Entering RCU idle mode may cause a deferred wake up of an RCU NOCB_GP kthread (rcuog) to be serviced.
Usually a wake up happening while running the idle task is spotted in one of the need_resched() checks carefully placed within the idle loop that can break to the scheduler. Unfortunately within acpi_idle_enter_bm() the call to rcu_idle_enter() is already beyond the last generic need_resched() check. The cpu idle implementation happens to be ok because it ends up calling mwait_idle_with_hints() or acpi_safe_halt() which both perform their own need_resched() checks. But the suspend to idle implementation doesn't so it may suspend the CPU with a resched request unhandled, leaving the task hanging. Fix this with performing a last minute need_resched() check after calling rcu_idle_enter(). Reported-by: Paul E. McKenney <[email protected]> Fixes: 1fecfdbb7acc (ACPI: processor: Take over RCU-idle for C3-BM idle) Cc: [email protected] Cc: Len Brown <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Rafael J. Wysocki <[email protected]> Cc: Thomas Gleixner <[email protected]> Cc: Ingo Molnar<[email protected]> Signed-off-by: Frederic Weisbecker <[email protected]> --- drivers/acpi/processor_idle.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) diff --git a/drivers/acpi/processor_idle.c b/drivers/acpi/processor_idle.c index d93e400940a3..c4939c49d972 100644 --- a/drivers/acpi/processor_idle.c +++ b/drivers/acpi/processor_idle.c @@ -604,8 +604,14 @@ static int acpi_idle_enter_bm(struct cpuidle_driver *drv, } rcu_idle_enter(); - - acpi_idle_do_entry(cx); + /* + * Last need_resched() check must come after rcu_idle_enter() + * which may wake up RCU internal tasks. mwait_idle_with_hints() + * and acpi_safe_halt() have their own checks but s2idle + * implementation doesn't. + */ + if (!need_resched()) + acpi_idle_do_entry(cx); rcu_idle_exit(); -- 2.25.1
