Re: [RFC][PATCH 2/4] sched: Make sched_class::set_cpus_allowed() unconditional

[Sasha Levin]

On 05/15/2015 11:43 AM, Peter Zijlstra wrote:
> Give every class a set_cpus_allowed() method, this enables some small
> optimization in the rt,dl implementation by avoiding a double
> cpumask_weight() call.
> 
> Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>

Hey Peter,

Here's the splat I mentioned earlier. I'm not sure if it's caused by this commit
or not, it's just that git blame took me here.

[392896.406248] WARNING: CPU: 9 PID: 51 at kernel/sched/core.c:1178 do_set_cpus_allowed+0x1cf/0x480()
[392896.407070] Modules linked in:
[392896.407408] CPU: 9 PID: 51 Comm: migration/9 Not tainted 4.2.0-rc6-next-20150817-sasha-00041-g9b6b2ab-dirty #2460
[392896.408376]  ffffffffab2c2ac0 ffff880477857948 ffffffffaafaf5a9 0000000000000000
[392896.409126]  ffff880477857988 ffffffffa1323246 ffffffffa13ab19f ffff880477284000
[392896.409874]  ffff8804781e15c0 ffff88006bc03d98 ffffffffae480d20 0000000000000001
[392896.410610] Call Trace:
[392896.410947] dump_stack (lib/dump_stack.c:52)
[392896.411460] warn_slowpath_common (kernel/panic.c:448)
[392896.412037] ? do_set_cpus_allowed (kernel/sched/sched.h:1050 (discriminator 9) kernel/sched/core.c:1180 (discriminator 9))
[392896.412629] warn_slowpath_null (kernel/panic.c:482)
[392896.413183] do_set_cpus_allowed (kernel/sched/sched.h:1050 (discriminator 9) kernel/sched/core.c:1180 (discriminator 9))
[392896.413778] cpuset_cpus_allowed_fallback (include/linux/rcupdate.h:911 kernel/cpuset.c:2381)
[392896.414415] ? cpuset_cpus_allowed_fallback (kernel/cpuset.c:2378)
[392896.415055] select_fallback_rq (kernel/sched/core.c:1589)
[392896.415636] ? put_prev_entity (kernel/sched/fair.c:3218)
[392896.416234] migration_call (kernel/sched/core.c:5230 kernel/sched/core.c:5492)
[392896.416786] notifier_call_chain (kernel/notifier.c:93)
[392896.417359] ? cpu_notify (kernel/cpu.c:334)
[392896.417863] __raw_notifier_call_chain (kernel/notifier.c:395)
[392896.419027] cpu_notify (include/linux/notifier.h:179 kernel/cpu.c:231 kernel/cpu.c:236)
[392896.419616] take_cpu_down (kernel/cpu.c:345)
[392896.420127] multi_cpu_stop (kernel/stop_machine.c:203)
[392896.420671] ? queue_stop_cpus_work (kernel/stop_machine.c:172)
[392896.421275] cpu_stopper_thread (kernel/stop_machine.c:440)
[392896.421849] ? cpu_stop_create (kernel/stop_machine.c:415)
[392896.422404] ? lock_release (kernel/locking/lockdep.c:3643)
[392896.422936] ? __raw_callee_save___pv_queued_spin_unlock (??:?)
[392896.423710] ? _raw_spin_unlock_irqrestore (./arch/x86/include/asm/paravirt.h:802 include/linux/spinlock_api_smp.h:162 kernel/locking/spinlock.c:191)
[392896.424354] ? trace_hardirqs_on_caller (kernel/locking/lockdep.c:2594 kernel/locking/lockdep.c:2636)
[392896.424978] ? _raw_spin_unlock_irqrestore (./arch/x86/include/asm/preempt.h:77 include/linux/spinlock_api_smp.h:163 kernel/locking/spinlock.c:191)
[392896.425610] ? cpu_stop_create (kernel/stop_machine.c:415)
[392896.426152] smpboot_thread_fn (kernel/smpboot.c:163 (discriminator 1))
[392896.426707] ? sort_range (kernel/smpboot.c:106)
[392896.427252] ? __kthread_parkme (kernel/kthread.c:165)
[392896.427817] ? sort_range (kernel/smpboot.c:106)
[392896.428332] kthread (kernel/kthread.c:209)
[392896.428862] ? kthread_stop (kernel/kthread.c:178)
[392896.429399] ? lock_release (kernel/locking/lockdep.c:3643)
[392896.429977] ? wait_for_completion (kernel/sched/completion.c:77 kernel/sched/completion.c:93 kernel/sched/completion.c:101 kernel/sched/completion.c:122)


Thanks,
Sasha
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/