- 28 May, 2020 11 commits
-
-
Ingo Molnar authored
Move the prototypes for sched_ttwu_pending() and send_call_function_single_ipi() into the newly created kernel/sched/smp.h header, to make sure they are all the same, and to architectures happy that use -Wmissing-prototypes. Signed-off-by:Ingo Molnar <mingo@kernel.org>
-
Peter Zijlstra authored
The recent commit: 90b5363a ("sched: Clean up scheduler_ipi()") got smp_call_function_single_async() subtly wrong. Even though it will return -EBUSY when trying to re-use a csd, that condition is not atomic and still requires external serialization. The change in ttwu_queue_remote() got this wrong. While on first reading ttwu_queue_remote() has an atomic test-and-set that appears to serialize the use, the matching 'release' is not in the right place to actually guarantee this serialization. The actual race is vs the sched_ttwu_pending() call in the idle loop; that can run the wakeup-list without consuming the CSD. Instead of trying to chain the lists, merge them. Signed-off-by:
Peter Zijlstra (Intel) <peterz@infradead.org> Signed-off-by:
-
Peter Zijlstra authored
In preparation of removing rq->wake_list, replace the !list_empty(rq->wake_list) with rq->ttwu_pending. This is not fully equivalent as this new variable is racy. Signed-off-by:
-
Peter Zijlstra authored
Currently irq_work_queue_on() will issue an unconditional arch_send_call_function_single_ipi() and has the handler do irq_work_run(). This is unfortunate in that it makes the IPI handler look at a second cacheline and it misses the opportunity to avoid the IPI. Instead note that struct irq_work and struct __call_single_data are very similar in layout, so use a few bits in the flags word to encode a type and stick the irq_work on the call_single_queue list. Signed-off-by:
-
Peter Zijlstra authored
Just like the ttwu_queue_remote() IPI, make use of _TIF_POLLING_NRFLAG to avoid sending IPIs to idle CPUs. [ mingo: Fix UP build bug. ] Signed-off-by:
-
Peter Zijlstra authored
This ensures flush_smp_call_function_queue() is strictly about call_single_queue. Signed-off-by:
-
Peter Zijlstra authored
The call_single_queue can contain (two) different callbacks, synchronous and asynchronous. The current interrupt handler runs them in-order, which means that remote CPUs that are waiting for their synchronous call can be delayed by running asynchronous callbacks. Rework the interrupt handler to first run the synchonous callbacks. Signed-off-by:
-
Peter Zijlstra authored
The recent commit: 90b5363a ("sched: Clean up scheduler_ipi()") got smp_call_function_single_async() subtly wrong. Even though it will return -EBUSY when trying to re-use a csd, that condition is not atomic and still requires external serialization. The change in kick_ilb() got this wrong. While on first reading kick_ilb() has an atomic test-and-set that appears to serialize the use, the matching 'release' is not in the right place to actually guarantee this serialization. Rework the nohz_idle_balance() trigger so that the release is in the IPI callback and thus guarantees the required serialization for the CSD. Fixes: 90b5363a ("sched: Clean up scheduler_ipi()") Reported-by:
Qian Cai <cai@lca.pw> Signed-off-by:
Frederic Weisbecker <frederic@kernel.org> Cc: mgorman@techsingularity.net Link: https://lore.kernel.org/r/20200526161907.778543557@infradead.org
-
Ingo Molnar authored
We are going to rely on the loosening of RCU callback semantics, introduced by this commit: 806f04e9: ("rcu: Allow for smp_call_function() running callbacks from idle") Signed-off-by:
-
Ingo Molnar authored
Signed-off-by: -
Peter Zijlstra authored
Current RCU hard relies on smp_call_function() callbacks running from interrupt context. A pending optimization is going to break that, it will allow idle CPUs to run the callbacks from the idle loop. This avoids raising the IPI on the requesting CPU and avoids handling an exception on the receiving CPU. Change rcu_is_cpu_rrupt_from_idle() to also accept task context, provided it is the idle task. Signed-off-by:
Paul E. McKenney <paulmck@kernel.org> Reviewed-by:
Joel Fernandes (Google) <joel@joelfernandes.org> Link: https://lore.kernel.org/r/20200527171236.GC706495@hirez.programming.kicks-ass.net
-
- 26 May, 2020 1 commit
-
-
Jens Axboe authored
Stefano reported a crash with using SQPOLL with io_uring: BUG: kernel NULL pointer dereference, address: 00000000000003b0 CPU: 2 PID: 1307 Comm: io_uring-sq Not tainted 5.7.0-rc7 #11 RIP: 0010:task_numa_work+0x4f/0x2c0 Call Trace: task_work_run+0x68/0xa0 io_sq_thread+0x252/0x3d0 kthread+0xf9/0x130 ret_from_fork+0x35/0x40 which is task_numa_work() oopsing on current->mm being NULL. The task work is queued by task_tick_numa(), which checks if current->mm is NULL at the time of the call. But this state isn't necessarily persistent, if the kthread is using use_mm() to temporarily adopt the mm of a task. Change the task_tick_numa() check to exclude kernel threads in general, as it doesn't make sense to attempt ot balance for kthreads anyway. Reported-by:
Stefano Garzarella <sgarzare@redhat.com> Signed-off-by:
Jens Axboe <axboe@kernel.dk> Signed-off-by:
-
- 25 May, 2020 2 commits
-
-
Mel Gorman authored
The previous commit: c6e7bd7a: ("sched/core: Optimize ttwu() spinning on p->on_cpu") avoids spinning on p->on_rq when the task is descheduling, but only if the wakee is on a CPU that does not share cache with the waker. This patch offloads the activation of the wakee to the CPU that is about to go idle if the task is the only one on the runqueue. This potentially allows the waker task to continue making progress when the wakeup is not strictly synchronous. This is very obvious with netperf UDP_STREAM running on localhost. The waker is sending packets as quickly as possible without waiting for any reply. It frequently wakes the server for the processing of packets and when netserver is using local memory, it quickly completes the processing and goes back to idle. The waker often observes that netserver is on_rq and spins excessively leading to a drop in throughput. This is a comparison of 5.7-rc6 against "sched: Optimize ttwu() spinning on p->on_cpu" and against this patch labeled vanilla, optttwu-v1r1 and localwakelist-v1r2 respectively. 5.7.0-rc6 5.7.0-rc6 5.7.0-rc6 vanilla optttwu-v1r1 localwakelist-v1r2 Hmean send-64 251.49 ( 0.00%) 258.05 * 2.61%* 305.59 * 21.51%* Hmean send-128 497.86 ( 0.00%) 519.89 * 4.43%* 600.25 * 20.57%* Hmean send-256 944.90 ( 0.00%) 997.45 * 5.56%* 1140.19 * 20.67%* Hmean send-1024 3779.03 ( 0.00%) 3859.18 * 2.12%* 4518.19 * 19.56%* Hmean send-2048 7030.81 ( 0.00%) 7315.99 * 4.06%* 8683.01 * 23.50%* Hmean send-3312 10847.44 ( 0.00%) 11149.43 * 2.78%* 12896.71 * 18.89%* Hmean send-4096 13436.19 ( 0.00%) 13614.09 ( 1.32%) 15041.09 * 11.94%* Hmean send-8192 22624.49 ( 0.00%) 23265.32 * 2.83%* 24534.96 * 8.44%* Hmean send-16384 34441.87 ( 0.00%) 36457.15 * 5.85%* 35986.21 * 4.48%* Note that this benefit is not universal to all wakeups, it only applies to the case where the waker often spins on p->on_rq. The impact can be seen from a "perf sched latency" report generated from a single iteration of one packet size: ----------------------------------------------------------------------------------------------------------------- Task | Runtime ms | Switches | Average delay ms | Maximum delay ms | Maximum delay at | ----------------------------------------------------------------------------------------------------------------- vanilla netperf:4337 | 21709.193 ms | 2932 | avg: 0.002 ms | max: 0.041 ms | max at: 112.154512 s netserver:4338 | 14629.459 ms | 5146990 | avg: 0.001 ms | max: 1615.864 ms | max at: 140.134496 s localwakelist-v1r2 netperf:4339 | 29789.717 ms | 2460 | avg: 0.002 ms | max: 0.059 ms | max at: 138.205389 s netserver:4340 | 18858.767 ms | 7279005 | avg: 0.001 ms | max: 0.362 ms | max at: 135.709683 s ----------------------------------------------------------------------------------------------------------------- Note that the average wakeup delay is quite small on both the vanilla kernel and with the two patches applied. However, there are significant outliers with the vanilla kernel with the maximum one measured as 1615 milliseconds with a vanilla kernel but never worse than 0.362 ms with both patches applied and a much higher rate of context switching. Similarly a separate profile of cycles showed that 2.83% of all cycles were spent in try_to_wake_up() with almost half of the cycles spent on spinning on p->on_rq. With the two patches, the percentage of cycles spent in try_to_wake_up() drops to 1.13% Signed-off-by:
Mel Gorman <mgorman@techsingularity.net> Signed-off-by:
-
Peter Zijlstra authored
Both Rik and Mel reported seeing ttwu() spend significant time on: smp_cond_load_acquire(&p->on_cpu, !VAL); Attempt to avoid this by queueing the wakeup on the CPU that owns the p->on_cpu value. This will then allow the ttwu() to complete without further waiting. Since we run schedule() with interrupts disabled, the IPI is guaranteed to happen after p->on_cpu is cleared, this is what makes it safe to queue early. Signed-off-by:
-
- 19 May, 2020 26 commits
-
-
Huaixin Chang authored
When users write some huge number into cpu.cfs_quota_us or cpu.rt_runtime_us, overflow might happen during to_ratio() shifts of schedulable checks. to_ratio() could be altered to avoid unnecessary internal overflow, but min_cfs_quota_period is less than 1 << BW_SHIFT, so a cutoff would still be needed. Set a cap MAX_BW for cfs_quota_us and rt_runtime_us to prevent overflow. Signed-off-by:
Huaixin Chang <changhuaixin@linux.alibaba.com> Signed-off-by:
Ben Segall <bsegall@google.com> Link: https://lkml.kernel.org/r/20200425105248.60093-1-changhuaixin@linux.alibaba.com
-
Muchun Song authored
The user_mode(task_pt_regs(tsk)) always return true for user thread, and false for kernel thread. So it means that the cpuacct.usage_sys is the time that kernel thread uses not the time that thread uses in the kernel mode. We can try get_irq_regs() first, if it is NULL, then we can fall back to task_pt_regs(). Signed-off-by:
Muchun Song <songmuchun@bytedance.com> Signed-off-by:
-
Gustavo A. R. Silva authored
The current codebase makes use of the zero-length array language extension to the C90 standard, but the preferred mechanism to declare variable-length types such as these ones is a flexible array member[1][2], introduced in C99: struct foo { int stuff; struct boo array[]; }; By making use of the mechanism above, we will get a compiler warning in case the flexible array does not occur last in the structure, which will help us prevent some kind of undefined behavior bugs from being inadvertently introduced[3] to the codebase from now on. Also, notice that, dynamic memory allocations won't be affected by this change: "Flexible array members have incomplete type, and so the sizeof operator may not be applied. As a quirk of the original implementation of zero-length arrays, sizeof evaluates to zero."[1] sizeof(flexible-array-member) triggers a warning because flexible array members have incomplete type[1]. There are some instances of code in which the sizeof operator is being incorrectly/erroneously applied to zero-length arrays and the result is zero. Such instances may be hiding some bugs. So, this work (flexible-array member conversions) will also help to get completely rid of those sorts of issues. This issue was found with the help of Coccinelle. [1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html [2] https://github.com/KSPP/linux/issues/21 [3] commit 76497732 ("cxgb3/l2t: Fix undefined behaviour") Signed-off-by:Gustavo A. R. Silva <gustavoars@kernel.org> Signed-off-by:
-
Vincent Guittot authored
update_tg_cfs_*() propagate the impact of the attach/detach of an entity down into the cfs_rq hierarchy and must keep the sync with the current pelt window. Even if we can't sync child cfs_rq and its group se, we can sync the group se and its parent cfs_rq with current position in the PELT window. In fact, we must keep them sync in order to stay also synced with others entities and group entities that are already attached to the cfs_rq. Signed-off-by:
Vincent Guittot <vincent.guittot@linaro.org> Signed-off-by:
-
Muchun Song authored
The cpuacct_charge() and cpuacct_account_field() are called with rq->lock held, and this means preemption(and IRQs) are indeed disabled, so it is safe to use __this_cpu_*() to allow for better code-generation. Signed-off-by:
-
Vincent Guittot authored
enqueue_task_fair jumps to enqueue_throttle label when cfs_rq_of(se) is throttled which means that se can't be NULL in such case and we can move the label after the if (!se) statement. Futhermore, the latter can be removed because se is always NULL when reaching this point. Signed-off-by:
Phil Auld <pauld@redhat.com> Link: https://lkml.kernel.org/r/20200513135502.4672-1-vincent.guittot@linaro.org
-
Peter Zijlstra authored
-
Vincent Guittot authored
Although not exactly identical, unthrottle_cfs_rq() and enqueue_task_fair() are quite close and follow the same sequence for enqueuing an entity in the cfs hierarchy. Modify unthrottle_cfs_rq() to use the same pattern as enqueue_task_fair(). This fixes a problem already faced with the latter and add an optimization in the last for_each_sched_entity loop. Fixes: fe61468b (sched/fair: Fix enqueue_task_fair warning) Reported-by Tao Zhou <zohooouoto@zoho.com.cn> Signed-off-by:
-
Pavankumar Kondeti authored
The intention of commit 96e74ebf ("sched/debug: Add task uclamp values to SCHED_DEBUG procfs") was to print requested and effective task uclamp values. The requested values printed are read from p->uclamp, which holds the last effective values. Fix this by printing the values from p->uclamp_req. Fixes: 96e74ebf ("sched/debug: Add task uclamp values to SCHED_DEBUG procfs") Signed-off-by:
Pavankumar Kondeti <pkondeti@codeaurora.org> Signed-off-by:
Valentin Schneider <valentin.schneider@arm.com> Tested-by:
-
Phil Auld authored
sched/fair: Fix enqueue_task_fair warning some more The recent patch, fe61468b (sched/fair: Fix enqueue_task_fair warning) did not fully resolve the issues with the rq->tmp_alone_branch != &rq->leaf_cfs_rq_list warning in enqueue_task_fair. There is a case where the first for_each_sched_entity loop exits due to on_rq, having incompletely updated the list. In this case the second for_each_sched_entity loop can further modify se. The later code to fix up the list management fails to do what is needed because se does not point to the sched_entity which broke out of the first loop. The list is not fixed up because the throttled parent was already added back to the list by a task enqueue in a parallel child hierarchy. Address this by calling list_add_leaf_cfs_rq if there are throttled parents while doing the second for_each_sched_entity loop. Fixes: fe61468b ("sched/fair: Fix enqueue_task_fair warning") Suggested-by:
Dietmar Eggemann <dietmar.eggemann@arm.com> Reviewed-by:
-
Thomas Gleixner authored
Same as rcu_is_watching() but without the preempt_disable/enable() pair inside the function. It is merked noinstr so it ends up in the non-instrumentable text section. This is useful for non-preemptible code especially in the low level entry section. Using rcu_is_watching() there results in a call to the preempt_schedule_notrace() thunk which triggers noinstr section warnings in objtool. Signed-off-by:
Thomas Gleixner <tglx@linutronix.de> Acked-by:
-
Thomas Gleixner authored
Interrupts and exceptions invoke rcu_irq_enter() on entry and need to invoke rcu_irq_exit() before they either return to the interrupted code or invoke the scheduler due to preemption. The general assumption is that RCU idle code has to have preemption disabled so that a return from interrupt cannot schedule. So the return from interrupt code invokes rcu_irq_exit() and preempt_schedule_irq(). If there is any imbalance in the rcu_irq/nmi* invocations or RCU idle code had preemption enabled then this goes unnoticed until the CPU goes idle or some other RCU check is executed. Provide rcu_irq_exit_preempt() which can be invoked from the interrupt/exception return code in case that preemption is enabled. It invokes rcu_irq_exit() and contains a few sanity checks in case that CONFIG_PROVE_RCU is enabled to catch such issues directly. Signed-off-by:
Alexandre Chartre <alexandre.chartre@oracle.com> Acked-by:
-
Paul E. McKenney authored
The rcu_nmi_enter_common() and rcu_nmi_exit_common() functions take an "irq" parameter that indicates whether these functions have been invoked from an irq handler (irq==true) or an NMI handler (irq==false). However, recent changes have applied notrace to a few critical functions such that rcu_nmi_enter_common() and rcu_nmi_exit_common() many now rely on in_nmi(). Note that in_nmi() works no differently than before, but rather that tracing is now prohibited in code regions where in_nmi() would incorrectly report NMI state. Therefore remove the "irq" parameter and inline rcu_nmi_enter_common() and rcu_nmi_exit_common() into rcu_nmi_enter() and rcu_nmi_exit(), respectively. Signed-off-by:
-
Thomas Gleixner authored
These functions are invoked from context tracking and other places in the low level entry code. Move them into the .noinstr.text section to exclude them from instrumentation. Mark the places which are safe to invoke traceable functions with instrumentation_begin/end() so objtool won't complain. Signed-off-by:
-
Peter Zijlstra authored
A few exceptions (like #DB and #BP) can happen at any location in the code, this then means that tracers should treat events from these exceptions as NMI-like. The interrupted context could be holding locks with interrupts disabled for instance. Similarly, #MC is an actual NMI-like exception. All of them use ist_enter() which only concerns itself with RCU, but does not do any of the other setup that NMIs need. This means things like: printk() raw_spin_lock_irq(&logbuf_lock); <#DB/#BP/#MC> printk() raw_spin_lock_irq(&logbuf_lock); are entirely possible (well, not really since printk tries hard to play nice, but the concept stands). So replace ist_enter() with nmi_enter(). Also observe that any nmi_enter() caller must be both notrace and NOKPROBE, or in the noinstr text section. Signed-off-by:
-
Peter Zijlstra authored
Convert #MC over to using task_work_add(); it will run the same code slightly later, on the return to user path of the same exception. Signed-off-by:
-
Thomas Gleixner authored
This is completely overengineered and definitely not an interface which should be made available to anything else than this particular MCE case. Signed-off-by:
-
Peter Zijlstra authored
If a tracer is invoked before in_nmi() becomes true, the tracer can no longer detect it is called from NMI context and behave correctly. Therefore change nmi_{enter,exit}() to use __preempt_count_{add,sub}() as the normal preempt_count_{add,sub}() have a (desired) function trace entry. This fixes a potential issue with the current code; when the function-tracer has stack-tracing enabled __trace_stack() will malfunction when it hits the preempt_count_add() function entry from NMI context. Suggested-by:Steven Rostedt (VMware) <rosted@goodmis.org> Signed-off-by:
Steven Rostedt (VMware) <rostedt@goodmis.org> Reviewed-by:
-
Peter Zijlstra authored
SuperH is the last remaining user of arch_ftrace_nmi_{enter,exit}(), remove it from the generic code and into the SuperH code. Signed-off-by: -
Peter Zijlstra authored
These functions are called {early,late} in nmi_{enter,exit} and should not be traced or probed. They are also puny, so 'inline' them. Reported-by: -
Peter Zijlstra authored
Since there are already a number of sites (ARM64, PowerPC) that effectively nest nmi_enter(), make the primitive support this before adding even more. Signed-off-by:
Marc Zyngier <maz@kernel.org> Acked-by:
Will Deacon <will@kernel.org> Cc: Michael Ellerman <mpe@ellerman.id.au> Link: https://lkml.kernel.org/r/20200505134100.864179229@linutronix.de
-
Frederic Weisbecker authored
When using nmi_enter() recursively, arch_nmi_enter() must also be recursion safe. In particular, it must be ensured that HCR_TGE is always set while in NMI context when in HYP mode, and be restored to it's former state when done. The current code fails this when interleaved wrong. Notably it overwrites the original hcr state on nesting. Introduce a nesting counter to make sure to store the original value. Signed-off-by:
-
Peter Zijlstra authored
It happens early in nmi_enter(), no tracing, probing or other funnies allowed. Specifically as nmi_enter() will be used in do_debug(), which would cause recursive exceptions when kprobed. Signed-off-by:
-
Petr Mladek authored
There is plenty of space in the printk_context variable. Reserve one byte there for the NMI context to be on the safe side. It should never overflow. The BUG_ON(in_nmi() == NMI_MASK) in nmi_enter() will trigger much earlier. Signed-off-by:
Petr Mladek <pmladek@suse.com> Signed-off-by:
-
Thomas Gleixner authored
Get the noinstr section and annotation markers to base the RCU parts on.
-
Thomas Gleixner authored
Some code pathes, especially the low level entry code, must be protected against instrumentation for various reasons: - Low level entry code can be a fragile beast, especially on x86. - With NO_HZ_FULL RCU state needs to be established before using it. Having a dedicated section for such code allows to validate with tooling that no unsafe functions are invoked. Add the .noinstr.text section and the noinstr attribute to mark functions. noinstr implies notrace. Kprobes will gain a section check later. Provide also a set of markers: instrumentation_begin()/end() These are used to mark code inside a noinstr function which calls into regular instrumentable text section as safe. The instrumentation markers are only active when CONFIG_DEBUG_ENTRY is enabled as the end marker emits a NOP to prevent the compiler from merging the annotation points. This means the objtool verification requires a kernel compiled with this option. Signed-off-by:
-
