2012 +0300

    Initialize `struct flock' per field in `sockets.c'
    
    sockets.c: In function 'generate_sockets':
    sockets.c:26:9: warning: missing initializer [-Wmissing-field-initializers]
    sockets.c:26:9: warning: (near initialization for 'fl.__unused') [-Wmissing-field-initializers]
    sockets.c: In function 'open_sockets':
    sockets.c:139:9: warning: missing initializer [-Wmissing-field-initializers]
    sockets.c:139:9: warning: (near initialization for 'fl.__unused') [-Wmissing-field-initializers]

commit ad2560c9e153fdcc01ecb30192080b46b4fc1d91
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Oct 25 13:36:23 2012 +0300

    Ignore the SPARC `perfctr' syscall
    
    The `perfctr' syscall is no longer provided by the SPARC kernels, so
    let's just ignore it. We did not have support for it in trinity anyway.

commit 841a32cddc7e3cd2a74ae3367a794dce9d0e2cad
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Oct 25 13:36:22 2012 +0300

    find_specific_proto() argument constness

commit 29b9d4805994618783789c90309394f8da532730
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Oct 25 13:36:21 2012 +0300

    Add PF_NFC to compat.h

commit 32423771134d211ff69045ad547b4ad7a121f39b
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Oct 25 13:36:20 2012 +0300

    Add missing PAGE_SHIFT and PTE_FILE_MAX_BITS for SPARC

commit 7b34f45b75ef9724755a6cefa9500ae80e435daf
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Oct 25 13:36:19 2012 +0300

    Fix missing header inclusions in `syscalls-sparc.h'

commit ea8c34f90edfab2bdf8d59199a7bfb3f4479c150
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Oct 25 13:36:18 2012 +0300

    Fix inotify syscalls in `syscalls-sparc.h'

commit c6e6afe3a29366a5907527e999e4cc7bd69f9153
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Oct 25 13:36:17 2012 +0300

    Pass more valid values for the first argument of `getrlimit()'

commit 048f2f5bff132db1ade88ac5a53b516aad9eda1a
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Oct 25 13:36:16 2012 +0300

    Wire kcmp for SuperH
    
    See linux commit cf7599322a6052dae7353f6d986eae06dc7759f5 ("sh: Wire up
    kcmp syscall.").

commit 281d0ff52b0223369a0171dc220a87205c0eae76
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Oct 25 13:36:15 2012 +0300

    Wire kcmp for MIPS
    
    See linux commit ce71d24cffcc2909a53582d66c5926a5a0398441 ("MIPS: Wire
    kcmp syscall.").

commit c099afbbbbb1973abf57996c770ce26f919c912a
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Oct 25 13:36:14 2012 +0300

    Use ARRAY_SIZE to determine syscall counts
    
    Instead of manually specifying the syscall count for each architecture,
    just take the ARRAY_SIZE of the syscall table.

commit 378110cf38daae9d568bd35345b68f062263c8ed
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Oct 25 13:36:13 2012 +0300

    Remove unused `NR_SYSCALLS' defines from `arch-syscalls.h'

commit 32d0d1de4bfeda587eda5f6fb402ef78b0125e01
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Oct 25 13:36:12 2012 +0300

    Hook SuperH `old_getrlimit()' syscall to `getrlimit()'
    
    Hook the `old_getrlimit()' syscall to `getrlimit()', as the syscall
    arguments are the same.

commit d6af1a51b2db259c613fc9eeb1e0bddeafe2ca8f
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Thu Oct 25 13:36:11 2012 +0300

    `cacheflush()' fuzzing on SuperH

commit 6d573d6ad23cd46d774930406df9d925a7177d48
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 22 09:43:38 2012 -0400

    Make sure MAP_UNINITIALIZED gets defined

commit 72d8089d995fe6fa22e40412dbfc10c89078e571
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 18 21:57:52 2012 -0400

    seed is unsigned

commit 25dbebcb94cb71846e7710a1494dc1724b11170c
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 18 17:53:42 2012 -0400

    document why that ANSI stripping is there so I don't forget again.

commit 4eadd2f490b052698743e94ebe7917063d8a3f0d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 18 17:51:58 2012 -0400

    Revert "remove unnecessary buffer building in output()"
    
    This reverts commit 7f6f26f1f84b5c3d6eb9e8b146d2d1ce32da459c.
    
    This *is* needed. We don't want the ANSI codes going to the logfiles.

commit 97b79f3ac0496bd0da51f9cd4f494f181972c962
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 18 17:51:09 2012 -0400

    tweak the 'reuse fd' logic.
    
    This was subtly broken, and now seems to work again.

commit 69451775eaf1342c1f9fbd23e5b42682171be17b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 18 15:35:31 2012 -0400

    Introduce NEED_ALARM flag
    
    This allows us to skip setting the alarm in cases where we know we won't hang.
    (Right now, everything that uses fd's is deemed 'might hang')

commit 5ea111001d271abf11e6830eea15884db2b330a7
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 18 15:35:00 2012 -0400

    log which syscall a process seems to be stuck in.

commit cd52ae3618c278bc4b937e1e40aa826d804a6c51
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 18 13:06:28 2012 -0400

    typo.

commit a2f2de0f9e3666ec9ab6691470d2359f73d7c013
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 18 12:23:33 2012 -0400

    allow ctrl-c to bail out of the filename scan

commit 272a9e460bb0fd0795142040d2ff731e5a89f096
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 18 12:10:26 2012 -0400

    remove unnecessary buffer building in output()
    
    The supression of ANSI codes happens in the macros now, so we don't
    need to do this stripping at output time.

commit 9012db1af1a6f3b1a58f5e4b4f637422f9d321ca
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 18 11:44:00 2012 -0400

    Add some missing flushes of stdout
    
    When a syscall was hung & killed, we never printed it out.

commit 8d65a283f7e4f4a46506e15ac95c118c1c73cd69
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 18 11:17:08 2012 -0400

    reseed when we fork a new child.
    
    Otherwise the newly spawned child does the same syscalls the previous one did.

commit 85dff451d2d10df6e7e2b1f5f3797a3c6ed55f22
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 17 17:59:09 2012 -0400

    remove unneeded signedness

commit 7cf9d548ac66fa24f311a6cebfa825cfcacb9f08
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 17 17:58:32 2012 -0400

    remove splint target
    
    this produces too much noise to be useful, for marginal wins.

commit 50e8268f329d048c00d2a1f2b9668385900a7097
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 17 17:48:45 2012 -0400

    Move this out of the function, and off the stack

commit be7dd8a59acace2d98396e6e45a40baa11226207
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 17 17:25:42 2012 -0400

    remove duplicate code

commit 26fdf9dfd5c612051229b324d2752ba0decf2865
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 17 17:22:36 2012 -0400

    do32bit is a per-child thing.

commit 19f2e130764f4fd4346eec3403903ece7660a414
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 17 16:43:35 2012 -0400

    maintain better counts of the syscalls done per child.

commit 2709b50ec0e4968a049079903f7422ffa7fff666
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 17 16:28:59 2012 -0400

    Revert "microoptimisation to exit faster when -N is near regeneration point."
    
    This reverts commit b8061965480e07e7451112e655355c59ee8f46e7.
    
    This looks suspect, and is probably unnecessary
    
    Conflicts:
            main.c

commit a016c1a200190968203eee270d9649bb5f2d29cd
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 17 16:14:54 2012 -0400

    this doesn't need to be a long long

commit a340fd2a143aa771ba117a90941db3d224d18a52
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 17 16:10:32 2012 -0400

    rename some variables

commit 77d281bdf11fba8b196b6f60617634470cd6e245
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 17 15:53:08 2012 -0400

    This is safe, because page_rand is COW'd

commit d97839ec8517e3e7c3fd75eff97d359a5aafbd7a
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 17 00:56:24 2012 -0400

    move some file related stuff to files.h

commit f2f3a10e7f1b0fc6fda0b0beba715197598f017f
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 17 00:46:30 2012 -0400

    timeval wrap workaround
    
    if a thread completes a syscall after we sample 'old', we can appear
    to have wrapped. Most the time, this shows up as a single second delta,
    but on occasion if the system is heavily loaded it can take a few more.
    This doesn't fix the problem entirely, but it should lessen it.

commit 9cabbb77438b88efaf9b1546b85eb71b5a5b5c10
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 17 00:43:23 2012 -0400

    Don't close non-open fd's

commit 7ee394f7029d9ff0c3511a2d9abaa792139d59a0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 16 21:55:50 2012 -0400

    Rework the filename tree walk.
    
    - Keep an index of every filename as we scan
    - kill off MAX_FDS & fds_left_to_create for better separation
      between sockets/files.
    - don't create so many fd's as before.
    
    Startup should now be a lot faster.

commit 88f32901eb1854971f55e22981b5d6f80a69f0a4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 16 16:16:45 2012 -0400

    factor out 'new seed' function

commit f32208a0fec13b5b1255274bff12ec81a962cb2a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 16 16:12:56 2012 -0400

    xor, not or.

commit cbe2f2407bbba512398402b462cb1ec838a73eef
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Oct 15 16:05:41 2012 +0300

    Support DESTDIR for installation staging directory
    
    In addition to INSTALL_PREFIX, support `DESTDIR', that is used by
    autotools for the purpose of installation staging directory.

commit 236d529a6454a7d22500d4998ba2f245fb90d891
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Oct 15 16:05:40 2012 +0300

    Fix off-by-one in NR_ARM_SYSCALLS
    
    In file included from arch-syscalls.h:24:0,
                     from tables.c:11:
    syscalls-arm.h:393:3: warning: excess elements in array initializer [enabled by default]
    syscalls-arm.h:393:3: warning: (near initialization for 'syscalls_arm') [enabled by default]

commit b4fb27e976e18c146b1c1e7b0b5493961f9e8606
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Oct 15 16:05:39 2012 +0300

    Comment out undefined syscall references in `syscalls-arm.h'

commit c6c6aec89abb7c735c261c9e12dae9050be2f353
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Oct 15 16:05:38 2012 +0300

    Change `syscall_fcntl64' to `syscall_fcntl' in ARM syscalls

commit 58a47c1cc489f8092edac9e6b470616b5fa5abb1
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Oct 15 16:05:37 2012 +0300

    Set all empty arguments lists `void'

commit 7c3fea471fbc68e4fbf678e51eb07e9e3f3afd46
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Oct 15 16:05:36 2012 +0300

    Add missing argument to `find_specific_proto()' prototype

commit 52ef652c04ce7fecbcac369ecb3911c2d1955307
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Oct 15 16:05:35 2012 +0300

    Do not declare `syscall_fcntl64'
    
    Remove the declaration for `syscall_fcntl64', as the struct was removed
    in commit 22b4b44ece ("fcntl64 is really just a wrapper for 32bit
    archs.").

commit 1cbcd91ee9398c238fcd7036a87afa0ec344c544
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Oct 15 16:05:34 2012 +0300

    SuperH support

commit a429325658eff3f21dafe72dc8f8954d95060c09
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Oct 15 16:05:33 2012 +0300

    Prefer NULL over zero in sanitise_ioctl_sg_io()

commit 58e24af039c8fb8295745b6e37dcbf102a197f5b
Author: Tommi Rantala <tt.rantala@gmail.com>
Date:   Mon Oct 15 16:05:32 2012 +0300

    Fix broken include statements in syscalls-ia64.h and syscalls-sparc.h

commit f49031830673557e80761effc2568f18dacb9fc2
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 16 10:05:06 2012 -0400

    factor out seed syslogging

commit 5334bdc4c40fe5f6b4684d7e22d0257b167e3d4b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 11 16:29:57 2012 -0400

    reduce getuid/getgid calls

commit a6a429650f1507e5488c3e98d88573a85383e885
Author: Dave Jones <davej@redhat.com>
Date:   Thu Oct 11 12:17:26 2012 -0400

    use correct type

commit f50f00d24dd85f3cb4f481ce0d5fbce38f5ca58e
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 8 13:48:52 2012 -0400

    nfc isn't in a lot of older toolchains, so add it to compat.
    
    caif should be the opposite.

commit 0b207f1c530c504dfe9be2cb79f5371d856b97a3
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 17:40:48 2012 -0400

    add 0xffff0000 as a 32bit 'interesting' value

commit 5eab053ef73022875a14670245fa6b9bf43e549e
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 17:36:35 2012 -0400

    remove unnecessary comment

commit ed82bac5a6b6c7cd37842352a632d23a51c4b314
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 17:31:51 2012 -0400

    change arg_len to sometimes return "number of (int/long/long long)"

commit f56c10d358554d95f6f58a8de7ab066b1389af72
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 17:13:18 2012 -0400

    instead of just localhost, make up some other IPv4 addresses

commit 2d8f023896b93b5740ec66b3a15716e047a35e0b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 15:43:24 2012 -0400

    move setsockopt values out to their own header

commit a4d0c0db440685ab1b7a7c88f10b1d34f0f8859d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 15:38:59 2012 -0400

    occasionally mangle the setsockopt option

commit f2804749c6f033629b3b6d036af15d63c29df2ea
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 15:35:33 2012 -0400

    finally, ALG support for setsockopt

commit 8279a33c9cf7d5bd83cabb924da2613e6b577ca6
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 15:34:55 2012 -0400

    CAIF support for setsockopt

commit 121d24fb3ee15b2541cbea4b3f6bec192732f2fb
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 15:31:59 2012 -0400

    iucv support for setsockopt

commit e94593c6a5897c9cf3a436f199d893024baa1464
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 15:28:09 2012 -0400

    RDS support for setsockopt

commit d4e739501d41893d090f5f1be178bc4fb5096a10
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 15:12:22 2012 -0400

    Bluetooth support for setsockopt
    
    ugh, messy

commit 048b23306632bb8cf285c5656efd77ae5ac08f19
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 14:47:12 2012 -0400

    pppo2ltp support for setsockopt

commit 3dfc107c577b8f36c7dcbc7d51f7f7c9069c617e
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 14:44:05 2012 -0400

    rxrpc support for setsockopt

commit 13508dcf998fd9bc3b78f60dc6c2f6b82281353d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 14:40:13 2012 -0400

    TIPC support for setsockopt

commit bdb065631542d6676906e9b1572e37330fc63646
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 14:34:58 2012 -0400

    netlink support for setsockopt

commit edf100f26a9988b642c4d537c039e12a0e4a1466
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 14:31:34 2012 -0400

    SOL_DCCP support for setsockopt

commit e8a51663d7aeffd3308f66800cd9584a16f8bfd6
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 14:28:30 2012 -0400

    SOL_LLC support for setsockopt

commit ad85dfdb1b7fadff3406bcec64fb83e1a5e8de74
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 14:23:48 2012 -0400

    IRDA support for setsockopt

commit ded5af61092dbddadeaa8076fce8670a9338e364
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 14:19:53 2012 -0400

    ATM support for setsockopt

commit 51bbfde85dba765622dccaf9710aee31b288bd7c
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 13:07:25 2012 -0400

    PACKET support for setsockopt

commit d284a8b64080221355197fcdafd98e146147765f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 13:03:44 2012 -0400

    X25 is boring

commit af49c317f1e40cbd81fd67e6a3816b8b1840b588
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 12:57:30 2012 -0400

    DECNET support for setsockopt

commit 8c81ad28bd5baaa31dc8509fa530c34ce1c84c91
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 12:52:27 2012 -0400

    ROSE support for setsockopt

commit 1a7e299a2acb7cefd51eb38bfe8063b08b1a7a55
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 12:50:38 2012 -0400

    netrom support for setsockopt

commit 00e84454b50e624abc8eb19b3486ae971b8f0754
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 12:49:02 2012 -0400

    atalk doesn't setsockopt

commit 9953e4db249a09ff4f7e8832d5f1e7fd98e383e4
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 12:47:32 2012 -0400

    AX25 support for setsockopt

commit 642e8b30f6080d3f1c9feb93a739fd406c3909db
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 12:43:21 2012 -0400

    IPX is boring.

commit 09861cba1779cc9655d2eeca250e76257c87655b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 12:41:38 2012 -0400

    these aren't bits, they're values. duh.

commit 45055619a0e0f1a9e626bca1f3d20d4acd305dee
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 12:35:38 2012 -0400

    RAW support for setsockopt

commit 903c068d2eae7a57813616d263ead864d2ab4731
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 12:32:19 2012 -0400

    SCTP support for setsockopt

commit 3231b07abde98a51b2d0b074dfdadf1aeed03ebd
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 12:23:16 2012 -0400

    icmpv6 support for setsockopt

commit 129245b2ade240e14d7ae53285f638e3f854f2ff
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 12:20:19 2012 -0400

    fix cut-n-paste screw up

commit 9b4139c3c0cf9f2b2c409051b376694f910340d3
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 12:19:41 2012 -0400

    SOL_IPV6 support for setsockopt

commit 1774fefbabf8ee232828699e0022f5026ce3079d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 12:13:21 2012 -0400

    udp/udplite support for setsockopt

commit 14d0ea345cb3e67641955b1b8f826b286127350b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 12:02:26 2012 -0400

    SOL_TCP support for setsockopt

commit df0ee14528a01a339e223737e67092840208dbd2
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 11:55:28 2012 -0400

    reorder

commit 25c13cbd3040e5fc31767785a3379623d432cbdb
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 11:55:01 2012 -0400

    move away from using the generic sanitise routines

commit 381f38bbdfe6acb6e80cc1cb1c82105c52aab6f5
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 11:50:42 2012 -0400

    sometimes disable the option

commit 1787111d0ce0520761a493b544be9b2284abe8d5
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 11:48:47 2012 -0400

    fallback case should just use single bit

commit 85bbd166869544c748f7f1ab87c08de6e727bb2e
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 11:46:02 2012 -0400

    SOL_IP socket options

commit ee3056f3ab04c0d30edce25cc787e44a1428dd38
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 11:36:38 2012 -0400

    support more levels in setsockopt

commit cd7c958191d80981aa511f0024469241a0a3ad06
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 5 11:34:35 2012 -0400

    more silence

commit a0c66feee3d35a0dd83c9a5462e2727d63d32dae
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 3 19:17:53 2012 -0400

    get the pid into the initial seed too

commit 64f8acdd8aa051c9679221efc233c142c5cdc5cb
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 3 15:51:26 2012 -0400

    sig is used.

commit 1131830d43c03d9ab777622caa7a5fedff7f183b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 3 12:45:59 2012 -0400

    Make syscall structs shared between children

commit bcd10d89487be21197112626cca0a7691601e8cc
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 3 09:35:58 2012 -0400

    abstract out shared alloc function

commit 90930c4b38152da097f7724bb396e501b427899d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 3 09:03:11 2012 -0400

    move some vars out of trinity.c

commit 43539b658bc71568e8390e43b663469ec5a9bf1c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 3 09:00:05 2012 -0400

    log where the shm is

commit e7d9b4783eb4911ffc75bfafe886d0115fca2c0c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 3 08:58:46 2012 -0400

    remove rand64
    
    just use random() in its place.

commit 2e330e7153869bfa4bb93a47e6244e4731d9371a
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 3 01:29:54 2012 -0400

    Avoid using 0 as a pidslot in the seed.
    
    Otherwise, childs 0 & 1 end up with the same seed.

commit cc21e513a63c3e689f553425cbadeb801e77dc73
Author: Dave Jones <davej@redhat.com>
Date:   Wed Oct 3 00:33:37 2012 -0400

    remove unused wip file
    
    it's going to be a log time before I get around to ioctl

commit b0dbfe91842c6cf89d03992b2e2168ea32045592
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 2 23:39:18 2012 -0400

    move log level into a parameter for output()
    
    The previous change broke logging for child processes.
    This isn't 100% the way I want it, but it's a step in the right
    direction, and at least gets things working again

commit 7c21b5f1f72e418f75d730e41b5853e63d75ce98
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 2 18:53:25 2012 -0400

    Move mapping setup to maps.c

commit b9d42d15577971bff212587b2221646892c6c0f3
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 2 18:47:37 2012 -0400

    remove old commented out code
    
    This isn't coming back. At least not like this.

commit 4a5595f02b4f3bc2d5fb3b045e5fecf8f479cc7f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 2 18:19:17 2012 -0400

    watchdog reseed improvements.
    
    - Reseed every 5 minutes instead of every 10 seconds.
      This is good enough for reproducability
    - While regenerating the fds etc, keep ticking.
    - Stop the reseed counter while we wait for the main thread to reseed.

commit 25d76caa9fbf22abae158298643d30d4f283c4cc
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 2 17:20:35 2012 -0400

    better describe the problem.

commit 5fd892c0b28a9a922573704daaa8840e556366bb
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 2 16:57:47 2012 -0400

    Change how 'quiet' works.
    
    This now allows multiple verbosity levels.
    
    0 = everything, even all the registers
    1 = Watchdog prints syscall count
    2 = Just the reseed values

commit f2271a791d4bc90f4025d5234c81b90e090c4a0b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 2 16:34:50 2012 -0400

    remove noisy and pointless output.

commit 22bd7acbe290c882fa94396d394e74fd5e97f5e1
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 2 15:51:17 2012 -0400

    remove pids from log file names.
    
    This just causes way too much clutter.

commit 1b2923bb5bbe6ce4976dcca826731bf0765494c6
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 2 14:17:05 2012 -0400

    propagate seed changes to child processes.
    
    If the parent reseeds, a long running child may still be using the old seed
    for some time.

commit 4ccc069cdf1d8ed336e1987b0fe2cf95872f183c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 2 12:26:36 2012 -0400

    Log the seed when we detect we oopsed.

commit ec2afc4f7277af507e7ea0dccab63a1b199d8703
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 2 12:24:58 2012 -0400

    reseed every 10 seconds.

commit 01ec0c3c8e3a13a8fd2952ccb31b561c46ecd10f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 2 11:24:00 2012 -0400

    Some debugging code for the bug where the child gets reparented.

commit c1ba13090678472c145b2b897ae7e25a7197d7e4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 2 11:19:45 2012 -0400

    go back to doing sequential test repeatedly for burn-in type testing

commit 9adc295dc2bf43ba79bcbe26ed5556ea9e27a45b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 2 11:19:08 2012 -0400

    move get_pid to pids.c

commit 32ff442b9fff8b647296385d7844207df407d814
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 2 11:17:15 2012 -0400

    Look at the whole pidmap when searching for random pid.
    
    Also handle empty pidslots correctly instead of returning -1

commit 4e94ca536b0f2b189fcf29ff6eee4bcc2e587dc0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Oct 2 11:13:50 2012 -0400

    Handle the case where we get called from the parent while no children around.

commit 973273381d3c9a080a36926db2c43ca05d0b5c22
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 1 22:36:30 2012 -0400

    avoid the thread kill syscalls

commit 8f4a21320e64115785037d1d3516989e64b03ff1
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 1 22:22:46 2012 -0400

    Fix seed=0
    
    Every child was doing the same syscall if seed was set to 0

commit b83530a9c099c9e8892ac44d65033f82b053f4da
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 1 22:22:18 2012 -0400

    ignore RT signals

commit 6e62195bff4c5604a5f21c20e0e9a55bd5811177
Author: Dave Jones <davej@redhat.com>
Date:   Fri Sep 28 15:11:29 2012 -0400

    Don't call rt_sigqueueinfo
    
    This confuses the hell out of trinity right now.

commit a55e085c3e7c60a61e1d28f4861d31a491aa6a91
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 27 18:59:36 2012 -0400

    fix warnings

commit b86dbb8064a86479a78e8cb6d678e01591e4a01e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 27 18:58:45 2012 -0400

    annotate some other syscalls that use sockaddr's

commit aab00ce29cfd23322383df3cc75227671a3f8770
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 27 18:50:50 2012 -0400

    change get_pid to only ever return one of our child pids
    
    (or 0, or 1 if non-root)
    
    sending signals to random processes is a dumb idea.
    This probably explains the bug where tmux was getting killed.

commit a7b41af32af0bf6897b45433474cce2a01991bdc
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 27 12:00:02 2012 -0400

    fix signedness warning
    
    Also add a fixme. Need a coherent story here.
    
    This is why we're randomly seeing tmux getting killed for eg.
    Perhaps we should just always return a child pid.

commit 2ead0869abbf07a0c2b8754196498f5cc6b78291
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 27 11:56:22 2012 -0400

    don't search the whole pidmap when picking a pid.
    
    Just look at the populated part.

commit fb6081d15c1898f0faa4fa22ac655c88572ccdab
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 27 11:55:58 2012 -0400

    Make sockaddr a generic type

commit ac93b2ff623f666b69214eb36d1e30baf558f3f6
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 27 11:55:06 2012 -0400

    max children is an even better variable than current running children.

commit 6acfaeed80a5aa731bb60c8a78eb299be1d30910
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 27 11:53:21 2012 -0400

    lower fd lifetime to number of child processes rather than always 64

commit 606875ccab7288d5c14f23f7461d6efa1d3e28eb
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 27 11:46:08 2012 -0400

    bump up the execcount corruption detector minimum.
    
    On a really fast machine, it's too easy to hit.
    This still might be too low for some machines, hence the FIXME.

commit b813f4a6874251314145d74dac4660b491694f66
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 27 11:38:12 2012 -0400

    Add missing string, to decode shm corruption on exit

commit 5bdcc0f97e6c1891aac198577860837835672cad
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 27 01:09:35 2012 -0400

    split up sockaddr generation into per packet family functions.

commit 56c3fc19c38abba2d0f1df63e0ba4c8997638561
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 27 00:40:52 2012 -0400

    split out sockaddr generation to its own file
    
    multiple syscalls will be able to use this.

commit bfd2fb903a4154f3d5d8d834d2c7d89415f0b3cd
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 18:10:12 2012 -0400

    support all the variants of PF_PPPOX in connect sanitise

commit 2b37e7edbc6f03d6b5834f947d20f21495ba76d5
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 16:27:26 2012 -0400

    don't count number of enabled syscalls.
    
    We only care if there are ANY still enabled.
    So if we find one, we can stop counting, and just return true.

commit 8c307b34099da7fdb77016847fd44fb9eedb5554
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 16:02:40 2012 -0400

    PPPOX has 4 possible protocols.

commit acd59ea83599bbcd98c1255ad48e575f6f9c41f6
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 15:58:19 2012 -0400

    bogus comment

commit b8449abf3271b68ed4a41ca2d433a6b5919e898d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 15:57:41 2012 -0400

    PHONET support for connect sanitise

commit 893b572bb7373c3efdc6d1c13022024cbe5ead89
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 15:55:17 2012 -0400

    ALG support for connect sanitise

commit a5acc0b27c0814d80d0b82f8467a12c770f832ba
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 15:52:40 2012 -0400

    CAIF support for connect sanitise

commit c2a2aeb7783e64f1fdfdc000a1ba5fe63fed22e9
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 15:44:43 2012 -0400

    TIPC support for connect sanitise

commit f0aea5778613da19ef1771cce20cab59697a0c75
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 15:38:06 2012 -0400

    PF_CAN support for connect sanitise

commit 81168f5309f0c769e56c3d2b40555f2659de9b35
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 15:35:07 2012 -0400

    LLC can be the same as NETBEUI for now.

commit ee6bce3b37ccf3b205014d51b7d3dded39b91d11
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 15:34:02 2012 -0400

    PPPOX support for connect sanitise

commit 1c039a44876c35bd62cfbe7221d795e1ad8ae30b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 15:27:44 2012 -0400

    PF_IRDA support for connect sanitise

commit 8c4300f29d1289b853c85b273bf94bf79d7fcb26
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 15:24:05 2012 -0400

    ATMSVC support for connect sanitise

commit cbff00c65ed172296d8dfb6083169bd6e8f9ee18
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 15:09:30 2012 -0400

    ECONET support for connect sanitise

commit ed3d6e36cbff0f392107230f5c69b5e8c9fba0b8
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 15:05:39 2012 -0400

    Add missing register setting

commit 8ed154cc47c7018babce0777c046e22a929ef412
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 15:05:03 2012 -0400

    PF_PACKET support for connect sanitise

commit ad65e7a4edb86357abee9f5cc4ca2f547671e3ce
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 15:01:42 2012 -0400

    NETBEUI support for connect sanitise

commit f85e1dc1534d51db378dcaa73bd4396d57cd2ae8
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 14:56:04 2012 -0400

    PF_DECNET support for connect sanitise

commit d2d90f68212d75a9ddbed574f640109acff419a8
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 14:49:45 2012 -0400

    This length should always be 7

commit 2ad2911e614e3d9a8d97c94999b3a986cd77baeb
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 14:48:38 2012 -0400

    ROSE support for connect sanitise

commit 94eb7c0d82371fb84b1d061f7543d2874455e7bc
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 14:42:27 2012 -0400

    PF_ATMPVC support for connect sanitise

commit 83977bea4a17201e158f8711fc9b6860d02c2309
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 14:35:34 2012 -0400

    PF_APPLETALK support for connect sanitise

commit 04503eb26354cc579f11e6c276ee551cbe8f4157
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 13:51:39 2012 -0400

    IPX support for connect sanitise

commit 544c03617cc271b842062053de7f0dc0cf89d034
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 13:44:12 2012 -0400

    PF_AX25 support for connect sanitise

commit b0593be5a9dda3342c9b27e7f9d15c74e533aae6
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 13:39:23 2012 -0400

    add some missing register setting.

commit 077266f66dbf9078223a80038bca403c5b3599d8
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 13:32:51 2012 -0400

    add placeholders for all packet families

commit 59264286ec7743241fa54ff72ce8f7e35c3e215d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 12:30:58 2012 -0400

    initial NFC support for connect sanitise

commit c711ed12389b3d32c423acb3a526ecfbd527fc65
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 12:27:01 2012 -0400

    netlink socket support for connect sanitise

commit 6634a771699fbd5047b7d5c83837e22ca9c25004
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 12:22:52 2012 -0400

    Add X25 support to connect sanitise

commit a4214636362b4365ca0e60af8a3820648b3011b8
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 26 12:16:10 2012 -0400

    Add ipv6 & unix sockets to connect() sanitise

commit d7c5b58d8132f625c4de8ea8a9119bce5433a822
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 25 14:41:46 2012 -0400

    Switch from SYSV shm to shared mmap.
    
    Because Android doesn't implement SYSV SHM.
    
    Based on a patch from Feng Zou <jef.zou@gmail.com>

commit 0e03d7ae0d2783357ad43f45a646212fdbc2c157
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 25 14:09:40 2012 -0400

    Add a check for potential shm corruption.
    
    This gets triggered occasionally. Still chasing it.

commit 79e3aef82367b6857478283585e820ca3899c4c7
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 25 14:03:03 2012 -0400

    regenerate the ARM syscall table.
    
    This introduces some compile breakage, but the old version had
    *runtime* breakage, which is worse.
    
    There can't be any gaps in the syscall table. It has to match the
    what the kernel expects _exactly_.  The missing syscalls that will
    be pointed out in the build failures either need to be pointed at
    existing entries (Sometimes ARM has syscalls which are the same
    as x86 etc, but in different arg orders -- see 7466ba9ccc), or if
    they are ARM specific, add them to syscalls/arm/

commit 76139908c714ca398cb33a5e90a4467823ea5ce1
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 25 13:39:15 2012 -0400

    clean up arm syscalls header
    
    - Remove ridiculous copyright notice.
    - We don't duplicate the GPL header in every file in trinity
    - Remove some noise from a perl s&r that went awry
    - Remove trailing whitespace

commit 044648bb9b66081030e699eeaddd7cf2719f063f
Author: Pekka Enberg <penberg@kernel.org>
Date:   Wed Sep 19 09:49:36 2012 +0300

    trinity: Add install target to Makefile
    
    This patch adds an 'install' target to Makefile that installs trinity to
    $HOME/bin by default. You can specify the installation target directory
    with INSTALL_PREFIX as follows:
    
      make install INSTALL_PREFIX=/usr/local
    
    Suggested-by: Sasha Levin <levinsasha928@gmail.com>
    Signed-off-by: Pekka Enberg <penberg@kernel.org>

commit a393b33d1fc081e83464c70afe44a0ccda16f3c4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 25 13:07:11 2012 -0400

    a basic sanitise routine for connect()
    
    Based on Eric Dumazet's reproducer for the SO_KEEPALIVE/Raw socket bug.

commit 1d18399de188cb6e7cf6428525e16174cca8f344
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 25 11:13:08 2012 -0400

    collapse syscalls/arm_fadvise64_64.c into fadvise64_64.c
    
    Again, this is just different arg order.
    Keep these in the same files so that they stay in sync from
    any future sanitise routines etc.

commit 97ed4bae527e5aa117719e3291b892879d2a0035
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 25 11:07:22 2012 -0400

    Remove powerpc/sync_file_range2.c
    
    1. It's not ppc specific. ARM also uses this.
    2. As it's identical to sync_file_range, but with different arg order,
       it can use the same sanitise routine.

commit 062e4642b84704c5344321e948f9bb2cd5a1b4e2
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 25 11:06:57 2012 -0400

    fix build. again.

commit 8a9efd05496565536ed57155e6f974d601a203e3
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 25 10:55:43 2012 -0400

    modify_ldt is x86 only.
    
    ugly fix for now, until I have a better way to deal with per-arch syscalls

commit 35792f21712721ce38ac908204d6d3b77b94cdc3
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 25 09:04:16 2012 -0400

    Mention SYSCALL_OFFSET in porting guide

commit f8db286260d9c5e405b04ea2646512c24bdb5cbc
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 25 08:54:37 2012 -0400

    support architectures with a non-zero syscall base in a cleaner way.

commit 23ee4d6151e6cae443a3a88fe86112d05015f071
Author: Maxin B. John <maxin.john@gmail.com>
Date:   Mon Sep 24 12:48:38 2012 -0400

    MIPS arch support for Trinity
    
    This is my first attempt to port Trinity System call fuzz tester to MIPS
    architecture (I am using a qemu emulation of MIPS - Malta).
    
    Compilation using Codesourcery MIPS Toolchain worked fine and I was able
    to run "Trinity" on the emulated MIPS platform.
    
    The "syscalls-mips.h" was derived from
    "arch/mips/kernel/scall32-o32.S" of 3.6.0-rc7 kernel. I have commented
    out some mips specific system calls like "cacheflush" as support for
    these system calls can be added to Trinity at a later stage.
    
    Signed-off-by: Maxin B. John <maxin.john@gmail.com>

commit b140cbce1cfafdfc48479a65027886aa1b45c6d7
Author: Dave Jones <davej@redhat.com>
Date:   Mon Sep 24 21:04:23 2012 -0400

    remove unnecessary allocation.
    
    This caused NR_CPUs*64 bytes to be leaked

commit c267c90f17da67a36de938240a7db3eae3df2c6a
Author: Dave Jones <davej@redhat.com>
Date:   Mon Sep 24 18:33:10 2012 -0400

    don't reseed children before we've waited for our pid to show up in the pidmap
    
    This occasionally led to spew about being unable to find our logfile,
    as the reseed code does an output()

commit 99d4d46256fba85eb311895b2ec98045931b1857
Author: Dave Jones <davej@redhat.com>
Date:   Mon Sep 24 16:30:52 2012 -0400

    make the main loop sleep when idle.
    
    Similar to what was removed in ad27756bef82bb2cb83771697c825d3921676963,
    but a level up the stack, where we have better visibility into what
    we're currently doing.

commit 0abdec27b6f32da35bb9d0a546601b1facc54061
Author: Dave Jones <davej@redhat.com>
Date:   Mon Sep 24 15:45:13 2012 -0400

    kill that warning
    
    (this is WIP anyway)

commit 11fd5d9b8254f11e40b044eead658044a0c823de
Author: Dave Jones <davej@redhat.com>
Date:   Mon Sep 24 15:43:59 2012 -0400

    beginnings of some sanitising for modify_ldt

commit 85ad3034c175bdef1845e906160244f96d51cb3f
Author: Dave Jones <davej@redhat.com>
Date:   Mon Sep 24 15:43:38 2012 -0400

    reseeding idea

commit 89242eba8f21b2685f93bf6004b78288a6313211
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 19 12:13:20 2012 -0400

    add commands list to msgctl

commit fbfe9edc671da53dd6c841586e379c2fffdeae7a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 18 20:53:55 2012 -0400

    constify ioctl list

commit f96e7d1b4890c86df15aea3244119ad0a83fad17
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 18 20:52:53 2012 -0400

    constify protocols list

commit f643c7647150ffbd0fae339c1c75384fbd9783bd
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 18 20:52:01 2012 -0400

    constify syscall tables

commit 282b8845b0b2748782cd38ed5ea557e1a27b748b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 18 20:41:11 2012 -0400

    remove more 'pid == 0' leftovers, and replace -1 uses with PIDSLOT_EMPTY

commit 15afb2fc1d9c2b27201f4d3c2eed930762d63458
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 18 20:35:05 2012 -0400

    time wrap bug.

commit f396d2adcd08c700bede20dc879309909e869da1
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 18 19:58:44 2012 -0400

    reparenting bug.

commit 4fa6ac6bb9228b63ea81d7f9df3e76e5861ddfbf
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 18 19:56:49 2012 -0400

    decode the exit reasons.

commit 2b50fbbfa86d026e978c9ac18f6640ca6389861b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 18 12:32:46 2012 -0400

    move the pid handling code to its own file.

commit 72155102192f5f536241eb99364c941ab8ffc459
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 18 12:25:14 2012 -0400

    Factor out the pidmap sanity check
    
    And call it when we're in handle_children as well as from the watchdog.
    
    This should fix a rare bug where we corrupt the pidmap just after reaching
    the maximum syscall count, and get stuck because pidmap_empty() never returns TRUE.

commit eb9e9752ffa79ae78ed8b05ced5c97fff4d9a547
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 18 12:22:54 2012 -0400

    This sleep should no longer be necessary.

commit fa6612268466a920b00cc36d6bfb5053ba9a5c19
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 18 12:22:30 2012 -0400

    Use define instead of -1

commit 2ed4d9c8e1bab68d2f3318c282d3ac51121b8a77
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 18 12:21:50 2012 -0400

    Remove another check for 0 pid.
    
    No longer possible.

commit fa6be354b7cc3e80492743e5a8abcd3fa2f048d8
Author: Dave Jones <davej@redhat.com>
Date:   Mon Sep 17 17:27:30 2012 -0400

    clear any potential SO_LINGER on sockets, so we don't hang in close on exit.

commit 7f92ef078ee9755bcd1eabb2a04eb02d3c1856d9
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 13 15:52:18 2012 -0400

    Mark the locks in the shm as volatile.
    
    Otherwise gcc optimises them into infinite loops.

commit 40c1ab01c47a6baeffb6c4eb0b2cbefc21db9cba
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 13 12:34:51 2012 -0400

    clean up some pid handling.
    
    '0' is no longer a valid pidslot state

commit 7350cf3b99ad7dda7ee620b50715ebbdf6d48b8d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 12 15:01:57 2012 -0400

    Do possible regeneration before handling child processes.

commit 00485e446f5dbce6695d7d2181ba654d8e888732
Author: Dave Jones <davej@redhat.com>
Date:   Fri Sep 7 14:25:11 2012 -0400

    VM 'victim' pages idea

commit df726603e08bc5319543dee3eccfcee7a93f3586
Author: Dave Jones <davej@redhat.com>
Date:   Fri Sep 7 10:42:48 2012 -0400

    TODO: Fix inappropriate page_rand use

commit 5e5903d1edc96a3d320930391e4ec29c6d45c0e7
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 6 23:26:45 2012 -0400

    Move some documentation around.

commit eb1134b68fa1e120ccb99025c67db80cd1773d60
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 5 17:32:53 2012 -0400

    Just run through the list once.

commit 321f40b0300ca32f066422902f5ae77f2079640b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 5 16:30:04 2012 -0400

    add pause-on-oops to TODO

commit dd75bdfdd7d4fcacfa8b936a18ca036ee6401bc1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 5 11:47:15 2012 -0400

    Some biarch=FALSE fixes.
    
    syscalls_32bit is only valid for use when biarch==TRUE

commit a4c70a95a3aaede1feaa3546718aec9154a85382
Author: Dave Jones <davej@redhat.com>
Date:   Fri Aug 31 01:00:10 2012 -0400

    remove unused 'retries' from shm.

commit c0c98a738c05d1841a5d1639c1e6d2e82a1ff20f
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 30 18:19:33 2012 -0400

    add argument to specify number of child processes

commit c1730988a35e1199012ad1c2ca72650fc6d06ea5
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 30 18:17:16 2012 -0400

    hide the output of the dependancy generation.

commit 46429f7a6af85aa577ac52f7fd75986f196a4a7a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 30 18:06:18 2012 -0400

    maintain the bugs-found list out of tree.
    
    I'll properly tabulate the web page, and linkify it later.

commit 1f94a5144aede410d4d63712dd59bf3d28ef2cc1
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 28 14:35:04 2012 -0400

    Start a new version.

commit 2caa97ac90684931b586b9cca99c18512fc14e2b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 22 21:39:06 2012 -0400

    3.6rc3 lockdep trace. tasklist_lock vs fasync

commit c9278beb231dcc15e5cbff7a72fe931a63624c2c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 8 14:13:49 2012 -0400

    Another bug found by Sasha

commit e282d8ef6314d9a1d06ba37ef3b2b59794b42e37
Author: Dave Jones <davej@redhat.com>
Date:   Mon Aug 6 12:51:44 2012 -0400

    missed a bug back in June

commit f1ddb936b93d5f347c342191b91b7c7c52144568
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 1 12:23:18 2012 -0400

    introduce a bool type.

commit 477455767b67e5c074b4c810196c05d854f23851
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 1 09:56:04 2012 -0400

    another mpolicy bug

commit 4f7157e4939496577cfa5a1a85e551caad2fcc07
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 27 10:22:40 2012 -0400

    introduce a define for the 'pidslot not found' result.

commit c3a1d86039ee81ab14fdf5731360c7633621925d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 27 10:20:44 2012 -0400

    remove unnecessary variable.

commit 694b001a2ccc82c811361d58b6df70e106b8c1fb
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 27 10:17:49 2012 -0400

    Fix dumb bug in logfile number searching.

commit 09e04b5cc0f7549f88f470bd6f9c465bfb516a80
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 24 17:13:41 2012 -0400

    posix timers bug

commit 76c0ecf87eb26a70c65d47110948b98803f8d45b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 20 16:35:44 2012 -0400

    Futex bug updates

commit 0fb7e4debdac470f34dc17da01e92d46889ef911
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 20 16:33:13 2012 -0400

    Sasha's latest roadkill

commit 1242efdaf0d51f5a8912218f57edd62b3cab37c2
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 20 15:15:29 2012 -0400

    remove futex ignorance
    
    there's a patch to fix this on lkml now

commit 36faed9a091e750ca33a82b68588aa97bb264ad0
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 19 23:46:01 2012 -0400

    move the make-it-fail flag to the shm
    
    Saves us rechecking it every time we fork a new child

commit de18cc50a04fcddd0e50ac815b505bd689840a61
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 19 23:41:19 2012 -0400

    update

commit 9fca480e0f29167cf9330c3f0132185fce8248c5
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 19 23:28:21 2012 -0400

    Move the shm sanity check to the watchdog
    
    It seems to make more sense to do this there instead.

commit 5e878f577a78bc5cd4543c9b5ff4766cab500e8e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 19 22:18:46 2012 -0400

    Rework the dependancy generation.
    
    - Don't delete the deps after each build, that removes the whole point of creating them.
    - Store the .d files in a hidden subdir so that they don't clutter the main src dir.

commit 7c4950c64f1458db90ade6f45cef1bf7b95c901f
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 19 21:35:19 2012 -0400

    rename nr_childs -> max_children

commit 2af269a80d84c65a92db3a28149f6d00f2e38faa
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 19 20:58:24 2012 -0400

    rename socks -> nr_sockets

commit c8b56cc70c81f21e5340a7e5e625ee547448a32e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 19 20:52:30 2012 -0400

    allocate a guard page after our zero page mappings.

commit f0170c23260dcbfbbb43b0965a4059d490a08ddd
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 19 20:47:21 2012 -0400

    invalidate the cached fd in the shm when we regenerate fd's

commit c7fa963bfc015dcf0865172be6749b2f33af46e8
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 19 20:41:10 2012 -0400

    oops, fix some compilation errors as a result of the last few changes.

commit 2155d62fe685b711c5e89ada355f92b2533a56b9
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 19 20:38:08 2012 -0400

    add a missing define conversion (harmless, but now wrong)

commit 2a6945c72daca7a9a4b1d585669606d84d51ca68
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 19 20:36:58 2012 -0400

    rename EXIT_FALSE -> STILL_RUNNING

commit 49f1d9672c10040a44a53fb575a182a4c88660e8
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 19 20:35:44 2012 -0400

    rename exit_now -> exit_reason

commit b7b91f1db7574f136ce6c1b8132e2397dde0a386
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 19 20:33:36 2012 -0400

    change exit_now from a boolean into an enum so we can track the reason we exited.

commit 5daa885af25eb5efbefb752eb20e55aa182a20f8
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 19 20:20:41 2012 -0400

    add another sanity check
    
    Another rare bug. Hopefully with debugging info I can figure out what happened
    prior to the fds disappearing.

commit c6c1cff2ba43e0f2629ba4a33dc11d4c870e9556
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 19 16:18:03 2012 -0400

    add some sanity checking to the shm.
    
    Very rarely, there's a case where we get nonsense pids in the shm.
    I've seen it happen twice now, so add some debugging to catch it
    so it can be properly debugged.
    
    If you see this happen, attach to a trinity process with gdb,
    and print out the values of shm->previous_syscallno, and shm->previous_a[1-6]
    Hopefully that will show the syscall that's doing the corrupting.

commit 0b370ea6c38bb3fc0767639f3af940f2e550d81d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 19 16:16:28 2012 -0400

    don't call kill.
    
    Even with ARG_PID being clever, we need more smarts here, which
    can come later.

commit e5facf1d61e3e989a7e909400f29b33a3927b54e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 19 14:59:16 2012 -0400

    Remove open-coded find_pid_slot

commit 2c334bd765ff06c27a2d1c1739094f3d86a0f4bd
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 18 13:49:04 2012 -0400

    add url

commit a3ebbc7a4ac8d64d32d75434357621e0480a90e5
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 17 23:45:25 2012 -0400

    remove some debug code that accidentally got committed.

commit 011a96375ff5d59382e42d5165ef76c3da7238c2
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 17 23:44:44 2012 -0400

    disable munmap for now

commit a3b9d0e4b43fa8c4200fb8728ea1907f59f83134
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 17 23:43:11 2012 -0400

    remove open coded ARG_NON_NULL_ADDRESS implementation

commit cf6f978336ddb806e88c1a0c49b159e3a58e5028
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 17 23:41:02 2012 -0400

    Don't set a mmap address.
    
    This causes segfaults right now

commit 220060b29c242d2c99ddb1ca69adf9a0315e0661
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 17 23:36:56 2012 -0400

    Make the mmap flag selection truly random.
    
    Instead of just 0-N.

commit bfd47f08222bdd3a023a02e28f3827ad708e5d03
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 17 23:04:02 2012 -0400

    output the reason for exiting.

commit 21e2b6238d10e51707bb1558311f673e6b7b1dac
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 17 08:39:16 2012 -0400

    sasha broke sched debug
    
    [sidenote: lkml.org fscking sucks]

commit 1ac68196f7d83631aaae52284fac41fe196bbb26
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 16 19:34:42 2012 -0400

    some bugs that need fixing

commit 2d3ff9c56d1629e21c51848dca85613c798d2dcf
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 16 18:54:41 2012 -0400

    Sasha found an overflow

commit aa1df4099900f7d7edcd0806162d76f051f332c7
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 16 16:48:28 2012 -0400

    debugging

commit d4a803a9b447d7597b30ad77bc749aef51741b4e
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 16 16:34:54 2012 -0400

    ugly workaround for a race.
    
    The pidmap really needs proper locking. I'll do it later.

commit 1948c94bbd71d5a6cacf52e577e335d1695b846d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 13 14:22:42 2012 -0400

    futex bug

commit 694e714c396bf428d992abe02eef9737634c138d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 13 14:00:16 2012 -0400

    setlease bug

commit c80d873c68e83a13b0dd49128703b4c0d0a553e1
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 13 12:17:28 2012 -0400

    return interesting numbers for unannotated arguments sometimes.

commit 4243e8964f5ac2bfff49e5dd1e60cee744b3c289
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 12 12:15:23 2012 -0400

    Give per-arch headers a common prefix

commit add782b6e3546e04fc7251d814972b615d7acac9
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 12 12:05:24 2012 -0400

    TODO: make the kernel addresses per-arch

commit bdae482658a8abe9f05dfd66052fe5d5bb02fc29
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 12 12:03:47 2012 -0400

    return the unmunged address 95% of the time.

commit 2e3032fd04232129a2bb89fe6a1e4cf273912365
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 11 21:37:45 2012 -0400

    sysvipc bug.

commit bba9a61affc63874c88b7a8a00d6998718989c28
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 11 16:48:03 2012 -0400

    more roadkill

commit e91c3062aee4bb910063283c8530322074d5f002
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 11 12:30:40 2012 -0400

    add some scripts I found handy during development

commit 530ebf3560ef88650c4cf517e6850c495402ddcf
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 11 12:16:22 2012 -0400

    only mmap/munmap a single page.
    
    This stops the segfaults we're seeing from happening
    when we try to mmap past the end of page_rand etc.
    
    We can improve this later to use larger sizes for /dev/zero maps etc.

commit 9f19b9dd10fdec99c2a124555d9eda1d5cfa560d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 11 12:04:46 2012 -0400

    sanitise madvise
    
    Limit it to a single page. Otherwise we segfault when we try to advise
    past the end of page_rand or similar.

commit 073299747af9c75b077689a5380c44da1689ec26
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 11 09:49:06 2012 -0400

    networking timer bug

commit d9ccb4a1bfa36ac6ae46f09108a0da11c0066f12
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 11 02:00:57 2012 -0400

    Mix in the pidslot into the seed.
    
    I accidentally committed a brainfart that reintroduced the "every child is exactly the same" bug.
    Do the srand in the child, and make the seed child-unique.

commit cf56a4fa8e3c7f2cc21813ae19a9d9f6f89ff49c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 10 22:40:08 2012 -0400

    debugging

commit d5ff7acda4729ccf661046585ebd4ff1a7a6affe
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 9 21:46:08 2012 -0400

    if we blocked while reading an fd, don't reuse that fd.
    
    There's lots of room for future improvement here, but this will suffice for now.

commit 7d40fb6b959931bd86bed9add17171283fde32e1
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 29 15:25:07 2012 -0400

    new bugs

commit 95b0657f4677c376d8368b0cedd8840d913c8e17
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 28 15:39:09 2012 -0400

    exit the children first, then the watchdog.
    
    This way, if any children get stuck while exiting, we can kill them.

commit f07c287de566c7acb0f6a68b28130785cf43addb
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 27 00:32:50 2012 -0400

    add some locking to the pid reaper.
    
    There was a rare situation where we had both the watchdog, and
    the main process ending up here if a pid had stopped responding for 60 seconds,
    and then woke up and exited at exactly the same moment the watchdog reaped it.
    It looked like this..
    
    [30662] Removing 1556 from pidmap.                      <- watchdog
    [30663] Something happened to pid 1556                  <- main
    [30663] ## Couldn't find pid slot for 1556
    slot0: 1460
    slot1: 1521
    slot2: 1557
    slot3: 1562
    slot4: 1524
    slot5: 1492
    slot6: -1
    slot7: 1538
    [30663] Waiting for watchdog (30662) to exit.
    [30663] Something happened to pid 1460
    [30663] got a signal from pid 1460 (Killed)
    
    Sometimes it happened the other way around..
    
    [8145] Created child 25504 [total:8/8]
    [8145] created enough children
    [8145] Something happened to pid 25504
    [8145] got a signal from pid 25504 (Segmentation fault)
    [8145] Removing 25504 from pidmap.                      <- main
    [8144] Removing 25504 from pidmap.                      <- watchdog
    [8145] Created child 25513 [total:7/8]
    [8145] ## Pid map was full!
    
    This simple locking should prevent it from happening.

commit ed5f978af5eae6b344544b413c4da2bd7e7b7bee
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 27 00:27:00 2012 -0400

    improve iovec code.
    
    - always alloc a whole page.
    - don't try to allocate zero iovecs

commit 00c39158443875d12fbca701b1d775d1e16715cf
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 22:11:37 2012 -0400

    we know this is ESRCH, so don't bother printing errno in this case.

commit b99586d9fc4e8f16cd0bde3f0d21d6b683dd4cf6
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 22:00:56 2012 -0400

    don't call reap_child if we can't find the pid slot

commit a791c90b816f94df22ec8734160e0cff0ff664de
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 21:58:10 2012 -0400

    common helper for dumping pid slots

commit 3a0e608b13cb84f221daece1e307f798b8d896db
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 21:16:35 2012 -0400

    switch 'wait for watchdog' to poll instead of sleeping.
    
    It seems there's a corner-case where we get children sleeping here
    after the watchdog has left the building.
    
    Until I figure out exactly what's going on, this band-aid should suffice.

commit 7e767c49f7830ef5140ab3e5f7af3de1b1fd8abc
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 20:38:07 2012 -0400

    pipes is only ever passed to pipe()
    
    so it doesn't need to be so big.

commit 2e752d37b20b70968047592685381d6a551e6bae
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 20:27:46 2012 -0400

    improve read/write
    
    - reduce read size to 1M
    - increase write size to 1M
    - don't try to allocate 0 bytes
    - in write, don't free filebuffer if null

commit 3c37aa67a32796ed048abad89b2a70f9311160e0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 17:34:47 2012 -0400

    tee only takes pipe fd's too.

commit 46e99dcac4822529b825c519df55f6b9b6276790
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 17:32:12 2012 -0400

    just in case, occasionally pass a non-pipe fd.

commit 2abe523b47fec5abf148aa3032571766f4214be2
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 17:31:22 2012 -0400

    vmsplice fd MUST be a pipe.

commit a0a494f4096a7b6ebd9bdba34b711985773d5153
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 17:27:14 2012 -0400

    sanitise routine for splice

commit dd47810b1205611a93146b444e7599da4ec2490c
Author: Brian Haley <brian.haley@hp.com>
Date:   Tue Jun 26 16:24:17 2012 -0400

    2.6.32 compat changes
    
    Another trinity compat patch.

commit b657b628a795b1ef71f7f2d57514a57a04278863
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 12:22:09 2012 -0400

    only reap if ESRCH

commit 4f5d184dfdf441b0da243ff4471afe63af92d070
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 11:54:49 2012 -0400

    more TODO

commit eb43880a3f9791d862a55c422ad3b6069c0a91a8
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 01:15:00 2012 -0400

    updates

commit 130f200230d4beed4cbcbc4cad41ed29270f0d82
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 00:56:31 2012 -0400

    fix case of monochrome short param

commit 83d1d11ac4e093aa8cf38bc91ddd396d88f5c69e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 00:56:07 2012 -0400

    remove long-ago removed option from help text

commit 2610d1f49f464be3a01d93b25bf9e75fbf6bed31
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 00:34:02 2012 -0400

    ignore window change signals.

commit 4c984f58f3a4e8dc71dfd12545d22ee5ca605758
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 00:31:17 2012 -0400

    clean up the ioctls & syscalls dirs too

commit 0df9867e88e164cfded5dae9a562377c71c8c93e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 00:30:13 2012 -0400

    delete the dependancy files after a successful link

commit 3ce004efcace3471736a171e81094d0b84507ffc
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 26 00:27:00 2012 -0400

    remove unnecessary memset

commit 7416a7ca27d26f5eb25482e1978e0bd4e8475b21
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 25 20:46:59 2012 -0400

    improve setsockopt sanitise
    
    pass random args for non-SOL_SOCKET.

commit a831f8f3a7d9605b5260f8a9cc55c6af166b3696
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 25 20:37:15 2012 -0400

    remove the extrafork code.
    
    it was only ever meant for debugging, and has outlived its usefulness
    (and is now broken with the new process model)

commit 9db8de2f5f95950bde68e0075274fbcbc3d394af
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 25 20:25:34 2012 -0400

    choose correct len for setsockopt based on the option set.

commit b1b522f67a0204499db2b6e927d1c2d847f1629d
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 25 20:15:35 2012 -0400

    add an exit check to the read sanitise too
    
    For the same reason as sanitise_write

commit 7625ff0fe567208324ca4b8173682049e1c78079
Author: Philippe De Swert <philippedeswert@gmail.com>
Date:   Sat Jun 23 14:06:17 2012 +0300

    Add arm_fadvise64_64 syscall support.
    
     This is basically the same as fadvise64 but has a different operand order.
    
    Signed-off-by: Philippe De Swert <philippedeswert@gmail.com>

commit 917d376d7e3969ceccbe3fcad892f5182d644c70
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 25 19:32:42 2012 -0400

    rework main loop.
    
    Split up the initialisation & the event handling.
    
    Now we also wait for both the main process and the watchdog to exit when we ctrl-c.

commit 7d9acc13ac401131ab8147734dbc674155ae0776
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 25 19:27:35 2012 -0400

    check for exit status in loop in write sanitise.
    
    Under memory pressure, we can spin here for a while, so explicitly check.

commit b752eb87ac1a8033f2680886584e1ab30b774b85
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 25 18:26:07 2012 -0400

    don't define ARG_UNUSED

commit 829156d18a20abeb8ed429cbbff941880ce6661f
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 25 18:25:24 2012 -0400

    factor out the watchdog startup and move to watchdog.c

commit 51eaf3d29e9e504e3512d58be9d9118d59288d4f
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 25 18:06:49 2012 -0400

    remove the need for ARG_ADDRESS2
    
    Now that the args are in the shm, we can just go hunting for them.

commit 499e022642ab59b5fda97215b711a777b5c230f3
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 25 17:34:23 2012 -0400

    introduce ARG_IOVEC/IOVECLEN

commit 7c8eceb662a87a3697c324708855eb3da61a90aa
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 25 17:34:03 2012 -0400

    print out the line number in BUG()'s

commit 083bed1b757d530539a11e183e6a6297976dda1a
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 25 14:19:44 2012 -0400

    reinstate the ctrl-c handler.

commit bc2152345e8fed0825183789c8196fe351849341
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 25 13:49:03 2012 -0400

    ignore ctrl-z for now.
    
    sometimes, a tmux attach will send this for some reason, and trinity shits itself.

commit 677cde8f0b16f2f480d5216628f052e3506c791e
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 25 13:10:41 2012 -0400

    avoid ptrace for now.
    
    This is going to need some more thought.
    There are a number of awkward cases where we end up with zombies.
    One example:
    
    child a ptraces child b
    child a segfaults
    child b waits forever for a PTRACE_DETACH/CONT
    
    attempting a PTRACE_DETACH from the watchdog gets -ESRCH, because
    it wasn't the pid that initiated the tracing.

commit 7520eb6c94b49ce216940e1ce936db4758844fd9
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 25 13:08:35 2012 -0400

    factor out child handling, and iterate over them all even if no events are pending.

commit 6544d46ad0eb2c123ce6cbbb0d5d46722f01ac54
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 25 13:07:40 2012 -0400

    use PTRACE_DETACH, not PTRACE_CONT
    
    Otherwise we may end up reparenting the child

commit 9ca0a2ceabf107bc1823588feaa1ba5e17c721ca
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 25 13:06:01 2012 -0400

    Make the watchdog zombie checker simpler.
    
    Don't screw around with ptrace, just search for the pid with getpgid.
    Only reap if the pid is no longer around.

commit c761b4086af55535da64ca2936a6a31ffe98f6e1
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 25 12:14:31 2012 -0400

    sanitise setsockopt

commit 083c786c80a13bd728599fd60de2f20d1ae91e9a
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jun 24 21:30:41 2012 -0400

    avoid set_tid_address
    
    Screwing with this confuses things horribly.

commit b60b183b770f983bb9513fcf8bc32cf7553ee706
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 22 15:56:38 2012 -0400

    handle (and ignore) sigalarm in child processing.

commit 697d1825a7b8f9ea2d193d48000eee9ce8b9fd9d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 22 15:11:33 2012 -0400

    Try the 'jump back to the child handler when the alarm goes off' thing again.
    
    A number of things have changed dramatically since I disabled this,
    so things may be good enough to reenable this feature, which
    should speed things up a little.

commit 4e7cd20c6f2de362ab0216918b3c1d4f48ee192a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 22 14:50:25 2012 -0400

    handle the case where ptrace succeeds explicitly

commit 309cbddf9ce4dfd53a5169112bf8946caaf33035
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 22 13:24:29 2012 -0400

    update the parentpid after forking.
    
    Since 01d0d494eb6777119a1a87983bf8eb580a591cd3, parentpid changes,
    so we need to take that into consideration both before and after
    the child handler.

commit f5ef0d10a2afd2613a5d9e2bbca025ba345057ae
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 22 12:44:27 2012 -0400

    mention the pid of the child that was stopped

commit f02fbb79815255d50c56ad147fe41d711874cb65
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 22 12:33:51 2012 -0400

    sleep when this "shouldn't happen" case happens.
    
    A recent change seems to have started to occasionally trigger this.

commit b5de5886ee36c4d08655a172552b5637f8d78816
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 22 12:32:46 2012 -0400

    handle signals when ptrace'd properly
    
    we don't need to send a CONT if the signal was a segfault or whatever.

commit 6bb9c51e26584047d8e05b1fce3d90366a0b5aa1
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 22 12:21:30 2012 -0400

    process signals from all children, not just the first one to wake up.
    
    I think the problem with the zombies sticking around too long was due
    to them being starved for attention in this function.
    
    If one pidslot kept exiting/forking, all the others weren't getting
    processed even though they were waiting for the parent to acknowledge
    they were about to exit.

commit 0db2e7f8c7eff8b0273753110c1fb793827fe6d0
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 22 12:03:26 2012 -0400

    add missing cases for generic sanitise

commit ecc85cbaef8167e0d3b4b4543fdc2f453fe3f160
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 22 12:01:43 2012 -0400

    move parentpid into shm

commit 865a9ebf8436a01b8e17161395298ac2621152e9
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 22 10:17:39 2012 -0400

    Sasha found a new tty bug

commit 5aa93e97d4fd18bd22e24b7191bd242449f4b5eb
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 21 19:38:56 2012 -0400

    gah, debugging left in the last commit actually broke some stuff.

commit 512cd69af232f2d934b030f6d92364267430f53f
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 21 19:36:30 2012 -0400

    sanity check the tables on startup for missing members.
    
    Fix up some missing ones.
    (recvfrom was the only one that was really bad).

commit 385c10ad927a9baadf275b770a99b05239b71414
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 21 18:59:11 2012 -0400

    log the previous syscall (for debugging purposes)

commit 73ccd10a28dab8f466c02abb3561b1287d49e6f1
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 21 16:52:31 2012 -0400

    fix segfaults in move_pages sanitise

commit e677ec6c013e6f8467cdd09ea01573f7fa6a6fa5
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 21 15:33:24 2012 -0400

    Fix infinite loop.
    
    If child1 exits because there's no more syscalls enabled, it sets exit_now,
    and the watchdog exits. After it has exited, child1 exits.
    In the meantime, child2 hasn't been scheduled. It then gets stuck in
    this loop with no way out.

commit d2a969058277833869dbb5981c6ac46e51aed779
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 21 13:18:40 2012 -0400

    Don't share a common parent between child processes & watchdog.
    
    This allows us to remove lots of code to special-case the watchdog
    in the child processing.
    
    process tree now looks like this..
    
    trinity─┬─trinity-main─┬─trinity-child0
            │              ├─trinity-child1
            │              ├─trinity-child2
            │              ├─trinity-child3
            │              ├─trinity-child4
            │              ├─trinity-child5
            │              ├─trinity-child6
            │              └─trinity-child7
            └─trinity-watchdog

commit bfed2595210aad5158306d94a58ef5c4b9d1de78
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 21 12:51:44 2012 -0400

    move the final wait for watchdog exit into the main loop
    
    This should be superfluous anyway, as the children shouldn't have exited.
    It's here purely as a sanity check in case it takes a while to exit.

commit 1d3c946adf3a8b5865b415aa2eab173e9d28da33
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 21 12:38:06 2012 -0400

    Make sure *all* children exit before the main loop terminates.

commit 38f81e60fa7cbcd626e88c8882eb311a4415c7f0
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 21 12:28:42 2012 -0400

    handle watchdog pid a a potential child.
    
    ugh. This might be better if the children and the watchdog didn't share
    a common parent.

commit 7503f09898d6342bab800238ca8d8d7a068ab444
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 21 12:05:40 2012 -0400

    Don't reap zombies.
    
    This is worse than leaving them running, as we end up essentially forkbombing
    if something goes wrong.  By leaving them, we cap the number of children
    that we fork.

commit 6d95adf1b3ba68608f52c81abbb24cf192beaf49
Author: Philippe De Swert <philippedeswert@gmail.com>
Date:   Wed Jun 20 15:50:17 2012 +0300

    Enable sync_file_range2 syscall for arm also
    
    Signed-off-by: Philippe De Swert <philippedeswert@gmail.com>

commit d37ac400420c8ad1ea238948aca8b301da54a551
Author: Philippe De Swert <philippedeswert@gmail.com>
Date:   Wed Jun 20 15:50:16 2012 +0300

    Add syscall_pciconfig_iobase (for arm and ppc)
    
    Signed-off-by: Philippe De Swert <philippedeswert@gmail.com>

commit 07e0adfa3a81da0b7981cbc70a2f001d5a2fb6e4
Author: Philippe De Swert <philippedeswert@gmail.com>
Date:   Wed Jun 20 15:50:15 2012 +0300

    No need anymore for assigning syscalls_32, BIARCH flag takes care of that
    
    Signed-off-by: Philippe De Swert <philippedeswert@gmail.com>

commit 3f40beb5bb4f0d11d7bc2bae4a296dd26d742da8
Author: Philippe De Swert <philippedeswert@gmail.com>
Date:   Wed Jun 20 15:50:14 2012 +0300

    clean-up syscalls-arm.h
    
    Signed-off-by: Philippe De Swert <philippedeswert@gmail.com>

commit cdb44b4a0d699abd98a8d5987dd2ee822b038f15
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 20 10:27:21 2012 -0400

    quit test harness if taint flag changes

commit d5df7b1bf56905f03d142dd43752bb610c2d4777
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 23:41:33 2012 -0400

    Move the total fail/success to the iterations printout rather than every syscall.

commit f0ce7e66311c884c9a52dc550ac7a38cc5e5b1dd
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 20:35:39 2012 -0400

    convert mmap to use ARG_OP

commit be7a0ebd5cd6c7bda99edbe49a7c32f6ad8116a6
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 18:48:03 2012 -0400

    remove unnecessary double return

commit bf81684a8675b9fe8a791935fa796c6e13f68ff2
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 18:34:43 2012 -0400

    handle the case of ENOSYS and -c correctly.
    
    When we have no more syscalls left to run, we got stuck in an infinite loop.
    Refactor the validation code accordingly.

commit 519b65f90ab4b3166fe9a514552e2b6428bd9b8d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 18:33:30 2012 -0400

    initialise the regeneration point to the correct value

commit 5260d073b5347a04dbb7be5a82ac39dc3654770b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 16:40:51 2012 -0400

    add dependancy files to ignore list

commit 95f9f44a5d33a0d49799f2eb7f061bdf8e88cc04
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 16:40:24 2012 -0400

    only log the "reached maximum" from the first child to hit it.

commit 56a1a5b8edd71d2d165021025b2cbeec2349af25
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Tue Jun 19 22:20:13 2012 +0200

    Makefile: take header file dependencies into account
    
    Shamelessly copied Scott McPeak's examples from:
    http://scottmcpeak.com/autodepend/autodepend.html
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit aaff35a7aad72ec53ec96f65de37d76333a56ece
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 16:27:10 2012 -0400

    note about futex & ENOSYS

commit bd49d3b191842b8374b855caa1d95503a6b4fc8f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 16:19:22 2012 -0400

    silence warning

commit 6c73cfcdaa23d68a3a3cb38187b1effdc37eff66
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 16:18:35 2012 -0400

    Introduce ARG_OP
    
    This is like ARG_LIST, but only a single value gets set.
    Used when we want to pick a 'command' argument for eg.

commit ea58721f7d3d8403d06451db2f31086044c6bc74
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 16:17:40 2012 -0400

    more seq harness improvements

commit d7195110e63d15b97f0ec61dac3c394b8b198382
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 15:26:48 2012 -0400

    don't log messages about EBUSY, just send a kill

commit 7ab7e0606e9d5aca82653af6fb5d80ca4edec072
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 15:24:30 2012 -0400

    missing pid arguments

commit 2a1add0107008655b5e33a47677343aee3063aef
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 14:55:29 2012 -0400

    Don't worry about ENOSYS on futex
    
    This could use some improving maybe. What a pain.

commit d3b344b638d217a71628a6745dec7ca8cc72ab01
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 14:51:07 2012 -0400

    log more information.

commit 0eef4ce6985319dd8144a75dcb5c1addf60b4b4b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 14:46:15 2012 -0400

    Abort if the watchdog segfaults.

commit 7c96bd6242830cb90211e5e8344194f8eb99ddb4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 14:20:21 2012 -0400

    microoptimisation to exit faster when -N is near regeneration point.

commit adefe6539c4286787c805e470f29c7bb54b04ecc
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 14:13:43 2012 -0400

    make sure there's always at least one syscall enabled.

commit b97347f0156dfe00d1ee9f8aa2f502e56264c9f0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 14:00:43 2012 -0400

    fix up sequential test harness
    
    - arg moved
    - drop mbind exclude
    - do 'N=just under the regeneration point' syscalls.

commit a9b93353b7780594e08f65dfac3e8b7ede96bcc2
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 13:57:53 2012 -0400

    argument moved

commit 2352e49de84efe430f84d3fffe7db15a49f90ae0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 13:55:30 2012 -0400

    remove mbind exclusion

commit 199cac96940b17471d2e5ce277c79766ecd4e6bd
Author: Philippe De Swert <philippedeswert@gmail.com>
Date:   Tue Jun 19 17:36:25 2012 +0200

    ARM support
    
    Signed-off-by: Philippe De Swert <philippedeswert@gmail.com>
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit dc6bc58559dbcf5e2d8eff3816897ee1dbba28ee
Author: Mikko Rapeli <mikko.rapeli@iki.fi>
Date:   Tue Jun 19 17:36:24 2012 +0200

    Copied i386.h to arm.h
    
    Content seems to be close enough.
    
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit 258d5dfcfb03c1c9911c2475d16b9e0c6a034298
Author: Philippe De Swert <philippedeswert@gmail.com>
Date:   Tue Jun 19 17:36:22 2012 +0200

    ARM syscalls
    
    Signed-off-by: Philippe De Swert <philippedeswert@gmail.com>
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit 18e40b8cfbbf0b773138745c821e803d8e9c92a1
Author: Philippe De Swert <philippedeswert@gmail.com>
Date:   Tue Jun 19 17:36:21 2012 +0200

    Disable -Werror since syscall 32 is not implemented yet for non-x86 archs
    
    Signed-off-by: Philippe De Swert <philippedeswert@gmail.com>
    Signed-off-by: Mikko Rapeli <mikko.rapeli@iki.fi>

commit b4b168ab8ff4796e2fe8c4fcceb7f613362eab8d
Author: Jan Stancek <jstancek@redhat.com>
Date:   Tue Jun 19 09:21:22 2012 -0400

    fix buffer overflow in log.c
    
    25 bytes is too short, bump it to 64.
    
     (gdb) bt
     #0  0x00000080f67d8610 in .raise () from /lib64/libc.so.6
     #1  0x00000080f67da678 in .abort () from /lib64/libc.so.6
     #2  0x00000080f681c368 in .__libc_message () from /lib64/libc.so.6
     #3  0x00000080f68b8fb4 in .__fortify_fail () from /lib64/libc.so.6
     #4  0x00000080f68b6464 in .__chk_fail () from /lib64/libc.so.6
     #5  0x00000080f68b56b0 in ._IO_str_chk_overflow () from /lib64/libc.so.6
     #6  0x00000080f6821490 in ._IO_default_xsputn () from /lib64/libc.so.6
     #7  0x00000080f67f0254 in ._IO_vfprintf () from /lib64/libc.so.6
     #8  0x00000080f68b576c in .__vsprintf_chk () from /lib64/libc.so.6
     #9  0x00000080f68b5674 in .__sprintf_chk () from /lib64/libc.so.6
     #10 0x0000000010006388 in sprintf (__fmt=0x1000e7f0 "trinity-%d-child%d.log", __s=<error reading variable: value has been optimized out>)
         at /usr/include/bits/stdio2.h:34
     #11 open_logfiles () at log.c:29
     #12 0x00000000100024e8 in main (argc=<optimized out>, argv=<optimized out>) at trinity.c:257
    
    Signed-off-by: Jan Stancek <jstancek@redhat.com>

commit 59bc3bbcd877c95ae267c27c0df1be1d4343d990
Author: Jan Stancek <jstancek@redhat.com>
Date:   Tue Jun 19 08:23:09 2012 -0400

    initialize max_nr_syscalls for powerpc
    
    Signed-off-by: Jan Stancek <jstancek@redhat.com>

commit 76d6343e23eea4d07c35dedbd072076f7b08d7e5
Author: Jan Stancek <jstancek@redhat.com>
Date:   Tue Jun 19 07:00:48 2012 -0400

    add sync_file_range2
    
    This is identical to sync_file_range with one exception,
    order of arguments is different. Define both in sync_file_range.c
    to use common sanitize code.
    
    Signed-off-by: Jan Stancek <jstancek@redhat.com>

commit 5eb32d23b8be8a98855067d1614e7c6ee7268dbe
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 19 12:56:19 2012 -0400

    backwards compat for ptrace

commit 487d269bb32da0725079829eec9f28b20f137fe3
Author: Jan Stancek <jstancek@redhat.com>
Date:   Tue Jun 19 06:14:04 2012 -0400

    ignore arch_prctl on powerpc
    
    Signed-off-by: Jan Stancek <jstancek@redhat.com>

commit 8afea4ce25c1fbf301cd86e67542188a0cda440d
Author: Jan Stancek <jstancek@redhat.com>
Date:   Tue Jun 19 06:03:02 2012 -0400

    fix syntax errros in syscalls-ppc.h
    
    Signed-off-by: Jan Stancek <jstancek@redhat.com>

commit 8f51986b44978cd7bc9febcfd845d69f1717ec7e
Author: Jan Stancek <jstancek@redhat.com>
Date:   Tue Jun 19 05:54:22 2012 -0400

    supress unused param warnings in syscall32()
    
    arches that do not define any code for syscall32()
    are failing to compile, because gcc complains
    about unused parameters. Supress these warnings.
    
    Signed-off-by: Jan Stancek <jstancek@redhat.com>

commit a387deb1a7a8749cd6765b67e037e40ffb949f38
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 18 12:29:37 2012 -0400

    3 more 'interesting' 64-bit values

commit f7491667fd7df6596847d73799508041fa6b3fe8
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 14 18:38:09 2012 -0400

    pause the children when regenerating fd's

commit de7a30ba1da8a742ad88577375cb0086f84434ee
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 14 18:23:41 2012 -0400

    oops, we want to generate sockets on startup.

commit 3c65a3647066fcb74dcf82e39f5422bfa69dd249
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 14 18:21:17 2012 -0400

    remove dead prototype

commit f56092e5eb9f86fb07371d9512347ab603f200c0
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 14 18:20:47 2012 -0400

    remove duplicate code that was closing sockets.

commit 9cd9a067c5537f5bcc4d1dc452151ca3634be864
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 14 18:19:51 2012 -0400

    don't regenerate sockets
    
    This is a pain in the ass, because the socket state keeps the fd's open
    for a while even after we close() them.

commit e3f76037e6d11356575daa3ee418cbc59e4e3d78
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 14 17:35:00 2012 -0400

    regenerate needs to close pipes/sockets too.

commit fe1a318ae456bde60a47ef1ab5098625379e3d06
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 14 16:03:47 2012 -0400

    disable the . .. passing for now
    
    it's a pain in the ass for debugging when it chmod's the sourcedir to something unreadable.

commit bf9a24f5c3f1ac00a9b0ed1ab23d642edeebc680
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 14 16:02:26 2012 -0400

    elaborate.

commit 6d280e6ab02867a68b4223ef2890d60951dd6965
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 14 15:37:05 2012 -0400

    regeneration should reopen pipes/sockets too

commit ff67a42436b00991003cb9992a24fe9b32ae0b55
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 14 15:35:52 2012 -0400

    reverse how the regeneration counts.
    
    Also move it to the right place. (main loop, instead of just on init)

commit d7203b1fdbb6b59c8c39d3136b38ebd5f701eec2
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 14 15:34:39 2012 -0400

    Change watchdog to send kill after 30s, and then reap only after 60s
    
    This gives processes time to exit cleanly, which should reduce zombies

commit 8b022e5c70c55b1ca126746ebe8a316333a0f0bd
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 14 15:34:00 2012 -0400

    increase regeneration point

commit 521c1af4647f80eeab5387f7a214e80f1641f3ed
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 14 14:07:54 2012 -0400

    handle more errors from watchdog ptracing

commit 30072ab9f5b1123f8e8d3796be7fa5bf23c122f6
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 14 12:26:24 2012 -0400

    always fflush stdout before we fork.

commit 30eacc8a734585ba4cac549e6310e744a4b49f22
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 14 11:39:36 2012 -0400

    Make the watchdog look for pids that have disappeared.
    
    If the oom-killer has taken away one of our children it could get confused.
    If we can't ptrace it, it's probably been oom-killed.

commit 23b2fd6fa8c7819aa0991f6d1e480d20599da6d0
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 14 11:17:06 2012 -0400

    don't reap in the watchdog.
    
    Let the child handler deal with the pidmap cleanup.

commit 25185bccd0b11cde94e9456141a6cc73e46b5614
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 14 11:10:34 2012 -0400

    Just reap a single child each pass through the watchdog.
    
    Reaping one child might free up enough resources that other
    children are blocked on, so they might continue.

commit e3995008789bb9fb5de461a468d79da5cc145d11
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 13 22:58:15 2012 -0400

    do ARG_ADDRESS half the time for ARG_ADDRESS2
    
    sometimes we'll want completely different addresses, not just
    same/mangled versions of what the first address was.

commit b05bc71edd56c2e81a22e8709536c1f07f883ad8
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 13 22:35:01 2012 -0400

    check for ARG_NON_NULL_ADDRESS too in ARG_ADDRESS2

commit 8747a95739ad88b09c2f83025d1af225369c08d1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 13 21:51:48 2012 -0400

    For ARG_ADDRESS2, munge the result of the previous ARG_ADDRESS

commit 077411ed900c1aeec266b716107189b25e048c90
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 13 21:23:35 2012 -0400

    Move the syscall number & arguments into per-child structs in the shm
    
    This is necessary for future work where one sanitise routine depends
    on the result from another.

commit 9259696348f98ef18c80e10811e5500d86ba64ef
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 13 18:51:09 2012 -0400

    There's no flags arg for mlock.
    
    Looks like I got carried away and copied mlockall

commit 7c133e13939b9570c92ef1cecef5e0deeabe073a
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 13 17:53:17 2012 -0400

    annotate syscalls that take a second address as an input
    
    No difference to ARG_ADDRESS yet.

commit 7c66ffd25629e18d365c0e60acaa269f534b7a29
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 13 17:30:44 2012 -0400

    add IA32_PAGE_OFFSET as two 32-bit interesting values

commit ef36409abe9a0faea50b21c271775165dabaa3a4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 12 16:12:27 2012 -0400

    another NFC bug

commit 662d6dfbbe2a2c9f557eb993e66472f8b37fe98c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 12 11:26:45 2012 -0400

    another bug. lkml.org continues to suck

commit 739056945235f78d1f8f34ac36b6cef1a8fcb4ee
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 11 15:48:24 2012 -0400

    sasha broke the block layer.
    
    (no lkml link yet, because it sucks, again)

commit 695314ebede358671b0ae03ba834fbb3f1db2b13
Author: Brian Haley <brian.haley@hp.com>
Date:   Mon Jun 11 14:17:27 2012 -0400

    Compat changes for 2.6.32/11.04 compile.
    
    Found these trying to compile trinity on an old 2.6.32/11.04 box.
    
    Signed-off-by: Brian Haley <brian.haley@hp.com>

commit 8ea0a5d638f8ee453a86fcfa76a714bd38e7344c
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 11 13:48:56 2012 -0400

    Fix the header. oops.
    
    (Three guesses where I copied this from)

commit ba788b979a394d15fb964cf55f7a50f601b7d4b5
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 11 13:47:45 2012 -0400

    add license text

commit 4cd2aa39ab9d8a76135e547df0ebe73d5fe86222
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 11 13:21:23 2012 -0400

    improve NFC support

commit 45ef5a6226364b7b1e956d4ba763664bce661498
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 11 11:33:19 2012 -0400

    busy week

commit 0cc7bdceea0fbe4299cb8751a289121b5a5ed979
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 11 09:43:48 2012 -0400

    patch exists for the xattr bug now.

commit 8c634d61b13b84e199e6d9a29d52ec92d89272a8
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 11 00:05:04 2012 -0400

    bit being shifted needs to be long

commit 3c2f96fec21cf1cd51da3cf33557bb2cd61e1dd9
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jun 10 23:48:54 2012 -0400

    another BUG conversion

commit e8bd93f35166c513517a63168287119303634009
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jun 10 23:45:26 2012 -0400

    report the 'should never happen' cases.
    
    This should save some head-scratching if I ever miss adding a case:

commit 748bea7c3da07cca50a81316b829fa6a2294cd16
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jun 10 23:37:52 2012 -0400

    rework the interesting number function
    
    introduce possibility for single bits.

commit 176d931e55c2feb5949f39fb8c20e947332b1965
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jun 10 22:25:42 2012 -0400

    only print out sockets that we successfully opened

commit a0d96561765f55c3622a0f5ba5603163ed2c2e5a
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jun 10 22:24:12 2012 -0400

    add a get_non_null_address function
    
    this removes the need for loops like we had in the mremap sanitise

commit 6ef4f44b764f8cdbee9c9d230993d48945ca060a
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jun 10 15:37:11 2012 -0400

    consume some randomness before each fork.
    
    Because the parent wasn't consuming any randomness, it was possible
    for all the child processes to end up doing the same syscalls, with
    the same values.  Side-effect of trinity no longer have per-child reseeding.

commit 65545f7e52c87fa2115b63b42f41b81be04789b5
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 8 10:49:59 2012 -0400

    Add support for logging to syslog.
    
    Based on patch from Jan Stancek <jstancek@redhat.com>

commit 1bffd8b1eac230320df52689264f6ea845f531c9
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 8 10:17:45 2012 -0400

    identify where the seed came from in the log

commit 6d7019dd91debcb50c7e2a19ac015ee31320b85c
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 8 10:12:25 2012 -0400

    don't reseed randomness in the child.
    
    This probably isn't making things any easier for reproducability if every thread
    has a different seed.

commit c8661eef4ac6c1b2787c76034d32e77344cf9bd5
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 7 11:00:16 2012 -0400

    Avoid mprotect for now.
    
    This needs some thought. We don't want to end up marking things like
    page_rand as read-only, or we'll segfault when we pass them to something else.
    
    Not sure what the answer to this is going to be yet.

commit cd7cf2f4f0f69c9a5fa6ea366b16d2b31f04fff5
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 7 10:21:15 2012 -0400

    don't mremap zero page

commit 360132b49acb4af4fd1922a73c00ad76dbb66d73
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 7 10:06:47 2012 -0400

    dump pidslots when we can't find a space.
    
    This still occasionally fails, so add some debugging.

commit ce87c50f6fe9e5a7d0e2a66530b73b910ebdc018
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 6 18:57:26 2012 -0400

    centralise the backward compatablity ifdefs

commit 8c14e3b68b38d0d3317a8db9c235beac82f7eaa2
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 6 18:12:46 2012 -0400

    Remove the spinning thing when generating sockets.
    
    It just uglies up the display, and it's pretty fast now anyway.

commit 0ef9ba586dc5d0fa0d9a10df5a4a7f8210a48f48
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 6 18:07:51 2012 -0400

    make the default case pick from just the known socket types

commit 1efdd2a59151b7e8a1bcef702897b7649e6d8baa
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 6 18:05:10 2012 -0400

    Add NFC sanitise

commit 4ccf59c2ffafad61ce992f076e271ac758fd5cf5
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 6 17:59:32 2012 -0400

    add NFC to search list

commit ad971695bfd20fe270de246cdeb52b8f0ce30a4e
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 6 17:59:06 2012 -0400

    Limit search to PF_MAX, not PROTO_MAX

commit 1829f6fa48d666625061a2b8acb47523c0b334f8
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 5 13:48:04 2012 -0400

    blacklist lsetxattr too

commit 9028b1a26c5611366d554bfcf9a12fd03441716b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 5 09:31:13 2012 -0400

    avoid sigaction for the same reason as signal()

commit a6e93039b30c40022e1b8b40f9223264f38b7e22
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 5 09:29:58 2012 -0400

    avoid signal()
    
    This causes all kinds of crazy shit to happen when if it actually succeeds.

commit bbaea03a4b3deee125fccdfaf9d2236da49cd6e3
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 5 08:21:44 2012 -0400

    Ignore SIGXCPU again

commit f15b6560b1b6fc8568a9d3ec3ca90c82ed518dd5
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 5 01:40:29 2012 -0400

    update

commit 82e369803ba2ad2527b41fbbba0c21e829276098
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 5 01:13:03 2012 -0400

    add url

commit 22d8042ec98ca27babb1448d7092e823b67ff044
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 5 00:43:03 2012 -0400

    mask off the setxattr calls too for now

commit 5e37d10a3d3f3336a30bb175c7a995a02f44194d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 5 00:37:13 2012 -0400

    32-bit/64-bit mix-up

commit 32c2a208d45e492fd8341adc2709202eccc37c4e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 5 00:33:49 2012 -0400

    output if a syscall is marked AVOID

commit f114e94c3107bb767b1aad4273daf160de685434
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 5 00:27:32 2012 -0400

    move debug statement

commit ae2b454a6153dd72bdc32f16c31e062e45204313
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 5 00:25:06 2012 -0400

    Make sure at least one syscall is enabled before we start main loop
    
    And adjust biarch path to only call per-arch functions if we have any enabled
    on that arch.

commit 70350cdcd5cf47bd28a9174d6ad5ff186ad88219
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 5 00:24:46 2012 -0400

    ignore the xattr syscalls for now until the selinux oopses are fixed

commit 472122abbcd37b13582150d35b2b14459ad5fda9
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 5 00:23:22 2012 -0400

    Proper detection of -ENOSYS

commit 571cb0f05ce67e0be5b016160f4644b064b0f1fa
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 23:45:15 2012 -0400

    standardise on output of 32-bit/64-bit

commit 203be55062e7ccb5db074473821c8b0c7c847cc6
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 23:44:04 2012 -0400

    remove unused function

commit d5d122732fff0670712458cfcedd790c5a2bc113
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 23:42:50 2012 -0400

    factor out the disable/enable printing

commit 9efe6ec4a7ef455f77e37daf48f92723230895b0
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 23:33:52 2012 -0400

    change boolean tests to use TRUE/FALSE instead of 1/0

commit ce08ea711ba934a57f6712a4ccaaefb848062683
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 23:25:41 2012 -0400

    output number of syscalls

commit f2cd521340d3d20c4e3a9883c4fe255892bac1e6
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 23:24:39 2012 -0400

    missing braces broke non-group behaviour.

commit 04b83f88a17ec22ea72c253113559ac1816c4154
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 23:11:49 2012 -0400

    list the entry point number

commit bf7a1ea3f39a5e10278d61a329905f2dc24763bd
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 23:07:34 2012 -0400

    Fix up the syscall group code
    
    On bi-arch, use the 32bit syscalls too.

commit a45b798e803b8eb63cd023b7adf063c35973d331
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 22:33:20 2012 -0400

    add 2MB as an interesting value.

commit 6f397bb43624fcac6921fc98cb7e562a45bcea65
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 22:26:48 2012 -0400

    remove pointless test

commit f7d89cb67fc1a70c294779f7ad08fce40d58884b
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 22:26:29 2012 -0400

    split out command line parsing to a separate file

commit cd057003ec7bba633922c31dccabf083b18a72f9
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 21:57:43 2012 -0400

    move syscall table management to its own file

commit 2c08cc987b56f198eea34706ae0581418d00824b
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 21:27:55 2012 -0400

    add an \0 string as a pathname.

commit 796f4cb40444f1b52b69547db6cefe927b871fdb
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 21:24:21 2012 -0400

    be silent about file open failure
    
    this happens with .  and .. sometimes (when trinity has screwed up dir perms)
    We still have them added to the filelist, so that's ok.

commit 3b2182619c3d2a18c01e086f729b61aeed4a7b17
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 21:22:15 2012 -0400

    Remove extraneous output

commit 7e4e363a11d9528d6d2873a5cd802b9da8e01d43
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 21:13:41 2012 -0400

    move find_specific_proto to sockets.c

commit 5a870acbfdf1992f83a9cc71edc5b223ada7b1ed
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 19:51:48 2012 -0400

    Make -x work again.
    
    Broken masking.  Also add a helper function to dump the table state.

commit adeab7899a8d035cd43779f1b31effb3c3f02d96
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 18:08:20 2012 -0400

    This can fail (usually -EPERM)
    
    Just ignore failures.

commit 072ec10ef9d535860928802ba2875b710c6ca6af
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 18:01:39 2012 -0400

    Make sure the tmp dir has the write perms.
    
    Sometimes trinity will screw it up, and test-random goes nuts.

commit 7148755060426f9f041978788ec0f75820a9994b
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 17:38:04 2012 -0400

    rework how -c and -x work
    
    * Now supports multiple -c's
    
    Instead of passing an explicit syscall # around, we just mark the syscall table
    entries as enabled/disabled as necessary.
    
    TODO: Compact the syscall table, so that we can lower the rand(nr_syscalls).
    Should reduce the number of retries in child_process() when passing -c

commit 1dc984eca747012f21f1101ae7d2133bb4150d93
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 16:34:18 2012 -0400

    compile fix
    
    not sure how I missed this.

commit 93aecc63a05d1910429a6ba563c0209708594e7d
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 15:57:43 2012 -0400

    Add kcmp syscall

commit 40d77071f9fa5d863d71759385feea0f4e35a8d3
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 15:45:12 2012 -0400

    Don't bother clearing out the output buffer
    
    We're over-writing it all anyway, so just null-terminate it
    before passing it to output()

commit bbefc02c8a2457ec7260a832cd564c578a77f42c
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 15:28:27 2012 -0400

    Add '.' '..' and '/' to the file list.

commit 0f9ea6c5fc77993e3a42c706bf5518c9e4013083
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 15:20:24 2012 -0400

    Remove double test. Add comments.

commit 5027cb2dfb83dd7b6db536e5a81b582b1163b158
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 14:22:38 2012 -0400

    Set some 'count' args to ARG_LEN

commit 407f20ff636a16dec3734aa0d311ac0ab41fdf65
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 14:20:33 2012 -0400

    typo

commit bbc8aa26dadc998e7e0d8dfe6b2d6ba3ab7a6115
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 14:19:32 2012 -0400

    the pipe code seems to work now, enable it.

commit 00117c0db2fc22105c4c3e8598491c7fd8da2801
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 4 14:19:07 2012 -0400

    Pass pathnames into link instead of random addresses

commit bd333c202ca226a25ece654c7db09214c2d54aac
Author: Fengguang Wu <fengguang.wu@intel.com>
Date:   Thu May 31 14:29:51 2012 +0800

    Fix segfault on --exclude some_syscall_with_large_index
    
    $ ./trinity -x get_robust_list
    Trinity v1.0  Dave Jones <davej@redhat.com> 2012
    [15094] Marking 64-bit syscall 274 (get_robust_list) as AVOID
    [15094] Marking 32-bit syscall 312 (get_robust_list) as AVOID
    [1]    15094 segmentation fault  ./trinity -x get_robust_list
    
    Signed-off-by: Fengguang Wu <fengguang.wu@intel.com>

commit 691205e4094c2c458aaaa50509ab8033bba65643
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 30 16:33:16 2012 -0400

    reorder

commit 1bd56cf3c1d4c176fb2ed1b89eeb8d4a678a59c4
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 30 12:30:04 2012 -0400

    updates for the buglist

commit 8b5958e08fee7506414955c14e215afdfb77f2d7
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 25 13:44:20 2012 -0400

    Fix pipe buffer setup

commit c55aafd4e253f44866a0b4e7bf0ca62e78fdcb4d
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 24 18:52:43 2012 -0400

    Don't give out the watchdog pid as ARG_PID

commit 1e119a16a3c2628cf60b825603dce31d7cf1fbd8
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 24 18:51:18 2012 -0400

    Set child thread names

commit 45f0895fca8ab17e36d66e488e363acf402da0e1
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 21:33:53 2012 -0400

    Use waitpid to wait for watchdog to die, not sleep.

commit 747d1fa432eb034eb1d88e4363268924823b4a51
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 19:41:22 2012 -0400

    -L shouldn't open log files.
    
    Also cleanup the shm afterwards.

commit 388f374b1ab904e68f150760e04ed337daef5bfd
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 19:31:02 2012 -0400

    whitespace

commit 32aa11a2ddb49bd887d4d33120ad96c47f04e1db
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 19:14:54 2012 -0400

    Fix the disabling/enabling of RLIMIT_CORE
    
    - Don't change rlim_max
    - Don't exit when something goes wrong (Zombies ensue)

commit 3fd8dc45c207ad8cb9f62995fd5954b7332d24a4
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 19:13:22 2012 -0400

    Sleep in children when waiting for the watchdog to exit.
    
    Also set exit_now when we reach the max syscall count.

commit 40cb2079b45730e4d0cf3c7caef07037e35faae6
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 19:11:04 2012 -0400

    only set shm->exit_now if it wasn't set already when the watchdog has quit (unexpectedly)

commit e82f56a6528c16b88b8de817ab449fe020b8319b
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 19:10:39 2012 -0400

    Log why the watchdog sets exit_now

commit b1b0c4c3e3a03590e1e9463e9f4c4bdb82fd7df8
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 18:21:47 2012 -0400

    Clear up the watchdog exiting.
    
    Make sure the watchdog is the first process to die.
    Also wait until it's started before we start forking children.

commit d9bbd3d92b327def5815dd2c0d143a01df862c9f
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 18:10:30 2012 -0400

    Move watchdog_pid into the shm

commit 33f930cbbc02357e334bc5d7b60aeb048ca9db3f
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 18:03:17 2012 -0400

    When -l off is passed, don't need to avoid shm->logfiles.

commit eec8aaad8a7fe58d61bf4c536afca0bdfbdcd6b6
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 17:45:04 2012 -0400

    factor out child checking.

commit 5b8f823b97ed4e61f30c70e7b08c2556000f087f
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 16:43:06 2012 -0400

    Remove \n to make test script work better

commit 4e3d0f0fa1ddfd6bae9a1c002135b826df1bbde9
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 16:23:15 2012 -0400

    distinguish between the watchdog exiting and children

commit b83af6fa8aeaede25894dc3343da7d53247a0629
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 16:22:43 2012 -0400

    Move exit_now into the shm

commit 5e9f7d95f526b8954c61237d848a568db29bdea3
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 16:06:02 2012 -0400

    Child needs to wait for parent to set the pidslot entry.

commit 351d31df3f7aa383e1a27919a7247f3d27f0dbd9
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 16:05:22 2012 -0400

    Want the watchdog to dump core if we segv.

commit d313d84b3d1ce01b45d45a40bd4836e9e623ade4
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 16:05:03 2012 -0400

    More debugging.

commit e324b48be6db4f2277e792dba1ba9aca6005b32c
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 15:39:51 2012 -0400

    Make some debug messages easier to grep for.

commit 3be1f4bb5c447521c1247a008c04a58c4c09b338
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 15:36:52 2012 -0400

    comment out some excessive watchdog debugging

commit a3c19deaed1d9d4a7236ea8fa40c62cdbf1763c8
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 15:35:46 2012 -0400

    output() calls from watchdog thread got to the parent log file.
    
    Also some extra debugging for when we can't find the pidslot

commit e3361c94879ccf61e12987e86d4aa2fb789cd073
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 14:55:36 2012 -0400

    Move the watchdog code to its own file

commit f91f784f922346caafb6c653eb489f647680615c
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 14:52:21 2012 -0400

    Make the watchdog be a separate process.
    
    Move the progress output there.
    This gives more accuracy, rather than each child potentially racing.

commit 30b7203f00754295bb85a3fe627a9e6a615c3711
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 13:41:26 2012 -0400

    Missing logfile debugging.

commit f315e698529992584e9a1f05268f802a5f3381a8
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 13:33:37 2012 -0400

    This comment isn't necessarily true if passed -N

commit afdd53b7e0a4e5de2cf68707931b8876d5bc08be
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 13:33:20 2012 -0400

    Last commit broke -N, fix it.

commit bd80474591a3035d01bdbe82172c348e492d8189
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 13:11:09 2012 -0400

    Move the regenerate decrement
    
    We were only regenerating every 25000 forks, rather than every 25k syscalls.

commit 8ef3f1b0654c6bd74e6580adcc249bfc771b2c21
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 13:08:37 2012 -0400

    Don't exit after 1000 syscalls.
    
    Just keep going. We'll eventually find a reason to exit some other way (ie, segfault)

commit d0b3c1525e97c3dd893209511bd212686eaf24df
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 12:55:36 2012 -0400

    Make the 64-bit only message the same as 32-bit only one.

commit 9f937b1a1a61f547e01dd93cc6da997e951e90ed
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 12:53:30 2012 -0400

    clean up find_specific_syscall some more.
    
    Always search both tables. Handle the case of 32-bit only syscalls better.

commit e62ad0509a0fb4ce46ae854a61f22f53942e3e9e
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 12:53:16 2012 -0400

    If we passed a 32bit only syscall as -c, we should only do 32bit calls.

commit 180b518f25f2e5d9228567c264358971b7d1bcc9
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 12:32:16 2012 -0400

    More specific cleanups.
    
    In the random case, instead of skipping, retry, rather than potentially
    incur the overhead of more forking.

commit 2f972f798086b2a52e51d5d375b275b8af7d885a
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 12:31:45 2012 -0400

    Second half of the prior accidental commit to clean up the 'specific' case.

commit e06ea6b2db46f3ce42889deb54cb158d20a2ee28
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 12:29:48 2012 -0400

    destroy the shm at the end of the cleanup.
    
    We need it around to close the logfiles etc.

commit ecf2deed14b1b805fa1dadc861ee5ad80efc3445
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 12:22:05 2012 -0400

    Check logfile was opened before trying to close it

commit f90a3a5eb0ca16aebc47a52239c1132eebfbbda5
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 11:34:16 2012 -0400

    Rename

commit bf5787eded8816853cc974125676579f660cb4fd
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 11:33:54 2012 -0400

    sequential test harness.

commit 5f5b85001644188b33cd53e07ffd7395c533cf05
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 11:33:12 2012 -0400

    Add a watchdog routine to ensure children are making progress.
    
    Only enabled in debug mode.

commit 337577460f28505664fd4b46f810ba2989040804
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 11:32:46 2012 -0400

    fix up the parallel syscall forkbomb test

commit 733dff53b1e4ffa771070a3a918b0deaca9f60cc
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 11:30:58 2012 -0400

    more bugs found
    
    (lkml.org seems to have issues tracking down recent threads,
     I'll add them later)

commit 248eea51649b52da4c7518161d5872e2a5c6630d
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 11:30:41 2012 -0400

    Don't print the headers out when we run -L

commit c3a60fd958731f335ec1afc5df620b9d751714fa
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 11:30:07 2012 -0400

    Print the syscall we had problems parsing

commit fdc1dc408ef3eec6cea6a54b80b51c13b6f34e5f
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 11:29:37 2012 -0400

    Make the parent segfault in debug mode too

commit f9a032d5a7c33e177b84564ccbd46aa3e5af155c
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 11:27:49 2012 -0400

    don't munmap 0

commit 9828a66e3caecc4daacfd8039199f8903af3dd81
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 11:24:57 2012 -0400

    Move log buffers onto stack
    
    multiple threads could write to the buffers when they were global.

commit 729970e5d4b8f56a9b40612d8c3141b0d651e8d6
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 11:24:24 2012 -0400

    Some error checking for log.c

commit a24f528cbbdc2a9cbe7a4aff9f710366d83d9dca
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 11:22:33 2012 -0400

    Enable segfaults when we are in debug mode.

commit 50c5072bc32813b5087fb666f2511527efdfefc0
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 23 11:07:57 2012 -0400

    Avoid brk
    
    just segfaults

commit d6b6197d4dafb805e2da5e85c8e9cc20398d6256
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 22 23:27:19 2012 -0400

    check for failed mallocs

commit 1c7f2e654c620392851a429f6558c3592a2b09e0
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 22 23:01:19 2012 -0400

    Add a buffer to the other side of the randomly alloc'd page.
    
    When we return an address near the end of the page, we might end up
    writing into unmapped memory, causing a segfault.

commit 8e57caef33f9886b6a4e2f99bb498916f20a12ed
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 22 23:00:26 2012 -0400

    bullet-proof the harness against glibc checking
    
    also for now, ignore mbind, until the kernel gets fixed.

commit b095954b79370aa1de7b7257906cae096755373d
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 22 22:59:46 2012 -0400

    when regenerating, close the fd before zeroing it.
    
    Should prevent racing with children

commit 13e36a941831eb07a6c7177494f0183b7499418f
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 22 22:59:14 2012 -0400

    Check for failure of fileno
    
    This would have saved us before if we were checking it.

commit b18f393131d73438e8138dbdee783e659a8c35e2
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 22 22:54:29 2012 -0400

    Fix off by one that caused a segv

commit 10a6972989d090cad8c519b5793d5d29fcc8ec2b
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 22 14:02:32 2012 -0400

    small optimisation for logging
    
    If we passed -m, we don't need to do the ANSI code stripping.

commit 1c51f58aaf029c5e25d903238fde3a34a2473d65
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 22 13:37:11 2012 -0400

    Fix the monochrome option.

commit 95ae151a091180afd39d30b60593e638e3580c20
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 23:26:59 2012 -0400

    ignore /dev/log

commit 991ce492ca7849f917d9e62335530c7908daf127
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 21:00:59 2012 -0400

    TODO updates

commit 4dcfe65efea7a8895ecab3289e73815bfa396198
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 20:53:31 2012 -0400

    clarify -N comment

commit 0ef855a75604d06a6307f56d17361a6834688147
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 20:50:53 2012 -0400

    remove --childcalls
    
    This was only really useful for debugging when rewriting the parent/child code.

commit e06c68ceae7bdd459a55b316ccad2c63a471040e
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 20:43:28 2012 -0400

    Document --victims in helptext
    
    Remove long-gone options.

commit 3ce973eada9faad5e17ee0336a7785ed0a2eb31a
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 20:41:00 2012 -0400

    rename nocolors to monochrome

commit 7d605c2ce6ae6c4011aa9f2457102cd763a6c2ac
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 20:38:14 2012 -0400

    Move usage() next to options
    
    Should be easier to keep in sync.
    
    Update, and sort.

commit 5ce191f1b0fa117ba718c79be4e376777e12c4c0
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 20:36:02 2012 -0400

    document --nocolor

commit 102e4929c49ff0846733a1d4c28b47e1c5151cce
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 20:35:08 2012 -0400

    Add --exclude option to skip syscalls.

commit 90457631295705e11abee49f047afcf91966e6b6
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 20:30:22 2012 -0400

    Make the 'specific syscall' code handle 32/64 bit tables.
    
    Before it assumed 64bit.

commit aec24d819590b8438d40441b04679e1aa8956411
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 19:50:06 2012 -0400

    sort

commit e8cd6ca886cb9aa9ae7608512afdaa5b821bd9af
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 18:55:12 2012 -0400

    fix two off-by-one's in switch statements

commit b208692f1d54d0e47a9f28478b6271f42e5b0bdb
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 18:54:01 2012 -0400

    More socket sanitisation. (AF_INET,INET6,UNIX,APPLETALK)
    
    Should speed up creation some more.

commit e8c27bf86f5537f8c4796c1aba7e5d69c3e97e7e
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 17:57:36 2012 -0400

    output the log files in monochrome.

commit b1c37d1bca49422a568b710f9c910bbd0c50737c
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 17:20:14 2012 -0400

    depend on the headers in subdirs too

commit f82c62a5c7a391cb6476ab498d7d240814544f72
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 17:18:44 2012 -0400

    flag the headers as dependancies

commit 326640577ade5c283d0eda99744bc597b0f0b74d
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 17:08:24 2012 -0400

    buildfix for F14
    
    (Because Linus hates gnome3)

commit 82013803ee41b7525a470307345a42d09ecc6efe
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 17:04:37 2012 -0400

    Add --nocolors option

commit 00c4a130c374f8059af7066466f001c105363a15
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 14:23:48 2012 -0400

    improve coredump limit restoration.

commit c2ceff1d8c15b226591e0dde3fbaa26800bc9d34
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 12:15:58 2012 -0400

    go back to the old way of masking signals.

commit f9ff55ed0094fe040c76e57aea6d0e8148f93eba
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 21 12:15:47 2012 -0400

    we only want coredumping disabled in the children.

commit 44b8609eec27fd4fb55cc431ae1952fba68a4bc7
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 18 15:25:28 2012 -0400

    latest bugs found.

commit c5e2eb69e5b74ff89959fb5b5b52a7ac411d39dd
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 17 15:19:54 2012 -0400

    remove some unnecessary code

commit 773fa99be8faa495150b5c2f3ceab0f82e0d1570
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 17 14:32:08 2012 -0400

    might as well sync when we sleep for a second

commit 99e0b4addd666a590bd25e371e507fe37889e3cf
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 17 14:29:25 2012 -0400

    sanity tests

commit 529759fd8adc9c2499da3d73581d95badb1e746c
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 17 14:25:23 2012 -0400

    avoid arch_prctl
    
    messing with segments is a quick way to segfault.

commit a98cbf29d90e43c6f6d04a3782c4342cafee55c0
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 17 14:01:12 2012 -0400

    ARG_RANGE without specifying the range causes segfaults.

commit 8905e1c07ecf6e3b2a3574b9fcbc2395c17e90bc
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 16 23:02:16 2012 -0400

    rewrite the dirwalker
    
    - now adds directories to the fd table
    - handle permissions on directories as well as files.

commit a7e39b7fb27cb0b1d51a5424588878883d72510b
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 16 22:14:51 2012 -0400

    improve file mode handling in dirwalker

commit 3d9580a64f8eb563acc4244f2b68ef60c35aab5c
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 16 21:47:20 2012 -0400

    add explicit continue in dirwalk.
    
    just for clarity, shouldn't change anything, as the rest of the loop is empty anyway.

commit b7f0914f023dcc132354f8a6e7c5cf0f215d47cb
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 16 18:53:18 2012 -0400

    Add an ability to pass our own 'victim' files.

commit f571d2b330e4ca377e90591c9319050fbb320a7c
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 15 16:46:02 2012 -0400

    maintain some statistics on how many syscalls a child did, and print them on exit

commit 2c1e1e6d4961fb0d5df47f1e4591aad810834a82
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 15 16:39:02 2012 -0400

    introduce find_pid_slot helper.

commit 36b19d4dd0b6182a07bba444148bd78c231bc4e7
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 15 16:25:10 2012 -0400

    things seem to work better right now if we just exit when we get SIGALRM

commit a647844ecf930aa094336425b0929b22ef4b5fbe
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 15 15:11:20 2012 -0400

    Duh, the alarm handler wasn't being set.

commit 6542596aeedb9b9ad0c9d907cf5f000c6017be54
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 9 13:19:57 2012 -0400

    Sasha tripped up something with execve

commit 7aa148770dbc9b4014491bc8d8c9ba8c81c85b84
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 9 00:22:06 2012 -0400

    constify the parts of the syscall struct that don't change

commit 4c9159ab3d09043b71fc5c000f9c664db540b570
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 9 00:18:05 2012 -0400

    build fix -- dupe variable

commit d890116a8ce9a81ce0ac4af3f810030d52c5aa44
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 22:10:11 2012 -0400

    avoid giving out pids for the trinity processes.
    
    This should stop children sending weird signals.

commit bf1949f3cb412fa92d5f2ac43b2f11c2d19cb4b2
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 22:09:42 2012 -0400

    print out pids in decimal.
    
    Also change the hex/dec cutoff to 8k

commit eeca1f4a41e19f6ce02ce0795226ec40bc0cc977
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 20:53:08 2012 -0400

    rewrite to not use goto

commit bc5d28f68b3590671fc78ee441257a00239a481a
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 20:35:53 2012 -0400

    ignore SIGXCPU

commit 8b26166b420206355f477f5245da0fe368cdb33f
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 20:35:35 2012 -0400

    remove debug printf

commit 6ecdbb567fcf71524a6a304fd5886f7d82d69d70
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 19:59:29 2012 -0400

    umount can take a pathname too

commit c593f382ef862a252019ad7191f6e64bd0f475c6
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 19:56:14 2012 -0400

    Add more ARG_PATHNAME annotations

commit 44acb04053e2091c3f9a3c66f3f5f72a0a312ffb
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 19:28:57 2012 -0400

    Don't ever return 0 as an fd.
    
    This happens sometimes, probably because of the currently disabled pipe fd code.

commit c6bc32b3623b06ac3da0e791102c0369f1105bde
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 19:21:32 2012 -0400

    make fallocate's offset be ARG_LEN

commit 2ba7591dd5176a6b1e02842c1aee909363c0e626
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 19:19:42 2012 -0400

    assign values to enum

commit e17ba3e6e6ffc78ca9264405ee341e7f42541d70
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 19:06:04 2012 -0400

    avoid handing out the logfile fd's

commit 0e29fa6a09f6e2b0d96af9fc5c2ccb147b381cb3
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 19:05:49 2012 -0400

    fds are signed

commit 3e12e6349d47e9c43bb03ee9d6fc80c9f05fc2d6
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 17:05:34 2012 -0400

    regenerate the filename list too

commit 348c463407bff4cb31a1f1d228f6c3035b50abcf
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 16:53:38 2012 -0400

    remove avoid from execve, annotate filename

commit f0839d5f436c1a63339bf99861c23a0c74a67403
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 16:43:35 2012 -0400

    Flesh out ARG_PATHNAME
    
    Gather some random paths during the fd scan, and hand them out when asked.

commit 778508f5e24107d0a39cc11227ad2b3591e16ea1
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 16:41:18 2012 -0400

    silence make-it-fail failures.
    
    Also, after the first EPERM failure, don't bother retrying.

commit 278587f44d706301cece7b22fdafc3e875cc130b
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 16:37:46 2012 -0400

    Fix inverted logic

commit 34e03c69b0ac142c1225567199ce3d4d124915b9
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 16:26:16 2012 -0400

    -c and biarch are a problem right now.

commit 087a465da3beaa69c26bd1b79dbe46a888bf8337
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 16:22:36 2012 -0400

    remove debug code.

commit bf471b07eca07c265d9e823ecd6e24720aec34dc
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 16:14:40 2012 -0400

    Fix -c

commit 8410c45bd7a6534b071598dfd881a3395e4ee425
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 15:55:45 2012 -0400

    Fix -L

commit 82dc142ca86350bd2023aa49704811f7357f18d5
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 15:29:01 2012 -0400

    annotate pathname arguments

commit 6874f635238775d1db4c1985c9d56257e5f29e9c
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 14:54:18 2012 -0400

    clean out the per-arch syscall objects too

commit f16444dcfb7604b0f48d63adf82d078d1a2c0122
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 13:52:19 2012 -0400

    return < page size as a length sometimes.

commit 040a7afacb7229d205d72ba474112adfa6173c20
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 13:15:03 2012 -0400

    introduce function to perform per-child initialisation
    
    should reduce some duplication in the 'extra fork' case.

commit 12b04e126284b3302fb7c344d0cb440160786e3b
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 13:11:53 2012 -0400

    simplify regenerate function

commit 0e3183439cfd951a30f8b2e033bb48ec71fe6f82
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 13:08:47 2012 -0400

    remove 'return 32bit' special case
    
    This is handled in the switch statement below.

commit 399c6defc32a0b6f45b992cf22246964f30d1eac
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 13:07:53 2012 -0400

    Improve the rand64 function
    
    Steal an idea from iknowthis to distribute the randomness a little.

commit f3c57813a41a49a0dc01f38cba660aaa3a4ec0d5
Author: Dave Jones <davej@redhat.com>
Date:   Tue May 8 11:24:54 2012 -0400

    make clean should just clean files we know about
    
    I got tired of it recursing into tmp/

commit 70538230d544ac2d56a02bd7d22665cc38fa4fd2
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 7 17:31:19 2012 -0400

    fix inverted logic

commit e8d2f0244fd515a574081738291a7bde5d375643
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 7 12:13:07 2012 -0400

    This finally got fixed

commit 1e12ebe94946f54772479415af019b25de1e8d62
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 7 12:02:12 2012 -0400

    Do some FP math sometimes in the children

commit 0df028f80cea1bb31360516b47c9c7f67ddae171
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 7 11:58:22 2012 -0400

    Move the child process code out to a separate file

commit c7d03ac50a668ec5c36c19c26223df3d10eb846d
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 7 10:17:55 2012 -0400

    Set /proc/self/make-it-fail in the child process
    
    Idea from Sasha to invoke the kernels fault-injection features when making syscalls.

commit 5e99f2fc70518f2417aaa05436c104a97d06638d
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 7 10:13:13 2012 -0400

    rename variable
    
    'b' is an awful name.

commit aa7da0ba29cf8454fc39da1f1c15f909c1fec8ce
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 7 10:01:39 2012 -0400

    split out the syscall tables to its own header

commit 3fc64c933a57b81162a4df4bc256ca0ce5218f2c
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 7 09:54:14 2012 -0400

    remove unused defines

commit 5ea9f5dcc881c280f8a114b5c38cd0ef52512945
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 7 09:52:49 2012 -0400

    split up trinity.h a little

commit 57ca25e9fcaa0f301b39e37a11b7a1c8ad706858
Author: Dave Jones <davej@redhat.com>
Date:   Sun May 6 23:07:44 2012 -0400

    disable the 32bit code by default for now
    
    It still has some issues, which seem to cause all processes to just lock up.
    They seem to wake up immediately when straced. Needs further debugging.

commit e22ba1a8780c8931e32f27e7a409982c5ef8ab84
Author: Dave Jones <davej@redhat.com>
Date:   Sun May 6 21:56:42 2012 -0400

    Bring back, and improve the 32bit emulation syscall code.
    
    Doing this randomly might cause some problems.
    
    Note, I recommend setting /proc/sys/debug/exception-trace to 0 while doing this,
    as trinity creates a ton of bad rt sigreturn frames right now, which I may or may not
    do something about later.

commit 894591e54b33882bbba6744134e2f69f7f74a2f4
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 3 23:13:31 2012 -0400

    AF_NETLINK sanitiser

commit 7771925ede73d2f44d2433028c51643b728fe6e2
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 3 12:25:25 2012 -0400

    add url (lkml.org seems to have broken this thread and lost the original post)

commit 7822c3b9cb21befba684bd0ad45e062746b9dab9
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 3 10:44:10 2012 -0400

    sasha found a bunch more bugs

commit dba5d381cd718ba031bfec140a9d57dbd9526650
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 30 10:10:18 2012 -0400

    remove dupe

commit 4b629e0ecffc523fc90807cada19a431caf73163
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 29 19:31:12 2012 -0400

    Add support for the new packetized pipes.

commit b3e44cdcb2b02cd6e52f29cc92783cb54bc9b9c1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 25 10:57:02 2012 -0400

    introduce some randomness into the fd lifetime rules

commit d226a5f319f9cc818d836f5569e76fa331b8140d
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 20 13:00:56 2012 -0400

    mm: divide by zero in percpu_pagelist_fraction_sysctl_handler()

commit 52015bdf46e486f95cfaa404f01d2646eb110dc3
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 19 14:33:28 2012 -0400

    Introduce fd lifetimes.
    
    Instead of getting a different fd every time, return the same one a few times.
    This might increase the chances of races appearing if two threads operate on
    the same fd.

commit fef1e1774f2b9d1dbe4b9fea3d0c02ddffbd2ea3
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 19 13:47:56 2012 -0400

    3.4-rc3: kernel BUG at mm/memory.c:1228!

commit c461952d41ebf04c5283022a86ac97da1d83a5b7
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 18 20:19:53 2012 -0400

    out of scope.

commit fbffc374869e41c332ef5e8bab186a3c3c69e74a
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 18 12:16:12 2012 -0400

    use defines instead of numbers

commit e07da1a111b72476ac1a6745d43afe41d35f303a
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 18 12:08:45 2012 -0400

    Disable dumping core

commit 0217276f86c2c5958dfdc5ef98e4691e33eed709
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 18 11:37:24 2012 -0400

    jbd: NULL dereference on chown()

commit 885911911814b815047b714b8590a0dc512c7406
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 18 11:26:54 2012 -0400

    move socket_fds into the shm

commit dc891500983c1abdfdd512a9c4e39aabd6f03c02
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 18 11:20:48 2012 -0400

    Move fds into the shm

commit db210e9322d3fa5d59e7461fa671065b422c965b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 18 11:20:01 2012 -0400

    disable logging in the test script to minimise disk io & maximise throughput

commit 7ee0f08dfd9e6f89c4a729ad8dc9fc47d8e5ed0d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 18 09:57:15 2012 -0400

    bdi_debug_stats_show oops.

commit a7e7844436f480a9c6c21e62f1501afa12f66271
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 17 13:38:01 2012 -0400

    [3.4-rc3] Thread overran stack, or stack corrupted

commit 34d18960653556c097e5620e7f15586b9d2e0994
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 17 09:52:39 2012 -0400

    Sasha broke RCU

commit 046bd9549ddc025b5d7912e17cf7874001084c68
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 16 08:28:18 2012 -0400

    Sasha broke 9p again.

commit 7661cd4b6673f8bbd3dc711baf389412120545f6
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 13 15:09:58 2012 -0400

    9p: disconnect channel when PCI device is removed

commit 7164790e6a4720367d9d1c27f9a93a53eb34f634
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 13 09:39:36 2012 -0400

    migrate_pages broke

commit fc758b171a493840a1f785cecafdfc3b0c362ad9
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 12 14:09:07 2012 -0400

    Disable the maps reading code for now.
    
    I'm suspicious that some of the self-corruption we're seeing is because of it.
    I think I'm going to flip this around to make it be a list of addresses to avoid
    rather than a list of addresses to use. But for now, just do nothing.

commit 88d5312dd431c5529f3f4e4db39f93dbadb88d56
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 12 13:39:22 2012 -0400

    Add a buffer page after our special pages, in case something scribbles past the end of the page.

commit 92dc9c38ec30ad19cd527d73689e99e5d63d7d56
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 12 13:36:25 2012 -0400

    Ftrace warn-on's

commit c8130fd2947649b0f05e5829964cbd760936ac4f
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 12 00:37:16 2012 -0400

    page align the pages we pass.

commit deb3abcea3811fbefb7cea7dd13a66e2e959095d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 12 00:03:26 2012 -0400

    Mask off the possible offsets to print the correct name.

commit 111267ea98299c386fc22a2790050ea1365a8777
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 12 00:02:53 2012 -0400

    Log the addresses of the pages we setup

commit d89074fb3ae91abb4f035651ada53386b7cb9eb7
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 11 20:57:33 2012 -0400

    Sometimes return the end of a page, not the beginning.

commit 7446728af58febc79f4c0c64c9ab562fa1cdc62b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 11 13:49:51 2012 -0400

    remove some dead code

commit 61f80ab4ddb064163419a975a1506bddf1dc39b5
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 11 13:46:52 2012 -0400

    enable a bunch more warnings (and fix them up)

commit 0d3423439c8a2ce4c578300e85e81d3e3d36f29b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 5 17:09:45 2012 -0400

    [PATCH] netfilter: ipvs: Verify that IP_VS protocol has been registered

commit d1377bcab043eaca1b12ec615c5f7ecfc8f0aba1
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 5 15:32:12 2012 -0400

    updates

commit c637878f0100f16e519165a68b88b06553cf4ebb
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 4 16:33:31 2012 -0400

    CPU softlockup due to smp_call_function()

commit a0bb9a09e0d0c01656609812f3fbdde45da909f1
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 2 22:02:09 2012 -0400

    kvm: RCU warning in async pf

commit 5881557f9b5a211b2cde65ee822b5ebe6e92fdb1
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 2 15:22:40 2012 -0400

    Sasha's latest findings

commit c15e7dd7cd9f512f7aae2c713885efb5add6c58f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 30 10:29:49 2012 -0400

    more Sasha discoveries

commit ee997c5215886c83255044acd9f952003ef9b63b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 28 20:56:26 2012 -0400

    this is believed to be the same problem

commit c7fff7ed995127816ed09bdba008605e7f1ad7db
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 28 18:26:17 2012 -0400

    annotate mode argument of fallocate

commit 9f562e3ad9749a75f0435e3a5837fa4e74d64597
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 28 14:38:50 2012 -0400

    Add https://lkml.org/lkml/2012/3/28/379

commit 497a0442f6b70e3bbcaf9af519236a232c2cab42
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 20 20:57:37 2012 -0400

    don't ptrace non-trinity processes.
    
    bad things happen when you ptrace tmux with trinity inside it.

commit dc7f2fd25ba35338bf6249450b687e215a0280b3
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 19 18:11:04 2012 -0400

    latest bugs found

commit e3bcce120d39517c96a8f4b42ceda34088ea7212
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 19 18:10:37 2012 -0400

    Revert "don't do anything with pts files."
    
    This reverts commit 83215d444ec76b2829abbb08330bfba6a739ad67.
    
    The value of finding tty bugs is worth putting up with the hassle of
    garbage on the screen.

commit 58c933f14635e8829cd2c6a438d0fd9180495bb3
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 15 19:50:55 2012 -0400

    resurrect an old idea of doing the actual syscall in yet another child.
    
    Another debugging idea. Some syscalls might be corrupting our
    internal state enough that we're segfaulting. If we segv in
    these 'throwaway' children, things aren't as severe.

commit 481de462cef9410f03825f0a0dbc85b407b9093a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 15 19:38:20 2012 -0400

    do as little work as possible in the signal handler.

commit 32bb57f617dd8f1f86907ad39403b121f657a1fd
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 15 19:21:44 2012 -0400

    avoid using libraries as mappings.
    
    This might be causing some of the stranger child segfaults if these
    get corrupted somehow

commit dc726e849da7466bb9de2da2e1463a1c3df321cc
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 15 18:51:27 2012 -0400

    don't flush stdout from sighandler.
    
    it segfaults a lot

commit 05d71eba522c3d0d0c6409a37fba9206601246be
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 15 18:18:37 2012 -0400

    don't use strsignal in the sighandler.
    
    This occasionally segfaults in some locale code in glibc.
    My guess is the environment has been corrupted, and glibc has no
    idea what language we're wanting translations in.
    
    Just use the array directly, and use english at all times.

commit 0e6a2e66abdabe476ef18f468ed04ca46ccbb581
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 15 17:48:15 2012 -0400

    don't do anything with pts files.
    
    dilemma. This is annoying because it causes scribbles over ssh sessions,
    but at the same time, we're now going to miss out on some potential tty bugs.

commit cb954de77d018de543682d3a84fae06a73aee0a1
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 15 17:31:57 2012 -0400

    Don't give out stdin/stdout/stderr to syscalls.
    
    things get messy when we randomly do a huge write to stdout,
    or read from stdin etc..

commit 5a2efad231de8d29f17dcb6e74d485f0d0c1504a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 15 15:54:27 2012 -0400

    on SIGALRM jump back to child

commit 9b107fa7bbd8a3c5492575196b977954ef87a04d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 15 15:52:57 2012 -0400

    clear & reset buffer after syscall.
    
    (mostly for debugging purposes)

commit f594f11afd521102500865f5e31b9cc48e17ca14
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 15 15:44:07 2012 -0400

    Mark prctl avoid for now.
    
    It might make sense to do something special for this later, rather
    than just calling it randomly.

commit 781874c61d58831799c8be9581f61d9d42f55845
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 15 14:29:03 2012 -0400

    log pids in the parent

commit bfcfe0d3ddfd587cb7eb7572ce8a1b2763847dd8
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 15 14:28:47 2012 -0400

    disable pipe generation for now, while debugging

commit 6cf06f802d80dbf70b01473e638cea394753895c
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 15 12:33:52 2012 -0400

    sound: BUG: unable to handle kernel irq event stamp: 2013419

commit 833215936b54087713988dc1089a03f34fc6ab4b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 15 12:12:01 2012 -0400

    use a null instead of a nasty cast to -1

commit fd9205055df0126613ac7324f0193ca06cacad96
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 15 12:09:32 2012 -0400

    The SHM must be private.
    
    We used to be able to share it between unrelated processes when
    we weren't using it for things like the pidmap.
    
    Now this causes state corruption, because there's no locking.

commit 5e7e8db2b97ddd9f2483138c676d7dbcf5360ac0
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 15 11:01:26 2012 -0400

    ntp: BUG: spinlock lockup on CPU#1

commit 483ee5d78bcee782b5429e7f672db8ba71f16a47
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 20:44:30 2012 -0400

    silence

commit e48c460518c4119b0b9d0a05fe4031465b69d074
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 20:43:14 2012 -0400

    fix up some of the scripts.
    
    remove dead args.

commit 3dbab2b187f49340a6f20c214cce6436b33e1b54
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 20:42:31 2012 -0400

    put the parent pid in the logfile names.

commit d46b4ff5ab749dc589f6d27090a31256a3f612b7
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 20:05:31 2012 -0400

    personality bug.

commit 4c5fc2b740d1afba736213740b4102cc398746fb
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 20:04:57 2012 -0400

    relax syncing even more

commit d10507062cc57afb5dc6970df25ce2ffe257ee4d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 19:31:13 2012 -0400

    bind child processes to a cpu

commit e905110467bdb3ce42a4b9f7a5b7343236e91e96
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 19:30:21 2012 -0400

    relax the fsyncing even more.
    
    Only sync the logs every hundred iterations

commit ceabb245e7958dbebf11b6941b39d25afe563820
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 19:21:56 2012 -0400

    actually, init this to 1, so we don't print out iterations: 0
    
    but move it to be with the rest of the shm init.

commit 3aee631bdc52458a5da0ee94c21e721163c6f1f1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 19:02:33 2012 -0400

    remove unnecessary initialisation

commit 31840be075c8824eec5ee44c9397850c2c3831c6
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 18:54:15 2012 -0400

    extra thread

commit 8242e3ee1dd805ed7e853ebef6957fd3f71f9b9e
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 18:50:51 2012 -0400

    print an extra \n to demark the new child in the log

commit f94da02cc13ba3dfda8177a1d6a5b88bd1ef1317
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 18:46:51 2012 -0400

    verbosify the count

commit 9231841a03880b27298d1f55404ca564a8598b8c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 18:42:14 2012 -0400

    relax the fsync'ing on the logs.
    
    in some rare cases, we might get incomplete logs, but
    this makes things so much faster.

commit b5f99ef124f4b5c74600b7690ba630d926692922
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 17:56:14 2012 -0400

    don't log the overall execcount. it's meaningless in the child.

commit 97ed9689b17ba44685ff250b25fb04e995154207
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 17:55:48 2012 -0400

    remove logfile locking now that there's no contention.

commit dac0a3827f4cfc981058cd35ae89aa686c13dfad
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 17:41:46 2012 -0400

    ntp overflow

commit a99385671d4daff45ee0409ed84b08760bd6fb17
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 17:33:14 2012 -0400

    Clear the shm on initialisation
    
    This fixes a stupid bug introduced in 63ce0518f0c1a611eac91b8254a0fc3b85a0b4b5
    that afaict was always there, but we got lucky before.

commit e75a23607c857f9457ed134f5fbd8addb2c066d3
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 17:13:15 2012 -0400

    open_logfiles doesn't take a parameter

commit 8a4bb1cac785925d88effd732c1df8d6de909020
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 17:13:00 2012 -0400

    move all shm init into same function

commit 6008a4389e894a1cd3fa6d5f13f28f1cf94b1cc8
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 16:34:34 2012 -0400

    find_logfile_handle can be static

commit a09fd7169cb380423dd3934348a11cd84b8f7385
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 16:26:35 2012 -0400

    Per-child logging.
    
    Due to contention on the logfile lock, we were effectively making trinity
    single threaded in many cases. By creating a per-child logfile,
    this shouldn't happen. (TODO: We might even be able to now not bother
    with the locking at all)

commit 4e37997fcb7b70b4170a7a2381ee490c440df484
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 15:48:42 2012 -0400

    break the main loop into two halves.

commit 5d891df376add1b15aecd35510ae7e2e001ddedc
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 15:40:16 2012 -0400

    compile fix
    
    check_tainted is also used on startup

commit 2a5a930191744009b627405ae87b84284dbea770
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 15:39:03 2012 -0400

    now that main_loop does everything, we want to stay in it, not return

commit 8108aa8170224aa06ec7079b8e43002a6eea685b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 15:37:18 2012 -0400

    move the fork/check loop into main_loop() where it belongs
    
    This has the side-effect that the taint checking should start working again.

commit cd699c3f1340650c9ead376aaa5be8699a45c7f4
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 15:19:43 2012 -0400

    another

commit 056087b619e6a26193cca2ff51f4ca977aed114e
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 15:08:54 2012 -0400

    start annotating more return types

commit a53457ac12ec117dc8ef566244e600c4df9b1d24
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 14 15:06:28 2012 -0400

    Sasha's latest roadkill

commit 3106b2905df612690ec8a7cd5f14df0f970e91a4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 13 23:56:56 2012 -0400

    annotate getpriority

commit 4f63470fead9d0f0c98d1ff8617d424fdb79a04c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 13 20:19:10 2012 -0400

    just set the nr childs to the nr CPUs

commit 1c96ad205c00c71ed8c5ff9cf70296953893e6ac
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 13 20:02:30 2012 -0400

    revert some accidental debugging

commit ecd7026a6f595aa0c5d8d152ad868786365d90a3
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 13 19:59:21 2012 -0400

    don't call alarm either

commit 2b83ab3015ffd046dbd34a79b7977eeaf13f2268
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 13 19:56:23 2012 -0400

    setitimer interferes with alarm(), so don't call it

commit bc3de078c036b443577aad558a77c72943bfccfc
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 13 17:12:49 2012 -0400

    TODO: fd generator rewrite

commit 28d8c3f35ade00461c038fd933d0d6051de5d2c9
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 13 16:19:51 2012 -0400

    the kernel seems to know more than the man pages hints at

commit ac491b8fd1426010a58cf62cb010b54b7b9afbbe
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 13 16:16:28 2012 -0400

    annotate unshare's flags

commit e27af51b1cd848c7cf9f4ba5ca90c02f529e711a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 13 12:19:08 2012 -0400

    ignore SIGILL

commit 998cd62221fefe3bace5f49e7a226ac717b2ebb8
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 13 12:10:26 2012 -0400

    Introduce a ->post method for syscalls
    
    Used for tearing down any allocations that may have occurred.
    First user: shm segments.
    
    We were hitting the shm limit pretty quickly when there were quite
    a few children running simultaneously.

commit 3349f45a6c382c20ab26995971ee1c399554d672
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 13 11:24:42 2012 -0400

    Remove the shm when the last attach point is detached

commit 0e0cbdceb685df572ffb8498a53dd0e0677eef83
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 12 23:46:59 2012 -0400

    tun/nsproxy oops

commit 6ece60275f0aef77877db6dbe65eaa95b4f31b15
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 12 21:09:51 2012 -0400

    sys_poll use after free

commit 26fa47e1de685f23c81c9b233d436c9683559a50
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 12 17:28:54 2012 -0400

    ignore SIGABRT.
    
    glibc can send us this if we corrupt the malloc pool with an mmap.

commit 6db86e57918466751ae5d0e72e0b4c783143bb1b
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 12 17:22:21 2012 -0400

    Release all child pids if all children have already exited.

commit 48b80d747a2fbd6baa8a21db86ee91854e350d33
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 12 17:17:34 2012 -0400

    ignore SIGBUS
    
    Otherwise we see stuff like this in the logs..
    
    trinity[29203] trap stack segment ip:36da87fc49 sp:7fffc355ac10 error:0
    
    which isn't too interesting

commit 86f393b10e92899d16929e03b5079c2806b23993
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 12 15:14:30 2012 -0400

    add a --debug flag to enable debugging info, and suppress by default

commit a62169c817257e8b167b390940fdfef2167b14e7
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 12 14:02:37 2012 -0400

    TODO

commit 7f46eb325dba872cea014a7a02f8f2a39cd5630c
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 12 13:40:18 2012 -0400

    by default, do 1000 syscalls per child.
    
    5 per child isn't really pushing things hard enough, and we waste
    too much time exiting/forking.

commit 7cd9649768fdae98e8cf0e8e474ac3c4d1cdbcdb
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 12 13:29:22 2012 -0400

    sanitise cmd arg of keyctl

commit de23149a46dacc07abafaf53188053a86977edad
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 12 00:12:30 2012 -0400

    compile fix (pid_t)

commit cedf69fbe9979fa8baaa0283ce40ea057fcece11
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 9 17:44:28 2012 -0500

    handle broken pipes

commit 1607c00c4590148b28939c3f1de2f7358bf0bc66
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 9 17:38:21 2012 -0500

    Rewrite signal handling entirely.
    
    * state machine variables about the child status/pidmap now in shm
    * Don't ignore SIGCHLD
    * Be explicit about which signals we choose to mask out (just SIGSEGV/SIGFPE for now)
    * Don't mask signals in child, we want them propagated up to the parent
    * parent now sleeps in waitpid() instead of spinning and sleep(1)'ing
    * no longer trap ctrl-c, we didn't do anything special with it.
    
    More improvement to come, but this is a clean place to stop for the week.

commit ef241a2b0e910884f6b7bed2fe43664899d57584
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 9 10:12:46 2012 -0500

    log the regeneration.

commit b169fbd7bc9062fe03bc45f76bacd24c35e0e534
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 9 02:19:45 2012 -0500

    handle WIFCONTINUED

commit 774b7f7c357ed23635e6e83b76d3e3c680839105
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 9 02:17:42 2012 -0500

    Send a kill after the SIGCONT
    
    It might be stuck for reasons other than ptrace.

commit 4f4146fd2b4e9402c95382d2bfda5b25c9cbb937
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 9 01:49:10 2012 -0500

    beginning of parent/child state machine
    
    The basic idea is that we now always keep n threads busy.
    Previously the parent would wait for the last thread to exit
    before spawning new ones. Now as soon as one dies, a new one
    takes its place.
    
    I had a version of this that worked better earlier this evening,
    but didn't commit it before I accidentally trashed up my source tree.
    In some ways this attempt is cleaner, but it's in need of some
    more debugging, which I'll attempt with a clearer head tomorrow.

commit 006864c2ab2e309035261a0804b4e06ee0026e4d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 8 23:53:43 2012 -0500

    factor out the regenerate code
    
    also, decrement the count in the children.

commit c7b3ef2e2af70ac3b2575632bcf48ad867c5257e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 8 23:51:58 2012 -0500

    set the return for setjmp just before we do the syscall.
    
    This is a better place to reenter.

commit 7df0ccbb7d357ff97cb03a26254c831e59cbddce
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 8 22:32:59 2012 -0500

    Turn off alarm after return from syscall.

commit 2f3a39401e78b51d3ac9fa3ae29bd8f3b8af8460
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 8 21:13:00 2012 -0500

    don't modify the syscalls_per_child parameter.
    
    create a local, and modify that instead.

commit b92e26d2a075fff63ce8982709c8db810fa0e592
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 8 18:55:50 2012 -0500

    log pid in signal handler

commit 4c47f9e8245cf143a77a174f57b327baf04b9ca6
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 8 18:15:17 2012 -0500

    change number of syscalls to do in child to a ulong

commit b63db6fad9b1cd5457679d163319f41ee2d9bb03
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 8 17:55:41 2012 -0500

    send the PTRACE_CONT, then waitpid.

commit 71ff685d65d8a02ef5ce3e43262ade41d346d0de
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 8 16:36:15 2012 -0500

    wait for ALL children to exit before exiting

commit c85930b6820218c04b3e62f5ab7774976fec5644
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 8 15:46:56 2012 -0500

    prevent heap corruption.
    
    The whole generate mappings thing needs to be rewritten tbh

commit 100c12cb296d48fa3ebace6fa85de3f0626ca54b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 8 15:00:34 2012 -0500

    further locking improvements

commit f3215ccf3f23b369f4c39d7e731dafece64568d3
Author: Dave Jones <davej@redhat.com>
Date:   Wed Mar 7 15:24:45 2012 -0500

    locking improvements

commit 936ec9c9f05f3e32ecc6efdf39f9c3defdd446ac
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 6 17:13:18 2012 -0500

    mask signals earlier

commit f0b78832b1b2e61087d11944d0ebb4a564159d7c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 6 17:05:15 2012 -0500

    avoid clone() again, until I figure out how to avoid creating zombie processes.

commit 74ff8f8a132dfad20c36e55b489813a9c4a5466f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 6 15:36:22 2012 -0500

    only do explicit log locking.
    
    locking doesn't nest, and output() unlocks.

commit fc24117142705c37351e61e33e7560d35cd86afc
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 6 15:21:49 2012 -0500

    Log the pid when reseeding

commit b4751024bf1b24e3fe7e2d66b2beae430877e9dc
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 6 15:14:28 2012 -0500

    change default logfile name.
    
    Now that we launch multiple threads, cpu0 is meaningless.

commit 8d5deff5ccd8a71c10021040fe334eff36cd4032
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 6 15:13:45 2012 -0500

    lock the logfile across both halves of the syscall output.

commit 59eae50fed4267997778fd754fbf4f6a4e586da2
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 6 14:52:00 2012 -0500

    Send PTRACE_CONT from the parent

commit 1c2923799f314f0b3e8da3eda3976bc5014e2e35
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 6 12:50:37 2012 -0500

    fork nr CPUs children

commit 133d41e793c8ccfd073a81035d10ce16d3a5de78
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 6 12:37:19 2012 -0500

    lock the logfile before writing.
    
    Multiple children writing to the same logfile without locking == mess.

commit a4c09645b468bd13c8bd1f8e4582f5bd7a2b58df
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 6 12:17:31 2012 -0500

    Launch five children from each process.

commit dda3d4558347f73c0de4e73272c335d3270888b7
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 6 12:11:38 2012 -0500

    factor out the child process code to a separate function.

commit 5093ecf46d2860f965ccd64268859f03b7294a43
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 6 12:01:20 2012 -0500

    add x64 PAGE_OFFSET to interesting addresses.
    
    Also reformat.

commit a05cefd6fa6a0c7f49c2703b4fdb6d1cf9b4b12e
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 5 23:08:55 2012 -0500

    remove -N1000 from test

commit 0337ed593170635804e35f1f8b70900e79fc45f9
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 5 18:55:59 2012 -0500

    -i is now the default.

commit 5cfbb0833f9c56d90974c96f179a8d312338881b
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 5 16:26:03 2012 -0500

    sanitise madvise flags

commit 7f85a149a4fba578f70584ffcc3d226f27b1830b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 2 01:32:19 2012 -0500

    Collapse do_syscall into do_syscall_from_child

commit 9be8a45546073fbe23fd82e2d1959d461a7dfb61
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 2 01:27:24 2012 -0500

    remove the bruteforce code.
    
    Just use '-c syscall' instead.

commit 580b78ac9b5c6925e8f20d5b2125d65a0c63d90f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 2 01:24:34 2012 -0500

    Remove unnecessary 'pick random syscall' duplicate.

commit 187936b2a456442b249e018fa38a28505fe04d09
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 2 01:21:05 2012 -0500

    Move the regeneration code to the same place.

commit 7ae5d755063c5255d17376e90573df77a604009b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 1 19:56:08 2012 -0500

    Update the help text

commit b47e1a8bf1df2cef2c9a0c54461ca694a68bb708
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 1 19:50:55 2012 -0500

    Rewrite a lot of the parent/child model.
    
    We had an extraneous fork() in there, and a whole bunch
    of code which was confused about where it belonged.

commit 072c1266001f74fef55cdff25bbf2911607fc6be
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 1 11:18:27 2012 -0500

    Do unconditional PTRACE_CONT before exiting child.
    
    If we randomly did a PTRACE_TRACEME, the child will hang on exit,
    and the parent doesn't know that it needs to do anything.

commit d25eb65f32c31db2859a43739346c3860c264907
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 28 23:01:00 2012 -0500

    fix inverted logic
    
    derp

commit 385475d7a6dab88c3c51e557d05ba29baec0358d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 28 19:51:27 2012 -0500

    Revert "Narrow the signal masking."
    
    This reverts commit d545d534bf543acede091c6c470d771bda3bc3d9.
    
    Turns out to clutter the logs too much.

commit a4e26987e47517aa3f9fbb71180b266d1cd6e700
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 28 19:45:02 2012 -0500

    add a switch to disable logging.

commit c2f58f284d39b23c49057c2b3774fc0c5c1ef5d8
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 28 18:30:11 2012 -0500

    sanitiser for move_pages

commit 051b2cdba543f0140a7c9ff1776282251d78d147
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 28 18:17:25 2012 -0500

    Narrow the signal masking.
    
    Makes things easier to debug when we get segfaults.

commit c7673689e4a5c76bce8a5f56f5f2beaab6c702d8
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 28 17:03:33 2012 -0500

    improve coverage of mmap sanitiser
    
    this seems to be occasionally causing malloc pool corruption.
    Probably needs some boundary checking like the maps code.

commit d485b2d2b1c56a5ef921ba37a932363322f0908d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 28 16:54:58 2012 -0500

    remove remnants of now unused PAGE_SIZE

commit a6f384e87bee277e4a5a8eb9cf8cdece09b129af
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 28 16:54:18 2012 -0500

    Consolidate PAGE_MASK.
    
    Make it use the page size we determine at runtime.

commit 05a1b400c60dd03bb832cde53b7ad87c928d6e97
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 28 15:59:27 2012 -0500

    lets throw remap_file_pages in with the vm group too

commit f85fb2a14b6ae16fd88b7640d068d2200c2e506a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 28 12:58:36 2012 -0500

    favor adding writable files to the fd table

commit c04d084441314f8ac0d1ff7c97d0841665187713
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 28 12:50:03 2012 -0500

    remove unnecessary initialisation

commit 8d62c7541d7add888b4b51982d4cd7210986b82e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 28 12:45:51 2012 -0500

    remove old cruft

commit f632e8783a076c7a4436a0895f7acb4c8be8155b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 28 12:37:32 2012 -0500

    remove the rotate mode.
    
    There's really no good reason to have this mode, now that coverage
    with random mode has improved so much.
    
    With random being the only mode left, remove the --mode code too.

commit 6a0464270d4e95cc122e97e6244e9f72d57f0fac
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 27 22:44:24 2012 -0500

    vm test script

commit ba69c163fa04282e7200050b3e65edfe9fb83968
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 27 22:35:48 2012 -0500

    annotate addr param of mlock/munlock

commit 5c74ff461ab5822c5833d52324deda85deee06a1
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 27 22:24:20 2012 -0500

    sanitise mlock[all]
    
    Passing 0 to this just EINVAL's.

commit 0c0def3b6f75ac167643ed448034650d55364c60
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 27 22:17:49 2012 -0500

    Call the right syscall when in group mode.
    
    Because we reconstruct the syscall table, we can't use a syscalls
    position in it as its syscall number.

commit 2be0e086724e3bbbf3488aa93e8d3f9e46ef43e3
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 27 21:56:23 2012 -0500

    new feature: syscall groups.
    
    -g vm will fuzz just vm related syscalls.

commit 9437f30fbf5ebec295b330b268a985d4bd0cc2f1
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 27 21:04:27 2012 -0500

    annotate VM syscalls as belong to GROUP_VM

commit 4451e9ca3b5dc536d95852ee7fea12a4cc3aa97c
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 27 20:34:14 2012 -0500

    Do multiple syscalls per child
    
    - By default do 5 syscalls per child forked
    - Change the -F parameter to override the number of syscalls.
      (The old "don't fork" behaviour of -F was broken, and no longer applies)

commit 07fcb368013ecd380f9c368be40aec0d046189c8
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 27 20:12:33 2012 -0500

    Sasha's roadkill of the day

commit 22fe002de0ab0012d70231ffe56486e3c93ab570
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 27 19:57:03 2012 -0500

    Fix up some splint warnings

commit b8603f8f2b06a9e7161b9b16ee32075c8364b819
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 27 19:08:08 2012 -0500

    split syscall.c into main.c
    
    The mainloop code should live in its own file as it doesn't do anything
    with syscalls directly.

commit ff7d2077cf27adf3449ad333d329347c3717c56d
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 27 17:12:31 2012 -0500

    log the pid of the child making the syscall.

commit f38deed53c3111ce2cbb75b9d3329347da29b7bd
Author: Dave Jones <davej@redhat.com>
Date:   Thu Feb 23 12:51:33 2012 -0500

    remove the retry logic from the mremap sanitise routine
    
    This caused more problems than it solved.

commit 919823d8523cf3ecbccd76e202dea15e7642d126
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 22 18:34:51 2012 -0500

    bind to cpu's in the forkbomb case
    
    it's hostile enough without the scheduler bouncing things around

commit 8d29c9a86771a0c722edf11adfdd938de47f10f1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 22 18:31:27 2012 -0500

    do file locking on the socket cachefile.

commit c0ad0c9d1448af034debb1e7bfae174f9bd5cc75
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 22 18:31:10 2012 -0500

    Fix an occasional fp exception on startup.

commit 6dbc0a9ffbc193671a88aeb2ecd2dbddd0acfaf2
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 22 17:07:49 2012 -0500

    more todo.

commit 3288a9bd9a6e6b2acc1ef12ca5baf9e5db2b9578
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 22 14:14:43 2012 -0500

    mark some strings const

commit f16082c205a58bb87c21d8966546b7c11e2f4158
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 22 14:03:27 2012 -0500

    add some infrastructure for making versioned releases.

commit bb4a83b6dd40dc9b2ba6925ed36841b4a8340b3d
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 22 12:26:25 2012 -0500

    rename the regeneration variable
    
    it regenerates more than just the fd's.

commit 18a1ee19b3a5ca9fb62415b4b4c18245ef505ffc
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 22 12:21:05 2012 -0500

    Fix memory leak.
    
    We can't call regenerate_random_page() from the parent, because we don't
    track the memory it may allocate, which leads to leaking massive amounts
    of memory when we regenerate every syscall.
    
    * Do the regeneration in the child.
    * Don't regenerate the page every syscall (unless we're fuzzing just one specific syscall)

commit 77430555c0c9419503d8483c7618c5cb61faa889
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 22 00:40:16 2012 -0500

    update the bugs found list

commit d508ca1be1ab3b7cd55d5754128b57cf19566388
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 22 00:29:26 2012 -0500

    update prototype to match Linus' change in faf309009e2e18d30c032b7d9479f29b91677c37

commit cc60f7c50b34e54310bb9c3004da76c88f81dc26
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 22 00:26:08 2012 -0500

    remove duplicate random socket code from sockets.c
    
    use the code from the sanitise function of socket()

commit c38096b25fb265f6d555f5564ac0e08d8392c9e8
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 22 00:15:50 2012 -0500

    add ->socket to socketcall sanitiser

commit 99b7a3f8eba0aa96d8bd295f5bc00da8ead6ee97
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 21 23:52:29 2012 -0500

    begin sanitise function for socketcall()
    
    needs fleshing out

commit e110db88939d2d549e6d4673c9cab71facd9ef23
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 21 23:52:18 2012 -0500

    sanitise function for socket()

commit 6125c0fa3cb1c1f8e1ceb26b37245fef41ad6029
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 21 23:11:28 2012 -0500

    remove annoying pragma's

commit 508debed39417f7887cb006d23c652e0f80d90c9
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 21 22:36:28 2012 -0500

    compile fix

commit d95f547c59fde01ccdafdaa0f041b392f2639655
Author: Dave Jones <davej@redhat.com>
Date:   Tue Feb 21 22:15:20 2012 -0500

    sendmsg sanitisation.
    
    * Fabricrate a fake msghdr struct to pass to the syscall
    
    * Beginning of some infrastructure here to allocate objects.
      For now, don't worry about freeing them.
      As the allocations all happen in a child process that exits,
      this shouldn't be a big deal.

commit d16ef8d036674fd4ac1f5233490f9200996ef36e
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 20 10:41:21 2012 -0500

    tty bug Sasha found

commit 7638f1c6acd3f9da3cdc3edc6e74505d85d19072
Author: Matt Alexander <mattalexander@mattalexander.com>
Date:   Wed Feb 8 18:17:01 2012 -0800

    Remove some noise when using --quiet with trinity
    
    cut down on some of the console noise when using the --quiet option.

commit c0f1718f30b10a39206f83a0d2e126a747657857
Author: Josh Boyer <jwboyer@redhat.com>
Date:   Tue Jan 24 15:36:08 2012 -0500

    Eliminate more hard coded PAGE_SIZE usage

commit 3115082ad4cfb3716906b6d8c3c277fe6631b846
Author: Josh Boyer <jwboyer@redhat.com>
Date:   Tue Jan 24 15:35:23 2012 -0500

    Add a sanitise function to mbind to eliminate hard coded PAGE_SIZE usage

commit 2404cc3cca266897643a96f3fa060ec835ff2313
Author: Dave Jones <davej@redhat.com>
Date:   Wed Feb 1 18:52:19 2012 -0500

    another bug that got found

commit d3036d20b050a5357633f611d9bc2c6cd54f8609
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 24 00:56:39 2012 -0500

    CAIF bug

commit c002d4e1d4fc1a40ece6f44ed227112d7ff96f88
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jan 22 16:25:17 2012 -0500

    yet another bug found.

commit 77f50f210008acefb47bf130810aff3fffef386d
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 19 21:54:11 2012 -0500

    remove commas from ->name

commit d72c408e34f5afeb4d290fd4aea37b5d7297b51e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 19 21:51:22 2012 -0500

    Add missing prototype.

commit a2c03e741c43342d0e4d0da318e8911e004beb6a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jan 19 21:48:58 2012 -0500

    Update syscall list as of 3.3rc1
    
    adds syscall_process_vm_readv and syscall_process_vm_writev, and
    also wires up some previous syscalls that got missed on x86[64]

commit 4ae8ee0635fa09b433b1861397674d8925f06a92
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 18 23:28:07 2012 -0500

    another bug Sasha found.

commit cbac8607025498b9e574fa317a0e2ecddcbbdf70
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 17 23:19:59 2012 -0500

    Handle ctrl-c properly during the file scan.

commit bf9238b6e6e4c77733171444a306f148cb0fc43b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 17 23:15:09 2012 -0500

    FD regeneration should happen in the parent, not the child.

commit f803f18492aab723ae3d73f2eb042ead233b69c9
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jan 14 16:48:28 2012 -0500

    Some random files tweaks
    
    - Add debugfs files to the mix
    - Decrease the likelyhood of a file being added
    - Keep scanning for files until we have enough.

commit 8c5f980e1e1e7664d1186e86b2c29c149f3fb736
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jan 14 12:29:52 2012 -0500

    make 0 an interesting 32 bit number too

commit af84988bb2f7184bf594d73b32bc0ef094adfa36
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jan 14 12:28:37 2012 -0500

    pass some 0-0xff bytes around.

commit a2bf21a3e55ddc242ab351270cbf5d6603a62137
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jan 14 12:20:53 2012 -0500

    add note about a gotcha for later

commit a4c77731f5ec7bea35fb5657d6bd5cdd6af5145d
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jan 14 12:17:30 2012 -0500

    TODO cleanup

commit ae70eb75a1b305a04fc7350e8d0b0450d7b4a2e8
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jan 14 12:06:14 2012 -0500

    sasha's bug of the day

commit 9784862326b31015f4793d827854189945408a09
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jan 11 16:57:02 2012 -0500

    another bug.

commit f3d08277323bfba06e18ef31ef91b0ff6f2b1963
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 3 20:02:09 2012 -0500

    futex hangs were possibly fixed.

commit f744371857143c73c1f3bedeae3f2c12c3b4051c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 3 19:59:52 2012 -0500

    Fix double increment.
    
    Spotted by Sasha Levin.

commit 5fce62b2872982fe621454f602f6f93dc64e1238
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jan 3 19:58:41 2012 -0500

    another bug found with trinity

commit 78f22be2a578692f8ee316f7582d5b4783da80b5
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 28 21:58:47 2011 -0500

    this also affects futex()

commit 577c0b9912c64229a8b647ec1382994c109b63c8
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 28 17:16:46 2011 -0500

    If the kernel was tainted on startup, run anyway.

commit 79ea327e0fb7f035acc5574a7a4e75cc1223419f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 27 13:12:32 2011 -0500

    Sasha found another bug.

commit c5a5a938d59930fd000856f7d29b8b0efe51860c
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 26 21:14:51 2011 -0500

    log timestamp

commit 372f33416feccaa6aacd41a66ae605ed97408eff
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 21 10:13:11 2011 -0500

    Another bug found. Fix pending.

commit 9767e44d1a58d9f2257df2c70f7af686ab33b5d5
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 16 13:56:32 2011 -0500

    Allow running as root with --dangerous
    
    Requested by Sasha and others, for testing virtual machines.

commit 97aa8094c4fe11f488fb33ca3eea63f4333773af
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 7 12:14:00 2011 -0500

    another bug found

commit 08a93d01d91c718fa7653ed7d672d24d5b4de983
Author: Dave Jones <davej@redhat.com>
Date:   Mon Dec 5 01:08:43 2011 -0500

    idea from Sasha to blacklist certain syscalls.

commit c168047c7fd2403336bb210ddc0af92b91311833
Author: Dave Jones <davej@redhat.com>
Date:   Sun Dec 4 17:52:14 2011 -0500

    Sasha found another bug with trinity

commit bffe00e610291aa4bd12dede9ae4672b85ffcb59
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 29 16:27:26 2011 -0500

    this was likely another trinity discovery

commit 6f6da91a16113a23ed3135ac166d789b5b4a6aa0
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 28 19:31:14 2011 -0500

    Note the bugs that Sasha Levin found using trinity.

commit 2b86cdc1f490d8769a8c36fbcea460c91ce31d02
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 17 17:42:01 2011 -0500

    merge two items that are related, and elaborate

commit 1b4685232280e473df8b88b685d22be7a93f2120
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 17 17:31:54 2011 -0500

    Something to think about for ARG_PID

commit 727b73e81e0a81b94d1f6614528f9f9e3727e132
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 17 17:29:27 2011 -0500

    annotate ptrace request types

commit 48db7580ade63882b9f525acb635a8ec43fd967e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 15 19:23:55 2011 -0500

    Fix ridiculously stupid bug that broke the actual syscall.
    
    This is up there in the top 10 'stupid bugs I wrote'.
    We were only ever fuzzing lstat for the last four months.

commit dd6a18e3fdd4b88eaf1f5f23279eaf4a9a6143c5
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 9 18:38:16 2011 -0500

    start annotating return values

commit b6964c88d91e1a46fc509204399cd5f488a7b333
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 9 18:37:48 2011 -0500

    decide what to do about 'dupe' syscalls.

commit c39b44a3c75184079d672b532e8ffbc53b93a543
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 9 18:14:24 2011 -0500

    fcntl64 is really just a wrapper for 32bit archs.
    
    given it takes the same arguments, just delete the fcntl64 definition.

commit dfaa7d4228cc0f872fbd82fefa8ebf3172ce46d1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 9 18:04:46 2011 -0500

    annotate arg parameter of fcntl64

commit a2da281937d1ef956b09a3adb6d82f3e4e1c5f69
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 9 16:01:33 2011 -0500

    sanitise the arguments of arch_prctl

commit cb1d2fd122c5784296bab9c6d4ca41a913583c65
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 9 16:00:56 2011 -0500

    avoid calling execve, in case it's successful.

commit 7045bfd01354240257c9d744094cd07155511185
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 9 14:38:46 2011 -0500

    Check /proc/sys/kernel/tainted after each syscall.
    Stop fuzzing if we find the kernel has become tainted.

commit cb0504e3c19fb7a3740ec7fd09244253fa0c0e21
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 9 12:16:36 2011 -0500

    remove dupe

commit 30908e27496ebf3c6789edd7b3138f3b34d9e233
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 9 12:14:54 2011 -0500

    Check tainted, and stop fuzzing

commit c46d07fa1adb3e48fa34993421bc879426bb5d6a
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 9 12:13:42 2011 -0500

    possible cleanup for later

commit b2f225715c03d19cd80ec647b5a134d0e7105e7c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 9 12:09:19 2011 -0500

    info about guard pages

commit 779eb3ba0a1de4db832fd89f9dc519116c714ecb
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 9 11:56:10 2011 -0500

    leak checking

commit c64ba0dc9360b196734d01fefbf1df188dabac07
Author: Vince Weaver <vweaver1@eecs.utk.edu>
Date:   Tue Nov 8 13:01:04 2011 -0500

    Add perf_event_open() sanitise
    
    I was looking to extend trinity to be better at finding perf_event bugs,
    as I spend a non-trivial amount of time bisecting kernel DoS caused by
    perf_event.
    
    Below is a start of a sanitise function.

commit 61fd1f3e22ef95b1dd2afbced89ec30bf6921bf4
Author: Vince Weaver <vweaver1@eecs.utk.edu>
Date:   Tue Nov 8 13:01:04 2011 -0500

    fix pid randomisation in generic-sanitise.
    
    I noticed, the get_pid() function in generic-sanitise.c
    seems to have too small a value for the rand()/% call, and
    the pid mask (if that's what it's supposed to be) is doing a 0 or
    1^16 setting, not masking at all.

commit e17a5243441dc3fc59e57420cc46c3ff15b9718f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 8 14:17:17 2011 -0500

    update to new syntax

commit 487ba33d7b5e195f0b026f129eda0f2abb385c82
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 1 19:30:48 2011 -0400

    speed up X25 packet creation

commit d739d45f936029ecf544387246448e7380b15678
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 18 20:48:27 2011 -0400

    make sure we always log before doing the syscall.

commit 5bc2b6932c0e205277c63f5332464062f84d2bf2
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 18 19:01:17 2011 -0400

    add a quiet mode.

commit c4689603d3d8b6ea671092c11c166a3c907fd84e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Aug 18 18:59:57 2011 -0400

    don't log the mode over and over

commit c3f66bffc53e8f6ca968da3c3dcedc0c72a8f720
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 28 15:22:37 2011 -0400

    fix typo

commit 5e4305386a73d93acf3ed99717258cfbaa1af184
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 28 15:18:30 2011 -0400

    update syscall list for x86-64 to match current Linus' tree.

commit d5fafe6c62d05ac2fa13d41d26a95a4eaabdbf61
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 25 02:53:59 2011 -0400

    use different data sizes in random page.

commit d7b39266af75f3c34068c4bb404e9927ded9d3d8
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 25 02:42:37 2011 -0400

    a new page to pass around. a page of pointers.

commit db909621e636bb59becc57d762e51a5af85eed06
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 25 02:42:24 2011 -0400

    some new bugs

commit 0abcc51d6adada1282c93847e06d691558cafabd
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jul 23 22:13:06 2011 -0400

    remove some other 32bit mode detritus

commit f2c8c97fd52f2608674739464725138e71567557
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jul 23 22:09:23 2011 -0400

    remove the whole --32bit path
    
    This is pointless. a 64bit process can't call the int 80h gate,
    it just behaves as if we syscall()'d.
    
    For fuzzing the 32bit emulation, we should build a 32bit binary and use that.

commit 5ea660d60d8c2d84082900b89c0869ecdd5068dd
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 20 21:42:46 2011 -0400

    use correct include

commit b21553518f28cc7c6ec20707a43278cd2a8b1e1a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 19 19:58:02 2011 -0400

    not pretty, but this works.

commit b0a0fd45a2d1c2b8ccc9f206469b09cacee8739c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 19 19:44:12 2011 -0400

    to fix on my flight back.

commit f5ea61f84e17aec9082b3325b8ad29c6077502b4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 19 19:36:03 2011 -0400

    socketcall annotation

commit 0f093ddb9d2de0ea910f168f34c77ec9bf4ed20d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 19 17:37:39 2011 -0400

    annotate 'op' from sys_futex

commit e4735b4aeea8910f204b9195b5016b07b3324c96
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 19 15:30:44 2011 -0400

    another fix for ubuntu 10.04 from Matt Alexander

commit 1745cdd3f201082703679f074424f5390484c543
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jul 19 13:37:53 2011 -0400

    another potential bug scenario to try and come up with.

commit b362ab5ca7f2639e9bea43e1ea41082c2879c014
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 15 17:02:38 2011 -0400

    annotate op field of epoll_ctl

commit 3880969d2a4d85ea55d206e79f7755bfe4d98a53
Author: Matt Alexander <mattalexander@mattalexander.com>
Date:   Thu Jul 14 17:24:54 2011 -0400

    document why some syscalls are AVOID

commit ab149fbe5df5f2d8d171b8b611972c6c0f227b0c
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 14 17:02:59 2011 -0400

    document -b bug.

commit 90a5b85865b0de8a43d64446943729283de1f9a5
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jul 14 16:56:07 2011 -0400

    Fix -c 0 parsing
    
    The check for 'was the parameter numeric' was pretty bogus, and failed if
    the number passed was 0.

commit 0dca256796f0d20a986d5aaef1dec8b974db9169
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 13 18:00:42 2011 -0400

    seed is unsigned

commit 778fb9d8e4158d9459f51a344e6d2bd6cb57df3e
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 13 17:59:41 2011 -0400

    actually, log it in both bases

commit 7b5a811ee5fe585b4845dbc9717c2a25ca7fd4ee
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 13 17:58:08 2011 -0400

    output seed in hex (and only when passed -s)

commit 633b7593f3cbeceb6d8536ed105b75949e252ffe
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 13 17:55:33 2011 -0400

    be defensive in output()
    
    this should hopefully prevent the previous bug from reoccuring.

commit d0943bfb2c9dc5c522bb6f50a1529a6e6de604e2
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 13 17:54:06 2011 -0400

    Don't use output() before we opened the logfile.

commit 29818c8aad93820af1aba75125989dd0abc11577
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 13 16:58:46 2011 -0400

    add missing format string.

commit 9a16dc0a5cb171ca3f0903d5c530fc92ffe9b244
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 13 16:57:48 2011 -0400

    another compile fix for Ubuntu 10.04 from Matt Alexander

commit 2438cfa90e4011e6fb3560a996fee593b056e123
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 13 16:57:41 2011 -0400

    missing include

commit 99a37694df7649d853bc1300b6080f880daddcb1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 13 14:21:26 2011 -0400

    more fixes for ubuntu 10.04

commit 9b43f63fc6c6295493f96ae3ed63014a89d7aea3
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jul 13 13:50:35 2011 -0400

    build fixes for Ubuntu 10.04

commit 4184041c853844ade96e3ed62e98159226faab0a
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 11 19:01:53 2011 -0400

    stupid off by one bugs

commit cb2d7e78c60e9cdaa5eaa8d5958bbbdac45d7bb9
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 11 18:54:14 2011 -0400

    no idea where this came from, but it's not around any more.

commit 282031480f50387e4a366133326650bbe2d3cda9
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jul 11 18:53:17 2011 -0400

    fix obvious build failure on ppc

commit fa341d9475ec3ab3409d8f492d90ba7c86bf2e1c
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jul 3 23:20:51 2011 -0400

    destroy contents of tmp dirs after exiting.

commit 7f862e0a81ea43ffbb500e7c27dba22a9b6830d5
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 30 16:17:09 2011 -0400

    ip route me harder bug finally got fixed

commit 0272f54b14a14ce9bb9994320e1a5d8e80820b69
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 27 18:40:32 2011 -0400

    Remove alignment special case for rotate mode
    
    This is done in the generic sanitise code.

commit aa312beed97d646c4612751d8b6c4ce66c347c87
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 27 18:38:09 2011 -0400

    Remove the remnants of the struct mode.
    
    This is better done in the sanitise code to pass structs
    that actually look semi-sensible.

commit dd4b44c591bfbc772b3bfff42a6ad2a76f69889d
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 27 18:30:10 2011 -0400

    exit if the option parser fails.

commit 1ce4bed3144a9b9cf19fa5d2cd4a6ababc7049cf
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 27 18:17:47 2011 -0400

    remove references to old modes that have been removed.

commit 69563773be4051d7569afc21356432d276eb5601
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 27 18:16:46 2011 -0400

    fix up some incorrect argument definitions

commit f92200869bba78987826895cf8cd69251e92d1f9
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 27 17:57:17 2011 -0400

    print type as hex

commit da6e34c9a52baea9b9525697a28bf8ab75a56415
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 27 17:55:06 2011 -0400

    speed up ipv6 socket creation

commit e8a0d2cc97c7283171f5ec1344e4a14984ce4640
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 27 17:36:45 2011 -0400

    display possible values for -P

commit d7e12ddb581b714eb1b44fa1910fdb8bb73f00a1
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 27 17:34:46 2011 -0400

    the --proto case is slow, so we want to use the cachefile where possible.

commit 43fabd57f60cbb64035db97019caf9d1d572ea4b
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 27 17:24:42 2011 -0400

    if we pass --proto, assume we only care about sockets.

commit b6faa181cbb913e28fb5c1b1f189f58bb69651c7
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 27 17:23:05 2011 -0400

    Reduce the range of the type arg, to speed up searching.

commit d72a9beb502392e540ba00e7f3f5e7528195f5b6
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 27 17:08:18 2011 -0400

    Address family should get passed in domain, not in proto.
    
    TODO: Limit type/proto based on passed domain.

commit df3f73b95fe12be09eaa7981c27b59eece07aac5
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 27 17:03:53 2011 -0400

    Fix segv when passing an unknown protocol.

commit bd5be619f0a3450d43b72ede2a64933c1e1d1fff
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jun 25 15:59:55 2011 -0400

    move the zeropage mappings further away from the shm.
    
    do this by creating a bunch of useless mmaps that will just take up VA space.

commit 930e744ebf6a7ddfc5b125ffa48de9f600fe08a6
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jun 25 15:51:20 2011 -0400

    more logging

commit 936c03e2dd8eeec8bf9d91fdeeb66ebb1bd17452
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jun 25 15:50:33 2011 -0400

    dumping should log.

commit 24ac75e830b443465e2fb4124df9d5045cf074b0
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jun 25 15:41:24 2011 -0400

    signed warning

commit 633d1c47f3f4fbf89e6537430325ad33a0635b41
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jun 25 15:40:41 2011 -0400

    log that we're skipping mappings.

commit 9dbc7a3f101a1963cdb063b6fba69e2cf0f287a8
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 24 11:45:11 2011 -0400

    fix 32bit warning

commit 7f66e1e9b90b42ab96f902d1223551c13656b038
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 20 13:08:12 2011 -0400

    generate some /dev/zero mappings.

commit 68fabd27558e91e498fe5ed06aa0f6f175454c7e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 16 11:03:33 2011 -0400

    if we pass -P, then ignore the cachefile.

commit ca9b0d2d915b17706a7de6e4f9c20d9c979340a5
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 15 13:10:20 2011 -0400

    close maps file once we've read it.

commit cd0f23ba0347beb84b439c2e62f2e419da9796e0
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jun 12 17:42:51 2011 -0400

    move some unfixed bugs to fixed, along with their sha1's.

commit fb0ff6b3f19f090aa3b53d1d85c955e93ac86e16
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 8 10:42:00 2011 -0400

    add a command line arg to force all sockets created to use a specific protocol.

commit 0254443a921ed03b86346e9386660874f7122a18
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 20 23:32:39 2011 -0400

    another bug found that I forgot to mention.

commit 6c037820a2aec1da9835ae73967792fe104f75f5
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 20 11:53:08 2011 -0400

    pass -1 to mmap 50% of the time when mapping MAP_ANONYMOUS

commit 0ff5729daa78741a6b6130c23d554272215b651b
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 18 18:48:37 2011 -0400

    Add SHA1's to fixed bugs, and bundle unfixed bugs together

commit f02428cc625b582d7848acffb12d8fb9ad9c927b
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 18 13:45:10 2011 -0400

    reset perms before doing 32bit random test.

commit 06d36ab2183fffdad9822973a7fa101c6ae2b3a6
Author: Dave Jones <davej@redhat.com>
Date:   Sat May 14 07:23:51 2011 -0400

    TODO: add versioning to syscalls

commit 764bccc52ee5e4f2ab6186eccdc8fcded1d6e483
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:44:09 2011 -0400

    All sanitisers now moved, so don't try building from sanitise/

commit c700bfdf979f077c629989d67f63e1f51a8aefef
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:43:13 2011 -0400

    move writre sanity into syscall def

commit 5cff51e343bd6ce4556110b2f4f211993ba673db
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:42:14 2011 -0400

    move vmsplice sanity into syscall def

commit 8a329d8f8a466214ed82018a04a2c02b8f82ef79
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:41:20 2011 -0400

    sync_file_range2 is ppc specific.
    
    additionally, it takes different arguments to sync_file_range, so
    can't use the same sanitise routine.

commit d243f6332faa1f336f9b4e96a9b5220263c5e23d
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:34:58 2011 -0400

    move sync_file_range sanity into syscall def

commit d9416fa2db1ef150d73b1bd6b6eb2b8ffa121496
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:33:34 2011 -0400

    move set_robust_list sanity into syscall def

commit bea6400fada62ae3ee5dd829ace896f9016370ac
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:32:32 2011 -0400

    move sendto sanity into syscall def

commit 5c3a8885164d5ea4bd7a978b19a5fc657a9f2304
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:30:11 2011 -0400

    move rt_sigprocmask sanity into syscall def

commit b1bf1c6b878470f3d0b70b38eff4bbf09073baa6
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:29:19 2011 -0400

    move rt_sigaction sanity into syscall def

commit 3a21729738e29f9599ac64600f0b39716edf3815
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:27:18 2011 -0400

    move remap_file_pages sanity into syscall def

commit 39502d603b0bf62415f02a9bd24f6faf081ff2a4
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:26:28 2011 -0400

    move read sanity into syscall def

commit d25046eb05df0e4b19324c90c3e9d3812744dd2c
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:23:31 2011 -0400

    move pwrite64 sanity into syscall def

commit 737ab29122da2ef322dc6f6703502ce2256de6a4
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:22:39 2011 -0400

    move pread64 sanity into syscall def

commit 2328af2c8b4c3ba0c50939befd9ccc2fe9ca478e
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:21:51 2011 -0400

    move mremap sanitise into syscall def

commit 4c6eb597d07579c103fd046b459845144e6126d3
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:20:34 2011 -0400

    move mprotect sanitise into syscall def

commit 2664d441b973fcd2c3b1286e151e1bee4207be34
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:19:18 2011 -0400

    move mmap sanitise into syscall def

commit 68026da53120b09ec9f6aed8a3a941e091b78551
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:17:36 2011 -0400

    move ioctl sanitise into syscall

commit 14267bbc1c233bc486e70ac6a8d4a898b54af154
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:14:23 2011 -0400

    Move fanotify_mark sanitise into syscall def

commit 463b132a83ac07ca6e4f4b05cd4608f88321e3fa
Author: Dave Jones <davej@redhat.com>
Date:   Fri May 13 14:09:29 2011 -0400

    Move the syscall definitions into proper structs instead of #include's
    
    This will allow for further cleanup, including the next stage:
    merging the sanitising into the syscall files.

commit 73320885f812487a6d3580b2989567f6c5dd26b1
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 12 16:32:38 2011 -0400

    skip adjacent mappings of the shm
    
    some syscalls will scribble on things > PAGE_SIZE, so the nearby
    mappings will cause us to scribble over the shm.

commit ba7273803a824a5657391da8ea23cc6cd0c5a87c
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 12 15:34:32 2011 -0400

    cast the pid we return in generic-sanity

commit eda6887724ef1c3c074b907c079715cee08eb309
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 12 15:10:51 2011 -0400

    regenerate mappings periodically.
    
    (only really useful with -F)

commit 8582b6c0b8850cc7e557b661ea314d3a7d1706d0
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 12 15:10:33 2011 -0400

    reset number of mappings on destruction

commit 970ec42f88e40dca65d5a8e3de5f8d8b66e8bfef
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 12 15:01:08 2011 -0400

    skip to EOL when we find the shm address

commit 288d0b09526d47a9d5c254fa6f4c2eb760bd6028
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 12 14:13:04 2011 -0400

    Log the owner of the mappings we pass.

commit c4fb4a415a066884c3d1490208db445bfb4f6316
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 12 13:15:35 2011 -0400

    log the start/end of each mapping

commit 7b492d3617b9879ada309b323ebce9631aabb7fd
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 12 13:05:05 2011 -0400

    fix a missing colour code setting.

commit 1f094a31b6e514a88058acc2eca327857cbeedaa
Author: Dave Jones <davej@redhat.com>
Date:   Thu May 5 01:05:59 2011 -0400

    various TODO.
    
    (gratuitous commit at 30,000 feet).

commit 5f587a48a17d91ae01fad3cc3ab9a97b190936e7
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 27 01:40:06 2011 -0400

    oops, deleted this by mistake in previous commit

commit 216d84d58099b831ce72564edbe2a72c9ada9302
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 27 01:38:36 2011 -0400

    document a bug that needs fixing

commit 03f687866a31bd89ee3f2c8dde55dea6a9953121
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 26 19:10:15 2011 -0400

    change the random page generation to return 'structs' more often.

commit f42d252e72fac2750f2c7300729fe0d54263edd6
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 26 19:01:23 2011 -0400

    every 25000 calls, regenerate the fd's.

commit 81fd88509eb018c1655298083fb12cc1da6a436c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 26 18:20:01 2011 -0400

    Change the pipe generation to use its own array

commit 43c96c7465fdfb2af49f26a744a48fa01b3e616b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 26 18:12:17 2011 -0400

    only do fd avoiding for file fd's, not sockets.

commit b538085732d4d7bbc16fe87885cd0163b0e8c993
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 26 18:08:54 2011 -0400

    Fix buffer overrun during file opening.
    
    This didn't happen every time, so was a pain to track down.
    We need to do the check before we try and open anything else
    and add to the array.

commit 87e26f830ebd4e0beec143d3349ef67d5282a5e8
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 26 03:33:03 2011 -0400

    fix another off-by-one. This time in -N

commit cd0fb478347f96c4bbc655debe2ebb400c23fc44
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 26 03:29:58 2011 -0400

    start counting execcount from 1. Fix off-by-one's

commit 923b4110ed4c5d88efe34546395ffe148a0c1dee
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 26 02:50:32 2011 -0400

    introduce sync_output() to flush tty & logfile.

commit bb0c5aded5dfe6173e6465f7437ceeb40f1f8f10
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 26 02:06:40 2011 -0400

    build with -D_FORTIFY_SOURCE=2 (add assorted small warning fixes)

commit 3dc821fceee8d8a9f1b3704af483edd3f4cb0f7c
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 25 21:03:02 2011 -0400

    print numbers < 1024 as decimal in the register dumper

commit 96e51081318b9684accc3f4c1e51c23eaccf2999
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 25 20:44:57 2011 -0400

    debug

commit 29d7030f2e85dac17b8740e6064ff030346e8fda
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 25 20:31:25 2011 -0400

    clock_nanosleep todo

commit 96e419030d4c7379de7ee47b768a07d4c9094149
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 25 20:02:37 2011 -0400

    output what the addresses point to in the register dumper

commit 9499c558e46f3f0a42a6fab6ac9f21fb7ba6a2ec
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 25 19:48:04 2011 -0400

    add CVE refs

commit 2aa997b7eb6a72975f0f4a3f7524d18ae55f1460
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 25 19:45:30 2011 -0400

    don't write past end of array when creating sockets.

commit adb3758dc88599d7be9affa95e07f207dee3f1fb
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 25 19:28:13 2011 -0400

    make the fd / socket buffers appropriately sized.

commit 8ae653c30959ae1ea2b515494e256c7df4fdbd9f
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 25 19:24:06 2011 -0400

    sockets code shouldn't mess with fd index. (which is for the files fd's only)

commit 508acff488b6eff9db7e089ea32b831728b8f7b3
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 25 17:27:48 2011 -0400

    avoid returning stdin/stdout/logfile when asked for a random fd
    
    otherwise we spew garbage to same.

commit 30a6de5e846ac47b09e3fa0ca9b322b422420464
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 25 17:21:12 2011 -0400

    remove duplicate print

commit 4c8604f19960d6206c0dd2b4b4f0992e79e684ab
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 25 17:18:23 2011 -0400

    log the syscall count

commit b55eceaffcfe756080672787c46709e69f447117
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 25 16:40:07 2011 -0400

    improved logging was done.
    
    Still some more improvements to come, but the TODO item can go

commit 46543df8a887f13aed591404c155837133216a12
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 25 15:17:26 2011 -0400

    several fd table fixes.
    
    - we were overwriting already existing fd entries
    - print the actual fd, not the array index.
    - fix a hard to hit segfault that happens when a file open failed.

commit 19039302ac1691938f1fafa6ebe5f483d4905b53
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 24 02:11:02 2011 -0400

    set timeouts for poll.

commit 93b3b689a914654debdd6327ea09dae588013581
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 24 02:01:08 2011 -0400

    close the logfile on exit

commit dd77d9228a553fc2b9a55606511fecdc2f682683
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 24 02:00:30 2011 -0400

    print the actual fd's

commit 7478feb06adb567086b68f745985377c5e54d2db
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 24 01:33:06 2011 -0400

    for now at least, don't close any of our fd's.
    
    later we might want to open something new in its place,
    but for now, lets keep it simple

commit 65798af14133e805bfd4eb1df15a58778e1955cf
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 24 01:22:32 2011 -0400

    fix tmp dir perms on each run

commit 52dce8b6858dab93ca0af0693177693805084bbd
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 24 00:57:30 2011 -0400

    some debug info.

commit b742183d26a23f9fc88f123fd1565e5139a4170d
Author: Dave Jones <davej@redhat.com>
Date:   Sat Apr 23 21:29:57 2011 -0400

    output socketcache used info to logfile

commit 277f81ed47167c0965e98c6a9f57de90df2649a0
Author: Dave Jones <davej@redhat.com>
Date:   Sat Apr 23 21:27:47 2011 -0400

    close the sockets when we exit.

commit efeea46fc663842a512ed767f6f133fe426d490a
Author: Dave Jones <davej@redhat.com>
Date:   Sat Apr 23 18:10:31 2011 -0400

    improve logfile code.
    
    log to stdout and a file, which allows to remove some duplicate outputs.
    This will later allow for --disable-logging to work, and also
    some other improvements.

commit b132e560b874c8e244b360acf0a0655b00af4a52
Author: Dave Jones <davej@redhat.com>
Date:   Sat Apr 23 18:10:23 2011 -0400

    some more bugs found.

commit 43329110ecb78ca7212ba006b546e19add9cda88
Author: Dave Jones <davej@redhat.com>
Date:   Sat Apr 23 18:10:14 2011 -0400

    some more TODO

commit 49805c43053b7d3244cbbe8db4803c117d7bd0a1
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 22 01:30:43 2011 -0400

    increase runtime of random tests
    
    startup costs are longer than runtime, so bump it up 10x

commit 94db3254e81a8843d1172461c1aeddf12ddb11a0
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 22 01:27:59 2011 -0400

    don't pause between syscalls any more.
    
    this was there for debugging yesterday, forgot to take it back out.

commit 82b92d6f23ea466fc29e0d88afabd3d5cffe48b3
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 21 22:30:32 2011 -0400

    use the same socket regeneration code for both codepaths.

commit 6ff67e6b48aeb6ff1b54968d1404d10255f7d062
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 21 22:20:23 2011 -0400

    several fixes for the stale socket handling code.
    
    - If we opened any sockets from the cachefile, close them before we start over.
    - after regeneration, don't drop through to the code that assumes
      cachefile success.

commit 159200e7af689ed1ad52019cb487c3b02ea38466
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 21 13:40:10 2011 -0400

    improve random socket generation.
    
    the randomness always seemed to favour some protocols a lot.
    (typically I saw a ton of CAN sockets all the time).
    
    Now the code goes through every domain sequentially, picking up
    to 5 random sockets from each. After we walk through all possible
    domains, we start over, until we have enough sockets.
    
    this should give better distribution.

commit 12927149e40cbd7d07eeb10d1721cc9c9c16b3e7
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 21 00:55:52 2011 -0400

    do rotate tests forever
    
    we get different randomness in the non-rotating regs, so it's
    worth doing.

commit a81ab3d1e7e5077f682a9557b446e68cf1bfc248
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 21 00:53:08 2011 -0400

    don't do nr processes > nr cpus.
    
    we just waste time overscheduling

commit 962bf6a95e3e867711bc938acdcc6505b6e7ba0e
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 20 18:23:31 2011 -0400

    remove some unneeded includes

commit 2cf0ff8bee26cad4f0f390d597d8c48e585eff11
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 20 18:20:43 2011 -0400

    remove the other redefinitions too

commit 9bced5a0c1998f3b6f976e910008248c0720a23c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 20 18:18:15 2011 -0400

    remove duplicate define

commit d154acfeb055f3f1d7780b4ce94659e606587c10
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 20 18:17:44 2011 -0400

    Split files.c into separate files / sockets code.

commit ee7de3d129c15be52d3e6ad60237e65aae94c974
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 19 23:04:07 2011 -0400

    recent network bugs

commit be5d82d2cb1df93b47e143c4be5d17cb2ea1fc67
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 19 02:02:20 2011 -0400

    play with the file/dir bias a little.

commit 621f5a5d9e177fe81ae4750d9d8f2c2a6608a1ef
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 19 00:08:24 2011 -0400

    limit mq_timedsend.h msg_prio to MQ_PRIO_MAX

commit 4793d9b66f8ee0d0ba4f9633b8eee80abae143d4
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 23:52:40 2011 -0400

    the mqdes args are file descriptors

commit d9b404bebc35ca866f1e8ea83e3451f43b602be0
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 23:25:00 2011 -0400

    parallelise

commit 04e8c4390486d2fe104a1463a4ec9df1291e709d
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 21:25:45 2011 -0400

    chmod the tmp dir before we chdir.

commit ab8e78af8f70dc645668b2b700b25e80207b4e71
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 21:24:42 2011 -0400

    Regenerate the sockets every 100 syscalls, just for extra randomness.

commit c616e77812e2c4f0121e54cffbf935ee1160a277
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 21:23:48 2011 -0400

    when passing -N and -F, we need to check the syscall count in another place.

commit 0b5b5976d8769782d83c81bce65fdf7e78efbd70
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 20:46:37 2011 -0400

    update the TODO some more.

commit 6c56e2d8be8edbfacf94b92e9530f263a2ebc70c
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 19:46:30 2011 -0400

    debugging info

commit d5a55f61538f4aa5e9779069bc5f620daf3047ad
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 19:45:49 2011 -0400

    chmod the tmp dir before we use it

commit 263be16dc358ca59224e52d55c8bbbd49d9015eb
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 19:20:47 2011 -0400

    regenerate socket cache if something doesn't look right.

commit df80ea7a77396020d8e69ba0f7a61e1d10f8ca0a
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 19:11:37 2011 -0400

    clear -F when in rotate mode.
    
    Because we can't carry on where we left off in the new child.

commit e382bfb07561ab784918fd5cabf3bdd210e8d094
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 18:59:08 2011 -0400

    this is also nearly upstream

commit d7fa7f978fbfb8f8722660f85c25cd446e25956f
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 18:58:19 2011 -0400

    some in-progress stuff.

commit 3a8d81bc97e35486bd4f9155aac72513caddadd1
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 17:02:06 2011 -0400

    improve the test scripts
    
    split up random/rotate
    use a 'logs' directory.

commit 02fed0015c47fbb29491d731851210b39fc6297d
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 17:01:06 2011 -0400

    add a new test
    
    this causes tremendous load on a system. Recommended only for
    systems with lots of processors.

commit 5ca8d022e9cdf133f664f67858a36913aa7b0074
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 16:56:09 2011 -0400

    use a different seed in every child, or we do the same syscalls over and over.

commit cc550750c59747df5fb73bb251ae969d651d9a83
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 16:44:48 2011 -0400

    move execcount into the shm.
    
    necessary due to the new forking code.

commit a60e384b7c2966520414f5512a68e472dbb52933
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 16:37:39 2011 -0400

    better handling of -F
    
    If we segfaulted when running in -F mode, we would exit the fuzzer.
    Now, we always have at least one child process for calling syscalls.

commit 1b3f74a512fc41cc19ea90a5067156b80cf8d7c4
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 15:37:29 2011 -0400

    abort fuzzing specific syscalls when passed numerically too.
    
    we already did this when we passed '-c exit_group' for eg.

commit a17efb4e60bdaadaa756bb72dc15564020c2afd6
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 12:26:06 2011 -0400

    as there are way more files in sys, do this after we scan proc

commit dcf96f2c6e684fd17f199087ec09bc8dbc975fbd
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 12:02:46 2011 -0400

    another param idea for later

commit 8316d51991f68b24e5fd007567384e07ec23090a
Author: Dave Jones <davej@redhat.com>
Date:   Mon Apr 18 11:00:31 2011 -0400

    improve the fd scanning.
    
    - Introduce randomness over whether an fd gets added.
      This mixes things up a little for each run.
    - Also add directories.

commit 2c3465fec955ee96447044b73b4635a3f0347fd8
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 17 14:22:24 2011 -0400

    destroy the shm on exit

commit 996ed757458dc829f71863566505feff470a6d9c
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:41:47 2011 -0400

    flags annotation for timerfd_settime

commit d36938ed99a71b9c064dca36713b30029780db88
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:40:39 2011 -0400

    annotation for timerfd_create

commit aef59e0f687c123e573c6a6236d756f5ade9b3bf
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:35:49 2011 -0400

    flags annotation for tee

commit 39812a29a24a40fdd1277c726e807f7970506d35
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:34:49 2011 -0400

    sync_file_range2 flags annotation

commit eb40df993735cff1c84522afa53d47e34710bebd
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:32:13 2011 -0400

    derp, build fix.

commit 8e9273b6638bf5be04fb7f33c2f65595b6a386f1
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:31:49 2011 -0400

    flags annotation for splice

commit ce6baf24c0c5efc65c5c7dff03dd2dfa76bfb2e7
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:29:21 2011 -0400

    flags annotation for signalfd4

commit 5a778d40947635d9016715fed00b306a9a4f4238
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:27:10 2011 -0400

    flags annotatio n for setxattr

commit 0c021cce3c5f80eb6175ce374e8d5c9cdceb6a3e
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:25:52 2011 -0400

    sendto flags annotation

commit d693c4c0fc2a4001483c1de0d471133a43ba3277
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:24:38 2011 -0400

    sendmsg flags annotation

commit 35bf54f0432d26c95245f8322edb1a3b0316ea42
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:22:29 2011 -0400

    flags annotation for remap_file_pages

commit e02e8d3f5a483821c8f11214e59da81dce349a30
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:20:20 2011 -0400

    recvmsg flags annotation

commit 2bc4e87c6cd8a57e0665beb4406c2d088e5ae560
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:19:08 2011 -0400

    flags annotation for recvmmsg

commit 8aee8872442dfc188e234bb8cdfc8fd70366a551
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:14:55 2011 -0400

    indentation

commit 0526b4dcaf196b9e01a45640cb8fb6c8f16aff81
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:11:11 2011 -0400

    flags annotate for pipe

commit a9811a4b8c0568b558d6178e3a34c6711b49d782
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:05:00 2011 -0400

    flags annotation for perf_event_open

commit 5a03f601bb6d110c92a3bc412a398bfc6c02c11b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:02:01 2011 -0400

    flags annotation for open_by_handle_at

commit 53752d86472dd767e9cd5a481834843d497d7aad
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:01:26 2011 -0400

    open flags annotation

commit 9f985c602c18828323481c14632e2553f059d9ec
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 23:00:24 2011 -0400

    openat flags annotation

commit 83867474c86d80e630f099207a43be543ec01a48
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 22:52:05 2011 -0400

    flags annotation for msync

commit 147ed33254b42a1c4ceb36c999a93ef06c6557af
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 22:50:21 2011 -0400

    annotate flags for move_pages

commit 22bf39dd9f53232c32d449a4e31cd1bc6aa56930
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 22:46:49 2011 -0400

    annotate mmap_pgoff

commit 418e220a0dafd4f9ff1833b5607e32f1a01c5713
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 22:44:07 2011 -0400

    annotate flags for mlockall

commit 1eeb6d60b1e8ab2971ccf2b20cbbf4183a149548
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 22:41:52 2011 -0400

    annotation for mbind

commit eb7cd3816742f339b825418af63ddcb2568f0cf5
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 22:33:51 2011 -0400

    annotate lsetxattr flags

commit c2bda894cad8e86bfe01be8b25b5db5c9f9be012
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 22:31:28 2011 -0400

    annotate flags for linkat

commit bf1a3fcdf5a9db55c0a29cfb6d378e8e9a8ac2e3
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 22:30:15 2011 -0400

    annotate kexec_load flags
    
    (completeness, this is a root-only syscall)

commit 92c966e424ab1610e43ecf49450718f16fe44c0a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 22:26:48 2011 -0400

    annotate inotify_init1 flags

commit 4549d25b721d02c933d4a7891a40e6f5a38ba2e6
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 22:24:21 2011 -0400

    annotate get_mempolicy flags

commit 1802d328ea9675d2d98bbe1a51f25dcb5b63d7de
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 21:49:45 2011 -0400

    annotate fsetxattr flags

commit d228539e68ec8f08b0ce9396599121fd7aad3e25
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 21:47:56 2011 -0400

    fanotify_init flags annotation

commit 38a0589e5b4dcd3e4071580b8b4eb4e05b431694
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 21:38:04 2011 -0400

    annotate eventfd2 flags

commit d836ce219617b61ee7183774f344683322a543f7
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 21:35:56 2011 -0400

    annotate epoll_create flags

commit 0991ac04a11ae3eaadf49647c509261b83ec979e
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 21:33:47 2011 -0400

    annotate dup3 flags

commit 07bc8dce960002e40eca017e387a50f1911a89e9
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 21:30:44 2011 -0400

    annotate clone.
    
    (this is an AVOID syscall, but do this for completeness)

commit fe7c50cca08bfa010236f00e8675e00770911d98
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 21:29:18 2011 -0400

    annotate clone2 flags

commit a2286e842dbbd6d13dc76cf673880a95fe90421a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 21:25:17 2011 -0400

    annotate clock_nanosleep flags

commit 48d76362bd0a473732676937be04e4ea18b14f49
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 20:35:18 2011 -0400

    add some 32bit modes to the test script

commit 59ff250ce957e37a08830b36ba633e5595796225
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 20:31:13 2011 -0400

    sanitise accept 4 flags

commit 584743dd60cdc929d72aeefef01031d4e25a15a8
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 20:30:39 2011 -0400

    increase logbuffer
    
    With all the ANSI codes etc, we easily blow past the end of the buffer.

commit ad2acae4d5588fea2b99aebb96eedaa898b83507
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 20:08:44 2011 -0400

    fix some include guards

commit a700b8e86bd67ea8874513dafeb23d8a2ea38d46
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 20:02:58 2011 -0400

    Remove the poison checking code.
    
    This was a misguided idea.
    The kernel will happily scribble on some pages, because it's expected
    that libc has setup things that way.

commit f0406296e494c4d18b49346565c913d5d3e16567
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 19:26:02 2011 -0400

    rewrite sync_file_range sanity
    
    this still always fails, but so did the old code, and that was
    a lot slower. Need to figure out which path it keeps getting -EINVAL from.

commit 6fd9ebfd60481e033456c95f34e9900e094f646e
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 19:09:30 2011 -0400

    poison debugging

commit 03cd84d59a4beefce196a41f203542c5cd1e846c
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 18:51:01 2011 -0400

    Generate a list of maps on startup
    
    Pick something random out of the linked list, instead of scanning
    for something over and over.

commit 915a9c26e8772151f83a885dad183acddbda64e1
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 17:58:41 2011 -0400

    don't stomp on the shm segment.
    
    It might be a better idea to generate a linked-list of maps on startup
    rather than keep doing this over and over.

commit b2297bb8f8cf916e090357dedbf9ed03f34418e1
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 17:32:23 2011 -0400

    improve logging output
    
    Render everything into a static string, and just writelog/printf that
    instead of having duplicate outputs for both.
    This keeps them both in sync, so that for eg, the logfile now has
    the annotated arguments.
    
    one downside: the ANSI codes now end up in the logfile.
    Unsure yet if I like that. I might strip them.

commit 69918d23cfbda689453e8a485808a998415de6d4
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 16:23:24 2011 -0400

    Parse /proc/self/maps and return a random map occasionally as ARG_ADDRESS

commit eb3b09a485f4313028d6bba94470589288d410ec
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 16:08:44 2011 -0400

    now done in the struct def

commit 3a7e53ff40435360039ccd447c24172e8787a404
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 15 14:40:43 2011 -0400

    mremap flags annotation

commit f8342b0356967c3d382ee3c52139b9acf0113158
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 14 22:07:19 2011 -0400

    nix that useless comment

commit 1e7195f318465fe16c592e3b86f4f3d2cd8d3352
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 14 22:06:56 2011 -0400

    clean up the rotate path

commit 097a4a6f6d196caf6734c8eb3b02acc5e5414110
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 14 21:42:55 2011 -0400

    dead fluff

commit 1bf3a826edd32dbcca6b7e89530ca6622f2a348b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 14 21:42:11 2011 -0400

    split main loop up

commit ec0b3e3d5a794a5638d1d99d42c12c98971f2f35
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 14 20:59:49 2011 -0400

    remove the capcheck feature.
    
    it was unfinished, and unlikely to be.

commit 48593cb86fa2f46ee603e130be27766fba5a22d1
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 14 19:59:19 2011 -0400

    another recent bug found

commit edd631c8a4d835f50be854838759e364c80af032
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 14 19:40:14 2011 -0400

    won't need this any more

commit a5ad1976f323374aa72610d6d09ccc85955245d9
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 12 21:06:28 2011 -0400

    rewrite the do_syscall path
    
    - move all the syscall avoiding into the same place
    - avoid NI_SYSCALL as well as AVOID_SYSCALL
    - don't bother calling syscalls with 0 args
    - make -c actually do the call regardless of what flags are set (AVOID etc)

commit a18f04033910825e779064c763895b2df008fc61
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 12 20:46:53 2011 -0400

    param TODO

commit ae332afa1fa2e4614aa54409f3224aeeb834f526
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 12 20:45:29 2011 -0400

    update TODO

commit a36b727bb5d48d1a1bc7cf2fe41e7ddb76985fce
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 12 20:40:36 2011 -0400

    return small values most the time in the 64bit path too

commit 9577d1bcad7140661bab1670525f4a5c6f205e57
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 12 20:24:02 2011 -0400

    clean up the max_fds usage in files.c
    
    parts of this were written so in a way that expected socket generation
    to happen after file opening.

commit 212e245ba2c1b5dfcd0180e51bee88e472504805
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 12 20:16:30 2011 -0400

    log how the file was opened.

commit 10d2029049d0cacf3936738af8f39d09ef0cd15b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 12 20:12:57 2011 -0400

    ignore the autogroup crap

commit 48a0d4d94c0edaeb26f6c2d259b4e117e1217d1e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 12 19:53:31 2011 -0400

    ignore all the cgroup junk that systemd created

commit e93628f8f0beb1251618e5ae80b57d8d477bec16
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 12 19:52:00 2011 -0400

    as O_RDONLY is 0, we can't OR it.
    Use another flag to signify that it was set.

commit 2cc44af19d2b3f086cc29affeb13a5c18de6b187
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 12 19:23:14 2011 -0400

    bias the 32bit values towards 16 bit values 70% of the time.
    
    Trying to increase the chance of a valid looking 'len'

commit 93d6350f92aa481d8f26863524b36b66c64583d2
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 12 19:06:40 2011 -0400

    Fix off by two bug in rand() usage

commit 5e569ac7cb6ee9acb57e8200b801c7249caaefac
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 12 18:16:58 2011 -0400

    the tty bug was also found using trinity

commit c8a3315f55681e484b38c29a2650455f09cee6f8
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 12 18:15:46 2011 -0400

    also add the atalk oops

commit 68d655c7fff73a860081b826656ee3317278a931
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 12 18:14:43 2011 -0400

    add another recent victim

commit 62e83a424e10a9098158dd56af8c0811358ba769
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 12 17:55:17 2011 -0400

    generate the sockets first
    
    this should solve the sporadic 'too many files' error while generating
    the sockets cachefile.

commit 9f9ea41a2c84eb89244f84773b6ec2b064f83b8c
Author: Jan Stancek <jstancek@redhat.com>
Date:   Tue Apr 12 10:38:04 2011 +0200

    syscalls: define constants if they do not exist
    
    Older installations are often missing new defines, such as
    MAP_UNINITIALIZED. This patch adds definitions, which are
    missing in RHEL5/6 i686/x86_64/ppc64/ia64.

commit 861fa5a85595266a346b64b496cffc0c3b9ed635
Author: Jan Stancek <jstancek@redhat.com>
Date:   Tue Apr 12 10:33:54 2011 +0200

    replace __64bit__ with __WORDSIZE
    
    I'm not sure, where __64bit__ comes from, but it's not present
    on RHEL6 x86_64, RHEL6 i686.
    
    Using __WORDSIZE should be as good, it has support in glibc and uClibc.

commit c7ae691124e97f5fbb19ae491d7854d23092ecff
Author: Jan Stancek <jstancek@redhat.com>
Date:   Tue Apr 12 10:23:40 2011 +0200

    i386: change PTE_FILE_MAX_BITS to 31
    
    32 overflows max range of long in ./sanitise/remap_file_pages.c:
    (1UL << PTE_FILE_MAX_BITS)

commit bf6109d43a1e251aa42f1e41e2c7b382a72393f5
Author: Jan Stancek <jstancek@redhat.com>
Date:   Mon Apr 11 14:50:58 2011 -0400

    i386 build fix

commit 409df898089112b345668b9699923da70bc026ab
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 10 20:12:10 2011 -0400

    Add ia64 syscalls.
    
    From: Jan Stancek <jstancek@redhat.com>

commit 1347d4136ff1382adf6715c806fc4b2cbb4d0195
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 10 20:11:08 2011 -0400

    ia64 build fixes from Jan Stancek <jstancek@redhat.com>
    
    (I just killed the macro, now that we have ni_syscall.h, use that instead)

commit e6ff863fda84adcdabdcf9cf1666af6c7793765e
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 10 20:10:58 2011 -0400

    Avoid pselect6

commit 99cd4ecc33172358150a37d77169f5ddd417a23f
Author: Dave Jones <davej@redhat.com>
Date:   Sun Apr 10 20:06:09 2011 -0400

    ppc build fixes from Jan Stancek <jstancek@redhat.com>
    
    (I modified some of this, but the bulk of it was done by Jan).

commit d891a89fe835d676a2f4020362c06d7d64dc0589
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 31 18:02:16 2011 -0400

    remove double increment

commit 7dcfafd868e2f62aeb7bfe9f5c2700ae8ed0d513
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 31 18:00:12 2011 -0400

    return trash in the random page sometimes

commit c19eeedf9e1af669f124ba76024f61cbccb13f1b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 31 17:12:27 2011 -0400

    improve the random page generator
    
    instead of just random crap, make it look kinda like a struct.

commit 4d1d47bdbb9093aa267a3622fc5db72fcae88077
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 31 17:12:10 2011 -0400

    arg2 of clock_adjtime is a range of clocks.

commit f5d5612abc851cf8cd3a70b53a4b7ecf6f06329e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 31 15:49:00 2011 -0400

    clarify how the rotate mode works

commit c228ce6935c6cef45dd5f2b68db702d97dad6202
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 31 15:48:02 2011 -0400

    fix off by one

commit 3b33bfb0f2e7539efe447a51c53c35f034613a4a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 31 15:46:02 2011 -0400

    Don't munge the nr_args in --rotate
    
    We're better off not doing this.
    Because we don't do 6-arg syscalls yet on 32bit, we were always
    doing no actual work when testing.

commit 0a2d697d1edd72c6f4216a83953da0faa182f529
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 31 13:55:20 2011 -0400

    change the socket creation to be more random.
    
    also bump up the number of fd's created.

commit 2f36b917331cc617ca649750f18fd1ecdb3063d5
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 31 13:13:37 2011 -0400

    make -c1 work again
    
    the work to make it parse syscall names broke the case where
    you just pass a syscall number.

commit 582ef025fc73b7ea933519543d9e70c25550cd3a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 31 11:19:56 2011 -0400

    fanotify.h isn't around on older distros
    
    so instead of including it, just #define the bits we care about.

commit f71059ed6ade821b93c4bf28b47f51be0a06202d
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 28 23:24:37 2011 -0400

    new syscalls for i386

commit 3bbadfe2eb2279bb2ac407bd3f2b007e3cfaa0ae
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 28 23:23:59 2011 -0400

    don't block on files

commit c69793f4d107cf95891282a527f058821c039e78
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 28 22:53:40 2011 -0400

    ignore tmp

commit 93af6d4f80f45474d7d081c1ec5b9da182bccecf
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 28 22:38:52 2011 -0400

    Add a missing syscall.
    Fix off by one in array size.

commit ad22539ca1f96a1e703909c16b5ba2a21ad4e06e
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 28 22:19:54 2011 -0400

    Update x86-64 for latest syscalls

commit 15515f1ee0c89cceda17dfad8b615843aabc129b
Author: Jan Stancek <jstancek@redhat.com>
Date:   Fri Mar 25 08:17:09 2011 -0400

    fix --logfile parameter being ignored
    
    Logfile was created before arguments are parsed, so anything passed
    to -l --logfile was always ignored. Arg parsing moved couple of lines up.

commit 0e2547fefcdc3abdc65b66dcf5b3959ef20d4689
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 22 13:14:56 2011 -0400

    mechanical conversion of sparc syscall table

commit 4891fd5f0888289708b61f895fb4e8fafdd1aedc
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 22 13:09:50 2011 -0400

    mechanical conversion of ppc syscall table

commit d10be7326f9bff983197290dbffa1a7a31d6f4c8
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 22 12:15:04 2011 -0400

    syscall commentary

commit 6d5c052b830944091f622de967fc429faf0e778a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 22 12:12:48 2011 -0400

    don't include these yet

commit 28b54fb7e8138a3571fe9221f19acdf77f038602
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 22 12:03:28 2011 -0400

    Fix build on non-x86

commit cfeb2bd2e7697a137b72089ef57ccd9ea20f112d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 22 12:02:34 2011 -0400

    document a bug that needs fixing later

commit a73b157b898580607df6eb36efe6c956a5631cd4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 22 11:52:04 2011 -0400

    add PPC constants.
    
    From: Jan Stancek <jstancek@redhat.com>

commit 8a9b19df1c4e9449196cb67524b6d3c63dba58af
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 22 11:50:39 2011 -0400

    SG_IO fuzzer. Work in progress.

commit 1baa520c48147adcf495c016da909fba05ee14f6
Author: Dave Jones <davej@redhat.com>
Date:   Fri Mar 18 00:19:53 2011 -0400

    mostly mechanical conversion of ia64 syscalls
    
    this needs checking/updating against the latest kernel.
    This should at least get it closer to compiling than it was before.

commit 72bef5e127463fd1393033961e5384c2d7b17dd7
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 15 20:53:58 2011 -0400

    open() TODO

commit 8ea025ee82a364a3dfa2e716428b2165fc02e6f5
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jan 29 18:47:31 2011 -0500

    make --list with in 32bit mode

commit 5695010ff03fb194d2c7d72faa9dc0b760b3d7e4
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jan 29 18:44:53 2011 -0500

    fix off by one

commit 1566fe31418835692ec826af62987fa5bc48bd33
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jan 29 18:41:33 2011 -0500

    32bit check

commit 98c12bc14e2c26bc302a639f8908b6ff33596863
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jan 29 18:38:20 2011 -0500

    add a primitive to sync the logfile
    
    and call it after we've been writing unsynced

commit ab321c084035b48aa2949163c720bfe87f21f4bf
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jan 29 18:24:15 2011 -0500

    make -c work when --32bit is specified.

commit 949f270a3af76a16410b70c150be9126d5458c45
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jan 29 18:04:03 2011 -0500

    document --32bit in help

commit 66eed49577df532daaa1ee09949c438c9650a961
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jan 10 20:24:49 2011 -0500

    fix logging.
    
    because we were closing/reopening the file with every logging attempt,
    we ended up writing half of it to one dir, and half to the other.
    
    Things will be a lot simpler if we just leave the file open.
    The fflush/fsync should ensure that things still hit the disk.

commit f9b854e6493541008ea345b01ac8acbc94cd62b4
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jan 10 20:11:50 2011 -0500

    beginnings of ioctl sanitiser.
    
    For now, we're focussing on the sg ioctls until things are fleshed out.
    (as such only ioctls/scsi-generic.h is included right now)
    
    (in order) plan of attack:
    - Flesh out sanitising for scsi-generic
    - Add more sanitisers for the rest of the ioctls currently in ioctls/
    - split up pending-ioctls into more per-ioctl files in ioctls/
      and then eventually be removed.

commit b9c6896e89426ff27fbf9a6a176bdc111a5d2b41
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 21:06:23 2010 -0500

    update

commit f1998d0161df9898b6cfbf12eca6f3e6b15a9b37
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 20:50:00 2010 -0500

    Don't let -c call AVOID_SYSCALL annotated syscalls

commit c4e0f16e97b14f02e37b9815a491199e2e7a5d73
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 20:30:36 2010 -0500

    Rename project to 'Trinity'.
    
    scrashme was too generic.

commit be332144c7453821252f776227f889bc0c65ddf4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 19:25:23 2010 -0500

    split the code that does the syscalls out to a separate file.
    
    This was getting complicated enough that it deserves its own place.
    
    (also clean up some variable types and names)

commit 0cfbe4410ed1d314bad7c85d28877d2bebe28df4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 18:33:12 2010 -0500

    some safety for when we run outside the harness.
    
    also put the logfile in the source dir, not the junk dir

commit fefcf664f9d84c7ac4075508e58256cf9f8900ea
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 18:29:41 2010 -0500

    test script improvements.
    
    - cpu binding
    - separate logfile per instance.

commit 05f57eb622a165bba3987aec2697876451789552
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 18:03:50 2010 -0500

    Add -L/--logfile to set alternative log file.

commit d4a57322b23b6e2dfdfb11b0cc642a82a10d4739
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 16:21:09 2010 -0500

    macroise

commit 7baa72874103f6055bb38c086d9cb1360c550ee4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 16:07:22 2010 -0500

    fix up the -z output.
    
    this code is attrocious. macros or functions for this is
    probably the way forward

commit d5e1374ada13d6a4da509176d672fa190bbc5396
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 15:51:07 2010 -0500

    Do some -z -i passes in the test script

commit edc64ec9125ed84373c0c9640e75d80daabc70c2
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 15:45:35 2010 -0500

    rt_sigaction improvements
    
    act/oact can be null sometimes.

commit af482361176dae0f9ec4e0dd84477f0d7af69fea
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 15:29:48 2010 -0500

    fanotify_init does a capable(CAP_SYS_ADMIN) check first thing

commit 2214757868b04c896421e4e997da23f000e6d607
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 15:25:30 2010 -0500

    annotate fanotify_mark mask field

commit c5540d8b8d7ca4e3cd6ff2ce377266c4dad63f99
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 15:21:33 2010 -0500

    annotate fanotify flags

commit 338c82d1cbca64091bf11667cd5df652f5c3b5ad
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 15:19:19 2010 -0500

    fix number of flags.

commit d5975c24084af9e00b8767b528fad9b409771b51
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 15:15:08 2010 -0500

    list annotation for splice flags

commit 81122f9d83a1ce503e19396c8817cc9d59d5376a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 15:09:48 2010 -0500

    use header

commit 7664e7e60e89259271d7391f750d73d5d6be9f68
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 15:06:06 2010 -0500

    another braino, this time in mremap sanitiser

commit 138550a2d11f33bde13f94bce8e3243b7577fe27
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 15:03:50 2010 -0500

    cut & paste braino in mmap sanitiser

commit f44f51c210331ba5eb17cb5213adcc297879a0eb
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 14:34:07 2010 -0500

    rename function

commit 2365800dd70eed6cefd4d018890e734fa9fdddc1
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 14:30:22 2010 -0500

    misc cleanup & small fixes
    
    move sanity check out to separate function
    don't reseed the randomness every syscall
    only seed with 'seed' if we passed -s
    log the randomness that was used.

commit 7c80a7d62bd735d697ed6aaa784cc53e226cdea5
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 14:21:02 2010 -0500

    remove more of the uid munging stuff.
    
    remnants of an older time when we cared about being run as root.

commit 2fb2d555b5841e3f99f5124fe89b72fd332a0730
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 14 13:34:14 2010 -0500

    Fail with an error when run as root.
    
    Also remove the pointless chroot call that will always fail when
    run as non-root.

commit 731c887db1e5d0f252b0c80941184ab3907667e1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 8 16:23:36 2010 -0500

    manufacture a 'struct' to pass to ioctl
    
    still just -EINVALs nearly always. Needs a rethink.

commit 34d46b2e6a597700f6ad73dbe10d82e0d4084459
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 8 15:36:16 2010 -0500

    Pass realistic cmd's to ioctl.
    
    Incomplete, but it's something to begin with.

commit 26c51c64e41d1f8857f9a5594db4343c57d27315
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 22:51:57 2010 -0500

    remove debug stuff

commit 0fbb1374e8dfa4538e1a1834a644c82bccfed1f3
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 22:50:14 2010 -0500

    cache the successful socket opens.

commit d6aeae244653a07e06552155b9ad7ee95664884a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 21:05:02 2010 -0500

    micro-optimise socket creation

commit a4fe906ce8a5e44b9be1ad821cca3ceec3396606
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 20:50:43 2010 -0500

    rewrite the socket generator.
    
    Doing this with rand() for all arguments sucked for a few reasons
    - lots of duplicates in fd's lowered coverage
    - it took forever when run with slab debugging on.
      (It still does, but this is at least deterministic now)
    
    Go back to what we did originally, going through every permutation
    and just storing the ones that work.
    
    print some output so we know what's happening.

commit 779760aafd4c11233564d1cab3182963d0f8c75f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 19:42:45 2010 -0500

    build fix

commit 5649c375ca39c8a04a71f94cab7ea618d78f7fea
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 19:33:52 2010 -0500

    return the random struct instead of mangled addresses in ioctl
    
    probably has more chance of success.

commit 1687bddcbc7d41fc545c8d85284451a7fae2863b
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 19:30:42 2010 -0500

    return a pointer to the random page sometimes

commit 479e14cb8e949cca81b635468d9a1a432e0f93fe
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 19:26:47 2010 -0500

    check poison for new buffers

commit e047666bac4d632bc0d8af2db7a59c3d9e791081
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 19:25:16 2010 -0500

    generate a random page every syscall

commit 34ac94760e7d378854357666066a0fb641a7e9a1
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 19:20:19 2010 -0500

    slightly improve ioctl. needs lots more work.

commit 44cbcf5a781b17bb406f7e87a3e770590250bf18
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 18:20:18 2010 -0500

    use libc defined signal limit

commit 64e3321f25adc3cc426f4284ba90e3d173e63c30
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 18:13:44 2010 -0500

    annotate sig argument for rt_sigaction

commit 385a2ebbfcfcbbec71fb5a8c907e1e2b2ad3ae08
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 18:03:33 2010 -0500

    ARG_LIST was only returning one bit. Or a few of them.

commit 9868ac5073081d431c2fd0775a5210d0cc40b784
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 18:00:06 2010 -0500

    annotate mprotect.
    
    remove prot randomiser, as it's now done by ARG_LIST

commit f9314061f45074e4503b743ab9edad77afffc82e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 17:56:01 2010 -0500

    finish annotating mmap

commit 79a1a599bd3a95cf59713b0064d3f34c4d336e43
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 17:41:40 2010 -0500

    add some page_size related numbers to 32 bit interesting numbers.

commit cfc8b4bf79afba23a2ea6daf855dd0903c9c8580
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 17:38:53 2010 -0500

    sometimes return a length that is < page sized

commit 31c021f8e0bc72dfbf9112aa0e1e2924a962e1ad
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 17:28:42 2010 -0500

    add dependant changeset

commit 374c8829a7248ec8231177402efbb43c840373a3
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 17:27:09 2010 -0500

    separate the generic sanitise routines from the per syscall functions.

commit d9a5903c20c9840787b3aa4f26d1360765aa1a39
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 16:58:55 2010 -0500

    sanitise mmap args

commit 4c30f8c885232fe8d7a1c62f15b896759f7db368
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 16:48:59 2010 -0500

    add CVE references

commit 4e9edcc920c5c7cebbc63d358ece008fef042ec0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 16:34:37 2010 -0500

    check all 3 poisoned pages. refactor.

commit 2e194b389947deda243aeedcb161dc089fce559e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 16:24:09 2010 -0500

    page size fixes

commit 7e84c020021a9d49d6957dfa1308fffd76799391
Author: Dave Jones <davej@redhat.com>
Date:   Tue Dec 7 16:19:51 2010 -0500

    improve userspace struct passing in get_address

commit 952131b53e1dc6704c73cfb3bd014e1ef5b5333a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 30 00:52:25 2010 -0500

    another one down

commit f1c2e1513ee88709b7b2c582d22667504d47ce48
Author: Dave Jones <davej@redhat.com>
Date:   Sat Nov 27 17:20:57 2010 -0500

    ignore

commit e1bce40652ca5791769854fa33be3d5da948e7b4
Author: Dave Jones <davej@redhat.com>
Date:   Sat Nov 27 17:19:46 2010 -0500

    clean todo

commit e559c7197505d1327cc9bfdf2b7c3e4288f4126a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 25 20:56:46 2010 -0500

    keep track of bugs discovered.

commit e2bb48ccc7c5ff21fdaa657e59a4335c63957adb
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 24 21:18:48 2010 -0500

    log the socket fd's as an fd count, not an array count.

commit a5b0830e8cc64e7f48a449edf3822eb9bb87038c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 24 21:04:36 2010 -0500

    move argtype gathering into fillargs
    
    reduce prototype some more.

commit 2a4c3796bc287bcf7e43ea170b4b9c877bb56726
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 24 17:53:35 2010 -0500

    make sure the log file hits the disk

commit f653e8f7937f0ef7a17b951274d9c38ad5b21925
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 24 17:39:11 2010 -0500

    remove some debug code that crept in.

commit 6138b5ecf226616253a7ca5526bb0a868240e6bd
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 24 17:38:32 2010 -0500

    Log the fd creation.

commit 17a6637df32a04b72b2a48fcfa909d8e7b33f1a1
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 24 17:07:38 2010 -0500

    hook up ARG_LIST support for fill_arg

commit 8036ab061f2607d31889602741ee6bfcc451ab5c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 24 16:58:01 2010 -0500

    move some of generic_sanitise into fill_arg
    
    this changes the function args to something that will be more useful
    for other argument types.

commit f4f6227ff58ddcff444ef7fce785f5a447a78d19
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 24 16:45:35 2010 -0500

    beginning of ARG_LIST support

commit fe18b6b75588b4d762067de21d8f04661ea3ad40
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 24 16:34:02 2010 -0500

    some includes use __unused as struct members, so define our local typedef as something else.

commit d1c502765aee1e488ca07c8274eabe7b80c14538
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 24 15:43:53 2010 -0500

    ARG_RANGE needs to be per-argument

commit bfd360ff6fd3a36723b9e56d5a2701f522b1bee4
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 24 00:06:30 2010 -0500

    add back an unsigned that got lost along the way.

commit 6123fa32c066de5a7ffb0fe220b1aec9b49f6dde
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 23 16:44:50 2010 -0500

    rough start to sanitiser for remap_file_pages

commit b13a96a959ba9666c11ff2cb520829fd6c5ab45a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 23 16:22:48 2010 -0500

    fd is provided generically.

commit 69bc4ac00e83a56e804984ca9262a60a3a49fe33
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 11 18:35:08 2010 -0500

    Reduce some of the extraneous output

commit cdd648f94b226788da8d5841dfa18cb4768414e6
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 11 18:24:25 2010 -0500

    dup takes an fd

commit 9032b0c5f0bacff985752ac9c685e6435383c9ad
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 11 18:22:34 2010 -0500

    A bunch of dumb bugs found with splint
    
    including a missing syscall on x86-64. Duh.

commit a10901a287269ff921eb677a0a81fe866d757e9e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 11 17:58:28 2010 -0500

    give a 50/50 chance whether the fd is a file or a socket.

commit bab2db11885065ebc41813902ac33f4394071261
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 11 17:44:23 2010 -0500

    nanosleep is boring, and just slows things down.

commit cbec1e5852b7682114d773ff70d0c3a8b6978fb7
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 11 17:34:31 2010 -0500

    limit random pids to pid_max

commit e8bfdff6d408f65b260ee60f98cc22d6d81d220c
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 11 17:31:26 2010 -0500

    Remove some sanity stuff that is now done (better) generically.

commit bc136f4f40d5b97504aa906b1d64d4f6f87442d7
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 11 17:28:49 2010 -0500

    ioctl sanitiser
    
    needs more work.

commit af3bd79b937627a4267176dbda470ba7cdfc30d3
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 11 17:00:08 2010 -0500

    Add back the pipe creation.
    Bias the number of sockets created.

commit 1024481da706e5bbda729a4252b724419b1e9643
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 11 16:35:26 2010 -0500

    Improve the random fd code.
    
    files code largely stolen from an ioctl fuzzer by Ilja van Sprundel.

commit 62817cf4d8a136cb617758a08c64a9e866659061
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 11 14:30:51 2010 -0500

    fix warning

commit a10cf60cfa37315489b01dbcafe4308f7f676172
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 11 14:30:16 2010 -0500

    add in some x86-64 addresses

commit 242d5172813a19e3fde569e15d4bbf2b0d2d66cb
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 11 14:15:49 2010 -0500

    Switch up the interesting value randomizer

commit b7babaa3f98de2b9ed8a32225975b82d1fecff74
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 11 01:24:32 2010 -0500

    Some more random fixes.
    
    Use rand64 where necessary. Elsewhere, just plain rand()
    Fix up some masking, and remove some redundant looping.

commit e7d3886e784b9af4d049c29111fb024cf8347b69
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 11 00:41:49 2010 -0500

    Count retries from --bruteforce in the summary

commit bac38d07d4964129bb83600b132ddf539b4b4f9b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 11 00:35:18 2010 -0500

    Revert ec05f803a334691af6daaed0c541a5e130dc15b7
    
    We want 64bit randomness for the registers, which is bigger than RAND_MAX

commit 834095302789b18ece3b2b7a51fadaac80fa6ba5
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 10 23:37:19 2010 -0500

    dumb random thinkos picked up by 'Ross' and others on lwn.net

commit 6eb9498d8e17e6d918bdd9eede9a34b9f4423ecf
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 10 22:38:51 2010 -0500

    More -c fixing

commit 793a55c130b0f9deff34ceb7c9e493221dfef247
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 10 22:27:45 2010 -0500

    repair -c

commit f966f46e92b8b38a3b086be645e33b8f314fdb26
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 10 22:12:52 2010 -0500

    oops, accidentally had the 'do the 32bit call' code commented out.

commit a91b6ea3ef40a0ccfa9f8d8ba9e4b71e9e194413
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 10 22:06:51 2010 -0500

    add syscall table for i386

commit 4681d213b46a94f00130d56b891bf99fee4db604
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 10 19:48:15 2010 -0500

    Split out the syscall structure into one syscall per header file.
    
    This will allow them to be #include'd in the correct order for the
    other architectures, without having to constantly keep all the other
    syscalls structs up to date.

commit 0266d8137084be870d8ba298c75deee8d997e833
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 10 16:47:16 2010 -0500

    First attempt at making 32bit syscalls work in 64bit.
    
    Danger: ugly hacks.
    
    Doesn't actually work properly yet either.
    - 6 arg 32bit syscalls missing.
    - needs structure updates

commit 6e322c8e333f3903fb2b7dd565fc62cbab361d12
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 10 15:54:39 2010 -0500

    remove references to dead functions

commit b52dfd68b60e5cf8b2d08708b21ecd2d16406cbe
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 8 17:59:40 2010 -0500

    document -F

commit f2dbd35760e83d0cf70af98ffbb18351a4f1c150
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 4 22:07:17 2010 -0400

    basic logging support

commit 7b6ee9f568e6d77ce926636a20b8e80283ef55ef
Author: Dave Jones <davej@redhat.com>
Date:   Thu Nov 4 21:18:21 2010 -0400

    syscall()'s retval is an int

commit 45e2bfb844f3a8ee1c283d92d3beaa6d7d4d205a
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 3 15:42:13 2010 -0400

    add a --nofork mode.

commit 40f13683edb968e8886a848bab022428de91d7dc
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 3 14:12:16 2010 -0400

    more todo

commit 6e0de874048a25e7741212e21ea3157be3563dde
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 3 13:59:26 2010 -0400

    add ARG_RANGE for a random range.

commit a5cdbae68790bbbc06c662f6017722c72ae9b2f2
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 3 13:44:03 2010 -0400

    pass pids to args that want them.

commit 194ac96e0742b9bb84ea06f6a1520d5433128531
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 3 11:28:44 2010 -0400

    Remove some done things, add a new thing

commit 6a0eb9090d31ff04d68636eb5a6b3f918ba060a6
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 3 11:13:29 2010 -0400

    missing proto

commit f1d13e22b8a7a79dae60c824e5a6bad21f2b791a
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 3 11:12:55 2010 -0400

    Revert "revert this for now"
    
    This reverts commit b005219e2d61d18b3998e79ed0405a57692cb755.

commit dec7c483176ef3fb7342d4f057b84f5673f2da12
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 3 11:04:41 2010 -0400

    fix the prot argument of mprotect if GROWSUP and GROWSDOWN are both set.

commit b005219e2d61d18b3998e79ed0405a57692cb755
Author: Dave Jones <davej@redhat.com>
Date:   Wed Nov 3 10:59:50 2010 -0400

    revert this for now

commit c2a577588eebc9aeadbf9f3744f2f009d65f2b56
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 2 15:56:50 2010 -0400

    increase the random range so we use all the switch cases.

commit 5a4bbeaeab1a0b160e4e451a75bdafad68e04431
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 2 15:52:57 2010 -0400

    fanotify_mark wants a 32bit mask

commit 673a610e235a8df7c112043e42c0d13792cfd7f8
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 2 15:23:44 2010 -0400

    Add all syscall argument names

commit e9b1b483541d5edf3b88dee6672846d307aaa9c2
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 2 14:21:28 2010 -0400

    fix the printout in the non -i case

commit 301fbdcbac51c03d6bd8bdfbe499faaeba830023
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 2 14:15:30 2010 -0400

    Some ARG todo's

commit 15b1087c7235cb965836dfbdfccf17745593c10e
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 2 11:56:14 2010 -0400

    merge the two post sanitised register printing routines.

commit 9449adf78dd99f1f665c5aa2321359d0d0bb0aa4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 2 11:49:49 2010 -0400

    don't print out unused registers in post sanitisation

commit d30d406a42f4377ee1f3eada793739a32a2f6eab
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 2 11:44:52 2010 -0400

    whitespace

commit dd623b8a276ed3e7ce2c6e85975d2eb561997169
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 2 11:43:10 2010 -0400

    beginning of argument name printing

commit cee1dc946beff95811f05d11f15e7af2dc32dbb3
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 2 11:25:25 2010 -0400

    more whitespace

commit 3962a001941a130d43cb5c68c5af81970f004858
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 2 11:21:55 2010 -0400

    things are getting cluttered in some modes, add a newline after each syscall

commit ec05f803a334691af6daaed0c541a5e130dc15b7
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 2 11:17:57 2010 -0400

    remove open-coded random long implementation

commit 46ff264ca047da96987f5ed64287f1f649596296
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 2 11:14:50 2010 -0400

    cleanup the mprotect sanitisation
    
    - it kept taking so long that the alarm would go off, and we
      wouldn't actually make the call.
    - fix up some rand() calls that actually need a long of random data.

commit 09243f0de29c9fba92ff06179fb72dca74a6a6bc
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 2 10:35:20 2010 -0400

    duh, we don't need to do it twice.

commit 7d9076a50e16f2410bb4dd3c5ca5a7afd6d40194
Author: Dave Jones <davej@redhat.com>
Date:   Tue Nov 2 10:32:40 2010 -0400

    reseed the random generator in the child
    
    otherwise every child gets the same sequence of random ints.

commit 05e36325741f47fd9def1973c6f793ab60280164
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 1 17:13:27 2010 -0400

    use defines for color codes

commit 7b00ef7beace11aed094f7c7516ef6ff3c2e6569
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 1 17:06:51 2010 -0400

    colorise the sanitized argument

commit f71a2095a59ef237c36538723fd4c133da95fbaa
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 1 16:56:14 2010 -0400

    Add ARG_ADDRESS
    
    This required a bunch of assorted cleanups, including some header
    file reorganisation.

commit e7c1cdfae77754a1cfa65787c8d70490f31d1333
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 1 15:04:40 2010 -0400

    Remove the length setting (now set generically)

commit 7f203bcd446a77e402a3f991c34e27b72a0241b3
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 1 14:59:19 2010 -0400

    always print out the pre/post sanitised options

commit 49d0662e12f12d276007b8ec0fcb0fe77b8fb4a6
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 1 14:04:48 2010 -0400

    missing comma

commit 42f54e58f6ed45b028cef407fe69aa2193e71f5a
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 1 13:41:23 2010 -0400

    Initial ARG_LEN support for passing interesting length arguments when -i is used.

commit d964f2ca7918e4996160b083059bf7e5bf5eb722
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 11 17:22:55 2010 -0400

    don't bruteforce ni_syscall

commit 1541bf6ea6ffff4e770afcdf708887ddce964205
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 11 17:20:01 2010 -0400

    ignore window changed signals

commit 786ad606426bb03c04c2f32a854b5d2ccd213f65
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 11 15:18:32 2010 -0400

    Add --bruteforce.
    Move on to another syscall if we fail 100 times. (Which we likely
    will until -i gets fleshed out some more)

commit a43419b0a45e47037cae7221fabd3ecfaf538347
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 11 14:56:02 2010 -0400

    exit child with the return code from the syscall

commit 337d2d4591eb3303c5a89211d1a4bc968ffc5c1d
Author: Dave Jones <davej@redhat.com>
Date:   Mon Oct 11 14:50:42 2010 -0400

    remove another done item, add another new one

commit 02b5b3ba658ffe087ff7e365e21fbcc8f3896e1f
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 1 15:54:42 2010 -0400

    All this is now done in the generic code.

commit d14594207ac9bfb8b409fc2d7dbbeed3497c3dc2
Author: Dave Jones <davej@redhat.com>
Date:   Fri Oct 1 15:31:39 2010 -0400

    Step 1 of generic sanitisation

commit edabfab235fdee5862264aef1641d2cb56948dee
Author: Dave Jones <davej@redhat.com>
Date:   Wed Sep 29 00:26:29 2010 -0400

    Print a summary of how many syscalls actually succeeded

commit d11ef1f01fd43ce39b81cc75b8f18c8cb45f2bc8
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 28 23:53:25 2010 -0400

    Don't bother running more than the number of CPUs that we have in parallel.
    
    We still run twice as many processes as possible, due to the two
    invocations per loop pass, but for now this is fine.

commit 7b9f3daa0f2651ce7ec799d34a1c973064309a5c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 28 23:51:28 2010 -0400

    Rename test-multi.sh
    
    It's no longer just for testing multiple in parallel

commit 941c9c4efba66ea91e32bd0e857e22dae48f9bc0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 28 23:50:46 2010 -0400

    Move the bounded runs outside the loop

commit 1e20b528ec10a075af0c0178d1b5498ae60b6d05
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 28 23:01:51 2010 -0400

    The beginnings of argument annotation.

commit 28d944b3bfcd38f546608db5af7e8837e327d7b6
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 28 22:48:26 2010 -0400

    add some comments to delineate the syscall list a little more obviously.

commit d539f9cd8b9ede89a394d838bbac500ce8e0fe59
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 28 22:24:50 2010 -0400

    colorise the result

commit 6772380861bf49499dc85aaa4b999fe99d633bb3
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 28 22:09:20 2010 -0400

    avoid vhangup.
    
    it has no inputs, so there's nothing we can fuzz, and it just confuses shit.

commit 9b6dff12993d5e75842fed6f5605c6a21c0d8f32
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 28 20:40:04 2010 -0400

    Add function prototypes of system calls as comments.

commit fa95f1704c4e7522bee2605833523e9f104fb06a
Author: Dave Jones <davej@redhat.com>
Date:   Tue Sep 28 20:01:52 2010 -0400

    comma separated syscalls idea.

commit 761c42fd5cf6179cb78f58968f4d49caf3e09fdd
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 9 22:21:32 2010 -0400

    Check during startup that all ni_syscall's actually return ENOSYS

commit 120299dbcbca129c377ed001e64a118777a2298c
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 9 21:55:07 2010 -0400

    an 'unbreak everything' todo.

commit 491e7ac30ecdc8f87f9fd76c196b25c08711e313
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 9 21:50:59 2010 -0400

    structure layout cleanup.
    
    Doing this for every architecture is going to be a pain in the ass.

commit 2381367623d39d9b03c55ff5843428e07ac3cc71
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 9 21:50:33 2010 -0400

    More done.

commit 6118b40140d757c06d8039d7c81b0554743f68bc
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 9 17:49:06 2010 -0400

    done.

commit 9d270f034be4a19713b2ba39a66751911f7d9aae
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 9 17:44:11 2010 -0400

    colorise the rotating value

commit de293ada59b9ea03cb1fd9027e3048e02155e126
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 9 17:09:47 2010 -0400

    Add a field to enumerate the number of arguments a syscall takes.
    (currently x86-64 only)

commit e2a3114f72bdd18cdf7c5784f06ee1fdf08f020a
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 9 15:57:51 2010 -0400

    test ni_syscalls on startup

commit 2f4899ccf3491c296413bc702ea76350486c12c2
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 9 15:29:38 2010 -0400

    Groups idea.

commit 4ab93473674ca564a405830a0b35f8b42fd51ef0
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 9 14:35:22 2010 -0400

    -c0 was broken.

commit 774c6dab30268e3c4e2aa3357bdb8420cf1b732b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 9 14:07:27 2010 -0400

    Check poison /after/ doing the syscall, duh.

commit 598800292b8715c3cfe32602f3ec6706079cd0de
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 18 20:10:25 2010 -0400

    more intelligence ideas.

commit 0e3059d089e365f9601d8e3c37be6238b572d10b
Author: Dave Jones <davej@redhat.com>
Date:   Wed Aug 18 19:57:05 2010 -0400

    a todo about OOMs.

commit 874bb3c3ad56cc8baaf748f1c705faad1472fd77
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 17 21:23:30 2010 -0400

    Update syscalls list for x86[64]

commit f169c7eb473d3fc7296657176d7825d7433b8dc6
Author: Dave Jones <davej@redhat.com>
Date:   Fri Sep 11 14:48:29 2009 -0400

    Background the other two tests too

commit ab11285134e9a7685c5b2b5710a903d2a5439433
Author: Dave Jones <davej@redhat.com>
Date:   Fri Sep 11 14:41:56 2009 -0400

    Fix parameters in test script

commit 48134d81587128b0cfcfe41436a40e5781948243
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 18 03:03:38 2009 -0400

    Fix some off by one errors

commit 7bcd7f53f390f50036276ae0ee8a323f2879b60f
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 18 03:01:09 2009 -0400

    Add --list for debugging purposes.

commit 457c91ea59d0cfee0d33daf3d0b01e812ef84611
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 18 02:50:52 2009 -0400

    Enhance poisoning.
    - Dump buffer when detecting corruption.
    - Determine if it's the pre or post buffer.
    
    Mess with alignment on -u

commit b6637a6f1aacbaabe0b2ad2c0ada759402dbd15d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 18 02:20:37 2009 -0400

    Add -u for passing userspace buffers as an address type.

commit 8d7bce56c16ac846376571b4f9bd2b85ba5beff1
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 18 01:47:07 2009 -0400

    Clean up option 'mode' setting.
    
    As a result, introduce a --mode parameter.
    MODE_STRUCT and MODE_REGVAL are killed off. They were never modes,
    they were types. Express them as such.
    This will now allow us to do new things, like rotating structs
    through regs.

commit 9a608d4858f1830d72810c9444797eab1e8b70a4
Author: Dave Jones <davej@redhat.com>
Date:   Mon Aug 17 22:58:06 2009 -0400

    Rename ZEROREGS to ROTATE

commit a64bee4dc53fa5fbac063a182ac0654a55454a59
Author: Dave Jones <davej@redhat.com>
Date:   Mon Aug 17 22:17:49 2009 -0400

    More TODO cleanups.
    
    Remove some unnecessary bits, expand on other bits.
    Reorder into priority

commit ef9e7e49bfd3350e91a792d6e73893d16b418324
Author: Dave Jones <davej@redhat.com>
Date:   Mon Aug 17 21:36:55 2009 -0400

    update TODO

commit 7b62341636e9e15841aed558e11f2dfab381f00b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jul 10 14:27:52 2009 -0400

    New syscalls as of 2.6.31rc2

commit cd86b71311532bd2e1fbf4ff63b634018dcabb99
Author: Dave Jones <davej@redhat.com>
Date:   Sun May 18 19:33:04 2008 -0400

    Update syscall lists to 2.6.26rc3

commit 1cb63e25cb3bc7ed90327f7159a3e26a77fb602e
Author: Dave Jones <davej@redhat.com>
Date:   Sun May 18 19:10:06 2008 -0400

    Don't test the pause syscall, it has no inputs, so it's pointless.

commit 98e44cb38214d2029b7ec34499c812f288dcc87f
Author: Dave Jones <davej@redhat.com>
Date:   Sun May 18 19:08:57 2008 -0400

    Make poisoning optional

commit 4bd763ed52b3a9245242e305f59484da033bab7e
Author: Dave Jones <davej@redhat.com>
Date:   Sun May 18 19:05:45 2008 -0400

    assorted sanitise_mremap improvements.

commit db365f2022bfcef1ea9d653e06ba7578ae6d20e0
Author: Dave Jones <davej@redhat.com>
Date:   Sun May 18 18:52:58 2008 -0400

    Set the alarm once. Also increase it slightly.

commit b4f0b5ad361cb45985761053a65741db50c68e1a
Author: Dave Jones <davej@redhat.com>
Date:   Sun May 18 15:58:04 2008 -0400

    pass useraddr to sanitise_sendto

commit 8d7cd0ed2632cf39870060d0e46c9b5d52a40b6a
Author: Dave Jones <davej@redhat.com>
Date:   Sat May 17 18:37:28 2008 -0400

    beginnings of userspace overrun checking.

commit 33471d6b92b1a62bbd5ee6c4b90fece0ae7c6b16
Author: Dave Jones <davej@redhat.com>
Date:   Sat May 17 18:28:18 2008 -0400

    Beginnings of some sendto fuzzing.

commit 96f9142ad24c479c0f97056fc1657ede517616ca
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 10 12:41:25 2008 -0400

    check for reg clobbering

commit 12c914b51f499876bc5814bf5d207506f5ed0686
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 27 20:48:56 2008 -0400

    Remove unnecessary (and bogus) include

commit acfd88f5ab184a57531da0beed1d414bb32b2284
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 11 21:54:28 2008 -0500

    -i and len TODO

commit b59f2dee197ed4e4dbb1fd85095fbb0b7e562f04
Author: Dave Jones <davej@redhat.com>
Date:   Mon Feb 11 21:50:50 2008 -0500

    --trashkernel TODO

commit 183702852f381aabf40cdc7800e5fbcbde717615
Author: Dave Jones <davej@redhat.com>
Date:   Sun Feb 10 22:41:57 2008 -0500

    sanitise_vmsplice improvements.
    - pass a pipe
    - limit iovec size

commit 0637fb168fb839952259b105dc1a603ee1ec8a67
Author: Dave Jones <davej@redhat.com>
Date:   Sun Feb 10 22:32:04 2008 -0500

    Accept a syscall name as an argument to -c

commit 94b810fc1abd5e83af9a69d928a2e9387e0ad886
Author: Dave Jones <davej@redhat.com>
Date:   Sun Feb 10 22:22:29 2008 -0500

    Return a pipe where necessary.

commit 3ddd1e25adcfa41e5521aa8317b3629e31748952
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jan 11 16:43:06 2008 -0500

    Pathnames changed.

commit e2eb47f581e78a7851a8047db52a73450cdf78bc
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 7 02:50:40 2007 -0400

    Update i386 syscalls as of 2.6.23rc2

commit a25ce97bcf3fe50ecbbc49d6e5fb8f9c2b641052
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 7 02:48:30 2007 -0400

    Update x86-64 syscalls as of 2.6.23rc2

commit 7ad3e01d3b5171839055c003c88cf7fcc5120b5c
Author: Dave Jones <davej@redhat.com>
Date:   Thu Apr 19 19:12:08 2007 -0400

    Compile fix. Duhhh.

commit 7683af011626a602fe06893cead7e3bd91de77ba
Author: Dave Jones <davej@redhat.com>
Date:   Sat Mar 31 04:49:07 2007 -0400

    remove unnecessary 'sys_' prefixes from names.

commit a23f8576b1231d80bf5fbdc77a48d62ee0100cd4
Author: Dave Jones <davej@redhat.com>
Date:   Sat Mar 31 04:44:02 2007 -0400

    lookup_dcookie checks CAP_SYS_ADMIN first thing

commit aa52b7c2fca9794e2052d03ca780137d3c9f8b3b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Mar 15 23:15:52 2007 -0400

    Typo

commit 01ccb3ae19fbd44737d8e60343a8e2d04b50551e
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jan 6 12:20:08 2007 -0500

    Don't repeatedly call ni_syscall

commit 6671e23f40e84afa8ac46d7f6267a1a10d9ec741
Author: Dave Jones <davej@redhat.com>
Date:   Sat Jan 6 12:14:42 2007 -0500

    More ideas

commit c19e2cb90863f9a3f850a444085daa548892bf4d
Author: Dave Jones <davej@redhat.com>
Date:   Sat Dec 30 23:36:43 2006 -0500

    Fuzz Xen some time.

commit b351d384299764fe865412cc4784c90f4041b3d7
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 29 17:15:28 2006 -0500

    More TODO

commit 56497cbc6774386b7b42a7ac7927d55be426e67b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 29 17:11:21 2006 -0500

    Make sure we always pass a struct of page size.

commit 12bdca00d1dd1624350d03e8a360a1e374c71680
Author: Randy Dunlap <rdunlap@xenotime.net>
Date:   Fri Dec 29 10:32:39 2006 -0800

    various nits
    
    For mode reg_val, pretty-print the register value that is being used.
    For mode -C (capability checks), print the syscall number also.
    Clarify -x usage message.
    Take -x parameter in hex or decimal.

commit 21251ebb0df65f4b534371696c0cb2e02b89bcc9
Author: Randy Dunlap <rdunlap@xenotime.net>
Date:   Fri Dec 29 09:59:38 2006 -0800

    logical function blocks
    
    Use functions as logical blocks for parsing command line args,
    preparing to run a test, and then running a test.

commit c8f0c8d8dbeb6b688c2f33e02819ade4ed48fa32
Author: Randy Dunlap <rdunlap@xenotime.net>
Date:   Fri Dec 29 09:59:18 2006 -0800

    modenames + fflush
    
    - check for argc==1 (call usage) before getopt() instead of after it;
    - print "signal:" before strsignal() output and add newline to it;
    - pretty-print test mode (opmode and structmode);
    - add fflush(stdout); after test mode:  I was seeing the test mode
      repeated in stdout when it was redirected to a disk file, like this
      (maybe parent/child file buffer sharing problem):
    
    scrashme mode: struct_fill, random
    11: munmap(0x50b010,0x50b010,0x50b010,0x50b010,0x50b010,0x50b010) = -1 Invalid argument
    scrashme mode: struct_fill, random
    12: brk(0x50b010,0x50b010,0x50b010,0x50b010,0x50b010,0x50b010) = 5287952
    scrashme mode: struct_fill, random
    13: rt_sigaction(0x50b010,0x50b010,0x50b010,0x50b010,0x50b010,0x50b010) = -1 Invalid argument

commit 4e2c74789541a6b85bd7e9b6a2b1de74440dffbf
Author: Dave Jones <davej@redhat.com>
Date:   Fri Dec 29 12:31:40 2006 -0500

    Remove unnecessary volatile.
    Noted by Randy Dunlap.

commit 7834b0f9176a57bbf1b9b8adc6980ccc2f268d86
Author: Dave Jones <davej@redhat.com>
Date:   Thu Dec 28 22:56:00 2006 -0500

    More todo's.

commit bb0fdce045285e240e937fc6fcbe7307cbc6c5a4
Author: Randy Dunlap <rdunlap@xenotime.net>
Date:   Fri Dec 15 19:21:56 2006 -0500

    Don't use max. syscallcount unless it is specified.  <bag it>

commit 801c22c790f16f9148a2500de4335040e8c008b0
Author: Randy Dunlap <rdunlap@xenotime.net>
Date:   Fri Dec 15 19:21:21 2006 -0500

    Change -f/-n to -Sxx (any hex value) or -Sr for junk/random values.

commit 5b3c9beb6dd8930e1cb787f755ce7977520e7d01
Author: Dave Jones <davej@redhat.com>
Date:   Wed Dec 13 14:23:09 2006 -0500

    TODO: struct options

commit 98162ace55b5f7e38999b5fbbb9b5618d71d8c08
Author: Randy Dunlap <rdunlap@xenotime.net>
Date:   Wed Dec 13 14:21:33 2006 -0500

    Add loop count option.

commit 03a832eb1688ef04d5141638728a8a236a91d642
Author: Randy Dunlap <rdunlap@xenotime.net>
Date:   Wed Dec 13 14:21:01 2006 -0500

    Minor nits.

commit 39ee707349c70b3edaafb149dba8ab4868a5155e
Author: Dave Jones <davej@redhat.com>
Date:   Mon Nov 6 19:36:01 2006 -0500

    .19 brings two new syscalls for i386

commit d141c4fc839298c2d048fe07369d31470397a8c5
Author: Dave Jones <davej@redhat.com>
Date:   Thu Sep 14 15:41:01 2006 -0400

    Add move_pages support for i386/x86_64

commit fd0f871c1e62210b74e4a8841e0d7921c8b2c66d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Aug 1 17:34:20 2006 -0400

    [SPARC]: Avoid sigsuspend.
    
    Signed-off-by: David S. Miller <davem@davemloft.net>

commit 8dc0dba91a911f436986b0c212b47fcc61a02315
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 16 12:45:50 2006 -0400

    Avoid clone2() on ia64

commit 7647fa8ef86c469fcfe36f6f0e488b27591b98b6
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 16 00:58:50 2006 -0400

    Avoid syscalls that fork/block on sparc.

commit 96e9d2d616babea2e6586bd4fd5f71157529c331
Author: David S. Miller <davem@sunset.davemloft.net>
Date:   Fri Jun 16 00:56:33 2006 -0400

    Initial sparc port.
    
    Signed-off-by: David S. Miller <davem@davemloft.net>

commit 67c4f13f9e2934160cc532b8437e7feac7018546
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 16 00:32:48 2006 -0400

    IA64 syscalls start at 1024 for god knows what reason.

commit 696a3afaf643a9f97b0cecce2ded062e0f7d8a4b
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 16 00:26:37 2006 -0400

    Remove unneeded checks now that this is done by .flags

commit 4b2818aef861dcbeb57b7787974926d1f419bb04
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 15 17:45:12 2006 -0400

    PAGE_MASK may not be in the headers, work it out by hand.

commit adacb829a1b02a704cdc866f4fee49ac55d7c55e
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 15 15:27:18 2006 -0400

    Clean up the per-arch 'skip this syscall' handling.

commit 13e4696a5d9dd46d5d80183fa35ce47dc149c91b
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 15 15:04:17 2006 -0400

    IA64 support from Jarod Wilson.

commit 15177e1ca5ee3778aa51980bb5eba599129bf96c
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 14 22:13:02 2006 -0400

    Crackmonkey jokes are so 2000.

commit e78e0391b17c866c23ce3cb90faf81d9440357b3
Author: Dave Jones <davej@redhat.com>
Date:   Wed Jun 14 22:11:27 2006 -0400

    Remove changelog.  I'll generate this from the SCM.

commit cab43598b71d84a2bee55f443de9f9b78a60eeb4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 13 12:53:05 2006 -0400

    Flush after doing the syscall
    (Stuff was getting lost, especially when running multiple copies
    at the same time)

commit b49f289f76593b7085b4e6c08422d782be04c7e2
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 12 14:17:20 2006 -0400

    i386 buildfix

commit bc628f1cc9358654d83bf732fec5edd827245f4d
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jun 11 23:34:15 2006 -0400

    Add some more -C flags for x86-64

commit 0c19d9475b55e6eff78a7b62f84a2033637c638a
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jun 11 23:05:59 2006 -0400

    -C:  check syscalls that call capable() return -EPERM.
    This calls every syscall that does a call to capable() as its
    first thing, and makes sure that it gets -EPERM back.
    
    (Also cleaned up the mode handling in scrashme.c as a result
     of some necessary refactoring for this feature)

commit de84e3d878cd3b1f948ab65f14c56a874835b66b
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jun 11 21:48:11 2006 -0400

    Vastly simplify the file handling for -i
    Now only uses a single file per-process.

commit e6137c108e7e360160974c88cce42f439c3dc184
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jun 11 21:08:21 2006 -0400

    sanitise set_robust_list

commit bceff9756be0e562f5ab96e0c444b9f7774db8f7
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jun 11 20:55:57 2006 -0400

    Duh, thinko

commit 1fd1fcca314446274afbe81a629089374dcecdbb
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jun 11 20:53:37 2006 -0400

    PPC build fix

commit 220c9b21ba7e64d6b0600a1a72aaaafe93228ace
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jun 11 20:49:54 2006 -0400

    Add in missing PPC syscalls.

commit 7c15c2ef0348688ac8b06f216b2dc3ca8968eb22
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jun 11 20:46:42 2006 -0400

    Add vmsplice sanitisation

commit f5fc3b51611eca82ec2fee6bb0c91ccda34fe131
Author: Dave Jones <davej@redhat.com>
Date:   Thu Jun 8 00:47:55 2006 -0400

    Add a todo item to skip ni_syscall entries

commit dd0d10ae635f5f603a11bb2f14cdc0e7014761bd
Author: Dave Jones <davej@redhat.com>
Date:   Tue Jun 6 08:46:38 2006 -0400

    Clarify

commit ab215d661c7b3cf8766a4f97abd6bc29f3b89019
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 5 19:43:47 2006 -0400

    TODO additions/cleanups

commit 5bfd6367607b2ab932be59554244a6e5b45cc750
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 5 19:11:19 2006 -0400

    Make -t the default.

commit 065f0f2029922b07628872adcb8a309dee5a8b06
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 5 00:14:12 2006 -0400

    Fix sigsuspend hang on ppc.

commit 2dbbaa979be4ed82dfe06d6724c2dc6590531c09
Author: Dave Jones <davej@redhat.com>
Date:   Mon Jun 5 00:12:33 2006 -0400

    More PowerPC fixes from Jarod.

commit f3b3987c19883fba6921d02e5464b35fbd887406
Author: Dave Jones <davej@redhat.com>
Date:   Sun Jun 4 21:18:50 2006 -0400

    I've changed so much, this is essentially a rewrite by now.

commit f13927f842c01b8e66f9aac45d7ba2e320ecc91a
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 2 18:26:19 2006 -0400

    PowerPC Support from Jarod Wilson.

commit 72ffb962fbfd08e76ef8c3552225f8141d737349
Author: Dave Jones <davej@redhat.com>
Date:   Fri Jun 2 16:45:44 2006 -0400

    Fix silly typo that broke the compile.
    Can you guess I only build x86-64 these days?

commit 70081d0d438f55ce2a156bf7133e4598da985530
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 31 17:44:45 2006 -0400

    Make the tmp dir if it doesn't exist
    Remove the fsck call after making. (I moved this to my local 'push' script)

commit 77252c73bb68b43705c07992a977a5a0697b3434
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 31 17:40:09 2006 -0400

    re-add the priv dropping in case we get run as root.

commit 304582036502ed2d0e1349db272436bf87cd78ac
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 31 17:31:07 2006 -0400

    Remove some more crap.

commit 19d55caa6f7667e9889b0ae68728b505f77dc66e
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 31 17:19:38 2006 -0400

    Remove some pointless checks that actually broke things.

commit 9fe47fd69a1b97211bbe7e2e2fa24a5ba44a662e
Author: Dave Jones <davej@redhat.com>
Date:   Mon May 29 12:31:45 2006 -0400

    Make some vars unsigned

commit 1b0333130051887acfc52d83db47a0f2589da363
Author: Dave Jones <davej@redhat.com>
Date:   Sun May 28 11:16:16 2006 -0400

    Add sys_vmsplice

commit a9b96dfd72f47007e75248020ef2e56cbfb8dec2
Author: Dave Jones <davej@redhat.com>
Date:   Sun May 21 20:47:19 2006 -0400

    Try multiple parameters in the example script.

commit 447d4cf1eb3bcd2b76f02b871c4b02708050de6d
Author: Dave Jones <davej@redhat.com>
Date:   Sun May 21 20:34:37 2006 -0400

    sync_file_range sanitisation

commit 28707439ea3f6cd5348fea7fcb0d5fb687380fbe
Author: Dave Jones <davej@redhat.com>
Date:   Sun May 21 15:51:40 2006 -0400

    Fix up prototype that sparse warned about

commit b7457d35ecabb592d04e16a835d47501b5aea80e
Author: Dave Jones <davej@redhat.com>
Date:   Sat May 20 23:59:58 2006 -0400

    Fix up implicit declaration

commit 1c8b24a4a2bca0f97502bcb245aef898a2ea7fde
Author: Dave Jones <davej@redhat.com>
Date:   Sat May 20 23:58:55 2006 -0400

    Fix up a bunch of splint warnings.

commit 84ab25db038978f8e718f07f20a2d0e6283ef4b5
Author: Dave Jones <davej@redhat.com>
Date:   Wed May 17 16:05:49 2006 -0400

    Add a missing flush, which fixes output when redirected.

commit 2b96d7f0a5b84be9e5eb3579e9de87b7b8447f97
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 21 02:17:32 2006 -0400

    Paranoia.

commit 490a5f77d8032f6054bb020403f7d62dd1cb7494
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 21 02:12:49 2006 -0400

    Fix several off-by-ones

commit 83fe4820d1d17d2d61a92d8390618033038c2135
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 21 02:05:28 2006 -0400

    make -z and -c work together.

commit 5ba8decaff35bfb2f5ca95f32fafd32db4a91440
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 21 01:54:22 2006 -0400

    Make debug stuff easier to turn on/off

commit 446d01fd8a57c7c5ce86a18f78bd5487a6b1d6f2
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 21 01:47:03 2006 -0400

    Add sanitise for sys_tee

commit 362849589a468924c966840bad6ecb303696d3f5
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 21 01:44:08 2006 -0400

    sys_splice changed its prototype.

commit d86dadeaac625a1b55aff4b9413a686c51662b57
Author: Dave Jones <davej@redhat.com>
Date:   Fri Apr 21 01:35:40 2006 -0400

    Add sys_tee

commit 1aae602776e938b0ef70a6f642257ae33aba2748
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 12 19:51:28 2006 -0500

    Compile fixes (Lose the _sys from sanitise function names)

commit ecda7b06ca3bf5e1b3fe88758e4041bed8244ddb
Author: Dave Jones <davej@redhat.com>
Date:   Wed Apr 12 19:51:09 2006 -0500

    chroot into tmp

commit 503c8f781f7c6264c1a034ccf52143654e3ef0fe
Author: Dave Jones <davej@redhat.com>
Date:   Sat Apr 8 20:34:56 2006 -0500

    Fix incorrect register setting, and loop if still 0.

commit 3f5ccfb748bc3a7f9ef691293781fd4dd2bc15e2
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 4 17:40:41 2006 -0500

    Improve file handling.
    Add primitive sanitise() to sys_read / sys_write

commit 7d46140b6f7830d6db45133670edab2fd5df80e6
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 4 16:26:45 2006 -0500

    Use some files of random junk as args instead.

commit 1db8acef436fa7ea9c48a38156937b1a590fb579
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 4 16:24:55 2006 -0500

    Add splice sanitise for i386 too

commit e6d8481ddc8428e6c6a843c8dad4f80bcde9c13c
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 4 16:24:02 2006 -0500

    Create some fd's to pass into syscalls when sanitised.

commit d6d24ed10e9896dba3c1290da520d8b807dff2d5
Author: Dave Jones <davej@redhat.com>
Date:   Tue Apr 4 15:49:43 2006 -0500

    Add framework for sanitising arguments before we do the syscall.
    (Added an empty splice() sanitiser for now, which doesn't do
     anything particularly useful)

commit 1aba452fd4cc98adb2e252a8159d31cd119947e6
Author: Dave Jones <davej@redhat.com>
Date:   Sat Apr 1 16:06:02 2006 -0500

    New syscalls.

commit 11d057b6eb2cd9ae54fd384202589bd8ba06a429
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 20 02:30:23 2006 -0500

    Make syscall naming per-arch.
    Now x86-64 resolves to the correct syscalls.

commit c17a4006f564bfccbb0628f65224ad5f1caf52df
Author: Dave Jones <davej@redhat.com>
Date:   Mon Mar 20 00:05:11 2006 -0500

    Update patterns

commit 293ebefef7e483a13422262ac014f7858136748d
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 7 02:51:25 2006 -0500

    Abstract some things out to per-arch header files to clean up main source.

commit 3ab5d38fc49a9202a324296c7a6a7a3bea145fb0
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 7 02:48:11 2006 -0500

    Add some comments

commit c0c0e0685b35f6af00d4a035c28da1de2254ad76
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 7 02:42:59 2006 -0500

    use ulong

commit 144ab1f3b86c3d5257edc76358d0a9fa979e7371
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 7 02:38:19 2006 -0500

    Use syscall() directly instead of fugly asm

commit 6dd7e798cd3b0472ad250b7b7dc204ccc28e9710
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 7 02:31:13 2006 -0500

    coding style nits

commit e0678c18b7b1e8ebbf58b754b25d704dad75f904
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 7 02:29:38 2006 -0500

    Make 64bit args work for -r too as well as -z

commit 18fda2f627190e483cf87a3b78b64677a64fb0bd
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 7 02:24:23 2006 -0500

    idea for patterns.

commit 1f8740b913c24d4fbfaacf5960063165d18a1caf
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 7 02:21:36 2006 -0500

    Pass 64 bit args.

commit 21efabfd67bd766e07108d4810677a8b22c3ca47
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 7 02:06:08 2006 -0500

    Add preliminary -k support.
    Needs more work.

commit 9eb9b4164675f1ce786305b6e08f246866ae91b4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 7 02:00:13 2006 -0500

    New ideas

commit e61e1b03eea92f5035ed332506bcec4d8ce13b84
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 7 01:59:08 2006 -0500

    Native x86-64 support.

commit fad7a0f168cc33968bf16732782ac6c05e0d0185
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 7 01:55:57 2006 -0500

    Added all the i386 syscalls as of 2.6.16rc3

commit d8a28d440cc2f34b31836b83ada19402d9a9b9a4
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 7 01:55:39 2006 -0500

    Do runs from a tmp dir.

commit 04dd193ad4e0432a85f9eb5274d167b0e5554025
Author: Dave Jones <davej@redhat.com>
Date:   Tue Mar 7 01:23:47 2006 -0500

    Initial import.
