Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.debian.kernel > #87880 > unrolled thread

Bug#1104269: linux-image-6.12.22-amd64: amdgpu flip_done and commit wait timeouts during normal use - causes session lockup

Back to article view | Back to linux.debian.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Bug#1104269: linux-image-6.12.22-amd64: amdgpu flip_done and commit wait timeouts during normal use - causes session lockup Mathias Gibbens <gibmat@debian.org> - 2025-06-05 17:50 +0200

#87880 — Bug#1104269: linux-image-6.12.22-amd64: amdgpu flip_done and commit wait timeouts during normal use - causes session lockup

[Multipart message — attachments visible in raw view] — view raw

  I think I have occasionally been encountering this (or a very
similar) issue. Like Abdurahman, I'm running trixie with KDE Plasma and
three times now in the past month the GPU has totally locked up. Two of
those times I had music playing in the background, and it continued
playing even after the GPU stopped responding.

  Maybe relevant is that each time I've experienced this, I have had a
QEMU VM running (via libvirt) that has had 3D acceleration pass-though
enabled. I run this VM on a daily basis, so it may not be directly
related, but it is a commonality in all three crashes.

  My GPU is a Sapphire NITRO+ RX 7900 XTX.

  I was able to capture the kernel error and call trace (attached) the
first two times this happened, back on May 7 & 8 with kernel 6.12.25.
This morning it happened again with kernel 6.12.27, but I failed to
grab the kernel log before rebooting the system.

  Back in May when this first happened, the only other relevant recent
change was an update of firmware-amd-graphics from 20241210-1 to
20250410-2. Because this issue didn't continue to occur for me, at the
time I wrote it off as a random issue. Having it happen again today
however indicates there is something, somewhere going wrong. Without a
reproducer, though, I don't know if there's much more I can provide at
this time.

Mathias

[toc] | [standalone]

Back to top | Article view | linux.debian.kernel

csiph-web