- 18 Jun, 2015 19 commits
-
-
Peter Zijlstra authored
The whole migrate_task{,s}() locking seems a little shaky, there's a lot of dropping an require happening. Pull the locking up into the callers as far as possible to streamline the lot. Signed-off-by:Peter Zijlstra (Intel) <peterz@infradead.org> Cc: ktkhai@parallels.com Cc: rostedt@goodmis.org Cc: juri.lelli@gmail.com Cc: pang.xunlei@linaro.org Cc: oleg@redhat.com Cc: wanpeng.li@linux.intel.com Cc: umgwanakikbuti@gmail.com Link: http://lkml.kernel.org/r/20150611124743.755256708@infradead.orgSigned-off-by:
Thomas Gleixner <tglx@linutronix.de>
-
Peter Zijlstra authored
In preparation to reworking set_cpus_allowed_ptr() move some code around. This also removes some superfluous #ifdefs and adds comments to some #endifs. text data bss dec hex filename 12211532 1738144 1081344 15031020 e55aec defconfig-build/vmlinux.pre 12211532 1738144 1081344 15031020 e55aec defconfig-build/vmlinux.post Signed-off-by:
-
Peter Zijlstra authored
We still have a few pending issues with the deadline code, one of which is that switching between scheduling classes can 'leak' CBS state. Close the hole by retaining the current CBS state when leaving SCHED_DEADLINE and unconditionally programming the deadline timer. The timer will then reset the CBS state if the task is still !SCHED_DEADLINE by the time it hits. If the task left SCHED_DEADLINE it will not call task_dead_dl() and we'll not cancel the hrtimer, leaving us a pending timer in free space. Avoid this by giving the timer a task reference, this avoids littering the task exit path for this rather uncommon case. In order to do this, I had to move dl_task_offline_migration() below the replenishment, such that the task_rq()->lock fully covers that. While doing this, I noticed that it (was) buggy in assuming a task is enqueued and or we need to enqueue the task now. Fixing this means select_task_rq_dl() might encounter an offline rq -- look into that. As a result this kills cancel_dl_timer() which included a rq->lock break. Fixes: 40767b0d ("sched/deadline: Fix deadline parameter modification handling") Cc: Wanpeng Li <wanpeng.li@linux.intel.com> Cc: Luca Abeni <luca.abeni@unitn.it> Cc: Juri Lelli <juri.lelli@arm.com> Signed-off-by:
-
Peter Zijlstra authored
Remove the direct {push,pull} balancing operations from switched_{from,to}_dl() / prio_changed_dl() and use the balance callback queue. Again, err on the side of too many reschedules; since too few is a hard bug while too many is just annoying. Signed-off-by: -
Peter Zijlstra authored
In order to be able to use pull_dl_task() from a callback, we need to do away with the return value. Since the return value indicates if we should reschedule, do this inside the function. Since not all callers currently do this, this can increase the number of reschedules due rt balancing. Too many reschedules is not a correctness issues, too few are. Signed-off-by:
-
Peter Zijlstra authored
Remove the direct {push,pull} balancing operations from switched_{from,to}_rt() / prio_changed_rt() and use the balance callback queue. Again, err on the side of too many reschedules; since too few is a hard bug while too many is just annoying. Signed-off-by: -
Peter Zijlstra authored
In order to be able to use pull_rt_task() from a callback, we need to do away with the return value. Since the return value indicates if we should reschedule, do this inside the function. Since not all callers currently do this, this can increase the number of reschedules due rt balancing. Too many reschedules is not a correctness issues, too few are. Signed-off-by:
-
Peter Zijlstra authored
In order to remove dropping rq->lock from the switched_{to,from}()/prio_changed() sched_class methods, run the balance callbacks after it. We need to remove dropping rq->lock because its buggy, suppose using sched_setattr()/sched_setscheduler() to change a running task from FIFO to OTHER. By the time we get to switched_from_rt() the task is already enqueued on the cfs runqueues. If switched_from_rt() does pull_rt_task() and drops rq->lock, load-balancing can come in and move our task @p to another rq. The subsequent switched_to_fair() still assumes @p is on @rq and bad things will happen. By using balance callbacks we delay the load-balancing operations {rt,dl}x{push,pull} until we've done all the important work and the task is fully set up. Furthermore, the balance callbacks do not know about @p, therefore they cannot get confused like this. Reported-by:Mike Galbraith <umgwanakikbuti@gmail.com> Signed-off-by:
-
Peter Zijlstra authored
Reduce duplicate logic; normalize_task() is a simplified version of __sched_setscheduler(). Parametrize the difference and collapse. This reduces the amount of check_class_changed() sites. Signed-off-by:
-
Peter Zijlstra authored
Generalize the post_schedule() stuff into a balance callback list. This allows us to more easily use it outside of schedule() and cross sched_class. Signed-off-by:
-
Thomas Gleixner authored
Merge sched/core and timers/core so we can apply the sched balancing patch queue, which depends on both.
-
Peter Zijlstra authored
Currently an hrtimer callback function cannot free its own timer because __run_hrtimer() still needs to clear HRTIMER_STATE_CALLBACK after it. Freeing the timer would result in a clear use-after-free. Solve this by using a scheme similar to regular timers; track the current running timer in hrtimer_clock_base::running. Suggested-by:
-
Peter Zijlstra authored
Introduce raw_write_seqcount_barrier(), a new construct that can be used to provide write barrier semantics in seqcount read loops instead of the usual consistency guarantee. raw_write_seqcount_barier() is equivalent to: raw_write_seqcount_begin(); raw_write_seqcount_end(); But avoids issueing two back-to-back smp_wmb() instructions. This construct works because the read side will 'stall' when observing odd values. This means that -- referring to the example in the comment below -- even though there is no (matching) read barrier between the loads of X and Y, we cannot observe !x && !y, because: - if we observe Y == false we must observe the first sequence increment, which makes us loop, until - we observe !(seq & 1) -- the second sequence increment -- at which time we must also observe T == true. Suggested-by:
Oleg Nesterov <oleg@redhat.com> Signed-off-by:
-
Peter Zijlstra authored
I'll shortly be introducing another seqcount primitive that's useful to provide ordering semantics and would like to use the write_seqcount_barrier() name for that. Seeing how there's only one user of the current primitive, lets rename it to invalidate, as that appears what its doing. While there, employ lockdep_assert_held() instead of assert_spin_locked() to not generate debug code for regular kernels. Signed-off-by:
-
Peter Zijlstra authored
A queued hrtimer that gets restarted (hrtimer_start*() while hrtimer_is_queued()) will briefly appear as unqueued/inactive, even though the timer has always been active, we just moved it. Close this hole by preserving timer->state in hrtimer_start_range_ns()'s remove_hrtimer() call. Reported-by:
-
Oleg Nesterov authored
I do not understand HRTIMER_STATE_MIGRATE. Unless I am totally confused it looks buggy and simply unneeded. migrate_hrtimer_list() sets it to keep hrtimer_active() == T, but this is not enough: this can fool, say, hrtimer_is_queued() in dequeue_signal(). Can't migrate_hrtimer_list() simply use HRTIMER_STATE_ENQUEUED? This fixes the race and we can kill STATE_MIGRATE. Signed-off-by:
-
John Stultz authored
In 0c4a5fc9 (Add leap-second timer edge testing to leap-a-day.c), we added a timer to the test which checks to make sure timers near the leapsecond edge behave correctly. However, the output generated from the timer uses ctime_r, which isn't async-signal safe, and should that signal land while the main test is using ctime_r to print its output, its possible for the test to deadlock on glibc internal locks. Thus this patch reworks the output to avoid using ctime_r in the signal handler. Signed-off-by:
John Stultz <john.stultz@linaro.org> Cc: Prarit Bhargava <prarit@redhat.com> Cc: Daniel Bristot de Oliveira <bristot@redhat.com> Cc: Richard Cochran <richardcochran@gmail.com> Cc: Jan Kara <jack@suse.cz> Cc: Jiri Bohac <jbohac@suse.cz> Cc: Shuah Khan <shuahkh@osg.samsung.com> Cc: Ingo Molnar <mingo@kernel.org> Link: http://lkml.kernel.org/r/1434565003-3386-1-git-send-email-john.stultz@linaro.orgSigned-off-by:
-
John Stultz authored
The fix in d1518326 (time: Move clock_was_set_seq update before updating shadow-timekeeper) was unfortunately incomplete. The main gist of that change was to do the shadow-copy update last, so that any state changes were properly duplicated, and we wouldn't accidentally have stale data in the shadow. Unfortunately in the main update_wall_time() logic, we update use the shadow-timekeeper to calculate the next update values, then while holding the lock, copy the shadow-timekeeper over, then call timekeeping_update() to do some additional bookkeeping, (skipping the shadow mirror). The bug with this is the additional bookkeeping isn't all read-only, and some changes timkeeper state. Thus we might then overwrite this state change on the next update. To avoid this problem, do the timekeeping_update() on the shadow-timekeeper prior to copying the full state over to the real-timekeeper. This avoids problems with both the clock_was_set_seq and next_leap_ktime being overwritten and possibly the fast-timekeepers as well. Many thanks to Prarit for his rigorous testing, which discovered this problem, along with Prarit and Daniel's work validating this fix. Reported-by:
Prarit Bhargava <prarit@redhat.com> Tested-by:
Daniel Bristot de Oliveira <bristot@redhat.com> Signed-off-by:
-
Viresh Kumar authored
CLOCK_EVT_MODE_* macros are present for backward compatibility (as most of the drivers are still using old ->set_mode() interface). These macro's shouldn't be used anymore in code, that is common to both driver interfaces, i.e. ->set_mode() and ->set_state_*(). Drivers implementing ->set_state_*() interface, which have their clkevt->mode set to 0 (clkevt device structures are normally globally defined), will not participate in suspend/resume as they will always be marked as UNUSED. Fix this by checking state of the clockevent device instead of mode, which is updated for both the interfaces. Fixes: ac34ad27 ("clockevents: Do not suspend/resume if unused") Signed-off-by:
Viresh Kumar <viresh.kumar@linaro.org> Cc: linaro-kernel@lists.linaro.org Cc: alexandre.belloni@free-electrons.com Cc: sylvain.rochet@finsecur.com Link: http://lkml.kernel.org/r/a1964eef6e8a47d02b1ff9083c6c91f73f0ff643.1434537215.git.viresh.kumar@linaro.orgSigned-off-by:
-
- 12 Jun, 2015 5 commits
-
-
John Stultz authored
Prarit reported an issue w/ timers around the leapsecond, where a timer set for Midnight UTC (00:00:00) might fire a second early right before the leapsecond (23:59:60 - though it appears as a repeated 23:59:59) is applied. So I've updated the leap-a-day.c test to integrate a similar test, where we set a timer and check if it triggers at the right time, and if the ntp state transition is managed properly. Reported-by:
-
John Stultz authored
Since the leapsecond is applied at tick-time, this means there is a small window of time at the start of a leap-second where we cross into the next second before applying the leap. This patch modified adjtimex so that the leap-second is applied on the second edge. Providing more correct leapsecond behavior. This does make it so that adjtimex()'s returned time values can be inconsistent with time values read from gettimeofday() or clock_gettime(CLOCK_REALTIME,...) for a brief period of one tick at the leapsecond. However, those other interfaces do not provide the TIME_OOP time_state return that adjtimex() provides, which allows the leapsecond to be properly represented. They instead only see a time discontinuity, and cannot tell the first 23:59:59 from the repeated 23:59:59 leap second. This seems like a reasonable tradeoff given clock_gettime() / gettimeofday() cannot properly represent a leapsecond, and users likely care more about performance, while folks who are using adjtimex() more likely care about leap-second correctness. Signed-off-by:
-
John Stultz authored
Currently, leapsecond adjustments are done at tick time. As a result, the leapsecond was applied at the first timer tick *after* the leapsecond (~1-10ms late depending on HZ), rather then exactly on the second edge. This was in part historical from back when we were always tick based, but correcting this since has been avoided since it adds extra conditional checks in the gettime fastpath, which has performance overhead. However, it was recently pointed out that ABS_TIME CLOCK_REALTIME timers set for right after the leapsecond could fire a second early, since some timers may be expired before we trigger the timekeeping timer, which then applies the leapsecond. This isn't quite as bad as it sounds, since behaviorally it is similar to what is possible w/ ntpd made leapsecond adjustments done w/o using the kernel discipline. Where due to latencies, timers may fire just prior to the settimeofday call. (Also, one should note that all applications using CLOCK_REALTIME timers should always be careful, since they are prone to quirks from settimeofday() disturbances.) However, the purpose of having the kernel do the leap adjustment is to avoid such latencies, so I think this is worth fixing. So in order to properly keep those timers from firing a second early, this patch modifies the ntp and timekeeping logic so that we keep enough state so that the update_base_offsets_now accessor, which provides the hrtimer core the current time, can check and apply the leapsecond adjustment on the second edge. This prevents the hrtimer core from expiring timers too early. This patch does not modify any other time read path, so no additional overhead is incurred. However, this also means that the leap-second continues to be applied at tick time for all other read-paths. Apologies to Richard Cochran, who pushed for similar changes years ago, which I resisted due to the concerns about the performance overhead. While I suspect this isn't extremely critical, folks who care about strict leap-second correctness will likely want to watch this. Potentially a -stable candidate eventually. Originally-suggested-by:
Richard Cochran <richardcochran@gmail.com> Reported-by:
-
John Stultz authored
Currently the leapsecond logic uses what looks like magic values. Improve this by defining SECS_PER_DAY and using that macro to make the logic more clear. Signed-off-by:
-
John Stultz authored
It was reported that 868a3e91 (hrtimer: Make offset update smarter) was causing timer problems after suspend/resume. The problem with that change is the modification to clock_was_set_seq in timekeeping_update is done prior to mirroring the time state to the shadow-timekeeper. Thus the next time we do update_wall_time() the updated sequence is overwritten by whats in the shadow copy. This patch moves the shadow-timekeeper mirroring to the end of the function, after all updates have been made, so all data is kept in sync. (This patch also affects the update_fast_timekeeper calls which were also problematically done prior to the mirroring). Reported-and-tested-by:
Jeremiah Mahler <jmmahler@gmail.com> Signed-off-by:
-
- 10 Jun, 2015 3 commits
-
-
Joe Perches authored
clocksource messages aren't prefixed in dmesg so it's a bit unclear what subsystem emits the messages. Use pr_fmt and pr_<level> to auto-prefix the messages appropriately. Miscellanea: o Remove "Warning" from KERN_WARNING level messages o Align "timekeeping watchdog: " messages o Coalesce formats o Align multiline arguments Signed-off-by:
Joe Perches <joe@perches.com> Cc: John Stultz <john.stultz@linaro.org> Link: http://lkml.kernel.org/r/1432579795.2846.75.camel@perches.comSigned-off-by:
-
Nicholas Mc Guire authored
To allow constant folding in usecs_to_jiffies() conditionally calls the HZ dependent _usecs_to_jiffies() helpers or, when gcc can not figure out constant folding, __usecs_to_jiffies, which is the renamed original usecs_to_jiffies() function. Signed-off-by:
Nicholas Mc Guire <hofrat@osadl.org> Cc: Masahiro Yamada <yamada.m@jp.panasonic.com> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Joe Perches <joe@perches.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Andrew Hunter <ahh@google.com> Cc: Paul Turner <pjt@google.com> Cc: Michal Marek <mmarek@suse.cz> Link: http://lkml.kernel.org/r/1432832996-12129-2-git-send-email-hofrat@osadl.orgSigned-off-by:
-
Nicholas Mc Guire authored
Refactor the usecs_to_jiffies conditional code part in time.c and jiffies.h putting it into conditional functions rather than #ifdefs to improve readability. This is analogous to the msecs_to_jiffies() cleanup in commit ca42aaf0 ("time: Refactor msecs_to_jiffies") Signed-off-by:
-
- 08 Jun, 2015 1 commit
-
-
Borislav Petkov authored
... in the !CONFIG_HIGH_RES_TIMERS case too. And thus fix warnings like this one: net/sched/sch_api.c: In function ‘psched_show’: net/sched/sch_api.c:1891:6: warning: format ‘%x’ expects argument of type ‘unsigned int’, but argument 6 has type ‘long int’ [-Wformat=] (u32)NSEC_PER_SEC / hrtimer_resolution); Signed-off-by:Borislav Petkov <bp@suse.de> Link: http://lkml.kernel.org/r/1433583000-32090-1-git-send-email-bp@alien8.deSigned-off-by:
-
- 07 Jun, 2015 6 commits
-
-
Rik van Riel authored
Changeset a43455a1 ("sched/numa: Ensure task_numa_migrate() checks the preferred node") fixes an issue where workloads would never converge on a fully loaded (or overloaded) system. However, it introduces a regression on less than fully loaded systems, where workloads converge on a few NUMA nodes, instead of properly staying spread out across the whole system. This leads to a reduction in available memory bandwidth, and usable CPU cache, with predictable performance problems. The root cause appears to be an interaction between the load balancer and NUMA balancing, where the short term load represented by the load balancer differs from the long term load the NUMA balancing code would like to base its decisions on. Simply reverting a43455a1 would re-introduce the non-convergence of workloads on fully loaded systems, so that is not a good option. As an aside, the check done before a43455a1 only applied to a task's preferred node, not to other candidate nodes in the system, so the converge-on-too-few-nodes problem still happens, just to a lesser degree. Instead, try to compensate for the impedance mismatch between the load balancer and NUMA balancing by only ever considering a lesser loaded node as a destination for NUMA balancing, regardless of whether the task is trying to move to the preferred node, or to another node. This patch also addresses the issue that a system with a single runnable thread would never migrate that thread to near its memory, introduced by 095bebf6 ("sched/numa: Do not move past the balance point if unbalanced"). A test where the main thread creates a large memory area, and spawns a worker thread to iterate over the memory (placed on another node by select_task_rq_fair), after which the main thread goes to sleep and waits for the worker thread to loop over all the memory now sees the worker thread migrated to where the memory is, instead of having all the memory migrated over like before. Jirka has run a number of performance tests on several systems: single instance SpecJBB 2005 performance is 7-15% higher on a 4 node system, with higher gains on systems with more cores per socket. Multi-instance SpecJBB 2005 (one per node), linpack, and stream see little or no changes with the revert of 095bebf6 and this patch. Reported-by:
Artem Bityutski <dedekind1@gmail.com> Reported-by:
Jirka Hladky <jhladky@redhat.com> Tested-by:
Rik van Riel <riel@redhat.com> Signed-off-by:
Mel Gorman <mgorman@suse.de> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: H. Peter Anvin <hpa@zytor.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/20150528095249.3083ade0@annuminas.surriel.comSigned-off-by:
Ingo Molnar <mingo@kernel.org>
-
Rik van Riel authored
Commit 095bebf6 ("sched/numa: Do not move past the balance point if unbalanced") broke convergence of workloads with just one runnable thread, by making it impossible for the one runnable thread on the system to move from one NUMA node to another. Instead, the thread would remain where it was, and pull all the memory across to its location, which is much slower than just migrating the thread to where the memory is. The next patch has a better fix for the issue that 095bebf6 tried to address. Reported-by:
-
Ben Segall authored
The optimized task selection logic optimistically selects a new task to run without first doing a full put_prev_task(). This is so that we can avoid a put/set on the common ancestors of the old and new task. Similarly, we should only call check_cfs_rq_runtime() to throttle eligible groups if they're part of the common ancestry, otherwise it is possible to end up with no eligible task in the simple task selection. Imagine: /root /prev /next /A /B If our optimistic selection ends up throttling /next, we goto simple and our put_prev_task() ends up throttling /prev, after which we're going to bug out in set_next_entity() because there aren't any tasks left. Avoid this scenario by only throttling common ancestors. Reported-by:
Mohammed Naser <mnaser@vexxhost.com> Reported-by:
Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Signed-off-by:
Ben Segall <bsegall@google.com> [ munged Changelog ] Signed-off-by:
-
Frederic Weisbecker authored
preempt.h has two seperate "#ifdef CONFIG_PREEMPT" sections: one to define preempt_enable() and another to define preempt_enable_notrace(). Lets gather both. Signed-off-by:
Frederic Weisbecker <fweisbec@gmail.com> Signed-off-by:
-
Frederic Weisbecker authored
preempt_schedule_context() is a tracing safe preemption point but it's only used when CONFIG_CONTEXT_TRACKING=y. Other configs have tracing recursion issues since commit: b30f0e3f ("sched/preempt: Optimize preemption operations on __schedule() callers") introduced function based preemp_count_*() ops. Lets make it available on all configs and give it a more appropriate name for its new position. Reported-by:
Fengguang Wu <fengguang.wu@intel.com> Signed-off-by:
-
Frederic Weisbecker authored
Since function tracing disables preemption, it needs a safe preemption point to use when preemption is re-enabled without worrying about tracing recursion. Ie: to avoid tracing recursion, that preemption point can't be traced (use of notrace qualifier) and it can't call any traceable function before that preemption point disables preemption itself, which disarms the recursion. preempt_schedule() was fine until commit: b30f0e3f ("sched/preempt: Optimize preemption operations on __schedule() callers") because PREEMPT_ACTIVE (which has the property to disable preemption and this disarm tracing preemption recursion) was set before calling any further function. But that commit introduced the use of preempt_count_add/sub() functions to set PREEMPT_ACTIVE and because these functions are called before preemption gets a chance to be disabled, we have a tracing recursion. preempt_schedule_context() is one of the possible preemption functions used by tracing. Its special purpose is to avoid tracing recursion against context tracking. Lets enhance this function to become more generally tracing safe by disabling preemption with raw accessors, such that no function is called before preemption gets disabled and disarm the tracing recursion. This function is going to become the specific tracing-safe preemption point in further commit. Reported-by:
-
- 02 Jun, 2015 6 commits
-
-
Thomas Gleixner authored
Merge branch 'clockevents/4.2' of http://git.linaro.org/people/daniel.lezcano/linux into timers/core Pull clockevents/clocksource changes from Daniel Lezcano: - Removed dead code in the files related to mach-msm for qcom (Stephen Boyd) - Cleaned up code for exynos_mct (Krzysztof Kozlowski) - Added the new timer lpc3220 (Joachim Eastwood) - Added the new timer STM32 and ARM system timer (Maxime Coquelin)
-
Thomas Gleixner authored
The only sensible way to make abuse of core internal fields obvious and easy to grep for. Signed-off-by: -
Thomas Gleixner authored
Signed-off-by: -
Thomas Gleixner authored
We want to rename dev->state, so provide proper get and set functions. Rename clockevents_set_state() to clockevents_switch_state() to avoid confusion. Signed-off-by: -
Viresh Kumar authored
Use accessor functions to check the state of clockevent devices in core code. Signed-off-by:
-
Viresh Kumar authored
Some clockevent drivers, once migrated to use per-state callbacks, need to check the state of the clockevent device in their callbacks or interrupt handler. Add accessor functions clockevent_state_*() to get this information. Signed-off-by:
-
