#14 races resulting in failed resume from S3
Opened by chrismurphy. Modified

Stumbled onto "resume from suspend to RAM not working properly with / on btrfs" on users@lists.fedoraproject where it was suggested to take it to 'kernel bugzilla' which is a separate issue #15.

That lead to me making the connection to this upstream linux-btrfs@ post and excerpt from Luis Chamberlain mcgrof@kernel.org

Indeed, but then again doing a a race against a filesystem in between
suspend/resume can cause all sorts of issues today, so btrfs is not the
only one with issue there. I had found issue with xfs as well [0]:
[0] https://lwn.net/Articles/735382/
I addressed a topic on this particular issue at LSFFMM in 2018, and here
are the notes from that session:
https://lore.kernel.org/linux-fsdevel/20180426212243.GA27853@wotan.suse.de/
I will pick up this work again now, but it will take a bit of time.

We probably should encourage this testing during test day next week to see if this is a conditional (local configuration specific) issue or if it's more widespread and what to do about it.


@kparal @adamwill @josef

Wow, nice detective work from Lukas. We should definitely include suspend to ram and possibly also hibernation in test cases for the next btrfs test day. @sumantrom Can you please do that?

Or do you have something more specific in mind, @chrismurphy ?

Yeah the freezer stuff clearly needs a bit of work, not just for btrfs. I remember being in this talk at LSF, but I don't remember the outcome. We should definitely test this a bunch. Obviously my laptop has been suspend/resuming normally, but the more hardware the better. If it's a widespread enough problem we can devote some time to it, it'll help xfs too.

Metadata Update from @ngompa:
- Issue tagged with: Kernel

Metadata Update from @ngompa:
- Issue assigned to josef

Metadata