)]}' { "log": [ { "commit": "d004e024058a0eaca097513ce62cbcf978913e0a", "tree": "94191734e099dd46b8998d813125efaa4a8a37d6", "parents": [ "367177e50189ab0a983b52f3d3f7eb2a1927db71" ], "author": { "name": "Hector Palacios", "email": "hector.palacios@digi.com", "time": "Mon Nov 14 11:15:25 2011 +0100" }, "committer": { "name": "John Stultz", "email": "john.stultz@linaro.org", "time": "Thu Nov 17 14:57:19 2011 -0800" }, "message": "timekeeping: add arch_offset hook to ktime_get functions\n\nktime_get and ktime_get_ts were calling timekeeping_get_ns()\nbut later they were not calling arch_gettimeoffset() so architectures\nusing this mechanism returned 0 ns when calling these functions.\n\nThis happened for example when running Busybox\u0027s ping which calls\nsyscall(__NR_clock_gettime, CLOCK_MONOTONIC, ts) which eventually\ncalls ktime_get. As a result the returned ping travel time was zero.\n\nCC: stable@kernel.org\nSigned-off-by: Hector Palacios \u003chector.palacios@digi.com\u003e\nSigned-off-by: John Stultz \u003cjohn.stultz@linaro.org\u003e\n" }, { "commit": "367177e50189ab0a983b52f3d3f7eb2a1927db71", "tree": "fed5e93eb35d002e21cb3c2fafbfe38062aaa266", "parents": [ "c2bc11113c50449f23c40b724fe410fc2380a8e9", "d65670a78cdbfae94f20a9e05ec705871d7cdf2b" ], "author": { "name": "Ingo Molnar", "email": "mingo@elte.hu", "time": "Fri Nov 11 08:10:42 2011 +0100" }, "committer": { "name": "Ingo Molnar", "email": "mingo@elte.hu", "time": "Fri Nov 11 08:10:42 2011 +0100" }, "message": "Merge branch \u0027formingo/3.2/tip/timers/core\u0027 of git://git.linaro.org/people/jstultz/linux into timers/core\n\nConflicts:\n\tkernel/time/timekeeping.c\n" }, { "commit": "d65670a78cdbfae94f20a9e05ec705871d7cdf2b", "tree": "3c16dedb75fa5ab6c0fdc072cc8d73df01d43a17", "parents": [ "e35f95b36e43f67a6f806172555a152c11ea0a78" ], "author": { "name": "John Stultz", "email": "john.stultz@linaro.org", "time": "Mon Oct 31 17:06:35 2011 -0400" }, "committer": { "name": "John Stultz", "email": "john.stultz@linaro.org", "time": "Thu Nov 10 11:27:08 2011 -0800" }, "message": "clocksource: Avoid selecting mult values that might overflow when adjusted\n\nFor some frequencies, the clocks_calc_mult_shift() function will\nunfortunately select mult values very close to 0xffffffff. This\nhas the potential to overflow when NTP adjusts the clock, adding\nto the mult value.\n\nThis patch adds a clocksource.maxadj value, which provides\nan approximation of an 11% adjustment(NTP limits adjustments to\n500ppm and the tick adjustment is limited to 10%), which could\nbe made to the clocksource.mult value. This is then used to both\ncheck that the current mult value won\u0027t overflow/underflow, as\nwell as warning us if the timekeeping_adjust() code pushes over\nthat 11% boundary.\n\nv2: Fix max_adjustment calculation, and improve WARN_ONCE\nmessages.\n\nv3: Don\u0027t warn before maxadj has actually been set\n\nCC: Yong Zhang \u003cyong.zhang0@gmail.com\u003e\nCC: David Daney \u003cddaney.cavm@gmail.com\u003e\nCC: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nCC: Chen Jie \u003cchenj@lemote.com\u003e\nCC: zhangfx \u003czhangfx@lemote.com\u003e\nCC: stable@kernel.org\nReported-by: Chen Jie \u003cchenj@lemote.com\u003e\nReported-by: zhangfx \u003czhangfx@lemote.com\u003e\nTested-by: Yong Zhang \u003cyong.zhang0@gmail.com\u003e\nSigned-off-by: John Stultz \u003cjohn.stultz@linaro.org\u003e\n" }, { "commit": "c2bc11113c50449f23c40b724fe410fc2380a8e9", "tree": "f5a6a5f9c199d6b3ed7a973dfb7556ad2af2bf5a", "parents": [ "e35f95b36e43f67a6f806172555a152c11ea0a78" ], "author": { "name": "John Stultz", "email": "john.stultz@linaro.org", "time": "Thu Oct 27 18:12:42 2011 -0700" }, "committer": { "name": "Ingo Molnar", "email": "mingo@elte.hu", "time": "Fri Oct 28 08:57:38 2011 +0200" }, "message": "time: Improve documentation of timekeeeping_adjust()\n\nAfter getting a number of questions in private emails about the\nmath around admittedly very complex timekeeping_adjust() and\ntimekeeping_big_adjust(), I figure the code needs some better\ncomments.\n\nHopefully the explanations are clear enough and don\u0027t muddy the\nwater any worse.\n\nStill needs documentation for ntp_error, but I couldn\u0027t recall\nexactly the full explanation behind the code that\u0027s there\n(although I do recall once working it out when Roman first\nproposed it). Given a bit more time I can probably work it out,\nbut I don\u0027t want to hold back this documentation until then.\n\nSigned-off-by: John Stultz \u003cjohn.stultz@linaro.org\u003e\nCc: Chen Jie \u003cchenj@lemote.com\u003e\nCc: Steven Rostedt \u003crostedt@goodmis.org\u003e\nLink: http://lkml.kernel.org/r/1319764362-32367-1-git-send-email-john.stultz@linaro.org\nSigned-off-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n" }, { "commit": "cbaa51524b3224813814607177a00c350ee35d12", "tree": "34f0fabff444e5dde2ee2b92d841abb97daf41b1", "parents": [ "3dcad5ff08f65ae30832220a0e0ee2eac3502a1a" ], "author": { "name": "John Stultz", "email": "john.stultz@linaro.org", "time": "Wed Jul 20 15:42:55 2011 -0700" }, "committer": { "name": "John Stultz", "email": "john.stultz@linaro.org", "time": "Wed Jul 20 15:42:55 2011 -0700" }, "message": "time: Fix stupid KERN_WARN compile issue\n\nTerribly embarassing. Don\u0027t know how I committed this, but its\nKERN_WARNING not KERN_WARN.\n\nThis fixes the following compile error:\nkernel/time/timekeeping.c: In function ‘__timekeeping_inject_sleeptime’:\nkernel/time/timekeeping.c:608: error: ‘KERN_WARN’ undeclared (first use in this function)\nkernel/time/timekeeping.c:608: error: (Each undeclared identifier is reported only once\nkernel/time/timekeeping.c:608: error: for each function it appears in.)\nkernel/time/timekeeping.c:608: error: expected ‘)’ before string constant\nmake[2]: *** [kernel/time/timekeeping.o] Error 1\n\nReported-by: Ingo Molnar \u003cmingo@elte.hu\u003e\nSigned-off-by: John Stultz \u003cjohn.stultz@linaro.org\u003e\n" }, { "commit": "cb33217b1b2523895eb328a0b13fb3b1c4000969", "tree": "f3ce72c0f6cb9a23411c2639acf300f1f453a3c0", "parents": [ "cb5de2f8d0306be38f9b377b8a5c56acca7dbc3d" ], "author": { "name": "John Stultz", "email": "john.stultz@linaro.org", "time": "Tue May 31 22:53:23 2011 -0700" }, "committer": { "name": "John Stultz", "email": "john.stultz@linaro.org", "time": "Tue Jun 21 16:55:37 2011 -0700" }, "message": "time: Avoid accumulating time drift in suspend/resume\n\nBecause the read_persistent_clock interface is usually backed by\nonly a second granular interface, each time we read from the persistent\nclock for suspend/resume, we introduce a half second (on average) of error.\n\nIn order to avoid this error accumulating as the system is suspended\nover and over, this patch measures the time delta between the persistent\nclock and the system CLOCK_REALTIME.\n\nIf the delta is less then 2 seconds from the last suspend, we compensate\nby using the previous time delta (keeping it close). If it is larger\nthen 2 seconds, we assume the clock was set or has been changed, so we\ndo no correction and update the delta.\n\nNote: If NTP is running, ths could seem to \"fight\" with the NTP corrected\ntime, where as if the system time was off by 1 second, and NTP slewed the\nvalue in, a suspend/resume cycle could undo this correction, by trying to\nrestore the previous offset from the persistent clock. However, without\nthis patch, since each read could cause almost a full second worth of\nerror, its possible to get almost 2 seconds of error just from the\nsuspend/resume cycle alone, so this about equal to any offset added by\nthe compensation.\n\nFurther on systems that suspend/resume frequently, this should keep time\ncloser then NTP could compensate for if the errors were allowed to\naccumulate.\n\nCredits to Arve Hjønnevåg for suggesting this solution.\n\nCC: Arve Hjønnevåg \u003carve@android.com\u003e\nCC: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nSigned-off-by: John Stultz \u003cjohn.stultz@linaro.org\u003e\n" }, { "commit": "cb5de2f8d0306be38f9b377b8a5c56acca7dbc3d", "tree": "9edd5bd1fd1e41be44026a37b2184200f6f90d50", "parents": [ "e08f6d4131ab964420f0bcabecc68d75fb49df79" ], "author": { "name": "John Stultz", "email": "john.stultz@linaro.org", "time": "Wed Jun 01 18:18:09 2011 -0700" }, "committer": { "name": "John Stultz", "email": "john.stultz@linaro.org", "time": "Tue Jun 21 16:55:36 2011 -0700" }, "message": "time: Catch invalid timespec sleep values in __timekeeping_inject_sleeptime\n\nArve suggested making sure we catch possible negative sleep time\nintervals that could be passed into timekeeping_inject_sleeptime.\n\nCC: Arve Hjønnevåg \u003carve@android.com\u003e\nCC: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nSigned-off-by: John Stultz \u003cjohn.stultz@linaro.org\u003e\n" }, { "commit": "99ee5315dac6211e972fa3f23bcc9a0343ff58c4", "tree": "6663d6ceaabcb9bac03193e2781cdbe6a139f70c", "parents": [ "b12a03ce4880bd13786a98db6de494a3e0123129" ], "author": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Wed Apr 27 14:16:42 2011 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Mon May 02 21:39:15 2011 +0200" }, "message": "timerfd: Allow timers to be cancelled when clock was set\n\nSome applications must be aware of clock realtime being set\nbackward. A simple example is a clock applet which arms a timer for\nthe next minute display. If clock realtime is set backward then the\napplet displays a stale time for the amount of time which the clock\nwas set backwards. Due to that applications poll the time because we\ndon\u0027t have an interface.\n\nExtend the timerfd interface by adding a flag which puts the timer\nonto a different internal realtime clock. All timers on this clock are\nexpired whenever the clock was set.\n\nThe timerfd core records the monotonic offset when the timer is\ncreated. When the timer is armed, then the current offset is compared\nto the previous recorded offset. When it has changed, then\ntimerfd_settime returns -ECANCELED. When a timer is read the offset is\ncompared and if it changed -ECANCELED returned to user space. Periodic\ntimers are not rearmed in the cancelation case.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nAcked-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Chris Friesen \u003cchris.friesen@genband.com\u003e\nTested-by: Kay Sievers \u003ckay.sievers@vrfy.org\u003e\nCc: \"Kirill A. Shutemov\" \u003ckirill@shutemov.name\u003e\nCc: Peter Zijlstra \u003cpeterz@infradead.org\u003e\nCc: Davide Libenzi \u003cdavidel@xmailserver.org\u003e\nReviewed-by: Alexander Shishkin \u003cvirtuoso@slind.org\u003e\nLink: http://lkml.kernel.org/r/%3Calpine.LFD.2.02.1104271359580.3323%40ionos%3E\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "b12a03ce4880bd13786a98db6de494a3e0123129", "tree": "2a8f2cf0d1d6749b4a33cc07c9b7839a4a139e9a", "parents": [ "942c3c5c329274fa6de5998cb911cf3d0a42d0b1" ], "author": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Mon May 02 16:48:57 2011 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Mon May 02 21:37:58 2011 +0200" }, "message": "hrtimers: Prepare for cancel on clock was set timers\n\nMake clock_was_set() unconditional and rename hres_timers_resume to\nhrtimers_resume. This is a preparatory patch for hrtimers which are\ncancelled when clock realtime was set.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "304529b1b6f8612ccbb4582e997051b48b94f4a4", "tree": "025b53989f39b463621e9a9eb8e903dc54dd87e1", "parents": [ "4175242c0dc1d92e821d2b45bc8fe06d4650a7b5" ], "author": { "name": "John Stultz", "email": "john.stultz@linaro.org", "time": "Fri Apr 01 14:32:09 2011 -0700" }, "committer": { "name": "John Stultz", "email": "john.stultz@linaro.org", "time": "Tue Apr 26 14:01:41 2011 -0700" }, "message": "time: Add timekeeping_inject_sleeptime\n\nSome platforms cannot implement read_persistent_clock, as\ntheir RTC devices are only accessible when interrupts are enabled.\nThis keeps them from being used by the timekeeping code on resume\nto measure the time in suspend.\n\nThe RTC layer tries to work around this, by calling do_settimeofday\non resume after irqs are reenabled to set the time properly. However,\nthis only corrects CLOCK_REALTIME, and does not properly adjust\nthe sleep time value. This causes btime in /proc/stat to be incorrect\nas well as making the new CLOCK_BOTTTIME inaccurate.\n\nThis patch resolves the issue by introducing a new timekeeping hook\nto allow the RTC layer to inject the sleep time on resume.\n\nThe code also checks to make sure that read_persistent_clock is\nnonfunctional before setting the sleep time, so that should the RTC\u0027s\nHCTOSYS option be configured in on a system that does support\nread_persistent_clock we will not increase the total_sleep_time twice.\n\nCC: Arve Hjønnevåg \u003carve@android.com\u003e\nCC: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nAcked-by: Arnd Bergmann \u003carnd@arndb.de\u003e\nSigned-off-by: John Stultz \u003cjohn.stultz@linaro.org\u003e\n" }, { "commit": "e1a85b2c519551d4792180cdab4074d7e99bf2c9", "tree": "ee53aed254043d06a7e30151af6500eab3bf0407", "parents": [ "f3c6ea1b06c71b43f751b36bd99345369fe911af" ], "author": { "name": "Rafael J. Wysocki", "email": "rjw@sisk.pl", "time": "Wed Mar 23 22:16:04 2011 +0100" }, "committer": { "name": "Rafael J. Wysocki", "email": "rjw@sisk.pl", "time": "Wed Mar 23 22:16:04 2011 +0100" }, "message": "timekeeping: Use syscore_ops instead of sysdev class and sysdev\n\nThe timekeeping subsystem uses a sysdev class and a sysdev for\nexecuting timekeeping_suspend() after interrupts have been turned off\non the boot CPU (during system suspend) and for executing\ntimekeeping_resume() before turning on interrupts on the boot CPU\n(during system resume). However, since both of these functions\nignore their arguments, the entire mechanism may be replaced with a\nstruct syscore_ops object which is simpler.\n\nSigned-off-by: Rafael J. Wysocki \u003crjw@sisk.pl\u003e\nReviewed-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "314ac37150011ebb398f522db528d2dbcc611189", "tree": "0731fced7713d3889bd783577673068a6bc93796", "parents": [ "abb3a4ea2e0ea7114a4475745da2f32bd9ad5b73" ], "author": { "name": "John Stultz", "email": "john.stultz@linaro.org", "time": "Mon Feb 14 18:43:08 2011 -0800" }, "committer": { "name": "John Stultz", "email": "john.stultz@linaro.org", "time": "Mon Feb 21 12:53:07 2011 -0800" }, "message": "time: Extend get_xtime_and_monotonic_offset() to also return sleep\n\nExtend get_xtime_and_monotonic_offset to\nget_xtime_and_monotonic_and_sleep_offset().\n\nCC: Jamie Lokier \u003cjamie@shareable.org\u003e\nCC: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nCC: Alexander Shishkin \u003cvirtuoso@slind.org\u003e\nCC: Arve Hjønnevåg \u003carve@android.com\u003e\nSigned-off-by: John Stultz \u003cjohn.stultz@linaro.org\u003e\n" }, { "commit": "abb3a4ea2e0ea7114a4475745da2f32bd9ad5b73", "tree": "ba78d928a1b58faf16c2507777b8430d54666f52", "parents": [ "e06383db9ec591696a06654257474b85bac1f8cb" ], "author": { "name": "John Stultz", "email": "john.stultz@linaro.org", "time": "Mon Feb 14 17:52:09 2011 -0800" }, "committer": { "name": "John Stultz", "email": "john.stultz@linaro.org", "time": "Mon Feb 21 12:53:05 2011 -0800" }, "message": "time: Introduce get_monotonic_boottime and ktime_get_boottime\n\nThis adds new functions that return the monotonic time since boot\n(in other words, CLOCK_MONOTONIC + suspend time).\n\nCC: Jamie Lokier \u003cjamie@shareable.org\u003e\nCC: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nCC: Alexander Shishkin \u003cvirtuoso@slind.org\u003e\nCC: Arve Hjønnevåg \u003carve@android.com\u003e\nSigned-off-by: John Stultz \u003cjohn.stultz@linaro.org\u003e\n" }, { "commit": "c528f7c6c208f1fae6b4025957173dec045e5f21", "tree": "86986bf39f82622ea81179dad3382fc4580875a8", "parents": [ "0061748dd2400d0bcd4d49d258db5d7b5d106ca0" ], "author": { "name": "John Stultz", "email": "john.stultz@linaro.org", "time": "Tue Feb 01 13:52:17 2011 +0000" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Wed Feb 02 15:28:18 2011 +0100" }, "message": "time: Introduce timekeeping_inject_offset\n\nThis adds a kernel-internal timekeeping interface to add or subtract\na fixed amount from CLOCK_REALTIME. This makes it so kernel users or\ninterfaces trying to do so do not have to read the time, then add an\noffset and then call settimeofday(), which adds some extra error in\ncomparision to just simply adding the offset in the kernel timekeeping\ncore.\n\nSigned-off-by: John Stultz \u003cjohn.stultz@linaro.org\u003e\nSigned-off-by: Richard Cochran \u003crichard.cochran@omicron.at\u003e\nLKML-Reference: \u003c20110201134419.584311693@linutronix.de\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "1e6d767924c74929c0cfe839ae8f37bcee9e544e", "tree": "4ace06971e2b3519e556bea2f7e3e999e860eedd", "parents": [ "7cf37e87dd2cfa17a64f28ea7f31eed4525f79e4" ], "author": { "name": "Richard Cochran", "email": "richard.cochran@omicron.at", "time": "Tue Feb 01 13:50:58 2011 +0000" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Wed Feb 02 15:28:11 2011 +0100" }, "message": "time: Correct the *settime* parameters\n\nBoth settimeofday() and clock_settime() promise with a \u0027const\u0027\nattribute not to alter the arguments passed in. This patch adds the\nmissing \u0027const\u0027 attribute into the various kernel functions\nimplementing these calls.\n\nSigned-off-by: Richard Cochran \u003crichard.cochran@omicron.at\u003e\nAcked-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nLKML-Reference: \u003c20110201134417.545698637@linutronix.de\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "f0af911a9dec9de702645182c8d269449e24d24b", "tree": "17cd5683ab3878479b59d9786102fcf699d3051a", "parents": [ "79ecaf0d15344d78904becf0f25de3fc9b49d430" ], "author": { "name": "Torben Hohn", "email": "torbenh@gmx.de", "time": "Thu Jan 27 15:59:10 2011 +0100" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Mon Jan 31 14:55:43 2011 +0100" }, "message": "time: Provide xtime_update()\n\nxtime_update() takes xtime_lock write locked and calls\ndo_timer(). Provided to replace the do_timer() calls in the\narchitecture code.\n\nSigned-off-by: Torben Hohn \u003ctorbenh@gmx.de\u003e\nCc: Peter Zijlstra \u003cpeterz@infradead.org\u003e\nCc: johnstul@us.ibm.com\nCc: yong.zhang0@gmail.com\nCc: hch@infradead.org\nLKML-Reference: \u003c20110127145910.23248.21379.stgit@localhost\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "79ecaf0d15344d78904becf0f25de3fc9b49d430", "tree": "1fb8b5f73c86dbb2636c9a6e3458212f0843e378", "parents": [ "48cf76f7104f655bbd48a75c7759dce82c3e1ab6" ], "author": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Mon Jan 31 11:07:54 2011 +0100" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Mon Jan 31 14:55:43 2011 +0100" }, "message": "time: Remove unused __get_wall_to_monotonic()\n\nNo users left. Remove it.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "48cf76f7104f655bbd48a75c7759dce82c3e1ab6", "tree": "2b3cf138fd277ba9c551d75ccbe575a40974560a", "parents": [ "fbad1ea94159a71bc0f68b00e57ae803606af9fb" ], "author": { "name": "Torben Hohn", "email": "torbenh@gmx.de", "time": "Thu Jan 27 15:59:05 2011 +0100" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Mon Jan 31 14:55:42 2011 +0100" }, "message": "time: Provide get_xtime_and_monotonic_offset()\n\nThe hrtimer code accesses timekeeping variables under\nxtime_lock. Provide a sensible accessor function and use it.\n\n[ tglx: Removed the conditionals, unused variable, fixed codingstyle\n \tand massaged changelog ]\n\nSigned-off-by: Torben Hohn \u003ctorbenh@gmx.de\u003e\nCc: Peter Zijlstra \u003cpeterz@infradead.org\u003e\nCc: johnstul@us.ibm.com\nCc: yong.zhang0@gmail.com\nCc: hch@infradead.org\nLKML-Reference: \u003c20110127145905.23248.30458.stgit@localhost\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "871cf1e5f2a17702f58539a3af8b18fc8666ad4c", "tree": "da80fbad2c79dee33d4ac559c8881ce4c77476b7", "parents": [ "70d1f365568e0cdbc9f4ab92428e1830fdb09ab0" ], "author": { "name": "Torben Hohn", "email": "torbenh@gmx.de", "time": "Thu Jan 27 15:58:55 2011 +0100" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Mon Jan 31 14:55:41 2011 +0100" }, "message": "time: Move do_timer() to kernel/time/timekeeping.c\n\ndo_timer() is primary timekeeping related. calc_global_load() is\ncalled from do_timer() as well, but that\u0027s more for historical\nreasons.\n\n[ tglx: Fixed up the calc_global_load() reject andmassaged changelog ]\n\nSigned-off-by: Torben Hohn \u003ctorbenh@gmx.de\u003e\nCc: Peter Zijlstra \u003cpeterz@infradead.org\u003e\nCc: johnstul@us.ibm.com\nCc: yong.zhang0@gmail.com\nCc: hch@infradead.org\nLKML-Reference: \u003c20110127145855.23248.56933.stgit@localhost\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "f9ee7f60d6f37ae0184812b4c59b3869f875768b", "tree": "6c30a7e94b90cfa3a43cda45f4bdb2f254f3ba10", "parents": [ "16c1020362083b320868c0deef492249089c3cd3", "1161ec944916069ceec21c487e30247d9ff22857", "76d1f7bfcd5872056902c5a88b5fcd5d4d00a7a9", "afa14e7c553ebe45844d76208f66017a43abd0e2", "7c46d8da09df22361d1d43465c4f1b06cecaf25f" ], "author": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Sat Jan 15 12:45:00 2011 -0800" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Sat Jan 15 12:45:00 2011 -0800" }, "message": "Merge branches \u0027core-fixes-for-linus\u0027, \u0027x86-fixes-for-linus\u0027, \u0027timers-fixes-for-linus\u0027 and \u0027perf-fixes-for-linus\u0027 of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip\n\n* \u0027core-fixes-for-linus\u0027 of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:\n rcu: avoid pointless blocked-task warnings\n rcu: demote SRCU_SYNCHRONIZE_DELAY from kernel-parameter status\n rtmutex: Fix comment about why new_owner can be NULL in wake_futex_pi()\n\n* \u0027x86-fixes-for-linus\u0027 of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:\n x86, olpc: Add missing Kconfig dependencies\n x86, mrst: Set correct APB timer IRQ affinity for secondary cpu\n x86: tsc: Fix calibration refinement conditionals to avoid divide by zero\n x86, ia64, acpi: Clean up x86-ism in drivers/acpi/numa.c\n\n* \u0027timers-fixes-for-linus\u0027 of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:\n timekeeping: Make local variables static\n time: Rename misnamed minsec argument of clocks_calc_mult_shift()\n\n* \u0027perf-fixes-for-linus\u0027 of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:\n tracing: Remove syscall_exit_fields\n tracing: Only process module tracepoints once\n perf record: Add \"nodelay\" mode, disabled by default\n perf sched: Fix list of events, dropping unsupported \u0027:r\u0027 modifier\n Revert \"perf tools: Emit clearer message for sys_perf_event_open ENOENT return\"\n perf top: Fix annotate segv\n perf evsel: Fix order of event list deletion\n" }, { "commit": "e2c18e49a0d4f822ffc29fb4958943beb1ff08b7", "tree": "227dbb544277f7908fa1666ce21a15e6b2282ccb", "parents": [ "025b40abe715d638e60516a657d354e8560c1a85" ], "author": { "name": "Alexander Gordeev", "email": "lasaine@lvk.cs.msu.su", "time": "Wed Jan 12 17:00:57 2011 -0800" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Thu Jan 13 08:03:21 2011 -0800" }, "message": "pps: capture MONOTONIC_RAW timestamps as well\n\nMONOTONIC_RAW clock timestamps are ideally suited for frequency\ncalculation and also fit well into the original NTP hardpps design. Now\nphase and frequency can be adjusted separately: the former based on\nREALTIME clock and the latter based on MONOTONIC_RAW clock.\n\nA new function getnstime_raw_and_real is added to timekeeping subsystem to\ncapture both timestamps at the same time and atomically.\n\nSigned-off-by: Alexander Gordeev \u003clasaine@lvk.cs.msu.su\u003e\nAcked-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Rodolfo Giometti \u003cgiometti@enneenne.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n" }, { "commit": "afa14e7c553ebe45844d76208f66017a43abd0e2", "tree": "2f1d13e996ba3db9be9d34b793edb432f6368bdd", "parents": [ "5fdade95f26372154459347dfb9f60721d22cfc7" ], "author": { "name": "H Hartley Sweeten", "email": "hartleys@visionengravers.com", "time": "Tue Jan 11 17:59:38 2011 -0600" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Wed Jan 12 12:23:13 2011 +0100" }, "message": "timekeeping: Make local variables static\n\nSigned-off-by: H Hartley Sweeten \u003chsweeten@visionengravers.com\u003e\nCc: John Stultz \u003cjohnstul@us.ibm.com\u003e\nLKML-Reference: \u003c0D753D10438DA54287A00B027084269764CE0E54B7@AUSP01VMBX24.collaborationhost.net\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n\n" }, { "commit": "a386b5af8edda1c742ce9f77891e112eefffc005", "tree": "8a74b512d751758144b219988f60d69a5fd028f9", "parents": [ "8af3c153baf95374eff20a37f00c59a295b52756" ], "author": { "name": "Kasper Pedersen", "email": "kkp2010@kasperkp.dk", "time": "Wed Oct 20 15:55:15 2010 -0700" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Thu Oct 21 17:30:05 2010 +0200" }, "message": "time: Compensate for rounding on odd-frequency clocksources\n\nWhen the clocksource is not a multiple of HZ, the clock will be off. For\nacpi_pm, HZ\u003d1000 the error is 127.111 ppm:\n\nThe rounding of cycle_interval ends up generating a false error term in\nntp_error accumulation since xtime_interval is not exactly 1/HZ. So, we\nsubtract out the error caused by the rounding.\n\nThis has been visible since 2.6.32-rc2\n\tcommit a092ff0f90cae22b2ac8028ecd2c6f6c1a9e4601\n\ttime: Implement logarithmic time accumulation\nThat commit raised NTP_INTERVAL_FREQ and exposed the rounding error.\n\ntesting tool: http://n1.taur.dk/permanent/testpmt.c\nAlso tested with ntpd and a frequency counter.\n\nSigned-off-by: Kasper Pedersen \u003ckkp2010@kasperkp.dk\u003e\nAcked-by: john stultz \u003cjohnstul@us.ibm.com\u003e\nCc: John Kacur \u003cjkacur@redhat.com\u003e\nCc: Clark Williams \u003cwilliams@redhat.com\u003e\nCc: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "c7dcf87a6881bf796faee83003163eb3de41a309", "tree": "1e4333e588ad6e8e69e310d8ae71798af1c0bdb5", "parents": [ "2be1f3a73dd02e38e181cf5abacb3d45a6a2d6b8" ], "author": { "name": "John Stultz", "email": "johnstul@us.ibm.com", "time": "Fri Aug 13 11:30:58 2010 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Fri Aug 13 12:03:24 2010 -0700" }, "message": "time: Workaround gcc loop optimization that causes 64bit div errors\n\nEarly 4.3 versions of gcc apparently aggressively optimize the raw\ntime accumulation loop, replacing it with a divide.\n\nOn 32bit systems, this causes the following link errors:\n\tundefined reference to `__umoddi3\u0027\n\tundefined reference to `__udivdi3\u0027\n\nThe gcc issue has been fixed in 4.4 and greater.\n\nThis patch replaces the accumulation loop with a do_div, as suggested\nby Linus.\n\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCC: Jason Wessel \u003cjason.wessel@windriver.com\u003e\nCC: Larry Finger \u003cLarry.Finger@lwfinger.net\u003e\nCC: Ingo Molnar \u003cmingo@elte.hu\u003e\nCC: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n" }, { "commit": "deda2e81961e96be4f2c09328baca4710a2fd1a0", "tree": "da1968f7e6791cb96735e8f2c602aa55e76fec21", "parents": [ "95f4572737da469b66ebfd1badd69d9adc340409" ], "author": { "name": "Jason Wessel", "email": "jason.wessel@windriver.com", "time": "Mon Aug 09 14:20:09 2010 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Thu Aug 12 09:53:39 2010 -0700" }, "message": "timekeeping: Fix overflow in rawtime tv_nsec on 32 bit archs\n\nThe tv_nsec is a long and when added to the shifted interval it can wrap\nand become negative which later causes looping problems in the\ngetrawmonotonic(). The edge case occurs when the system has slept for\na short period of time of ~2 seconds.\n\nA trace printk of the values in this patch illustrate the problem:\n\nftrace time stamp: log\n43.716079: logarithmic_accumulation: raw: 3d0913 tv_nsec d687faa\n43.718513: logarithmic_accumulation: raw: 3d0913 tv_nsec da588bd\n43.722161: logarithmic_accumulation: raw: 3d0913 tv_nsec de291d0\n46.349925: logarithmic_accumulation: raw: 7a122600 tv_nsec e1f9ae3\n46.349930: logarithmic_accumulation: raw: 1e848980 tv_nsec 8831c0e3\n\nThe kernel starts looping at 46.349925 in the getrawmonotonic() due to\nthe negative value from adding the raw value to tv_nsec.\n\nA simple solution is to accumulate into a u64, and then normalize it\nto a timespec_t.\n\nSigned-off-by: Jason Wessel \u003cjason.wessel@windriver.com\u003e\n [ Reworked variable names and simplified some of the code. - John ]\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nCc: H. Peter Anvin \u003chpa@zytor.com\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n" }, { "commit": "0fb86b06298b6cd3205cac2e68a499f269282dac", "tree": "3adcd4149fbe643ea8cae88578e169d02b7a9e5f", "parents": [ "8ab4351a4c888016620f43bde605b3d0964af339" ], "author": { "name": "John Stultz", "email": "johnstul@us.ibm.com", "time": "Tue Jul 13 17:56:26 2010 -0700" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Tue Jul 27 12:40:55 2010 +0200" }, "message": "timekeeping: Make xtime and wall_to_monotonic static\n\nThis patch makes xtime and wall_to_monotonic static, as planned in\nDocumentation/feature-removal-schedule.txt. This will allow for\nfurther cleanups to the timekeeping core.\n\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nLKML-Reference: \u003c1279068988-21864-10-git-send-email-johnstul@us.ibm.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n\n" }, { "commit": "8ab4351a4c888016620f43bde605b3d0964af339", "tree": "9d92ee7fbf9391bfcfdc1e49fecc31b9fa4e03c0", "parents": [ "9f31f5774961a735687fee17953ab505b3df3abf" ], "author": { "name": "John Stultz", "email": "johnstul@us.ibm.com", "time": "Tue Jul 13 17:56:25 2010 -0700" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Tue Jul 27 12:40:55 2010 +0200" }, "message": "hrtimer: Cleanup direct access to wall_to_monotonic\n\nProvides an accessor function to replace hrtimer.c\u0027s\ndirect access of wall_to_monotonic.\n\nThis will allow wall_to_monotonic to be made static as\nplanned in Documentation/feature-removal-schedule.txt\n\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nLKML-Reference: \u003c1279068988-21864-9-git-send-email-johnstul@us.ibm.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "7615856ebfee52b080c22d263ca4debbd0df0ac1", "tree": "beb06e60af7560c28d5bad41f03a5238636e91fc", "parents": [ "06d518e3dfb25334282c7e38b4d7a4eada215f6d" ], "author": { "name": "John Stultz", "email": "johnstul@us.ibm.com", "time": "Tue Jul 13 17:56:23 2010 -0700" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Tue Jul 27 12:40:54 2010 +0200" }, "message": "timkeeping: Fix update_vsyscall to provide wall_to_monotonic offset\n\nupdate_vsyscall() did not provide the wall_to_monotoinc offset,\nso arch specific implementations tend to reference wall_to_monotonic\ndirectly. This limits future cleanups in the timekeeping core, so\nthis patch fixes the update_vsyscall interface to provide\nwall_to_monotonic, allowing wall_to_monotonic to be made static\nas planned in Documentation/feature-removal-schedule.txt\n\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nCc: Anton Blanchard \u003canton@samba.org\u003e\nCc: Paul Mackerras \u003cpaulus@samba.org\u003e\nCc: Tony Luck \u003ctony.luck@intel.com\u003e\nLKML-Reference: \u003c1279068988-21864-7-git-send-email-johnstul@us.ibm.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "592913ecb87a9e06f98ddb55b298f1a66bf94c6b", "tree": "0f63f5535fe517faf418e0f7e69800fcf1e4d8db", "parents": [ "ce3bf7ab22527183634a76512d9854a38615e4d5" ], "author": { "name": "John Stultz", "email": "johnstul@us.ibm.com", "time": "Tue Jul 13 17:56:20 2010 -0700" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Tue Jul 27 12:40:54 2010 +0200" }, "message": "time: Kill off CONFIG_GENERIC_TIME\n\nNow that all arches have been converted over to use generic time via\nclocksources or arch_gettimeoffset(), we can remove the GENERIC_TIME\nconfig option and simplify the generic code.\n\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nLKML-Reference: \u003c1279068988-21864-4-git-send-email-johnstul@us.ibm.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "ce3bf7ab22527183634a76512d9854a38615e4d5", "tree": "79e48c7c84618d968664bd660c75e03fc98e7d11", "parents": [ "8c73626ab28527b7eb7f3061c027fbfe530c488c" ], "author": { "name": "John Stultz", "email": "johnstul@us.ibm.com", "time": "Tue Jul 13 17:56:19 2010 -0700" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Tue Jul 27 12:40:53 2010 +0200" }, "message": "time: Implement timespec_add\n\nAfter accidentally misusing timespec_add_safe, I wanted to make sure\nwe don\u0027t accidently trip over that issue again, so I created a simple\ntimespec_add() function which we can use to replace the instances\nof timespec_add_safe() that don\u0027t want the overflow detection.\n\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nLKML-Reference: \u003c1279068988-21864-3-git-send-email-johnstul@us.ibm.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n\n" }, { "commit": "dbb6be6d5e974c42bbecd183effaa0df69e1dd8b", "tree": "5735cb47e70853d057a9881dd0ce44b83e88fa63", "parents": [ "6a867a395558a7f882d041783e4cdea6744ca2bf", "b57f95a38233a2e73b679bea4a5453a1cc2a1cc9" ], "author": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Mon May 10 11:59:37 2010 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Mon May 10 14:20:42 2010 +0200" }, "message": "Merge branch \u0027linus\u0027 into timers/core\n\nReason: Further posix_cpu_timer patches depend on mainline changes\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "6a867a395558a7f882d041783e4cdea6744ca2bf", "tree": "dfe350df25fba5fec6f7e1088b04d6b03f0974b3", "parents": [ "9ca7d8e6834c40a99622bbe4a88aaf64313ae43c" ], "author": { "name": "John Stultz", "email": "johnstul@us.ibm.com", "time": "Tue Apr 06 14:30:51 2010 -0700" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Tue Apr 13 12:43:42 2010 +0200" }, "message": "time: Remove xtime_cache\n\nWith the earlier logarithmic time accumulation patch, xtime will now\nalways be within one \"tick\" of the current time, instead of possibly\nhalf a second off.\n\nThis removes the need for the xtime_cache value, which always stored the\ntime at the last interrupt, so this patch cleans that up removing the\nxtime_cache related code.\n\nThis patch also addresses an issue with an earlier version of this change,\nwhere xtime_cache was normalizing xtime, which could in some cases be\nnot valid (ie: tv_nsec \u003d\u003d NSEC_PER_SEC). This is fixed by handling\nthe edge case in update_wall_time().\n\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Petr Titěra \u003cP.Titera@century.cz\u003e\nLKML-Reference: \u003c1270589451-30773-1-git-send-email-johnstul@us.ibm.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "830ec0458c390f29c6c99e1ff7feab9e36368d12", "tree": "81a344892bd7696a42d87b76189b59f4e1af71ef", "parents": [ "15365c108ea27598e265f8c13e7051d99ca5b0b9" ], "author": { "name": "John Stultz", "email": "johnstul@us.ibm.com", "time": "Thu Mar 18 14:47:30 2010 -0700" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Tue Mar 23 16:41:01 2010 +0100" }, "message": "time: Fix accumulation bug triggered by long delay.\n\nThe logarithmic accumulation done in the timekeeping has some overflow\nprotection that limits the max shift value. That means it will take\nmore then shift loops to accumulate all of the cycles. This causes\nthe shift decrement to underflow, which causes the loop to never exit.\n\nThe simplest fix would be simply to do a:\n\tif (shift)\n\t\tshift--;\n\nHowever that is not optimal, as we know the cycle offset is larger\nthen the interval \u003c\u003c shift, the above would make shift drop to zero,\nthen we would be spinning for quite awhile accumulating at interval\nchunks at a time.\n\nInstead, this patch only decreases shift if the offset is smaller\nthen cycle_interval \u003c\u003c shift. This makes sure we accumulate using\nthe largest chunks possible without overflowing tick_length, and limits\nthe number of iterations through the loop.\n\nThis issue was found and reported by Sonic Zhang, who also tested the fix.\nMany thanks your explanation and testing!\n\nReported-by: Sonic Zhang \u003csonic.adi@gmail.com\u003e\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nTested-by: Sonic Zhang \u003csonic.adi@gmail.com\u003e\nLKML-Reference: \u003c1268948850-5225-1-git-send-email-johnstul@us.ibm.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "e56425b135a8892d1e71ad5bb605d12c10efeb32", "tree": "9bb899e1b6b45a125bcc21554f3673878d932035", "parents": [ "786f8ba2e9449a7f01ec6bc35838d0a335921061", "6622e670b26fbaa9c4ae7772a4d2d0abd3414b51" ], "author": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Mon Mar 01 08:48:25 2010 -0800" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Mon Mar 01 08:48:25 2010 -0800" }, "message": "Merge branch \u0027timers-for-linus\u0027 of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip\n\n* \u0027timers-for-linus\u0027 of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:\n posix-timers.c: Don\u0027t export local functions\n clocksource: start CMT at clocksource resume\n clocksource: add suspend callback\n clocksource: add argument to resume callback\n ntp: Cleanup xtime references in ntp.c\n ntp: Make time_esterror and time_maxerror static\n" }, { "commit": "c93d89f3dbf0202bf19c07960ca8602b48c2f9a0", "tree": "1534d49d8baeab231dc9f3693780a605d5c32d2e", "parents": [ "6339204ecc2aa2067a99595522de0403f0854bb8" ], "author": { "name": "Jason Wang", "email": "jasowang@redhat.com", "time": "Wed Jan 27 19:13:40 2010 +0800" }, "committer": { "name": "Avi Kivity", "email": "avi@redhat.com", "time": "Tue Feb 09 19:20:15 2010 +0200" }, "message": "Export the symbol of getboottime and mmonotonic_to_bootbased\n\nExport getboottime and monotonic_to_bootbased in order to let them\ncould be used by following patch.\n\nCc: stable@kernel.org\nSigned-off-by: Jason Wang \u003cjasowang@redhat.com\u003e\nSigned-off-by: Marcelo Tosatti \u003cmtosatti@redhat.com\u003e\n" }, { "commit": "c54a42b19fbaae4e9f212322ecca25a6bc95c1ba", "tree": "2c929d3cde3b4ae864d5f5a50bdda688fd5293eb", "parents": [ "17622339af2536b32cf29699ddd4ba0fe79a61d5" ], "author": { "name": "Magnus Damm", "email": "damm@opensource.se", "time": "Tue Feb 02 14:41:41 2010 -0800" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Fri Feb 05 14:54:10 2010 +0100" }, "message": "clocksource: add suspend callback\n\nAdd a clocksource suspend callback. This callback can be used by the\nclocksource driver to shutdown and perform any kind of late suspend\nactivities even though the clocksource driver itself is a non-sysdev\ndriver.\n\nOne example where this is useful is to fix the sh_cmt.c platform driver\nthat today suspends using the platform bus and shuts down the clocksource\ntoo early.\n\nWith this callback in place the sh_cmt driver will suspend using the\nclocksource and clockevent hooks and leave the platform device pm\ncallbacks unused.\n\nSigned-off-by: Magnus Damm \u003cdamm@opensource.se\u003e\nCc: Paul Mundt \u003clethal@linux-sh.org\u003e\nCc: john stultz \u003cjohnstul@us.ibm.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "83f57a11d84460dfe2afdb5a8bc759953428e38b", "tree": "b8506a6b61eb23e3d203fb248776c7eab9270c2d", "parents": [ "0e2c8b8f55072a98b99e7bdad55c912084d6a526" ], "author": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Tue Dec 22 14:10:37 2009 -0800" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Tue Dec 22 14:10:37 2009 -0800" }, "message": "Revert \"time: Remove xtime_cache\"\n\nThis reverts commit 7bc7d637452383d56ba4368d4336b0dde1bb476d, as\nrequested by John Stultz. Quoting John:\n\n \"Petr Titěra reported an issue where he saw odd atime regressions with\n 2.6.33 where there were a full second worth of nanoseconds in the\n nanoseconds field.\n\n He also reviewed the time code and narrowed down the problem: unhandled\n overflow of the nanosecond field caused by rounding up the\n sub-nanosecond accumulated time.\n\n Details:\n\n * At the end of update_wall_time(), we currently round up the\n sub-nanosecond portion of accumulated time when storing it into xtime.\n This was added to avoid time inconsistencies caused when the\n sub-nanosecond portion was truncated when storing into xtime.\n Unfortunately we don\u0027t handle the possible second overflow caused by\n that rounding.\n\n * Previously the xtime_cache code hid this overflow by normalizing the\n xtime value when storing into the xtime_cache.\n\n * We could try to handle the second overflow after the rounding up, but\n since this affects the timekeeping\u0027s internal state, this would further\n complicate the next accumulation cycle, causing small errors in ntp\n steering. As much as I\u0027d like to get rid of it, the xtime_cache code is\n known to work.\n\n * The correct fix is really to include the sub-nanosecond portion in the\n timekeeping accessor function, so we don\u0027t need to round up at during\n accumulation. This would greatly simplify the accumulation code.\n Unfortunately, we can\u0027t do this safely until the last three\n non-GENERIC_TIME arches (sparc32, arm, cris) are converted (those\n patches are in -mm) and we kill off the spots where arches set xtime\n directly. This is all 2.6.34 material, so I think reverting the\n xtime_cache change is the best approach for now.\n\n Many thanks to Petr for both reporting and finding the issue!\"\n\nReported-by: Petr Titěra \u003cP.Titera@century.cz\u003e\nRequested-by: john stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Ingo Molnar \u003cmingo@elte.hu\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n" }, { "commit": "fbf07eac7bf21c262143194181bd97c5d18b8ceb", "tree": "c9228b7d272f1e39472da28032605ecd1860e688", "parents": [ "60d8ce2cd6c283132928c11f3fd57ff4187287e0", "8629ea2eaba8ca0de2e38ce1b4a825e16255976e" ], "author": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Tue Dec 08 19:28:09 2009 -0800" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Tue Dec 08 19:28:09 2009 -0800" }, "message": "Merge branch \u0027timers-for-linus-urgent\u0027 of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip\n\n* \u0027timers-for-linus-urgent\u0027 of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:\n hrtimer: Fix /proc/timer_list regression\n itimers: Fix racy writes to cpu_itimer fields\n timekeeping: Fix clock_gettime vsyscall time warp\n" }, { "commit": "60d8ce2cd6c283132928c11f3fd57ff4187287e0", "tree": "36d08a2ead7a7d8c3c081d484215ccca00bf6aab", "parents": [ "849e8dea099aafa56db9e74b580b0d858b956533", "feae3203d711db0a9965300ee6d592257fdaae4f" ], "author": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Tue Dec 08 19:27:08 2009 -0800" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Tue Dec 08 19:27:08 2009 -0800" }, "message": "Merge branch \u0027timers-for-linus\u0027 of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip\n\n* \u0027timers-for-linus\u0027 of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:\n timers, init: Limit the number of per cpu calibration bootup messages\n posix-cpu-timers: optimize and document timer_create callback\n clockevents: Add missing include to pacify sparse\n x86: vmiclock: Fix printk format\n x86: Fix printk format due to variable type change\n sparc: fix printk for change of variable type\n clocksource/events: Fix fallout of generic code changes\n nohz: Allow 32-bit machines to sleep for more than 2.15 seconds\n nohz: Track last do_timer() cpu\n nohz: Prevent clocksource wrapping during idle\n nohz: Type cast printk argument\n mips: Use generic mult/shift factor calculation for clocks\n clocksource: Provide a generic mult/shift factor calculation\n clockevents: Use u32 for mult and shift factors\n nohz: Introduce arch_needs_cpu\n nohz: Reuse ktime in sub-functions of tick_check_idle.\n time: Remove xtime_cache\n time: Implement logarithmic time accumulation\n" }, { "commit": "0696b711e4be45fa104c12329f617beb29c03f78", "tree": "96292fdc4ab443d6218077d363548532df6cbd8b", "parents": [ "a9366e61b03f55a6e009e687ad10e706714c9907" ], "author": { "name": "Lin Ming", "email": "ming.m.lin@intel.com", "time": "Tue Nov 17 13:49:50 2009 +0800" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Tue Nov 17 11:52:34 2009 +0100" }, "message": "timekeeping: Fix clock_gettime vsyscall time warp\n\nSince commit 0a544198 \"timekeeping: Move NTP adjusted clock multiplier\nto struct timekeeper\" the clock multiplier of vsyscall is updated with\nthe unmodified clock multiplier of the clock source and not with the\nNTP adjusted multiplier of the timekeeper.\n\nThis causes user space observerable time warps:\nnew CLOCK-warp maximum: 120 nsecs, 00000025c337c537 -\u003e 00000025c337c4bf\n\nAdd a new argument \"mult\" to update_vsyscall() and hand in the\ntimekeeping internal NTP adjusted multiplier.\n\nSigned-off-by: Lin Ming \u003cming.m.lin@intel.com\u003e\nCc: \"Zhang Yanmin\" \u003cyanmin_zhang@linux.intel.com\u003e\nCc: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nCc: Benjamin Herrenschmidt \u003cbenh@kernel.crashing.org\u003e\nCc: Tony Luck \u003ctony.luck@intel.com\u003e\nLKML-Reference: \u003c1258436990.17765.83.camel@minggr.sh.intel.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "98962465ed9e6ea99c38e0af63fe1dcb5a79dc25", "tree": "f3f69ad8f6cd47e72a75da6de49eb3402f15cd9b", "parents": [ "529eaccd900a59724619b4a6ef6579fd518d5218" ], "author": { "name": "Jon Hunter", "email": "jon-hunter@ti.com", "time": "Tue Aug 18 12:45:10 2009 -0500" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Fri Nov 13 20:46:24 2009 +0100" }, "message": "nohz: Prevent clocksource wrapping during idle\n\nThe dynamic tick allows the kernel to sleep for periods longer than a\nsingle tick, but it does not limit the sleep time currently. In the\nworst case the kernel could sleep longer than the wrap around time of\nthe time keeping clock source which would result in losing track of\ntime.\n\nPrevent this by limiting it to the safe maximum sleep time of the\ncurrent time keeping clock source. The value is calculated when the\nclock source is registered.\n\n[ tglx: simplified the code a bit and massaged the commit msg ]\n\nSigned-off-by: Jon Hunter \u003cjon-hunter@ti.com\u003e\nCc: John Stultz \u003cjohnstul@us.ibm.com\u003e\nLKML-Reference: \u003c1250617512-23567-2-git-send-email-jon-hunter@ti.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "d43c36dc6b357fa1806800f18aa30123c747a6d1", "tree": "339ce510073ecbe9b3592008f7dece7b277035ef", "parents": [ "69585dd69e663a40729492c7b52eb82477a2027a" ], "author": { "name": "Alexey Dobriyan", "email": "adobriyan@gmail.com", "time": "Wed Oct 07 17:09:06 2009 +0400" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Sun Oct 11 11:20:58 2009 -0700" }, "message": "headers: remove sched.h from interrupt.h\n\nAfter m68k\u0027s task_thread_info() doesn\u0027t refer to current,\nit\u0027s possible to remove sched.h from interrupt.h and not break m68k!\nMany thanks to Heiko Carstens for allowing this.\n\nSigned-off-by: Alexey Dobriyan \u003cadobriyan@gmail.com\u003e\n" }, { "commit": "7bc7d637452383d56ba4368d4336b0dde1bb476d", "tree": "f1c8e1e103a900850342d46de27ebdf1e7d07f78", "parents": [ "a092ff0f90cae22b2ac8028ecd2c6f6c1a9e4601" ], "author": { "name": "john stultz", "email": "johnstul@us.ibm.com", "time": "Fri Oct 02 16:24:15 2009 -0700" }, "committer": { "name": "Ingo Molnar", "email": "mingo@elte.hu", "time": "Mon Oct 05 13:52:02 2009 +0200" }, "message": "time: Remove xtime_cache\n\nWith the prior logarithmic time accumulation patch, xtime will now\nalways be within one \"tick\" of the current time, instead of\npossibly half a second off.\n\nThis removes the need for the xtime_cache value, which always\nstored the time at the last interrupt, so this patch cleans that up\nremoving the xtime_cache related code.\n\nThis is a bit simpler, but still could use some wider testing.\n\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nAcked-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nReviewed-by: John Kacur \u003cjkacur@redhat.com\u003e\nCc: Clark Williams \u003cwilliams@redhat.com\u003e\nCc: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nCc: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nLKML-Reference: \u003c1254525855.7741.95.camel@localhost.localdomain\u003e\nSigned-off-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n" }, { "commit": "a092ff0f90cae22b2ac8028ecd2c6f6c1a9e4601", "tree": "79ec451d0bcdf6c08e0bc210b4beed694fbbf4a9", "parents": [ "8a0382f6fceaf0c6479e582e1054f36333ea3d24" ], "author": { "name": "john stultz", "email": "johnstul@us.ibm.com", "time": "Fri Oct 02 16:17:53 2009 -0700" }, "committer": { "name": "Ingo Molnar", "email": "mingo@elte.hu", "time": "Mon Oct 05 13:51:48 2009 +0200" }, "message": "time: Implement logarithmic time accumulation\n\nAccumulating one tick at a time works well unless we\u0027re using NOHZ.\nThen it can be an issue, since we may have to run through the loop\na few thousand times, which can increase timer interrupt caused\nlatency.\n\nThe current solution was to accumulate in half-second intervals\nwith NOHZ. This kept the number of loops down, however it did\nslightly change how we make NTP adjustments. While not an issue\nwith NTPd users, as NTPd makes adjustments over a longer period of\ntime, other adjtimex() users have noticed the half-second\ngranularity with which we can apply frequency changes to the clock.\n\nFor instance, if a application tries to apply a 100ppm frequency\ncorrection for 20ms to correct a 2us offset, with NOHZ they either\nget no correction, or a 50us correction.\n\nNow, there will always be some granularity error for applying\nfrequency corrections. However with users sensitive to this error\nhave seen a 50-500x increase with NOHZ compared to running without\nNOHZ.\n\nSo I figured I\u0027d try another approach then just simply increasing\nthe interval. My approach is to consume the time interval\nlogarithmically. This reduces the number of times through the loop\nneeded keeping latency down, while still preserving the original\ngranularity error for adjtimex() changes.\n\nFurther, this change allows us to remove the xtime_cache code\n(patch to follow), as xtime is always within one tick of the\ncurrent time, instead of the half-second updates it saw before.\n\nAn earlier version of this patch has been shipping to x86 users in\nthe RedHat MRG releases for awhile without issue, but I\u0027ve reworked\nthis version to be even more careful about avoiding possible\noverflows if the shift value gets too large.\n\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nAcked-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nReviewed-by: John Kacur \u003cjkacur@redhat.com\u003e\nCc: Clark Williams \u003cwilliams@redhat.com\u003e\nCc: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nCc: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nLKML-Reference: \u003c1254525473.7741.88.camel@localhost.localdomain\u003e\nSigned-off-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n" }, { "commit": "36d47481b3824b661b464077db95d447984df799", "tree": "7ae9dc060872f879403e78ab1b1abd7f4ada4ed0", "parents": [ "0ceb4c3e3f1ccaf121851e33c3ea269b8ad0f219" ], "author": { "name": "Hiroshi Shimamoto", "email": "h-shimamoto@ct.jp.nec.com", "time": "Tue Aug 25 15:08:30 2009 +0900" }, "committer": { "name": "Ingo Molnar", "email": "mingo@elte.hu", "time": "Tue Aug 25 09:09:02 2009 +0200" }, "message": "timekeeping: Fix invalid getboottime() value\n\nDon\u0027t use timespec_add_safe() with wall_to_monotonic, because\nwall_to_monotonic has negative values which will cause overflow\nin timespec_add_safe(). That makes btime in /proc/stat invalid.\n\nSigned-off-by: Hiroshi Shimamoto \u003ch-shimamoto@ct.jp.nec.com\u003e\nCc: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nCc: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Daniel Walker \u003cdwalker@fifo99.com\u003e\nLKML-Reference: \u003c4A937FDE.4050506@ct.jp.nec.com\u003e\nSigned-off-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n" }, { "commit": "da15cfdae03351c689736f8d142618592e3cebc3", "tree": "497fe3f77e27fa9cf0a484422c7bc382031df1bd", "parents": [ "8cab02dc3c58a12235c6d463ce684dded9696848" ], "author": { "name": "john stultz", "email": "johnstul@us.ibm.com", "time": "Wed Aug 19 19:13:34 2009 -0700" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Fri Aug 21 21:43:46 2009 +0200" }, "message": "time: Introduce CLOCK_REALTIME_COARSE\n\nAfter talking with some application writers who want very fast, but not\nfine-grained timestamps, I decided to try to implement new clock_ids\nto clock_gettime(): CLOCK_REALTIME_COARSE and CLOCK_MONOTONIC_COARSE\nwhich returns the time at the last tick. This is very fast as we don\u0027t\nhave to access any hardware (which can be very painful if you\u0027re using\nsomething like the acpi_pm clocksource), and we can even use the vdso\nclock_gettime() method to avoid the syscall. The only trade off is you\nonly get low-res tick grained time resolution.\n\nThis isn\u0027t a new idea, I know Ingo has a patch in the -rt tree that made\nthe vsyscall gettimeofday() return coarse grained time when the\nvsyscall64 sysctrl was set to 2. However this affects all applications\non a system.\n\nWith this method, applications can choose the proper speed/granularity\ntrade-off for themselves.\n\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Andi Kleen \u003candi@firstfloor.org\u003e\nCc: nikolag@ca.ibm.com\nCc: Darren Hart \u003cdvhltc@us.ibm.com\u003e\nCc: arjan@infradead.org\nCc: jonathan@jonmasters.org\nLKML-Reference: \u003c1250734414.6897.5.camel@localhost.localdomain\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n\n" }, { "commit": "23970e389e9cee43c4b41023935e1417271708b2", "tree": "d97d299cf74c44cf7ceb4c04c82dafc451fbbfbb", "parents": [ "d4f587c67fc39e0030ddd718675e252e208da4d7" ], "author": { "name": "Martin Schwidefsky", "email": "schwidefsky@de.ibm.com", "time": "Fri Aug 14 15:47:32 2009 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Sat Aug 15 10:55:47 2009 +0200" }, "message": "timekeeping: Introduce read_boot_clock\n\nAdd the new function read_boot_clock to get the exact time the system\nhas been started. For architectures without support for exact boot\ntime a new weak function is added that returns 0. Use the exact boot\ntime to initialize wall_to_monotonic, or xtime if the read_boot_clock\nreturned 0.\n\nSigned-off-by: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nCc: Ingo Molnar \u003cmingo@elte.hu\u003e\nAcked-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Daniel Walker \u003cdwalker@fifo99.com\u003e\nLKML-Reference: \u003c20090814134811.296703241@de.ibm.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "d4f587c67fc39e0030ddd718675e252e208da4d7", "tree": "2222681c57966cc6b8404afdff3ab1a09ac7ea69", "parents": [ "75c5158f70c065b9704b924503d96e8297838f79" ], "author": { "name": "Martin Schwidefsky", "email": "schwidefsky@de.ibm.com", "time": "Fri Aug 14 15:47:31 2009 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Sat Aug 15 10:55:46 2009 +0200" }, "message": "timekeeping: Increase granularity of read_persistent_clock()\n\nThe persistent clock of some architectures (e.g. s390) have a\nbetter granularity than seconds. To reduce the delta between the\nhost clock and the guest clock in a virtualized system change the \nread_persistent_clock function to return a struct timespec.\n\nSigned-off-by: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nCc: Ingo Molnar \u003cmingo@elte.hu\u003e\nAcked-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Daniel Walker \u003cdwalker@fifo99.com\u003e\nLKML-Reference: \u003c20090814134811.013873340@de.ibm.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "75c5158f70c065b9704b924503d96e8297838f79", "tree": "74b02ba1f13aaf8292bd472a8a197ac900ff20e7", "parents": [ "2ba2a3054fdffc8e6452f4ee120760322a6fbd43" ], "author": { "name": "Martin Schwidefsky", "email": "schwidefsky@de.ibm.com", "time": "Fri Aug 14 15:47:30 2009 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Sat Aug 15 10:55:46 2009 +0200" }, "message": "timekeeping: Update clocksource with stop_machine\n\nupdate_wall_time calls change_clocksource HZ times per second to check\nif a new clock source is available. In close to 100% of all calls\nthere is no new clock. Replace the tick based check by an update done\nwith stop_machine.\n\nSigned-off-by: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nCc: Ingo Molnar \u003cmingo@elte.hu\u003e\nAcked-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Daniel Walker \u003cdwalker@fifo99.com\u003e\nLKML-Reference: \u003c20090814134810.711836357@de.ibm.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "2ba2a3054fdffc8e6452f4ee120760322a6fbd43", "tree": "1ce2f3b323cb332d2b3d5ba4930ea10bb5d9e27e", "parents": [ "0a54419836254a27baecd9037103171bcbabaf67" ], "author": { "name": "Martin Schwidefsky", "email": "schwidefsky@de.ibm.com", "time": "Fri Aug 14 15:47:29 2009 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Sat Aug 15 10:55:46 2009 +0200" }, "message": "timekeeping: Add timekeeper read_clock helper functions\n\nAdd timekeeper_read_clock_ntp and timekeeper_read_clock_raw and use\nthem for getnstimeofday, ktime_get, ktime_get_ts and getrawmonotonic.\n\nSigned-off-by: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nCc: Ingo Molnar \u003cmingo@elte.hu\u003e\nAcked-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Daniel Walker \u003cdwalker@fifo99.com\u003e\nLKML-Reference: \u003c20090814134810.435105711@de.ibm.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "0a54419836254a27baecd9037103171bcbabaf67", "tree": "ad4dc7ca1fc974a53c7fc2e972558bcb6b786afe", "parents": [ "23ce72117c714baab794e66c8daf343bf6a912bf" ], "author": { "name": "Martin Schwidefsky", "email": "schwidefsky@de.ibm.com", "time": "Fri Aug 14 15:47:28 2009 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Sat Aug 15 10:55:46 2009 +0200" }, "message": "timekeeping: Move NTP adjusted clock multiplier to struct timekeeper\n\nThe clocksource structure has two multipliers, the unmodified multiplier\nclock-\u003emult_orig and the NTP corrected multiplier clock-\u003emult. The NTP\nmultiplier is misplaced in the struct clocksource, this is private\ninformation of the timekeeping code. Add the mult field to the struct\ntimekeeper to contain the NTP corrected value, keep the unmodifed\nmultiplier in clock-\u003emult and remove clock-\u003emult_orig.\n\nSigned-off-by: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nCc: Ingo Molnar \u003cmingo@elte.hu\u003e\nAcked-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Daniel Walker \u003cdwalker@fifo99.com\u003e\nLKML-Reference: \u003c20090814134810.149047645@de.ibm.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "23ce72117c714baab794e66c8daf343bf6a912bf", "tree": "ef2cfc1ffbf07c344ab28a6f5f2357e6b2c9a99d", "parents": [ "155ec60226ae0ae2aadaa57c951a58a359331030" ], "author": { "name": "Martin Schwidefsky", "email": "schwidefsky@de.ibm.com", "time": "Fri Aug 14 15:47:27 2009 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Sat Aug 15 10:55:46 2009 +0200" }, "message": "timekeeping: Add xtime_shift and ntp_error_shift to struct timekeeper\n\nThe xtime_nsec value in the timekeeper structure is shifted by a few\nbits to improve precision. This happens to be the same value as the\nclock-\u003eshift. To improve readability add xtime_shift to the timekeeper\nand use it instead of the clock-\u003eshift. Likewise add ntp_error_shift\nand replace all (NTP_SCALE_SHIFT - clock-\u003eshift) expressions.\n\nSigned-off-by: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nCc: Ingo Molnar \u003cmingo@elte.hu\u003e\nAcked-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Daniel Walker \u003cdwalker@fifo99.com\u003e\nLKML-Reference: \u003c20090814134809.871899606@de.ibm.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "155ec60226ae0ae2aadaa57c951a58a359331030", "tree": "fdee05f7b587f8d49cdd277abdbe44212279a4ba", "parents": [ "c55c87c892c1875deace0c8fc28787335277fdf2" ], "author": { "name": "Martin Schwidefsky", "email": "schwidefsky@de.ibm.com", "time": "Fri Aug 14 15:47:26 2009 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Sat Aug 15 10:55:46 2009 +0200" }, "message": "timekeeping: Introduce struct timekeeper\n\nAdd struct timekeeper to keep the internal values timekeeping.c needs\nin regard to the currently selected clock source. This moves the\ntimekeeping intervals, xtime_nsec and the ntp error value from struct\nclocksource to struct timekeeper. The raw_time is removed from the\nclocksource as well. It gets treated like xtime as a global variable.\nEventually xtime raw_time should be moved to struct timekeeper.\n\n[ tglx: minor cleanup ]\n\nSigned-off-by: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nCc: Ingo Molnar \u003cmingo@elte.hu\u003e\nAcked-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Daniel Walker \u003cdwalker@fifo99.com\u003e\nLKML-Reference: \u003c20090814134809.613209842@de.ibm.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "f1b82746c1e93daf24e1ab9bfbd39bcdb2e7018b", "tree": "1403e3662dd3b32cb8b17218bfdd9a640061f654", "parents": [ "1be396794897f80bfc8774719ba60309a9e3d374" ], "author": { "name": "Martin Schwidefsky", "email": "schwidefsky@de.ibm.com", "time": "Fri Aug 14 15:47:21 2009 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Sat Aug 15 10:55:46 2009 +0200" }, "message": "clocksource: Cleanup clocksource selection\n\nIf a non high-resolution clocksource is first set as override clock\nand then registered it becomes active even if the system is in one-shot\nmode. Move the override check from sysfs_override_clocksource to the\nclocksource selection. That fixes the bug and simplifies the code. The\ncheck in clocksource_register for double registration of the same\nclocksource is removed without replacement.\n\nTo find the initial clocksource a new weak function in jiffies.c is\ndefined that returns the jiffies clocksource. The architecture code\ncan then override the weak function with a more suitable clocksource,\ne.g. the TOD clock on s390.\n\n[ tglx: Folded in a fix from John Stultz ]\n\nSigned-off-by: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nAcked-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Daniel Walker \u003cdwalker@fifo99.com\u003e\nLKML-Reference: \u003c20090814134808.388024160@de.ibm.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "1be396794897f80bfc8774719ba60309a9e3d374", "tree": "f0e11ffe7b7cc2cf58129cf3ea1c543de0da9898", "parents": [ "a0f7d48bfb95a4c5172a2756dbc4b82afc8e9ae4" ], "author": { "name": "Martin Schwidefsky", "email": "schwidefsky@de.ibm.com", "time": "Fri Aug 14 15:47:20 2009 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Sat Aug 15 10:55:45 2009 +0200" }, "message": "timekeeping: Move reset of cycle_last for tsc clocksource to tsc\n\nchange_clocksource resets the cycle_last value to zero then sets it to\na value read from the clocksource. The reset to zero is required only\nfor the TSC clocksource to make the read_tsc function work after a\nresume. The reason is that the TSC read function uses cycle_last to\ndetect backwards going TSCs. In the resume case cycle_last contains\nthe TSC value from the last update before the suspend. On resume the\nTSC starts counting from 0 again and would trip over the cycle_last\ncomparison.\n\nThis is subtle and surprising. Move the reset to a resume function in\nthe tsc code.\n\nSigned-off-by: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nAcked-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nAcked-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Daniel Walker \u003cdwalker@fifo99.com\u003e\nLKML-Reference: \u003c20090814134808.142191175@de.ibm.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "a0f7d48bfb95a4c5172a2756dbc4b82afc8e9ae4", "tree": "b82580eb76f99ee1352f399975050630ca21936a", "parents": [ "31089c13bcb18d2cd2a3ddfbe3a28666346f237e" ], "author": { "name": "Martin Schwidefsky", "email": "schwidefsky@de.ibm.com", "time": "Fri Aug 14 15:47:19 2009 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Sat Aug 15 10:55:45 2009 +0200" }, "message": "timekeeping: Remove clocksource inline functions\n\nThe three inline functions clocksource_read, clocksource_enable and\nclocksource_disable are simple wrappers of an indirect call plus the\ncopy from and to the mult_orig value. The functions are exclusively\nused by the timekeeping code which has intimate knowledge of the\nclocksource anyway. Therefore remove the inline functions. No\nfunctional change.\n\nSigned-off-by: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nAcked-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Daniel Walker \u003cdwalker@fifo99.com\u003e\nLKML-Reference: \u003c20090814134807.903108946@de.ibm.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "31089c13bcb18d2cd2a3ddfbe3a28666346f237e", "tree": "9db4868a4c524efd6f927a555e43c009a70243a5", "parents": [ "4cd1993f0046fbc765dbf20af90966f5661e3789" ], "author": { "name": "John Stultz", "email": "johnstul@us.ibm.com", "time": "Fri Aug 14 15:47:18 2009 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Sat Aug 15 10:55:45 2009 +0200" }, "message": "timekeeping: Introduce timekeeping_leap_insert\n\nMove the adjustment of xtime, wall_to_monotonic and the update of the\nvsyscall variables to the timekeeping code.\n\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nSigned-off-by: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nLKML-Reference: \u003c20090814134807.609730216@de.ibm.com\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "a40f262cc21fbfd781bbddcc40b16b83a75f5f34", "tree": "1600f58173be05f561bcad045bca4014ab635d6f", "parents": [ "951ed4d36b77ba9fe1ea08fc3c59d8bb6c9bda32" ], "author": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Tue Jul 07 13:00:31 2009 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Tue Jul 07 13:00:31 2009 +0200" }, "message": "timekeeping: Move ktime_get() functions to timekeeping.c\n\nThe ktime_get() functions for GENERIC_TIME\u003dn are still located in\nhrtimer.c. Move them to time/timekeeping.c where they belong.\n\nLKML-Reference: \u003cnew-submission\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "951ed4d36b77ba9fe1ea08fc3c59d8bb6c9bda32", "tree": "870456aff2c513c00608f5416d1b7b440fa5f963", "parents": [ "faf80d62e44dc627efb741f48db50c1858d1667c" ], "author": { "name": "Martin Schwidefsky", "email": "schwidefsky@de.ibm.com", "time": "Tue Jul 07 11:27:28 2009 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Tue Jul 07 12:47:33 2009 +0200" }, "message": "timekeeping: optimized ktime_get[_ts] for GENERIC_TIME\u003dy\n\nThe generic ktime_get function defined in kernel/hrtimer.c is suboptimial\nfor GENERIC_TIME\u003dy:\n\n 0) | ktime_get() {\n 0) | ktime_get_ts() {\n 0) | getnstimeofday() {\n 0) | read_tod_clock() {\n 0) 0.601 us | }\n 0) 1.938 us | }\n 0) | set_normalized_timespec() {\n 0) 0.602 us | }\n 0) 4.375 us | }\n 0) 5.523 us | }\n\nOverall there are two read_seqbegin/read_seqretry loops and a lot of\nunnecessary struct timespec calculations. ktime_get returns a nano second\nvalue which is the sum of xtime, wall_to_monotonic and the nano second\ndelta from the clock source.\n\nktime_get can be optimized for GENERIC_TIME\u003dy. The new version only calls\nclocksource_read:\n\n 0) | ktime_get() {\n 0) | read_tod_clock() {\n 0) 0.610 us | }\n 0) 1.977 us | }\n\nIt uses a single read_seqbegin/readseqretry loop and just adds everthing\nto a nano second value.\n\nktime_get_ts is optimized in a similar fashion.\n\n[ tglx: added WARN_ON(timekeeping_suspended) as in getnstimeofday() ]\n\nSigned-off-by: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nAcked-by: john stultz \u003cjohnstul@us.ibm.com\u003e\nLKML-Reference: \u003c20090707112728.3005244d@skybase\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "cf9fe114e3b37e14fc8434d5abb192e35df551b1", "tree": "0f82879295dc792f9df1a3ce79e143a3c073510f", "parents": [ "c1d0d32a603ed06377f404adf2c538de33bb3634", "991ec02cdca33b03a132a0cacfe6f0aa0be9aa8d" ], "author": { "name": "Paul Mundt", "email": "lethal@linux-sh.org", "time": "Thu Jun 11 09:01:14 2009 +0300" }, "committer": { "name": "Paul Mundt", "email": "lethal@linux-sh.org", "time": "Thu Jun 11 09:01:14 2009 +0300" }, "message": "Merge branch \u0027master\u0027 of git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6\n" }, { "commit": "dce48a84adf1806676319f6f480e30a6daa012f9", "tree": "79151f5d31d9c3dcdc723ab8877cb943b944890e", "parents": [ "2ff799d3cff1ecb274049378b28120ee5c1c5e5f" ], "author": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Sat Apr 11 10:43:41 2009 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Fri May 15 15:32:45 2009 +0200" }, "message": "sched, timers: move calc_load() to scheduler\n\nDimitri Sivanich noticed that xtime_lock is held write locked across\ncalc_load() which iterates over all online CPUs. That can cause long\nlatencies for xtime_lock readers on large SMP systems. \n\nThe load average calculation is an rough estimate anyway so there is\nno real need to protect the readers vs. the update. It\u0027s not a problem\nwhen the avenrun array is updated while a reader copies the values.\n\nInstead of iterating over all online CPUs let the scheduler_tick code\nupdate the number of active tasks shortly before the avenrun update\nhappens. The avenrun update itself is handled by the CPU which calls\ndo_timer().\n\n[ Impact: reduce xtime_lock write locked section ]\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nAcked-by: Peter Zijlstra \u003cpeterz@infradead.org\u003e\n" }, { "commit": "7d27558c4138ac6b3684dea35c2f4379b940a7dd", "tree": "c10f5cee4dda1e6bc7977d016db020377f97feab", "parents": [ "a25cbd045a2ffc42787d4dbcbb9c7118f5f42732" ], "author": { "name": "john stultz", "email": "johnstul@us.ibm.com", "time": "Fri May 01 13:10:26 2009 -0700" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Sat May 02 11:45:15 2009 +0200" }, "message": "timekeeping: create arch_gettimeoffset infrastructure\n\nSome arches don\u0027t supply their own clocksource. This is mainly the\ncase in architectures that get their inter-tick times by reading the\ncounter on their interval timer. Since these timers wrap every tick,\nthey\u0027re not really useful as clocksources. Wrapping them to act like\none is possible but not very efficient. So we provide a callout these\narches can implement for use with the jiffies clocksource to provide\nfiner then tick granular time.\n\n[ Impact: ease the migration to generic time keeping ]\n\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "4614e6adafa2c5e6c3a9c245af2807fa7bc5117a", "tree": "e4bb5eeaa252ede053abc6b3c4d15c4127ce7a32", "parents": [ "8e19608e8b5c001e4a66ce482edc474f05fb7355" ], "author": { "name": "Magnus Damm", "email": "damm@igel.co.jp", "time": "Tue Apr 21 12:24:02 2009 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Tue Apr 21 13:41:47 2009 -0700" }, "message": "clocksource: add enable() and disable() callbacks\n\nAdd enable() and disable() callbacks for clocksources.\n\nThis allows us to put unused clocksources in power save mode. The\nfunctions clocksource_enable() and clocksource_disable() wrap the\ncallbacks and are inserted in the timekeeping code to enable before use\nand disable after switching to a new clocksource.\n\nSigned-off-by: Magnus Damm \u003cdamm@igel.co.jp\u003e\nAcked-by: John Stultz \u003cjohnstul@us.ibm.com\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": "1c5745aa380efb6417b5681104b007c8612fb496", "tree": "263b81742f7973c0dd86efe414a79a324c0769e4", "parents": [ "6a94cb73064c952255336cc57731904174b2c58f" ], "author": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Mon Dec 22 23:05:28 2008 +0100" }, "committer": { "name": "Ingo Molnar", "email": "mingo@elte.hu", "time": "Wed Dec 31 09:53:21 2008 +0100" }, "message": "sched_clock: prevent scd-\u003eclock from moving backwards, take #2\n\nRedo:\n\n 5b7dba4: sched_clock: prevent scd-\u003eclock from moving backwards\n\nwhich had to be reverted due to s2ram hangs:\n\n ca7e716: Revert \"sched_clock: prevent scd-\u003eclock from moving backwards\"\n\n... this time with resume restoring GTOD later in the sequence\ntaken into account as well.\n\nThe \"timekeeping_suspended\" flag is not very nice but we cannot call into\nGTOD before it has been properly resumed and the scheduler will run very\nearly in the resume sequence.\n\nCc: \u003cstable@kernel.org\u003e\nSigned-off-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n" }, { "commit": "6c9bacb41c10ba84ff68f238e234d96f35fb64f7", "tree": "220e4bde083339abd25cb26b973585b9aa19ab80", "parents": [ "eccdaeafaea3ed115068ba55d01f22e486e5437d" ], "author": { "name": "john stultz", "email": "johnstul@us.ibm.com", "time": "Mon Dec 01 18:34:41 2008 -0800" }, "committer": { "name": "Ingo Molnar", "email": "mingo@elte.hu", "time": "Thu Dec 04 08:43:02 2008 +0100" }, "message": "time: catch xtime_nsec underflows and fix them\n\nImpact: fix time warp bug\n\nAlex Shi, along with Yanmin Zhang have been noticing occasional time\ninconsistencies recently. Through their great diagnosis, they found that\nthe xtime_nsec value used in update_wall_time was occasionally going\nnegative. After looking through the code for awhile, I realized we have\nthe possibility for an underflow when three conditions are met in\nupdate_wall_time():\n\n 1) We have accumulated a second\u0027s worth of nanoseconds, so we\n incremented xtime.tv_sec and appropriately decrement xtime_nsec.\n (This doesn\u0027t cause xtime_nsec to go negative, but it can cause it\n to be small).\n\n 2) The remaining offset value is large, but just slightly less then\n cycle_interval.\n\n 3) clocksource_adjust() is speeding up the clock, causing a\n corrective amount (compensating for the increase in the multiplier\n being multiplied against the unaccumulated offset value) to be\n subtracted from xtime_nsec.\n\nThis can cause xtime_nsec to underflow.\n\nUnfortunately, since we notify the NTP subsystem via second_overflow()\nwhenever we accumulate a full second, and this effects the error\naccumulation that has already occured, we cannot simply revert the\naccumulated second from xtime nor move the second accumulation to after\nthe clocksource_adjust call without a change in behavior.\n\nThis leaves us with (at least) two options:\n\n1) Simply return from clocksource_adjust() without making a change if we\n notice the adjustment would cause xtime_nsec to go negative.\n\nThis would work, but I\u0027m concerned that if a large adjustment was needed\n(due to the error being large), it may be possible to get stuck with an\never increasing error that becomes too large to correct (since it may\nalways force xtime_nsec negative). This may just be paranoia on my part.\n\n2) Catch xtime_nsec if it is negative, then add back the amount its\n negative to both xtime_nsec and the error.\n\nThis second method is consistent with how we\u0027ve handled earlier rounding\nissues, and also has the benefit that the error being added is always in\nthe oposite direction also always equal or smaller then the correction\nbeing applied. So the risk of a corner case where things get out of\ncontrol is lessened.\n\nThis patch fixes bug 11970, as tested by Yanmin Zhang\nhttp://bugzilla.kernel.org/show_bug.cgi?id\u003d11970\n\nReported-by: alex.shi@intel.com\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nAcked-by: \"Zhang, Yanmin\" \u003cyanmin_zhang@linux.intel.com\u003e\nTested-by: \"Zhang, Yanmin\" \u003cyanmin_zhang@linux.intel.com\u003e\nSigned-off-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n" }, { "commit": "c465a76af658b443075d6efee1c3131257643020", "tree": "63c28c9fab02dedec7f03cee4a3ef7fe4dc1c072", "parents": [ "2d42244ae71d6c7b0884b5664cf2eda30fb2ae68", "1b02469088ac7a13d7e622b618b7410d0f1ce5ec", "fb02fbc14d17837b4b7b02dbb36142c16a7bf208", "d40e944c25fb4642adb2a4c580a48218a9f3f824", "1508487e7f16d992ad23cabd3712563ff912f413", "322acf6585f3c4e82ee32a246b0483ca0f6ad3f4" ], "author": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Mon Oct 20 13:14:06 2008 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Mon Oct 20 13:14:06 2008 +0200" }, "message": "Merge branches \u0027timers/clocksource\u0027, \u0027timers/hrtimers\u0027, \u0027timers/nohz\u0027, \u0027timers/ntp\u0027, \u0027timers/posixtimers\u0027 and \u0027timers/debug\u0027 into v28-timers-for-linus\n" }, { "commit": "5cd1c9c5cf30d4b33df3d3f74d8142f278d536b7", "tree": "6d74ddeff7e9a044d961d36db13071c158f0557a", "parents": [ "eb3f938fd6292dc79f43a5fe14784b044776e9f0" ], "author": { "name": "Roman Zippel", "email": "zippel@linux-m68k.org", "time": "Mon Sep 22 14:42:43 2008 -0700" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Wed Sep 24 17:33:13 2008 +0200" }, "message": "timekeeping: fix rounding problem during clock update\n\nDue to a rounding problem during a clock update it\u0027s possible for readers\nto observe the clock jumping back by 1nsec. The following simplified\nexample demonstrates the problem:\n\ncycle\txtime\n0\t0\n1000\t999999.6\n2000\t1999999.2\n3000\t2999998.8\n...\n\n1500 \u003d\t1499999.4\n\u003d\t0.0 + 1499999.4\n\u003d\t999999.6 + 499999.8\n\nWhen reading the clock only the full nanosecond part is used, while\ntimekeeping internally keeps nanosecond fractions. If the clock is now\nupdated at cycle 1500 here, a nanosecond is missing due to the truncation.\n\nThe simple fix is to round up the xtime value during the update, this also\nchanges the distance to the reference time, but the adjustment will\nautomatically take care that it stays under control.\n\nSigned-off-by: Roman Zippel \u003czippel@linux-m68k.org\u003e\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "2d42244ae71d6c7b0884b5664cf2eda30fb2ae68", "tree": "947e86ec6e2d7362daa9a170a352c035f3618d64", "parents": [ "9a055117d3d9cb562f83f8d4cd88772761f4cab0" ], "author": { "name": "John Stultz", "email": "johnstul@us.ibm.com", "time": "Wed Aug 20 16:37:30 2008 -0700" }, "committer": { "name": "Ingo Molnar", "email": "mingo@elte.hu", "time": "Thu Aug 21 09:50:24 2008 +0200" }, "message": "clocksource: introduce CLOCK_MONOTONIC_RAW\n\nIn talking with Josip Loncaric, and his work on clock synchronization (see\nbtime.sf.net), he mentioned that for really close synchronization, it is\nuseful to have access to \"hardware time\", that is a notion of time that is\nnot in any way adjusted by the clock slewing done to keep close time sync.\n\nPart of the issue is if we are using the kernel\u0027s ntp adjusted\nrepresentation of time in order to measure how we should correct time, we\ncan run into what Paul McKenney aptly described as \"Painting a road using\nthe lines we\u0027re painting as the guide\".\n\nI had been thinking of a similar problem, and was trying to come up with a\nway to give users access to a purely hardware based time representation\nthat avoided users having to know the underlying frequency and mask values\nneeded to deal with the wide variety of possible underlying hardware\ncounters.\n\nMy solution is to introduce CLOCK_MONOTONIC_RAW. This exposes a\nnanosecond based time value, that increments starting at bootup and has no\nfrequency adjustments made to it what so ever.\n\nThe time is accessed from userspace via the posix_clock_gettime() syscall,\npassing CLOCK_MONOTONIC_RAW as the clock_id.\n\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nSigned-off-by: Roman Zippel \u003czippel@linux-m68k.org\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n" }, { "commit": "9a055117d3d9cb562f83f8d4cd88772761f4cab0", "tree": "a822607ebb46491e3b480d11c136c5bc8585e38c", "parents": [ "1aa5dfb751d275ae7117d3b73ac423b4a46f2a73" ], "author": { "name": "Roman Zippel", "email": "zippel@linux-m68k.org", "time": "Wed Aug 20 16:37:28 2008 -0700" }, "committer": { "name": "Ingo Molnar", "email": "mingo@elte.hu", "time": "Thu Aug 21 09:50:24 2008 +0200" }, "message": "clocksource: introduce clocksource_forward_now()\n\nTo keep the raw monotonic patch simple first introduce\nclocksource_forward_now(), which takes care of the offset since the last\nupdate_wall_time() call and adds it to the clock, so there is no need\nanymore to deal with it explicitly at various places, which need to make\nsignificant changes to the clock.\n\nThis is also gets rid of the timekeeping_suspend_nsecs, instead of\nwaiting until resume, the value is accumulated during suspend. In the end\nthere is only a single user of __get_nsec_offset() left, so I integrated\nit back to getnstimeofday().\n\nSigned-off-by: Roman Zippel \u003czippel@linux-m68k.org\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n" }, { "commit": "7dffa3c673fbcf835cd7be80bb4aec8ad3f51168", "tree": "63264208ed97f18a74a5a7cd2e100cc2c4e13449", "parents": [ "8383c42399f394a89bd6c2f03632c53689bdde7a" ], "author": { "name": "Roman Zippel", "email": "zippel@linux-m68k.org", "time": "Thu May 01 04:34:41 2008 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Thu May 01 08:03:59 2008 -0700" }, "message": "ntp: handle leap second via timer\n\nRemove the leap second handling from second_overflow(), which doesn\u0027t have to\ncheck for it every second anymore. With CONFIG_NO_HZ this also makes sure the\nleap second is handled close to the full second. Additionally this makes it\npossible to abort a leap second properly by resetting the STA_INS/STA_DEL\nstatus bits.\n\nSigned-off-by: Roman Zippel \u003czippel@linux-m68k.org\u003e\nCc: john stultz \u003cjohnstul@us.ibm.com\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": "8383c42399f394a89bd6c2f03632c53689bdde7a", "tree": "c3f3a42a546a4afe3746e5894bcd425dc5a3d0bb", "parents": [ "7fc5c78409479d826341b103bdf734cb4fb02436" ], "author": { "name": "Roman Zippel", "email": "zippel@linux-m68k.org", "time": "Thu May 01 04:34:39 2008 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Thu May 01 08:03:59 2008 -0700" }, "message": "ntp: remove current_tick_length()\n\ncurrent_tick_length used to do a little more, but now it just returns\ntick_length, which we can also access directly at the few places, where it\u0027s\nneeded.\n\nSigned-off-by: Roman Zippel \u003czippel@linux-m68k.org\u003e\nCc: john stultz \u003cjohnstul@us.ibm.com\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": "7fc5c78409479d826341b103bdf734cb4fb02436", "tree": "e68d8edddd633a9ed410b304684343afacf36175", "parents": [ "153b5d054ac2d98ea0d86504884326b6777f683d" ], "author": { "name": "Roman Zippel", "email": "zippel@linux-m68k.org", "time": "Thu May 01 04:34:38 2008 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Thu May 01 08:03:59 2008 -0700" }, "message": "ntp: rename TICK_LENGTH_SHIFT to NTP_SCALE_SHIFT\n\nAs TICK_LENGTH_SHIFT is used for more than just the tick length, the name\nisn\u0027t quite approriate anymore, so this renames it to NTP_SCALE_SHIFT.\n\nSigned-off-by: Roman Zippel \u003czippel@linux-m68k.org\u003e\nCc: john stultz \u003cjohnstul@us.ibm.com\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": "d8bb6f4c1670c8324e4135c61ef07486f7f17379", "tree": "d53b676621b3bd6bb4d39a4b22588b58e1a7ea45", "parents": [ "f1326973262382150c26bf4dfccd0fce310c4a9c" ], "author": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Tue Apr 01 19:45:18 2008 +0200" }, "committer": { "name": "Ingo Molnar", "email": "mingo@elte.hu", "time": "Sat Apr 19 19:19:55 2008 +0200" }, "message": "x86: tsc prevent time going backwards\n\nWe already catch most of the TSC problems by sanity checks, but there\nis a subtle bug which has been in the code forever. This can cause\ntime jumps in the range of hours.\n\nThis was reported in:\n http://lkml.org/lkml/2007/8/23/96\nand\n http://lkml.org/lkml/2008/3/31/23\n\nI was able to reproduce the problem with a gettimeofday loop test on a\ndual core and a quad core machine which both have sychronized\nTSCs. The TSCs seems not to be perfectly in sync though, but the\nkernel is not able to detect the slight delta in the sync check. Still\nthere exists an extremly small window where this delta can be observed\nwith a real big time jump. So far I was only able to reproduce this\nwith the vsyscall gettimeofday implementation, but in theory this\nmight be observable with the syscall based version as well.\n\nCPU 0 updates the clock source variables under xtime/vyscall lock and\nCPU1, where the TSC is slighty behind CPU0, is reading the time right\nafter the seqlock was unlocked.\n\nThe clocksource reference data was updated with the TSC from CPU0 and\nthe value which is read from TSC on CPU1 is less than the reference\ndata. This results in a huge delta value due to the unsigned\nsubtraction of the TSC value and the reference value. This algorithm\ncan not be changed due to the support of wrapping clock sources like\npm timer.\n\nThe huge delta is converted to nanoseconds and added to xtime, which\nis then observable by the caller. The next gettimeofday call on CPU1\nwill show the correct time again as now the TSC has advanced above the\nreference value.\n\nTo prevent this TSC specific wreckage we need to compare the TSC value\nagainst the reference value and return the latter when it is larger\nthan the actual TSC value.\n\nI pondered to mark the TSC unstable when the readout is smaller than\nthe reference value, but this would render an otherwise good and fast\nclocksource unusable without a real good reason.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nSigned-off-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n" }, { "commit": "92896bd9fd75b1c993b92874d339a8088bb75560", "tree": "2bd8bc5a8c8ee10a4b1fd62804281d2b486067f0", "parents": [ "d2532dd20a126020de407c1c2476a75b53fce7ac" ], "author": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Mon Mar 24 11:07:15 2008 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@linux-foundation.org", "time": "Mon Mar 24 11:07:15 2008 -0700" }, "message": "Don\u0027t \u0027printk()\u0027 while holding xtime lock for writing\n\nThe printk() can deadlock because it can wake up klogd(), and\ntask enqueueing will try to read the time in order to set a hrtimer.\n\nReported-by: Marcin Slusarz \u003cmarcin.slusarz@gmail.com\u003e\nDebugged-by: Peter Zijlstra \u003ca.p.zijlstra@chello.nl\u003e\nCc: Ingo Molnar \u003cmingo@elte.hu\u003e\nCc: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n" }, { "commit": "10a398d04c4a1fc395840f4d040493375f562302", "tree": "02096fe81b70e2baf8db35ebc9d303a1b13029c3", "parents": [ "a79017660ea4597ec489fab3b5aaf71dd776dfc7" ], "author": { "name": "Roman Zippel", "email": "zippel@linux-m68k.org", "time": "Tue Mar 04 15:14:26 2008 -0800" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Sun Mar 09 08:42:57 2008 +0100" }, "message": "time: remove obsolete CLOCK_TICK_ADJUST\n\nThe first version of the ntp_interval/tick_length inconsistent usage patch was\nrecently merged as bbe4d18ac2e058c56adb0cd71f49d9ed3216a405\n\nhttp://git.kernel.org/gitweb.cgi?p\u003dlinux/kernel/git/torvalds/linux-2.6.git;a\u003dcommit;h\u003dbbe4d18ac2e058c56adb0cd71f49d9ed3216a405\n\nWhile the fix did greatly improve the situation, it was correctly pointed out\nby Roman that it does have a small bug: If the users change clocksources after\nthe system has been running and NTP has made corrections, the correctoins made\nagainst the old clocksource will be applied against the new clocksource,\ncausing error.\n\nThe second attempt, which corrects the issue in the NTP_INTERVAL_LENGTH\ndefinition has also made it up-stream as commit\ne13a2e61dd5152f5499d2003470acf9c838eab84\n\nhttp://git.kernel.org/gitweb.cgi?p\u003dlinux/kernel/git/torvalds/linux-2.6.git;a\u003dcommit;h\u003de13a2e61dd5152f5499d2003470acf9c838eab84\n\nRoman has correctly pointed out that CLOCK_TICK_ADJUST is calculated\nbased on the PIT\u0027s frequency, and isn\u0027t really relevant to non-PIT\ndriven clocksources (that is, clocksources other then jiffies and pit).\n\nThis patch reverts both of those changes, and simply removes\nCLOCK_TICK_ADJUST.\n\nThis does remove the granularity error correction for users of PIT and Jiffies\nclocksource users, but the granularity error but for the majority of users, it\nshould be within the 500ppm range NTP can accommodate for.\n\nFor systems that have granularity errors greater then 500ppm, the\n\"ntp_tick_adj\u003d\" boot option can be used to compensate.\n\n[johnstul@us.ibm.com: provided changelog]\n[mattilinnanvuori@yahoo.com: maek ntp_tick_adj static]\nSigned-off-by: Roman Zippel \u003czippel@linux-m68k.org\u003e\nAcked-by: john stultz \u003cjohnstul@us.ibm.com\u003e\nSigned-off-by: Matti Linnanvuori \u003cmattilinnanvuori@yahoo.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nCc: mingo@elte.hu\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "3eb056764dd806bbe84eb604e45e7470feeaafd8", "tree": "97a8fe2123d2c9df2bcb48052ac33d445a00f64c", "parents": [ "cf4fc6cb76e50b01666e28a9f4b2e6fbcbb96d5f" ], "author": { "name": "Li Zefan", "email": "lizf@cn.fujitsu.com", "time": "Fri Feb 08 04:19:25 2008 -0800" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@woody.linux-foundation.org", "time": "Fri Feb 08 09:22:29 2008 -0800" }, "message": "time: fix typo in comments\n\nFix typo in comments.\n\nBTW: I have to fix coding style in arch/ia64/kernel/time.c also, otherwise\ncheckpatch.pl will be complaining.\n\nSigned-off-by: Li Zefan \u003clizf@cn.fujitsu.com\u003e\nCc: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nCc: Ingo Molnar \u003cmingo@elte.hu\u003e\nCc: john stultz \u003cjohnstul@us.ibm.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n" }, { "commit": "cf4fc6cb76e50b01666e28a9f4b2e6fbcbb96d5f", "tree": "ac1e027b7e759822014bd3a72ababd1fa1052198", "parents": [ "0b858e6ff9a38b987a83d22e6f2a2f621c80608d" ], "author": { "name": "Li Zefan", "email": "lizf@cn.fujitsu.com", "time": "Fri Feb 08 04:19:24 2008 -0800" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@woody.linux-foundation.org", "time": "Fri Feb 08 09:22:29 2008 -0800" }, "message": "timekeeping: rename timekeeping_is_continuous to timekeeping_valid_for_hres\n\nFunction timekeeping_is_continuous() no longer checks flag\nCLOCK_IS_CONTINUOUS, and it checks CLOCK_SOURCE_VALID_FOR_HRES now. So rename\nthe function accordingly.\n\nSigned-off-by: Li Zefan \u003clizf@cn.fujitsu.com\u003e\nCc: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nCc: Ingo Molnar \u003cmingo@elte.hu\u003e\nCc: john stultz \u003cjohnstul@us.ibm.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n" }, { "commit": "1001d0a9ee74a468077dfd4da0565174e88de26b", "tree": "8775592b23e8049700ff98a7bdbcf9c802a0f127", "parents": [ "3588a085cd52ef080bf72df772378e1ba6bb292f" ], "author": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Fri Feb 01 17:45:13 2008 +0100" }, "committer": { "name": "Ingo Molnar", "email": "mingo@elte.hu", "time": "Fri Feb 01 17:45:13 2008 +0100" }, "message": "timekeeping: update xtime_cache when time(zone) changes\n\nxtime_cache needs to be updated whenever xtime and or wall_to_monotic\nare changed. Otherwise users of xtime_cache might see a stale (and in\nthe case of timezone changes utterly wrong) value until the next\nupdate happens.\n\nFixup the obvious places, which miss this update.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nAcked-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nTested-by: Dhaval Giani \u003cdhaval@linux.vnet.ibm.com\u003e\nSigned-off-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n" }, { "commit": "bbe4d18ac2e058c56adb0cd71f49d9ed3216a405", "tree": "bc81b3a04cc01fed11ac47673475bf4019cc013f", "parents": [ "37a47db8d7f0f38dac5acf5a13abbc8f401707fa" ], "author": { "name": "john stultz", "email": "johnstul@us.ibm.com", "time": "Wed Jan 30 13:30:03 2008 +0100" }, "committer": { "name": "Ingo Molnar", "email": "mingo@elte.hu", "time": "Wed Jan 30 13:30:03 2008 +0100" }, "message": "NTP: correct inconsistent ntp interval/tick_length usage\n\nI recently noticed on one of my boxes that when synched with an NTP\nserver, the drift value reported for the system was ~283ppm. While in\nsome cases, clock hardware can be that bad, it struck me as unusual as\nthe system was using the acpi_pm clocksource, which is one of the more\ntrustworthy and accurate clocksources on x86 hardware.\n\nI brought up another system and let it sync to the same NTP server, and\nI noticed a similar 280some ppm drift.\n\nIn looking at the code, I found that the acpi_pm\u0027s constant frequency\nwas being computed correctly at boot-up, however once the system was up,\neven without the ntp daemon running, the clocksource\u0027s frequency was\nbeing modified by the clocksource_adjust() function.\n\nDigging deeper, I realized that in the code that keeps track of how much\nthe clocksource is skewing from the ntp desired time, we were using\ndifferent lengths to establish how long an time interval was.\n\nThe clocksource was being setup with the following interval:\n\tNTP_INTERVAL_LENGTH \u003d NSEC_PER_SEC/NTP_INTERVAL_FREQ\n\nWhile the ntp code was using the tick_length_base value:\n\ttick_length_base ~\u003d (tick_usec * NSEC_PER_USEC * USER_HZ)\n\t\t\t\t\t/NTP_INTERVAL_FREQ\n\nThe subtle difference is:\n\t(tick_usec * NSEC_PER_USEC * USER_HZ) !\u003d NSEC_PER_SEC\n\nThis difference in calculation was causing the clocksource correction\ncode to apply a correction factor to the clocksource so the two\nintervals were the same, however this results in the actual frequency of\nthe clocksource to be made incorrect. I believe this difference would\naffect all clocksources, although to differing degrees depending on the\nclocksource resolution.\n\nThe issue was introduced when my HZ free ntp patch landed in 2.6.21-rc1,\nso my apologies for the mistake, and for not noticing it until now.\n\nThe following patch, corrects the clocksource\u0027s initialization code so\nit uses the same interval length as the code in ntp.c. After applying\nthis patch, the drift value for the same system went from ~283ppm to\nonly 2.635ppm.\n\nI believe this patch to be good, however it does affect all arches and\nI\u0027ve only tested on x86, so some caution is advised. I do think it would\nbe a likely candidate for a stable 2.6.24.x release.\n\nAny thoughts or feedback would be appreciated.\n\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nSigned-off-by: Ingo Molnar \u003cmingo@elte.hu\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "efd9ac8630e89b9ee7ce64008bd7783952374f37", "tree": "2b06184b9241770aff309a9940f6e400371003a4", "parents": [ "1d76c2622813fbc692b0d323028cfef9ee36051a" ], "author": { "name": "Geert Uytterhoeven", "email": "Geert.Uytterhoeven@sonycom.com", "time": "Wed Jan 30 13:30:01 2008 +0100" }, "committer": { "name": "Ingo Molnar", "email": "mingo@elte.hu", "time": "Wed Jan 30 13:30:01 2008 +0100" }, "message": "time: fold __get_realtime_clock_ts() into getnstimeofday()\n\n - getnstimeofday() was just a wrapper around __get_realtime_clock_ts()\n - Replace calls to __get_realtime_clock_ts() by calls to getnstimeofday()\n - Fix bogus reference to get_realtime_clock_ts(), which never existed\n\nSigned-off-by: Geert Uytterhoeven \u003cGeert.Uytterhoeven@sonycom.com\u003e\nCc: john stultz \u003cjohnstul@us.ibm.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Ingo Molnar \u003cmingo@elte.hu\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "af5ca3f4ec5cc4432a42a73b050dd8898ce8fd00", "tree": "3e5a3081b2802547f10da72c0026b4929d0e287b", "parents": [ "528a4bf1d5ffed310d26fc1d82d45c02949f71cf" ], "author": { "name": "Kay Sievers", "email": "kay.sievers@vrfy.org", "time": "Thu Dec 20 02:09:39 2007 +0100" }, "committer": { "name": "Greg Kroah-Hartman", "email": "gregkh@suse.de", "time": "Thu Jan 24 20:40:40 2008 -0800" }, "message": "Driver core: change sysdev classes to use dynamic kobject names\n\nAll kobjects require a dynamically allocated name now. We no longer\nneed to keep track if the name is statically assigned, we can just\nunconditionally free() all kobject names on cleanup.\n\nSigned-off-by: Kay Sievers \u003ckay.sievers@vrfy.org\u003e\nSigned-off-by: Greg Kroah-Hartman \u003cgregkh@suse.de\u003e\n\n" }, { "commit": "ba2a631b14fe8a9c122801c648c49a543d020d8d", "tree": "7f89f02d314c9aec6959066a1285fcd16f5bc2b1", "parents": [ "f44ec6f3f89889a469773b1fd894f8fcc07c29cf" ], "author": { "name": "Adrian Bunk", "email": "bunk@kernel.org", "time": "Tue Oct 16 23:27:16 2007 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@woody.linux-foundation.org", "time": "Wed Oct 17 08:42:53 2007 -0700" }, "message": "kernel/time/timekeeping.c: cleanups\n\n- remove the no longer required __attribute__((weak)) of xtime_lock\n- remove the following no longer used EXPORT_SYMBOL\u0027s:\n - xtime\n - xtime_lock\n\nSigned-off-by: Adrian Bunk \u003cbunk@kernel.org\u003e\nCc: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nCc: john stultz \u003cjohnstul@us.ibm.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n" }, { "commit": "f20bf6125605acbbc7eb8c9420d7221c91aa83eb", "tree": "ce3dd753431dd32b346aa19f2f7b6bb84bb1f22e", "parents": [ "8636225e8d23e6079d43f56faf415bd71b589dc0" ], "author": { "name": "Ingo Molnar", "email": "mingo@elte.hu", "time": "Tue Oct 16 16:09:20 2007 +0200" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@woody.linux-foundation.org", "time": "Tue Oct 16 10:01:50 2007 -0700" }, "message": "time: introduce xtime_seconds\n\nimprove performance of sys_time(). sys_time() returns time in seconds,\nbut it does so by calling do_gettimeofday() and then returning the\ntv_sec portion of the GTOD time. But the data structure \"xtime\", which\nis updated by every timer/scheduler tick, already offers HZ granularity\ntime.\n\nthe patch improves the sysbench oltp macrobenchmark by 4-5% on an AMD\ndual-core system:\n\nv2.6.23:\n\n#threads\n\n 1: transactions: 4073 (407.23 per sec.)\n 2: transactions: 8530 (852.81 per sec.)\n 3: transactions: 8321 (831.88 per sec.)\n 4: transactions: 8407 (840.58 per sec.)\n 5: transactions: 8070 (806.74 per sec.)\n\nv2.6.23 + sys_time-speedup.patch:\n\n 1: transactions: 4281 (428.09 per sec.)\n 2: transactions: 8910 (890.85 per sec.)\n 3: transactions: 8659 (865.79 per sec.)\n 4: transactions: 8676 (867.34 per sec.)\n 5: transactions: 8532 (852.91 per sec.)\n\nand by 4-5% on an Intel dual-core system too:\n\n2.6.23:\n\n 1: transactions: 4560 (455.94 per sec.)\n 2: transactions: 10094 (1009.30 per sec.)\n 3: transactions: 9755 (975.36 per sec.)\n 4: transactions: 9859 (985.78 per sec.)\n 5: transactions: 9701 (969.72 per sec.)\n\n2.6.23 + sys_time-speedup.patch:\n\n 1: transactions: 4779 (477.84 per sec.)\n 2: transactions: 10103 (1010.14 per sec.)\n 3: transactions: 10141 (1013.93 per sec.)\n 4: transactions: 10371 (1036.89 per sec.)\n 5: transactions: 10178 (1017.50 per sec.)\n\n(the more CPUs the system has, the more speedup this patch gives for\nthis particular workload.)\n\nSigned-off-by: Ingo Molnar \u003cmingo@elte.hu\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n" }, { "commit": "6a669ee8a790487b7ec1edda762d39615a78264b", "tree": "487270be1b035d75622f0f9b1e34f1c6b2cfbeae", "parents": [ "3be9095063885d482b87d3875ea7f28e635882d0" ], "author": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Sun Sep 16 15:36:43 2007 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@chaos.(none)", "time": "Sun Sep 16 15:36:43 2007 +0200" }, "message": "timekeeping: Prevent time going backwards on resume\n\nTimekeeping resume adjusts xtime by adding the slept time in seconds and\nresets the reference value of the clock source (clock-\u003ecycle_last).\nclock-\u003ecycle last is used to calculate the delta between the last xtime\nupdate and the readout of the clock source in __get_nsec_offset(). xtime\nplus the offset is the current time. The resume code ignores the delta\nwhich had already elapsed between the last xtime update and the actual\ntime of suspend. If the suspend time is short, then we can see time\ngoing backwards on resume.\n\nSuspend:\noffs_s \u003d clock-\u003eread() - clock-\u003ecycle_last;\nnow \u003d xtime + offs_s;\ntimekeeping_suspend_time \u003d read_rtc();\n\nResume:\nsleep_time \u003d read_rtc() - timekeeping_suspend_time;\nxtime.tv_sec +\u003d sleep_time;\nclock-\u003ecycle_last \u003d clock-\u003eread();\noffs_r \u003d clock-\u003eread() - clock-\u003ecycle_last;\nnow \u003d xtime + offs_r;\n\nif sleep_time_seconds \u003d\u003d 0 and offs_r \u003c offs_s, then time goes\nbackwards.\n\nFix this by storing the offset from the last xtime update and add it to\nxtime during resume, when we reset clock-\u003ecycle_last:\n\nsleep_time \u003d read_rtc() - timekeeping_suspend_time;\nxtime.tv_sec +\u003d sleep_time;\nxtime +\u003d offs_s;\t/* Fixup xtime offset at suspend time */\nclock-\u003ecycle_last \u003d clock-\u003eread();\noffs_r \u003d clock-\u003eread() - clock-\u003ecycle_last;\nnow \u003d xtime + offs_r;\n\nThanks to Marcelo for tracking this down on the OLPC and providing the\nnecessary details to analyze the root cause.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nCc: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Tosatti \u003cmarcelo@kvack.org\u003e\n\n" }, { "commit": "3be9095063885d482b87d3875ea7f28e635882d0", "tree": "c29f9a90a244032515c4f830e1d73d4dd2a3f1ac", "parents": [ "d0174640eedc1cd756754f03afe2dbb3d56de74e" ], "author": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Sun Sep 16 15:36:43 2007 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@chaos.(none)", "time": "Sun Sep 16 15:36:43 2007 +0200" }, "message": "timekeeping: access rtc outside of xtime lock\n\nLockdep complains about the access of rtc in timekeeping_suspend\ninside the interrupt disabled region of the write locked xtime lock.\nMove the access outside.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nCc: John Stultz \u003cjohnstul@us.ibm.com\u003e\n\n" }, { "commit": "17c38b7490b3f0300c7812aefdae2ddda7ab4112", "tree": "0f7a9ee0c691aef3497030e38c3ba8e3c13a985c", "parents": [ "2c6b47de17c75d553de3e2fb426d8298d2074585" ], "author": { "name": "john stultz", "email": "johnstul@us.ibm.com", "time": "Tue Jul 24 18:38:34 2007 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@woody.linux-foundation.org", "time": "Wed Jul 25 10:17:44 2007 -0700" }, "message": "Cache xtime every call to update_wall_time\n\nThis avoids xtime lag seen with dynticks, because while \u0027xtime\u0027 itself\nis still not updated often, we keep a \u0027xtime_cache\u0027 variable around that\ncontains the approximate real-time that _is_ updated each time we do a\n\u0027update_wall_time()\u0027, and is thus never off by more than one tick.\n\nIOW, this restores the original semantics for \u0027xtime\u0027 users, as long as\nyou use the proper abstraction functions (ie \u0027current_kernel_time()\u0027 or\n\u0027get_seconds()\u0027 depending on whether you want a timespec or just the\nseconds field).\n\n[ Updated Patch. As penance for my sins I\u0027ve also yanked another #ifdef\n that was added to avoid the xtime lag w/ hrtimers. ]\n\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Ingo Molnar \u003cmingo@elte.hu\u003e\nCc: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n" }, { "commit": "2c6b47de17c75d553de3e2fb426d8298d2074585", "tree": "c8edc3d727d85cb3e1c043583c350a04a133e1a3", "parents": [ "0de085bb474f64e4fdb2f1ff3268590792648c7b" ], "author": { "name": "john stultz", "email": "johnstul@us.ibm.com", "time": "Tue Jul 24 17:47:43 2007 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@woody.linux-foundation.org", "time": "Wed Jul 25 10:09:20 2007 -0700" }, "message": "Cleanup non-arch xtime uses, use get_seconds() or current_kernel_time().\n\nThis avoids use of the kernel-internal \"xtime\" variable directly outside\nof the actual time-related functions. Instead, use the helper functions\nthat we already have available to us.\n\nThis doesn\u0027t actually change any behaviour, but this will allow us to\nfix the fact that \"xtime\" isn\u0027t updated very often with CONFIG_NO_HZ\n(because much of the realtime information is maintained as separate\noffsets to \u0027xtime\u0027), which has caused interfaces that use xtime directly\nto get a time that is out of sync with the real-time clock by up to a\nthird of a second or so.\n\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Ingo Molnar \u003cmingo@elte.hu\u003e\nCc: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n" }, { "commit": "c36c282b88963d0957368a443168588e62301fda", "tree": "6343887ae42a65635a61b4ad99fd7f3e8dd24758", "parents": [ "f4fbfb0dda5577075a049eec7fb7ad38abca1912", "1f564ad6d4182859612cbae452122e5eb2d62a76" ], "author": { "name": "Tony Luck", "email": "tony.luck@intel.com", "time": "Fri Jul 20 11:26:47 2007 -0700" }, "committer": { "name": "Tony Luck", "email": "tony.luck@intel.com", "time": "Fri Jul 20 11:26:47 2007 -0700" }, "message": "Pull ia64-clocksource into release branch\n" }, { "commit": "1f564ad6d4182859612cbae452122e5eb2d62a76", "tree": "ef98ce12ed64853437c35a81e29f0d1c09a4393b", "parents": [ "0aa366f351d044703e25c8425e508170e80d83b1" ], "author": { "name": "Bob Picco", "email": "bob.picco@hp.com", "time": "Wed Jul 18 15:51:28 2007 -0700" }, "committer": { "name": "Tony Luck", "email": "tony.luck@intel.com", "time": "Fri Jul 20 11:23:02 2007 -0700" }, "message": "[IA64] remove time interpolator\n\nRemove time_interpolator code (This is generic code, but\nonly user was ia64. It has been superseded by the\nCONFIG_GENERIC_TIME code).\n\nSigned-off-by: Bob Picco \u003cbob.picco@hp.com\u003e\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nSigned-off-by: Peter Keilty \u003cpeter.keilty@hp.com\u003e\nSigned-off-by: Tony Luck \u003ctony.luck@intel.com\u003e\n" }, { "commit": "71120f183bff04ba4f7ba3cc554202061912d548", "tree": "3b8281d6d3c6034aab4c3fa1c15dfd95faf2c17f", "parents": [ "10146801e8b960d7473e350b60458d9d5a2be1cf" ], "author": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Thu Jul 19 01:49:16 2007 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@woody.linux-foundation.org", "time": "Thu Jul 19 10:04:52 2007 -0700" }, "message": "timekeeping: fixup shadow variable argument\n\nclocksource_adjust() has a clock argument, which shadows the file global clock\nvariable. Fix this up.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nCc: john stultz \u003cjohnstul@us.ibm.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n" }, { "commit": "7c3f1a573237b90ef331267260358a0ec4ac9079", "tree": "f308047c4f40022d4b8f7226fba73b067dcd2d70", "parents": [ "35ef63f635aa0e414ad6cdb2a4092e1caf99272c" ], "author": { "name": "Tomas Janousek", "email": "tjanouse@redhat.com", "time": "Sun Jul 15 23:39:41 2007 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@woody.linux-foundation.org", "time": "Mon Jul 16 09:05:41 2007 -0700" }, "message": "Introduce boot based time\n\nThe commits\n\n 411187fb05cd11676b0979d9fbf3291db69dbce2 (GTOD: persistent clock support)\n c1d370e167d66b10bca3b602d3740405469383de (i386: use GTOD persistent clock\n support)\n\nchanged the monotonic time so that it no longer jumps after resume, but it\u0027s\nnot possible to use it for boot time and process start time calculations then.\n Also, the uptime no longer increases during suspend.\n\nI add a variable to track the wall_to_monotonic changes, a function to get the\nreal boot time and a function to get the boot based time from the monotonic\none.\n\n[akpm@linux-foundation.org: remove exports, add comment]\nSigned-off-by: Tomas Janousek \u003ctjanouse@redhat.com\u003e\nCc: Tomas Smetana \u003ctsmetana@redhat.com\u003e\nCc: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nCc: Ingo Molnar \u003cmingo@elte.hu\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n" }, { "commit": "d10ff3fb62bd38415c0f7be3d75d107e1f67e59a", "tree": "402df109059cb64f88fb26c0cf5eb86b8c9fdae3", "parents": [ "705962ccc9d21a08b74b6b6e1d3cf10f98968a67" ], "author": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Mon May 14 11:10:02 2007 +0200" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@woody.linux-foundation.org", "time": "Mon May 14 12:13:11 2007 -0700" }, "message": "timekeeping fix patch got mis-applied\n\nThe time keeping code move to kernel/time/timekeeping.c broke the\nclocksource resume logic patch, which got applied to the old file by a\nfuzzy application. Fix it up and move the clocksource_resume() call to\nthe appropriate place.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n[ tssk, tssk, everybody should use --fuzz\u003d0 ]\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n" }, { "commit": "8524070b7982d76258942275908b7434cfcab4b4", "tree": "6e63c45c3b9ff6a86ad32b1de7adf48889eb0bfc", "parents": [ "329c8d84ca1946c037d9859dc251b56d8b1b4630" ], "author": { "name": "john stultz", "email": "johnstul@us.ibm.com", "time": "Tue May 08 00:27:59 2007 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@woody.linux-foundation.org", "time": "Tue May 08 11:15:06 2007 -0700" }, "message": "Move timekeeping code to timekeeping.c\n\nMove the timekeeping code out of kernel/timer.c and into\nkernel/time/timekeeping.c. I made no cleanups or other changes in transit.\n\n[akpm@linux-foundation.org: build fix]\nSigned-off-by: John Stultz \u003cjohnstul@us.ibm.com\u003e\nCc: Ingo Molnar \u003cmingo@elte.hu\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" } ] }