- 22 Apr, 2013 6 commits
-
-
Frederic Weisbecker authored
The scheduler IPI is used by the scheduler to kick full dynticks CPUs asynchronously when more than one task are running or when a new timer list timer is enqueued. This way the destination CPU can decide to restart the tick to handle this new situation. Now let's call that kick in the scheduler IPI. (Reusing the scheduler IPI rather than implementing a new IPI was suggested by Peter Zijlstra a while ago) Signed-off-by:Frederic Weisbecker <fweisbec@gmail.com> Cc: Chris Metcalf <cmetcalf@tilera.com> Cc: Christoph Lameter <cl@linux.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Gilad Ben Yossef <gilad@benyossef.com> Cc: Hakan Akkan <hakanakkan@gmail.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Kevin Hilman <khilman@linaro.org> Cc: Li Zhong <zhong@linux.vnet.ibm.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Thomas Gleixner <tglx@linutronix.de>
-
Frederic Weisbecker authored
Provide a new helper to be called from the full dynticks engine before stopping the tick in order to make sure we don't stop it when there is more than one task running on the CPU. This way we make sure that the tick stays alive to maintain fairness. Signed-off-by: -
Frederic Weisbecker authored
Kick the tick on full dynticks CPUs when they get more than one task running on their queue. This makes sure that local fairness is maintained by the tick on the destination. This is done regardless of these tasks' class. We should be able to be more clever in the future depending on these. eg: a CPU that runs a SCHED_FIFO task doesn't need to maintain fairness against local pending tasks of the fair class. But keep things simple for now. Signed-off-by: -
Frederic Weisbecker authored
Provide a new helper that help full dynticks CPUs to prevent from stopping their tick in case there are events in the local rotation list. This way we make sure that perf_event_task_tick() is serviced on demand. Signed-off-by: -
Frederic Weisbecker authored
Kick the current CPU's tick by sending it a self IPI when an event is queued on the rotation list and it is the first element inserted. This makes sure that perf_event_task_tick() works on full dynticks CPUs. Signed-off-by: -
Frederic Weisbecker authored
The test that checks if a CPU can stop its tick from posix CPU timers angle was mistakenly inverted. What we want is to prevent the tick from being stopped as long as the current CPU's task runs a posix CPU timer. Fix this. Signed-off-by:
-
- 21 Apr, 2013 2 commits
-
-
Ingo Molnar authored
Merge branch 'timers/nohz-posix-timers-v2' of git://git.kernel.org/pub/scm/linux/kernel/git/frederic/linux-dynticks into timers/nohz Pull posix cpu timers handling on full dynticks from Frederic Weisbecker. Signed-off-by:
Ingo Molnar <mingo@kernel.org>
-
Ingo Molnar authored
Merge branch 'timers/nohz-reviewed' of git://git.kernel.org/pub/scm/linux/kernel/git/frederic/linux-dynticks into timers/nohz Pull full dynticks timekeeping and RCU improvements from Frederic Weisbecker. Signed-off-by:
-
- 19 Apr, 2013 5 commits
-
-
Frederic Weisbecker authored
Bring a new helper that the full dynticks infrastructure can call in order to know if it can safely stop the tick from the posix cpu timers subsystem point of view. Signed-off-by: -
Frederic Weisbecker authored
Kick the full dynticks CPUs when a posix cpu timer is enqueued by way of a standard call to posix_cpu_timer_set() or set_process_cpu_timer(). This also include rescheduled firing timers. This way they can re-evaluate the state of (and possibly restart) their tick against the new expiry modification. Signed-off-by: -
Frederic Weisbecker authored
Provide a new kernel config that defaults all CPUs to be part of the full dynticks range, except the boot one for timekeeping. This default setting is overriden by the nohz_full= boot option if passed by the user. This is helpful for those who don't need a finegrained range of full dynticks CPU and also for automated testing. Suggested-by:
Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by:
-
Frederic Weisbecker authored
We need full dynticks CPU to also be RCU nocb so that we don't have to keep the tick to handle RCU callbacks. Make sure the range passed to nohz_full= boot parameter is a subset of rcu_nocbs= The CPUs that fail to meet this requirement will be excluded from the nohz_full range. This is checked early in boot time, before any CPU has the opportunity to stop its tick. Suggested-by:
Steven Rostedt <rostedt@goodmis.org> Reviewed-by:
-
Frederic Weisbecker authored
The timekeeping job must be able to run early on boot because there may be some pre-SMP (and thus pre-initcalls ) components that rely on it. The IO-APIC is one such users as it tests the timer health by watching jiffies progression. Given that it happens before we know the initial online set, we can't rely on it to select a timekeeper. We need one before SMP time otherwise we simply crash on boot. To fix this and keep things simple for now, force the boot CPU outside of the full dynticks range in any case and do this early on kernel parameter parsing time. We might want a trickier solution later, expecially for aSMP architectures that need to assign housekeeping tasks to arbitrary low power CPUs. But it's still first pass KISS time for now. Reviewed-by:
-
- 18 Apr, 2013 1 commit
-
-
Frederic Weisbecker authored
Provide two new helpers in order to notify the full dynticks CPUs about some internal system changes against which they may reconsider the state of their tick. Some practical examples include: posix cpu timers, perf tick and sched clock tick. For now the notifying handler, implemented through IPIs, is a stub that will be implemented when we get the tick stop/restart infrastructure in. Signed-off-by:
-
- 15 Apr, 2013 5 commits
-
-
Paul E. McKenney authored
Adaptive-ticks CPUs inform RCU when they enter kernel mode, but they do not necessarily turn the scheduler-clock tick back on. This state of affairs could result in RCU waiting on an adaptive-ticks CPU running for an extended period in kernel mode. Such a CPU will never run the RCU state machine, and could therefore indefinitely extend the RCU state machine, sooner or later resulting in an OOM condition. This patch, inspired by an earlier patch by Frederic Weisbecker, therefore causes RCU's force-quiescent-state processing to check for this condition and to send an IPI to CPUs that remain in that state for too long. "Too long" currently means about three jiffies by default, which is quite some time for a CPU to remain in the kernel without blocking. The rcu_tree.jiffies_till_first_fqs and rcutree.jiffies_till_next_fqs sysfs variables may be used to tune "too long" if needed. Reported-by:
Josh Triplett <josh@joshtriplett.org> Signed-off-by:
-
Frederic Weisbecker authored
Remove the "single task" statement from CONFIG_NO_HZ_FULL title. The constraint can be invalidated when tasks from other sched classes than SCHED_FAIR are running. Moreover it's possible that hrtick join the party in the future. Also add a line about the dependency on SMP. Reported-by:
-
Frederic Weisbecker authored
Rename CONFIG_PERIODIC_HZ to CONFIG_HZ_PERIODIC in order to stay consistent with other tick implementation entries: CONFIG_HZ_PERIODIC CONFIG_NO_HZ_IDLE CONFIG_NO_HZ_FULL Reported-by:
-
Frederic Weisbecker authored
"Extended nohz" was used as a naming base for the full dynticks API and Kconfig symbols. It reflects the fact the system tries to stop the tick in more places than just idle. But that "extended" name is a bit opaque and vague. Rename it to "full" makes it clearer what the system tries to do under this config: try to shutdown the tick anytime it can. The various constraints that prevent that to happen shouldn't be considered as fundamental properties of this feature but rather technical issues that may be solved in the future. Reported-by:
-
Frederic Weisbecker authored
In order to enforce backward compatibility with older config files, we want the new dynticks-idle Kconfig entry to default its value to the one of the old CONFIG_NO_HZ symbol if present. Namely we want: config NO_HZ # old obsolete dynticks idle symbol bool config NO_HZ_IDLE # new dynticks idle symbol default NO_HZ However Kconfig prevents this to work if the old symbol is not visible. And this is currently the case because NO_HZ lacks a title in order to show it in make oldconfig and alike. To fix this, bring a minimal title and help text to the obsolete Kconfig entry that explains its purpose. This makes the "defaulting" to work. Reported-by:
-
- 03 Apr, 2013 3 commits
-
-
Frederic Weisbecker authored
Given that we apply a few restrictions on the full dynticks CPUs range (keep an online timekeeper oustide the range, then in the future have the range be an RCU nocb CPUs subset), let's print the final resulting range of full dynticks CPUs to the user so that he knows what's really going to run. Signed-off-by: -
Frederic Weisbecker authored
Now the user has the choice between three implementations of the timer tick: * Static periodic tick * Idle dynticks * Full dynticks At least for now, these are mutually exclusive choices, so let's rely on the proper Kconfig feature to display these to the user. A new entry CONFIG_NO_HZ_IDLE is created and the old CONFIG_NO_HZ maps to it for config file backward compatibility. The old name was too general now that we have more granular dynticks implementations. While at it, add some explanation to help the user on his decision between the 3 entries. Reported-by:
-
Frederic Weisbecker authored
We are planning to convert the dynticks Kconfig options layout into a choice menu. The user must be able to easily pick any of the following implementations: constant periodic tick, idle dynticks, full dynticks. As this implies a mutual exclusion, the two dynticks implementions need to converge on the selection of a common Kconfig option in order to ease the sharing of a common infrastructure. It would thus seem pretty natural to reuse CONFIG_NO_HZ to that end. It already implements all the idle dynticks code and the full dynticks depends on all that code for now. So ideally the choice menu would propose CONFIG_NO_HZ_IDLE and CONFIG_NO_HZ_EXTENDED then both would select CONFIG_NO_HZ. On the other hand we want to stay backward compatible: if CONFIG_NO_HZ is set in an older config file, we want to enable CONFIG_NO_HZ_IDLE by default. But we can't afford both at the same time or we run into a circular dependency: 1) CONFIG_NO_HZ_IDLE and CONFIG_NO_HZ_EXTENDED both select CONFIG_NO_HZ 2) If CONFIG_NO_HZ is set, we default to CONFIG_NO_HZ_IDLE We might be able to support that from Kconfig/Kbuild but it may not be wise to introduce such a confusing behaviour. So to solve this, create a new CONFIG_NO_HZ_COMMON option which gathers the common code between idle and full dynticks (that common code for now is simply the idle dynticks code) and select it from their referring Kconfig. Then we'll later create CONFIG_NO_HZ_IDLE and map CONFIG_NO_HZ to it for backward compatibility. Signed-off-by:
-
- 02 Apr, 2013 1 commit
-
-
Frederic Weisbecker authored
The full dynticks feature only shows up when all its Kconfig dependencies are met (RCU nocbs, RCU user mode, ...) This is far from being user friendly as those who want to activate this feature need to look into the Kconfig files and iterate through each dependency then activate these by hand in order to show and select the full dynticks Kconfig option. So process the other way around: show up the Kconfig option if the minimal low level dependencies are met and activate the high level ones when we enable the feature. Note there is one exception in the picture: CONFIG_VIRT_CPU_ACCOUNTING_GEN is part of a Kconfig choice menu and it appears we can't select it from another Kconfig selection when it's under such layout. So for now this particular item stays as a passive dependency. Reported-by:
-
- 21 Mar, 2013 3 commits
-
-
Frederic Weisbecker authored
Wake up a CPU when a timer list timer is enqueued there and the target is part of the full dynticks range. Sending an IPI to it makes it reconsidering the next timer to program on top of recent updates. This may later be improved by checking if the tick is really stopped on the target. This would need some careful synchronization though. So deal with such optimization later and start simple. Signed-off-by: -
Frederic Weisbecker authored
This way the full nohz CPUs can safely run with the tick stopped with a guarantee that somebody else is taking care of the jiffies and GTOD progression. Once the duty is attributed to a CPU, it won't change. Also that CPU can't enter into dyntick idle mode or be hot unplugged. This may later be improved from a power consumption POV. At least we should be able to share the duty amongst all CPUs outside the full dynticks range. Then the duty could even be shared with full dynticks CPUs when those can't stop their tick for any reason. But let's start with that very simple approach first. Signed-off-by:
-
Frederic Weisbecker authored
For extreme usecases such as Real Time or HPC, having the ability to shutdown the tick when a single task runs on a CPU is a desired feature: * Reducing the amount of interrupts improves throughput for CPU-bound tasks. The CPU is less distracted from its real job, from an execution time and from the cache point of views. * This also improve latency response as we have less critical sections. Start with introducing a very simple interface to define full dynticks CPU: use a boot time option defined cpumask through the "nohz_extended=" kernel parameter. CPUs that are part of this range will have their tick shutdown whenever possible: provided they run a single task and they don't do kernel activity that require the periodic tick. These details will be later documented in Documentation/* An online CPU must be kept outside this range to handle the timekeeping. Suggested-by:
-
- 18 Mar, 2013 2 commits
-
-
Peter Zijlstra authored
Thomas noted that we do the wakeup preemption check after the wakeup trace point, this means the tracepoint cannot test/report this decision; which is rather important for latency sensitive workloads. Therefore move the tracepoint after doing the preemption check. Suggested-by:
Thomas Gleixner <tglx@linutronix.de> Signed-off-by:
Peter Zijlstra <a.p.zijlstra@chello.nl> Acked-by:
Paul Turner <pjt@google.com> Cc: Mike Galbraith <efault@gmx.de> Link: http://lkml.kernel.org/r/1363254519.26965.9.camel@laptopSigned-off-by:
-
Ingo Molnar authored
Merge branch 'sched/core' of git://git.kernel.org/pub/scm/linux/kernel/git/frederic/linux-dynticks into sched/core Pull CPU runtime stats/accounting fixes from Frederic Weisbecker: " Some users are complaining that their threadgroup's runtime accounting freezes after a week or so of intense cpu-bound workload. This set tries to fix the issue by reducing the risk of multiplication overflow in the cputime scaling code. " Stanislaw Gruszka further explained the historic context and impact of the bug: " Commit 0cf55e1e start to use scalling for whole thread group, so increase chances of hitting multiplication overflow, depending on how many CPUs are on the system. We have multiplication utime * rtime for one thread since commit b27f03d4. Overflow will happen after: rtime * utime > 0xffffffffffffffff jiffies if thread utilize 100% of CPU time, that gives: rtime > sqrt(0xffffffffffffffff) jiffies ritme > sqrt(0xffffffffffffffff) / (24 * 60 * 60 * HZ) days For HZ 100 it will be 497 days for HZ 1000 it will be 49 days. Bug affect only users, who run CPU intensive application for that long period. Also they have to be interested on utime,stime values, as bug has no other visible effect as making those values incorrect. " Signed-off-by:
-
- 14 Mar, 2013 1 commit
-
-
Andrei Epure authored
The min_vruntime variable actually stores the maximum value. The added comment was taken from place_entity function. Signed-off-by:
Andrei Epure <epure.andrei@gmail.com> Cc: peterz@infradead.org Link: http://lkml.kernel.org/r/1363115544-1964-1-git-send-email-epure.andrei@gmail.comSigned-off-by:
-
- 13 Mar, 2013 2 commits
-
-
Frederic Weisbecker authored
Some users have reported that after running a process with hundreds of threads on intensive CPU-bound loads, the cputime of the group started to freeze after a few days. This is due to how we scale the tick-based cputime against the scheduler precise execution time value. We add the values of all threads in the group and we multiply that against the sum of the scheduler exec runtime of the whole group. This easily overflows after a few days/weeks of execution. A proposed solution to solve this was to compute that multiplication on stime instead of utime: 62188451 ("cputime: Avoid multiplication overflow on utime scaling") The rationale behind that was that it's easy for a thread to spend most of its time in userspace under intensive CPU-bound workload but it's much harder to do CPU-bound intensive long run in the kernel. This postulate got defeated when a user recently reported he was still seeing cputime freezes after the above patch. The workload that triggers this issue relates to intensive networking workloads where most of the cputime is consumed in the kernel. To reduce much more the opportunities for multiplication overflow, lets reduce the multiplication factors to the remainders of the division between sched exec runtime and cputime. Assuming the difference between these shouldn't ever be that large, it could work on many situations. This gets the same results as in the upstream scaling code except for a small difference: the upstream code always rounds the results to the nearest integer not greater to what would be the precise result. The new code rounds to the nearest integer either greater or not greater. In practice this difference probably shouldn't matter but it's worth mentioning. If this solution appears not to be enough in the end, we'll need to partly revert back to the behaviour prior to commit 0cf55e1e ("sched, cputime: Introduce thread_group_times()") Back then, the scaling was done on exit() time before adding the cputime of an exiting thread to the signal struct. And then we'll need to scale one-by-one the live threads cputime in thread_group_cputime(). The drawback may be a slightly slower code on exit time. Signed-off-by:
-
Frederic Weisbecker authored
Provide an extended version of div64_u64() that also returns the remainder of the division. We are going to need this to refine the cputime scaling code. Signed-off-by:
-
- 11 Mar, 2013 2 commits
-
-
Andrei Epure authored
Signed-off-by:
-
Li Zefan authored
All warnings: In file included from kernel/sched/core.c:85:0: kernel/sched/sched.h:1036:39: warning: 'struct sched_domain' declared inside parameter list kernel/sched/sched.h:1036:39: warning: its scope is only this definition or declaration, which is probably not what you want It's because struct sched_domain is defined inside #if CONFIG_SMP, while update_group_power() is declared unconditionally. Fix this warning by declaring update_group_power() only if CONFIG_SMP=n. Build tested with CONFIG_SMP enabled and then disabled. Reported-by:
Fengguang Wu <fengguang.wu@intel.com> Signed-off-by:
Li Zefan <lizefan@huawei.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/5137F4BA.2060101@huawei.comSigned-off-by:
-
- 08 Mar, 2013 1 commit
-
-
Ingo Molnar authored
Merge branch 'sched/cputime' of git://git.kernel.org/pub/scm/linux/kernel/git/frederic/linux-dynticks into sched/core Pull cputime changes from Frederic Weisbecker: * Generalize exception handling * Fix race in context tracking state restore on return from exception and irq exit kernel preemption * Fix cputime scaling in full dynticks accounting dynamic off-case * Fix default Kconfig value Signed-off-by:
-
- 07 Mar, 2013 5 commits
-
-
Frederic Weisbecker authored
Until we provide the nohz_mask boot parameter, keeping the context tracking probes disabled by default is pointless since what we want is to runtime test this code anyway. It's furthermore confusing for the users which don't expect the probes to be off when they select RCU user mode or full dynticks cputime accounting. Let's enable these probes selftests by default for now. Suggested: Steven Rostedt <rostedt@goodmis.org> Signed-off-by: -
Frederic Weisbecker authored
The full dynticks cputime accounting is able to account either using the tick or the context tracking subsystem. This way the housekeeping CPU can keep the low overhead tick based solution. This latter mode has a low jiffies resolution granularity and need to be scaled against CFS precise runtime accounting to improve its result. We are doing this for CONFIG_TICK_CPU_ACCOUNTING, now we also need to expand it to full dynticks accounting dynamic off-case as well. Signed-off-by: -
Frederic Weisbecker authored
From the context tracking POV, preempt_schedule_irq() behaves pretty much like an exception: It can be called anytime and schedule another task. But currently it doesn't restore the context tracking state of the preempted code on preempt_schedule_irq() return. As a result, if preempt_schedule_irq() is called in the tiny frame between user_enter() and the actual return to userspace, we resume userspace with the wrong context tracking state. Fix this by using exception_enter/exit() which are a perfect fit for this kind of issue. Signed-off-by: -
Frederic Weisbecker authored
On exception exit, we restore the previous context tracking state based on the regs of the interrupted frame. Iff that frame is in user mode as stated by user_mode() helper, we restore the context tracking user mode. However there is a tiny chunck of low level arch code after we pass through user_enter() and until the CPU eventually resumes userspace. If an exception happens in this tiny area, exception_enter() correctly exits the context tracking user mode but exception_exit() won't restore it because of the value returned by user_mode(regs). As a result we may return to userspace with the wrong context tracking state. To fix this, change exception_enter() to return the context tracking state prior to its call and pass this saved state to exception_exit(). This restores the real context tracking state of the interrupted frame. (May be this patch was suggested to me, I don't recall exactly. If so, sorry for the missing credit). Signed-off-by: -
Frederic Weisbecker authored
Exceptions handling on context tracking should share common treatment: on entry we exit user mode if the exception triggered in that context. Then on exception exit we return to that previous context. Generalize this to avoid duplication across archs. Signed-off-by:
-
- 06 Mar, 2013 1 commit
-
-
Li Zefan authored
It's already declared in include/linux/sched.h Signed-off-by:
-
