https://github.com/torvalds/linux
Revision f841b682baef90ee144df8b12e2c76aa460717c1 authored by Chengming Zhou on 07 December 2022, 12:40:23 UTC, committed by Peter Zijlstra on 27 December 2022, 11:44:00 UTC
We encounter perf warnings when using cgroup events like: cd /sys/fs/cgroup mkdir test perf stat -e cycles -a -G test Which then triggers: WARNING: CPU: 0 PID: 690 at kernel/events/core.c:849 perf_cgroup_switch+0xb2/0xc0 Call Trace: <TASK> __schedule+0x4ae/0x9f0 ? _raw_spin_unlock_irqrestore+0x23/0x40 ? __cond_resched+0x18/0x20 preempt_schedule_common+0x2d/0x70 __cond_resched+0x18/0x20 wait_for_completion+0x2f/0x160 ? cpu_stop_queue_work+0x9e/0x130 affine_move_task+0x18a/0x4f0 WARNING: CPU: 0 PID: 690 at kernel/events/core.c:829 ctx_sched_in+0x1cf/0x1e0 Call Trace: <TASK> ? ctx_sched_out+0xb7/0x1b0 perf_cgroup_switch+0x88/0xc0 __schedule+0x4ae/0x9f0 ? _raw_spin_unlock_irqrestore+0x23/0x40 ? __cond_resched+0x18/0x20 preempt_schedule_common+0x2d/0x70 __cond_resched+0x18/0x20 wait_for_completion+0x2f/0x160 ? cpu_stop_queue_work+0x9e/0x130 affine_move_task+0x18a/0x4f0 The above two warnings are not complete here since I remove other unimportant information. The problem is caused by the perf cgroup events tracking: CPU0 CPU1 perf_event_open() perf_event_alloc() account_event() account_event_cpu() atomic_inc(perf_cgroup_events) __perf_event_task_sched_out() if (atomic_read(perf_cgroup_events)) perf_cgroup_switch() // kernel/events/core.c:849 WARN_ON_ONCE(cpuctx->ctx.nr_cgroups == 0) if (READ_ONCE(cpuctx->cgrp) == cgrp) // false return perf_ctx_lock() ctx_sched_out() cpuctx->cgrp = cgrp ctx_sched_in() perf_cgroup_set_timestamp() // kernel/events/core.c:829 WARN_ON_ONCE(!ctx->nr_cgroups) perf_ctx_unlock() perf_install_in_context() cpu_function_call() __perf_install_in_context() add_event_to_ctx() list_add_event() perf_cgroup_event_enable() ctx->nr_cgroups++ cpuctx->cgrp = X We can see from above that we wrongly use percpu atomic perf_cgroup_events to check if we need to perf_cgroup_switch(), which should only be used when we know this CPU has cgroup events enabled. The commit bd2756811766 ("perf: Rewrite core context handling") change to have only one context per-CPU, so we can just use cpuctx->cgrp to check if this CPU has cgroup events enabled. So percpu atomic perf_cgroup_events is not needed. Fixes: bd2756811766 ("perf: Rewrite core context handling") Signed-off-by: Chengming Zhou <zhouchengming@bytedance.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Tested-by: Ravi Bangoria <ravi.bangoria@amd.com> Link: https://lkml.kernel.org/r/20221207124023.66252-1-zhouchengming@bytedance.com
1 parent e2d3714
Tip revision: f841b682baef90ee144df8b12e2c76aa460717c1 authored by Chengming Zhou on 07 December 2022, 12:40:23 UTC
perf/core: Fix cgroup events tracking
perf/core: Fix cgroup events tracking
Tip revision: f841b68
File | Mode | Size |
---|---|---|
Documentation | ||
LICENSES | ||
arch | ||
block | ||
certs | ||
crypto | ||
drivers | ||
fs | ||
include | ||
init | ||
io_uring | ||
ipc | ||
kernel | ||
lib | ||
mm | ||
net | ||
rust | ||
samples | ||
scripts | ||
security | ||
sound | ||
tools | ||
usr | ||
virt | ||
.clang-format | -rw-r--r-- | 20.0 KB |
.cocciconfig | -rw-r--r-- | 59 bytes |
.get_maintainer.ignore | -rw-r--r-- | 151 bytes |
.gitattributes | -rw-r--r-- | 62 bytes |
.gitignore | -rw-r--r-- | 2.0 KB |
.mailmap | -rw-r--r-- | 24.6 KB |
.rustfmt.toml | -rw-r--r-- | 369 bytes |
COPYING | -rw-r--r-- | 496 bytes |
CREDITS | -rw-r--r-- | 99.3 KB |
Kbuild | -rw-r--r-- | 2.5 KB |
Kconfig | -rw-r--r-- | 555 bytes |
MAINTAINERS | -rw-r--r-- | 681.0 KB |
Makefile | -rw-r--r-- | 69.5 KB |
README | -rw-r--r-- | 727 bytes |
![swh spinner](/static/img/swh-spinner.gif)
Computing file changes ...