Skip to content

Add linux 5.10.22 #2047

Merged
merged 12 commits into from
Mar 12, 2021
Merged

Add linux 5.10.22 #2047

merged 12 commits into from
Mar 12, 2021

Conversation

donald
Copy link
Collaborator

@donald donald commented Mar 10, 2021

No description provided.

Create bee file with

    cp linux-5.4.97-368.bee linux-5.10.21-373.bee
Update patch URL for 5.10.21 kernel.
Remove patch, which is no longer required for kernel 5.10.21.
Remove patch, which is no longer required for kernel 5.10.21.
Remove patch, which is no longer required for kernel 5.10.21.
Remove patch, which is no longer quired for kernel 5.1.21.
Update patch URL for 5.10.21 kernel.
@donald donald changed the title Add linux 5.10.21 Add linux 5.10.22 Mar 10, 2021
Add patch, which should avoid spurious xfs deadlocks

xfs: Wake CIL push waiters more reliably

Threads, which committed items to the CIL, wait in the xc_push_wait
waitqueue when used_space in the push context goes over a limit. These
threads need to be woken when the CIL is pushed.

The CIL push worker tries to avoid the overhead of calling wake_all()
when there are no waiters waiting. It does so by checking the same
condition which caused the waits to happen. This, however, is
unreliable, because ctx->space_used can actually decrease when items are
recommitted. If the value goes below the limit while some threads are
already waiting but before the push worker gets to it, these threads are
not woken.

Always wake all CIL push waiters. Test with waitqueue_active() as an
optimization. This is possible, because we hold the xc_push_lock
spinlock, which prevents additions to the waitqueue.
smartpqi: Set host_tagset = 1

Set host_tagset = 1 as suggested in [1] to avoid the problem of
controller command overrun under certain load conditions.

[1] https://lore.kernel.org/linux-scsi/20210120184548.20219-1-mwilck@suse.com/T/#u
Create bee file with

    cp nvidia_linux-5.4.97-368-460.39-0.bee nvidia_linux-5.10.22-373-460.39-0.bee
@donald
Copy link
Collaborator Author

donald commented Mar 10, 2021

Tests are running on deadbird, claptrap, done, gone and theinternet.

@thomas
Copy link
Collaborator

thomas commented Mar 10, 2021 via email

@donald
Copy link
Collaborator Author

donald commented Mar 10, 2021

Yes, go ahead. But if you want to test the change, you'd need to pause and restart it. I've run though the high I/O load test on deadbird ( smartpqi + xfs). I'll start another test on it to pause and continue the mdcheck in a loop. If that is okay now, we need to decide, how we continue. We'd like to restart the regular mdcheck on all nodes, I assume.

@thomas
Copy link
Collaborator

thomas commented Mar 10, 2021 via email

@donald donald marked this pull request as ready for review March 11, 2021 18:15
@donald donald merged commit c30a411 into master Mar 12, 2021
Sign in to join this conversation on GitHub.
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

2 participants