)]}'
{
  "log": [
    {
      "commit": "626cf236608505d376e4799adb4f7eb00a8594af",
      "tree": "eb0421fec1a7fed05cd9ea785bd18b5f40c80971",
      "parents": [
        "5cde7656d0dd222170eb0250bd1f70c9018fd438"
      ],
      "author": {
        "name": "Hans Verkuil",
        "email": "hans.verkuil@cisco.com",
        "time": "Fri Mar 23 15:02:27 2012 -0700"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@linux-foundation.org",
        "time": "Fri Mar 23 16:58:38 2012 -0700"
      },
      "message": "poll: add poll_requested_events() and poll_does_not_wait() functions\n\nIn some cases the poll() implementation in a driver has to do different\nthings depending on the events the caller wants to poll for.  An example\nis when a driver needs to start a DMA engine if the caller polls for\nPOLLIN, but doesn\u0027t want to do that if POLLIN is not requested but instead\nonly POLLOUT or POLLPRI is requested.  This is something that can happen\nin the video4linux subsystem among others.\n\nUnfortunately, the current epoll/poll/select implementation doesn\u0027t\nprovide that information reliably.  The poll_table_struct does have it: it\nhas a key field with the event mask.  But once a poll() call matches one\nor more bits of that mask any following poll() calls are passed a NULL\npoll_table pointer.\n\nAlso, the eventpoll implementation always left the key field at ~0 instead\nof using the requested events mask.\n\nThis was changed in eventpoll.c so the key field now contains the actual\nevents that should be polled for as set by the caller.\n\nThe solution to the NULL poll_table pointer is to set the qproc field to\nNULL in poll_table once poll() matches the events, not the poll_table\npointer itself.  That way drivers can obtain the mask through a new\npoll_requested_events inline.\n\nThe poll_table_struct can still be NULL since some kernel code calls it\ninternally (netfs_state_poll() in ./drivers/staging/pohmelfs/netfs.h).  In\nthat case poll_requested_events() returns ~0 (i.e.  all events).\n\nVery rarely drivers might want to know whether poll_wait will actually\nwait.  If another earlier file descriptor in the set already matched the\nevents the caller wanted to wait for, then the kernel will return from the\nselect() call without waiting.  This might be useful information in order\nto avoid doing expensive work.\n\nA new helper function poll_does_not_wait() is added that drivers can use\nto detect this situation.  This is now used in sock_poll_wait() in\ninclude/net/sock.h.  This was the only place in the kernel that needed\nthis information.\n\nDrivers should no longer access any of the poll_table internals, but use\nthe poll_requested_events() and poll_does_not_wait() access functions\ninstead.  In order to enforce that the poll_table fields are now prepended\nwith an underscore and a comment was added warning against using them\ndirectly.\n\nThis required a change in unix_dgram_poll() in unix/af_unix.c which used\nthe key field to get the requested events.  It\u0027s been replaced by a call\nto poll_requested_events().\n\nFor qproc it was especially important to change its name since the\nbehavior of that field changes with this patch since this function pointer\ncan now be NULL when that wasn\u0027t possible in the past.\n\nAny driver accessing the qproc or key fields directly will now fail to compile.\n\nSome notes regarding the correctness of this patch: the driver\u0027s poll()\nfunction is called with a \u0027struct poll_table_struct *wait\u0027 argument.  This\npointer may or may not be NULL, drivers can never rely on it being one or\nthe other as that depends on whether or not an earlier file descriptor in\nthe select()\u0027s fdset matched the requested events.\n\nThere are only three things a driver can do with the wait argument:\n\n1) obtain the key field:\n\n\tevents \u003d wait ? wait-\u003ekey : ~0;\n\n   This will still work although it should be replaced with the new\n   poll_requested_events() function (which does exactly the same).\n   This will now even work better, since wait is no longer set to NULL\n   unnecessarily.\n\n2) use the qproc callback. This could be deadly since qproc can now be\n   NULL. Renaming qproc should prevent this from happening. There are no\n   kernel drivers that actually access this callback directly, BTW.\n\n3) test whether wait \u003d\u003d NULL to determine whether poll would return without\n   waiting. This is no longer sufficient as the correct test is now\n   wait \u003d\u003d NULL || wait-\u003e_qproc \u003d\u003d NULL.\n\n   However, the worst that can happen here is a slight performance hit in\n   the case where wait !\u003d NULL and wait-\u003e_qproc \u003d\u003d NULL. In that case the\n   driver will assume that poll_wait() will actually add the fd to the set\n   of waiting file descriptors. Of course, poll_wait() will not do that\n   since it tests for wait-\u003e_qproc. This will not break anything, though.\n\n   There is only one place in the whole kernel where this happens\n   (sock_poll_wait() in include/net/sock.h) and that code will be replaced\n   by a call to poll_does_not_wait() in the next patch.\n\n   Note that even if wait-\u003e_qproc !\u003d NULL drivers cannot rely on poll_wait()\n   actually waiting. The next file descriptor from the set might match the\n   event mask and thus any possible waits will never happen.\n\nSigned-off-by: Hans Verkuil \u003chans.verkuil@cisco.com\u003e\nReviewed-by: Jonathan Corbet \u003ccorbet@lwn.net\u003e\nReviewed-by: Al Viro \u003cviro@zeniv.linux.org.uk\u003e\nCc: Davide Libenzi \u003cdavidel@xmailserver.org\u003e\nSigned-off-by: Hans de Goede \u003chdegoede@redhat.com\u003e\nCc: Mauro Carvalho Chehab \u003cmchehab@infradead.org\u003e\nCc: David Miller \u003cdavem@davemloft.net\u003e\nCc: Eric Dumazet \u003ceric.dumazet@gmail.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n"
    },
    {
      "commit": "25985edcedea6396277003854657b5f3cb31a628",
      "tree": "f026e810210a2ee7290caeb737c23cb6472b7c38",
      "parents": [
        "6aba74f2791287ec407e0f92487a725a25908067"
      ],
      "author": {
        "name": "Lucas De Marchi",
        "email": "lucas.demarchi@profusion.mobi",
        "time": "Wed Mar 30 22:57:33 2011 -0300"
      },
      "committer": {
        "name": "Lucas De Marchi",
        "email": "lucas.demarchi@profusion.mobi",
        "time": "Thu Mar 31 11:26:23 2011 -0300"
      },
      "message": "Fix common misspellings\n\nFixes generated by \u0027codespell\u0027 and manually reviewed.\n\nSigned-off-by: Lucas De Marchi \u003clucas.demarchi@profusion.mobi\u003e\n"
    },
    {
      "commit": "dac36dd87de10d1fd81dc7b7a40256cb2e965abc",
      "tree": "5ad8afda6dbc12a7e315c57b6ec0d66b334a62eb",
      "parents": [
        "747fecab32e47180e2668c2b22b25752d371e636"
      ],
      "author": {
        "name": "Namhyung Kim",
        "email": "namhyung@gmail.com",
        "time": "Fri Dec 10 01:57:07 2010 +0900"
      },
      "committer": {
        "name": "Jiri Kosina",
        "email": "jkosina@suse.cz",
        "time": "Fri Dec 10 15:06:43 2010 +0100"
      },
      "message": "poll: fix a typo in comment\n\nConvert duplicated sys_poll to select. As Kosaki suggests, sys_poll() and\nsys_select() are now hidden by SYSCALL_DEFINEx() macros so it would be\nbetter to use plain select/poll syscall name.\n\nSigned-off-by: Namhyung Kim \u003cnamhyung@gmail.com\u003e\nReviewed-by: KOSAKI Motohiro \u003ckosaki.motohiro@jp.fujitsu.com\u003e\nSigned-off-by: Jiri Kosina \u003cjkosina@suse.cz\u003e\n"
    },
    {
      "commit": "95aac7b1cd224f568fb83937044cd303ff11b029",
      "tree": "24c08cfe031ecc0549cf3f7900e992a27044c4ad",
      "parents": [
        "231f3d393f63f6e3b505afa179999bba491d0f08"
      ],
      "author": {
        "name": "Shawn Bohrer",
        "email": "shawn.bohrer@gmail.com",
        "time": "Wed Oct 27 15:34:54 2010 -0700"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@linux-foundation.org",
        "time": "Wed Oct 27 18:03:18 2010 -0700"
      },
      "message": "epoll: make epoll_wait() use the hrtimer range feature\n\nThis make epoll use hrtimers for the timeout value which prevents\nepoll_wait() from timing out up to a millisecond early.\n\nThis mirrors the behavior of select() and poll().\n\nSigned-off-by: Shawn Bohrer \u003cshawn.bohrer@gmail.com\u003e\nCc: Al Viro \u003cviro@zeniv.linux.org.uk\u003e\nAcked-by: Davide Libenzi \u003cdavidel@xmailserver.org\u003e\nCc: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n"
    },
    {
      "commit": "9ff99339447de403a46be5e3f23d0c794d540b06",
      "tree": "8668d1b54543e19a42550e7f7332444e68661ec6",
      "parents": [
        "2edf5e49800846a2b2b6461d99cdae18067c440f"
      ],
      "author": {
        "name": "Dave Young",
        "email": "hidave.darkstar@gmail.com",
        "time": "Wed Mar 10 15:24:10 2010 -0800"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@linux-foundation.org",
        "time": "Fri Mar 12 15:53:11 2010 -0800"
      },
      "message": "sysctl extern cleanup: poll\n\nExtern declarations in sysctl.c should be moved to their own header file,\nand then include them in relavant .c files.\n\nMove epoll_table extern declaration to linux/poll.h\n\nSigned-off-by: Dave Young \u003chidave.darkstar@gmail.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n"
    },
    {
      "commit": "a99bbaf5ee6bad1aca0c88ea65ec6e5373e86184",
      "tree": "2b0314d68b9e0a76e8f4fb60865a1d56e138833a",
      "parents": [
        "5e5027bd26ed4df735d29e66cd5c1c9b5959a587"
      ],
      "author": {
        "name": "Alexey Dobriyan",
        "email": "adobriyan@gmail.com",
        "time": "Sun Oct 04 16:11:37 2009 +0400"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@linux-foundation.org",
        "time": "Sun Oct 04 15:05:10 2009 -0700"
      },
      "message": "headers: remove sched.h from poll.h\n\nSigned-off-by: Alexey Dobriyan \u003cadobriyan@gmail.com\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n"
    },
    {
      "commit": "4938d7e0233a455f04507bac81d0886c71529537",
      "tree": "6e0c6f0e8623ef44270890519f08b98e9db8afee",
      "parents": [
        "02d5341ae53d32681241b27a40397475caef1c83"
      ],
      "author": {
        "name": "Eric Dumazet",
        "email": "dada1@cosmosbay.com",
        "time": "Tue Jun 16 15:33:36 2009 -0700"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@linux-foundation.org",
        "time": "Tue Jun 16 19:47:48 2009 -0700"
      },
      "message": "poll: avoid extra wakeups in select/poll\n\nAfter introduction of keyed wakeups Davide Libenzi did on epoll, we are\nable to avoid spurious wakeups in poll()/select() code too.\n\nFor example, typical use of poll()/select() is to wait for incoming\nnetwork frames on many sockets.  But TX completion for UDP/TCP frames call\nsock_wfree() which in turn schedules thread.\n\nWhen scheduled, thread does a full scan of all polled fds and can sleep\nagain, because nothing is really available.  If number of fds is large,\nthis cause significant load.\n\nThis patch makes select()/poll() aware of keyed wakeups and useless\nwakeups are avoided.  This reduces number of context switches by about 50%\non some setups, and work performed by sofirq handlers.\n\nSigned-off-by: Eric Dumazet \u003cdada1@cosmosbay.com\u003e\nAcked-by: David S. Miller \u003cdavem@davemloft.net\u003e\nAcked-by: Andi Kleen \u003cak@linux.intel.com\u003e\nAcked-by: Ingo Molnar \u003cmingo@elte.hu\u003e\nAcked-by: Davide Libenzi \u003cdavidel@xmailserver.org\u003e\nCc: Christoph Lameter \u003ccl@linux-foundation.org\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n"
    },
    {
      "commit": "5f820f648c92a5ecc771a96b3c29aa6e90013bba",
      "tree": "0445b45fa33072d37b32c6ef592a4d0c102e05cc",
      "parents": [
        "67ec7d3ab779ad9001ef57a6b4cfdf80ac9f9acc"
      ],
      "author": {
        "name": "Tejun Heo",
        "email": "htejun@gmail.com",
        "time": "Tue Jan 06 14:40:59 2009 -0800"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@linux-foundation.org",
        "time": "Tue Jan 06 15:59:12 2009 -0800"
      },
      "message": "poll: allow f_op-\u003epoll to sleep\n\nf_op-\u003epoll is the only vfs operation which is not allowed to sleep.  It\u0027s\nbecause poll and select implementation used task state to synchronize\nagainst wake ups, which doesn\u0027t have to be the case anymore as wait/wake\ninterface can now use custom wake up functions.  The non-sleep restriction\ncan be a bit tricky because -\u003epoll is not called from an atomic context\nand the result of accidentally sleeping in -\u003epoll only shows up as\ntemporary busy looping when the timing is right or rather wrong.\n\nThis patch converts poll/select to use custom wake up function and use\nseparate triggered variable to synchronize against wake up events.  The\nonly added overhead is an extra function call during wake up and\nnegligible.\n\nThis patch removes the one non-sleep exception from vfs locking rules and\nis beneficial to userland filesystem implementations like FUSE, 9p or\npeculiar fs like spufs as it\u0027s very difficult for those to implement\nnon-sleeping poll method.\n\nWhile at it, make the following cosmetic changes to make poll.h and\nselect.c checkpatch friendly.\n\n* s/type * symbol/type *symbol/\t\t   : three places in poll.h\n* remove blank line before EXPORT_SYMBOL() : two places in select.c\n\nOleg: spotted missing barrier in poll_schedule_timeout()\nDavide: spotted missing write barrier in pollwake()\n\nSigned-off-by: Tejun Heo \u003ctj@kernel.org\u003e\nCc: Eric Van Hensbergen \u003cericvh@gmail.com\u003e\nCc: Ron Minnich \u003crminnich@sandia.gov\u003e\nCc: Ingo Molnar \u003cmingo@elte.hu\u003e\nCc: Christoph Hellwig \u003chch@infradead.org\u003e\nSigned-off-by: Miklos Szeredi \u003cmszeredi@suse.cz\u003e\nCc: Davide Libenzi \u003cdavidel@xmailserver.org\u003e\nCc: Brad Boyer \u003cflar@allandria.com\u003e\nCc: Al Viro \u003cviro@zeniv.linux.org.uk\u003e\nCc: Roland McGrath \u003croland@redhat.com\u003e\nCc: Mauro Carvalho Chehab \u003cmchehab@infradead.org\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nCc: Davide Libenzi \u003cdavidel@xmailserver.org\u003e\nCc: Oleg Nesterov \u003coleg@redhat.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n"
    },
    {
      "commit": "8ff3e8e85fa6c312051134b3953e397feb639f51",
      "tree": "526886377ca92a62f030c25bc8f91f13a6f991eb",
      "parents": [
        "be5dad20a55e054a35dac7f6f5f184dc72b379b4"
      ],
      "author": {
        "name": "Arjan van de Ven",
        "email": "arjan@linux.intel.com",
        "time": "Sun Aug 31 08:26:40 2008 -0700"
      },
      "committer": {
        "name": "Arjan van de Ven",
        "email": "arjan@linux.intel.com",
        "time": "Fri Sep 05 21:35:03 2008 -0700"
      },
      "message": "select: switch select() and poll() over to hrtimers\n\nWith lots of help, input and cleanups from Thomas Gleixner\n\nThis patch switches select() and poll() over to hrtimers.\n\nThe core of the patch is replacing the \"s64 timeout\" with a\n\"struct timespec end_time\" in all the plumbing.\n\nBut most of the diffstat comes from using the just introduced helpers:\n\tpoll_select_set_timeout\n\tpoll_select_copy_remaining\n\ttimespec_add_safe\nwhich make manipulating the timespec easier and less error-prone.\n\nSigned-off-by: Arjan van de Ven \u003carjan@linux.intel.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n"
    },
    {
      "commit": "b773ad40aca5bd755ba886620842f16e8fef6d75",
      "tree": "7065661f6559aff70244f2805ced48f894c9b38c",
      "parents": [
        "df0cc0539b4127bd02f64de2c335b4af1fdb3845"
      ],
      "author": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Sun Aug 31 08:16:57 2008 -0700"
      },
      "committer": {
        "name": "Arjan van de Ven",
        "email": "arjan@linux.intel.com",
        "time": "Fri Sep 05 21:34:59 2008 -0700"
      },
      "message": "select: add poll_select_set_timeout() and poll_select_copy_remaining() helpers\n\nThis patch adds 2 helpers that will be used for the hrtimer based select/poll:\n\npoll_select_set_timeout() is a helper that takes a timeout (as a second, nanosecond\npair) and turns that into a \"struct timespec\" that represents the absolute end time.\nThis is a common operation in the many select() and poll() variants and needs various,\ncommon, sanity checks.\n\npoll_select_copy_remaining() is a helper that takes care of copying the remaining\ntime to userspace, as select(), pselect() and ppoll() do. This function comes in\nboth a natural and a compat implementation (due to datastructure differences).\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nSigned-off-by: Arjan van de Ven \u003carjan@linux.intel.com\u003e\n"
    },
    {
      "commit": "a2dcb44c3c5a8151d2d9f6ac8ad0789efcdbe184",
      "tree": "8711fc210dc58372a859e6a6cf78dc638d768600",
      "parents": [
        "bf7da7bcfb38409b4cdea34b0905bdf344f1b36d"
      ],
      "author": {
        "name": "Al Viro",
        "email": "viro@zeniv.linux.org.uk",
        "time": "Wed Apr 23 14:05:15 2008 -0400"
      },
      "committer": {
        "name": "Al Viro",
        "email": "viro@zeniv.linux.org.uk",
        "time": "Thu May 01 13:07:28 2008 -0400"
      },
      "message": "[PATCH] make osf_select() use core_sys_select()\n\n... instead of open-coding it\n\nSigned-off-by: Al Viro \u003cviro@zeniv.linux.org.uk\u003e\n"
    },
    {
      "commit": "dd23aae4f5edf4e1dbd8f7f8013a754ba3253f48",
      "tree": "c7babf45f8132876e8a715f0327cf480c05c6131",
      "parents": [
        "3210f0ecdba6a81c3f8efe6f442d2e1f57db98f9"
      ],
      "author": {
        "name": "Alexey Dobriyan",
        "email": "adobriyan@gmail.com",
        "time": "Tue Sep 11 15:23:55 2007 -0700"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@woody.linux-foundation.org",
        "time": "Tue Sep 11 17:21:20 2007 -0700"
      },
      "message": "Fix select on /proc files without -\u003epoll\n\nTaneli Vähäkangas \u003cvahakang@cs.helsinki.fi\u003e reported that commit\n786d7e1612f0b0adb6046f19b906609e4fe8b1ba aka \"Fix rmmod/read/write races\nin /proc entries\" broke SBCL + SLIME combo.\n\nThe old code in do_select() used DEFAULT_POLLMASK, if couldn\u0027t find\n-\u003epoll handler.  The new code makes -\u003epoll always there and returns 0 by\ndefault, which is not correct.  Return DEFAULT_POLLMASK instead.\n\nSteps to reproduce:\n\n\tinstall emacs, SBCL, SLIME\n\temacs\n\tM-x slime\tin *inferior-lisp* buffer\n\t[watch it doing \"Connecting to Swank on port X..\"]\n\nPlease, apply before 2.6.23.\n\nP.S.: why SBCL can\u0027t just read(2) /proc/cpuinfo is a mystery.\n\nSigned-off-by: Alexey Dobriyan \u003cadobriyan@gmail.com\u003e\nCc: T Taneli Vahakangas \u003cvahakang@cs.helsinki.fi\u003e\nCc: Oleg Nesterov \u003coleg@tv-sign.ru\u003e\nCc: \"Eric W. Biederman\" \u003cebiederm@xmission.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n"
    },
    {
      "commit": "f23f6e08c47acbdd20e9c49a79da8c404ea168e1",
      "tree": "00c86da440e4399b4dd04e9f967acd324e0fdf78",
      "parents": [
        "d7fe0f241dceade9c8d4af75498765c5ff7f27e6"
      ],
      "author": {
        "name": "Al Viro",
        "email": "viro@zeniv.linux.org.uk",
        "time": "Fri Oct 20 15:17:02 2006 -0400"
      },
      "committer": {
        "name": "Al Viro",
        "email": "viro@zeniv.linux.org.uk",
        "time": "Mon Dec 04 02:00:36 2006 -0500"
      },
      "message": "[PATCH] severing poll.h -\u003e mm.h\n\nSigned-off-by: Al Viro \u003cviro@zeniv.linux.org.uk\u003e\n"
    },
    {
      "commit": "70674f95c0a2ea694d5c39f4e514f538a09be36f",
      "tree": "906d109fafc5eafff6a90c8d866e0525fdaf6783",
      "parents": [
        "b02389e98a7b64ad5cd4823740defa8821f30bbd"
      ],
      "author": {
        "name": "Andi Kleen",
        "email": "ak@suse.de",
        "time": "Tue Mar 28 01:56:33 2006 -0800"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@g5.osdl.org",
        "time": "Tue Mar 28 09:16:04 2006 -0800"
      },
      "message": "[PATCH] Optimize select/poll by putting small data sets on the stack\n\nOptimize select and poll by a using stack space for small fd sets\n\nThis brings back an old optimization from Linux 2.0.  Using the stack is\nfaster than kmalloc.  On a Intel P4 system it speeds up a select of a\nsingle pty fd by about 13% (~4000 cycles -\u003e ~3500)\n\nIt also saves memory because a daemon hanging in select or poll will\nusually save one or two less pages.  This can add up - e.g.  if you have 10\ndaemons blocking in poll/select you save 40KB of memory.\n\nI did a patch for this long ago, but it was never applied.  This version is\na reimplementation of the old patch that tries to be less intrusive.  I\nonly did the minimal changes needed for the stack allocation.\n\nThe cut off point before external memory is allocated is currently at\n832bytes.  The system calls always allocate this much memory on the stack.\n\nThese 832 bytes are divided into 256 bytes frontend data (for the select\nbitmaps of the pollfds) and the rest of the space for the wait queues used\nby the low level drivers.  There are some extreme cases where this won\u0027t\nwork out for select and it falls back to allocating memory too early -\nespecially with very sparse large select bitmaps - but the majority of\nprocesses who only have a small number of file descriptors should be ok.\n[TBD: 832/256 might not be the best split for select or poll]\n\nI suspect more optimizations might be possible, but they would be more\ncomplicated.  One way would be to cache the select/poll context over\nmultiple system calls because typically the input values should be similar.\n Problem is when to flush the file descriptors out though.\n\nSigned-off-by: Andi Kleen \u003cak@suse.de\u003e\nCc: Eric Dumazet \u003cdada1@cosmosbay.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n"
    },
    {
      "commit": "9f72949f679df06021c9e43886c9191494fdb007",
      "tree": "f4d76ed281b34e195db7741b69a7d095e168a864",
      "parents": [
        "36a7878a224c18aa4a5e098dc93d19cf5601462b"
      ],
      "author": {
        "name": "David Woodhouse",
        "email": "dwmw2@infradead.org",
        "time": "Wed Jan 18 17:44:05 2006 -0800"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@g5.osdl.org",
        "time": "Wed Jan 18 19:20:30 2006 -0800"
      },
      "message": "[PATCH] Add pselect/ppoll system call implementation\n\nThe following implementation of ppoll() and pselect() system calls\ndepends on the architecture providing a TIF_RESTORE_SIGMASK flag in the\nthread_info.\n\nThese system calls have to change the signal mask during their\noperation, and signal handlers must be invoked using the new, temporary\nsignal mask. The old signal mask must be restored either upon successful\nexit from the system call, or upon returning from the invoked signal\nhandler if the system call is interrupted. We can\u0027t simply restore the\noriginal signal mask and return to userspace, since the restored signal\nmask may actually block the signal which interrupted the system call.\n\nThe TIF_RESTORE_SIGMASK flag deals with this by causing the syscall exit\npath to trap into do_signal() just as TIF_SIGPENDING does, and by\ncausing do_signal() to use the saved signal mask instead of the current\nsignal mask when setting up the stack frame for the signal handler -- or\nby causing do_signal() to simply restore the saved signal mask in the\ncase where there is no handler to be invoked.\n\nThe first patch implements the sys_pselect() and sys_ppoll() system\ncalls, which are present only if TIF_RESTORE_SIGMASK is defined. That\n#ifdef should go away in time when all architectures have implemented\nit. The second patch implements TIF_RESTORE_SIGMASK for the PowerPC\nkernel (in the -mm tree), and the third patch then removes the\narch-specific implementations of sys_rt_sigsuspend() and replaces them\nwith generic versions using the same trick.\n\nThe fourth and fifth patches, provided by David Howells, implement\nTIF_RESTORE_SIGMASK for FR-V and i386 respectively, and the sixth patch\nadds the syscalls to the i386 syscall table.\n\nThis patch:\n\nAdd the pselect() and ppoll() system calls, providing core routines usable by\nthe original select() and poll() system calls and also the new calls (with\ntheir semantics w.r.t timeouts).\n\nSigned-off-by: David Woodhouse \u003cdwmw2@infradead.org\u003e\nCc: Michael Kerrisk \u003cmtk-manpages@gmx.net\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n"
    },
    {
      "commit": "1da177e4c3f41524e886b7f1b8a0c1fc7321cac2",
      "tree": "0bba044c4ce775e45a88a51686b5d9f90697ea9d",
      "parents": [],
      "author": {
        "name": "Linus Torvalds",
        "email": "torvalds@ppc970.osdl.org",
        "time": "Sat Apr 16 15:20:36 2005 -0700"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@ppc970.osdl.org",
        "time": "Sat Apr 16 15:20:36 2005 -0700"
      },
      "message": "Linux-2.6.12-rc2\n\nInitial git repository build. I\u0027m not bothering with the full history,\neven though we have it. We can create a separate \"historical\" git\narchive of that later if we want to, and in the meantime it\u0027s about\n3.2GB when imported into git - space that would just make the early\ngit days unnecessarily complicated, when we don\u0027t have a lot of good\ninfrastructure for it.\n\nLet it rip!\n"
    }
  ]
}