Hello,
On Tue, Apr 02, 2024 at 10:08:11AM +0200, Linux regression tracking (Thorsten Leemhuis) wrote:
Hi stable team, there is a report that the recent backport of 5797b1c18919cd ("workqueue: Implement system-wide nr_active enforcement for unbound workqueues") [from Tejun] to 6.6.y (as 5a70baec2294) broke hibernate for a user. 6.6.24-rc1 did not fix this problem; reverting the culprit does.
With kernel 6.6.23 hibernating usually hangs here: the display stays on but the mouse pointer does not move and the keyboard does not work. But SysRq REISUB does reboot. Sometimes it seems to hibernate: the computer powers down and can be waked up and the previous display comes visible, but it is stuck there.
See https://bugzilla.kernel.org/show_bug.cgi?id=218658 for details. Note, you have to use bugzilla to reach the reporter, as I sadly[1] can not CCed them in mails like this.
Side note: there is a mainline report about problems due to 5797b1c18919cd ("workqueue: Implement system-wide nr_active enforcement for unbound workqueues") as well, but it's about "nohz_full=0 prevents kernel from booting": https://bugzilla.kernel.org/show_bug.cgi?id=218665; will forward that separately to Tejun.
Ciao, Thorsten (wearing his 'the Linux kernel's regression tracker' hat)
Sorry about late reply but that commit is not for -stable. It does fix an undesirable behavior from an earlier commit, so it has the Fixes tag but it has a series of dependencies that need to be backported together which would be far too invasive for -stable, thus no Cc: stable. I didn't know a Fixes tag automatically triggers backport to -stable. I will keep that in mind for future.
Thanks.