Hi Richard,
On 27/07/16 10:08, Richard Cochran wrote:
> On the tear-down path, the dead CPU callback for the timers was
> misplaced within the 'cpuhp_state' enumeration. There is a hidden
> dependency between the timers and block multiqueue. The timers
> callback must happen before the block multiqueue callback otherwise a
> RCU stall occurs.
>
> This patch fixes the order within the hot plug state machine.
>
> Signed-off-by: Richard Cochran <rcoch...@linutronix.de>
> Fixes: 24f73b99716a ("timers/core: Convert to hotplug state machine")
>
> Cc: John Stultz <john.stu...@linaro.org>
> Cc: Jon Hunter <jonath...@nvidia.com>
> Cc: Linus Torvalds <torva...@linux-foundation.org>
> Cc: Oleg Nesterov <o...@redhat.com>
> Cc: Peter Zijlstra <pet...@infradead.org>
> Cc: Rasmus Villemoes <li...@rasmusvillemoes.dk>
> Cc: Sebastian Andrzej Siewior <bige...@linutronix.de>
> Cc: Thomas Gleixner <t...@linutronix.de>
> Cc: r...@linutronix.de
> ---
> include/linux/cpuhotplug.h | 2 +-
> kernel/cpu.c | 15 ++++++++++-----
> 2 files changed, 11 insertions(+), 6 deletions(-)
>
> diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
> index 6d405db..242bf53 100644
> --- a/include/linux/cpuhotplug.h
> +++ b/include/linux/cpuhotplug.h
> @@ -20,9 +20,9 @@ enum cpuhp_state {
> CPUHP_PROFILE_PREPARE,
> CPUHP_X2APIC_PREPARE,
> CPUHP_SMPCFD_PREPARE,
> - CPUHP_TIMERS_DEAD,
> CPUHP_RCUTREE_PREP,
> CPUHP_NOTIFY_PREPARE,
> + CPUHP_TIMERS_DEAD,
> CPUHP_BRINGUP_CPU,
> CPUHP_AP_IDLE_DEAD,
> CPUHP_AP_OFFLINE,
> diff --git a/kernel/cpu.c b/kernel/cpu.c
> index 67f4943..ec12b72 100644
> --- a/kernel/cpu.c
> +++ b/kernel/cpu.c
> @@ -1208,11 +1208,6 @@ static struct cpuhp_step cpuhp_bp_states[] = {
> .startup = smpcfd_prepare_cpu,
> .teardown = smpcfd_dead_cpu,
> },
> - [CPUHP_TIMERS_DEAD] = {
> - .name = "timers dead",
> - .startup = NULL,
> - .teardown = timers_dead_cpu,
> - },
> [CPUHP_RCUTREE_PREP] = {
> .name = "RCU-tree prepare",
> .startup = rcutree_prepare_cpu,
> @@ -1229,6 +1224,16 @@ static struct cpuhp_step cpuhp_bp_states[] = {
> .skip_onerr = true,
> .cant_stop = true,
> },
> + /*
> + * On the tear-down path, timers_dead_cpu() must be invoked
> + * before blk_mq_queue_reinit_notify() from notify_dead(),
> + * otherwise a RCU stall occurs.
> + */
> + [CPUHP_TIMERS_DEAD] = {
> + .name = "timers dead",
> + .startup = NULL,
> + .teardown = timers_dead_cpu,
> + },
> /* Kicks the plugged cpu into life */
> [CPUHP_BRINGUP_CPU] = {
> .name = "cpu:bringup",
>
This fixes the RCU stalls I was seeing during suspend on Tegra and so ...
Tested-by: Jon Hunter <jonath...@nvidia.com>
Cheers!
Jon
--
nvpublic
