Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
Merge tag 'timers-urgent-2024-03-17' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull timer fix from Ingo Molnar:
"Fix timer migration bug that can result in long bootup delays and
other oddities"
* tag 'timers-urgent-2024-03-17' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
timer/migration: Remove buggy early return on deactivation
-20
-20
kernel/time/timer_migration.c
-20
kernel/time/timer_migration.c
···
751
751
752
752
first_childevt = evt = data->evt;
753
753
754
-
/*
755
-
* Walking the hierarchy is required in any case when a
756
-
* remote expiry was done before. This ensures to not lose
757
-
* already queued events in non active groups (see section
758
-
* "Required event and timerqueue update after a remote
759
-
* expiry" in the documentation at the top).
760
-
*
761
-
* The two call sites which are executed without a remote expiry
762
-
* before, are not prevented from propagating changes through
763
-
* the hierarchy by the return:
764
-
* - When entering this path by tmigr_new_timer(), @evt->ignore
765
-
* is never set.
766
-
* - tmigr_inactive_up() takes care of the propagation by
767
-
* itself and ignores the return value. But an immediate
768
-
* return is required because nothing has to be done in this
769
-
* level as the event could be ignored.
770
-
*/
771
-
if (evt->ignore && !remote)
772
-
return true;
773
-
774
754
raw_spin_lock(&group->lock);
775
755
776
756
childstate.state = 0;