amdgpu is crashing again on Linux 7.2.0.

It wasn’t happening on rc7 and it failed twice so far today with page fault (src_id:0 ring:24 vmid:7 pasid:n) on two different processes when I was doing nothing special. If it’s a bad release for amd it won’t be the first time. Or it could be my gpu dying. Either way it’s pretty annoying.

  • just_another_person@lemmy.world
    link
    fedilink
    arrow-up
    2
    ·
    20 hours ago

    Not sure what “RC7” is, can you clarify?

    If you weren’t previously having issues, then just revert back to whatever version wasn’t giving you problems.

    • kbal@fedia.ioOP
      link
      fedilink
      arrow-up
      3
      ·
      20 hours ago

      rc7 was the final release candidate build for 7.2, the seventh in the usual series of them. They’re meant to fix bugs, not introduce new ones. I often run them, for fun and testing. One expects bugs like this — assuming it is a bug — in the rc builds but normally they’re meant to be fixed before the final release which happened this week.

      Switching back to a working kernel is what I did for today, yeah. I have no particular reason to need 7.2 right now and don’t have time to try and track down the bug.

      • just_another_person@lemmy.world
        link
        fedilink
        arrow-up
        1
        ·
        12 hours ago

        That’s not how software works. RC is beta software. It includes warnings that there MIGHT be issues running it. Sounds like you found some.

        Just revert to the known good work version if you’re not able to debug yourself.

        • kbal@fedia.ioOP
          link
          fedilink
          arrow-up
          1
          ·
          12 hours ago

          You appear to have misunderstood what I said there. It was not in rc7.

            • kbal@fedia.ioOP
              link
              fedilink
              arrow-up
              1
              ·
              45 minutes ago

              Hopefully it won’t matter too much in practical terms. More distributions have lately started using very new kernel versions and substantial numbers of people might therefore be hit by it. But the volume of patches to amdgpu is so high that I have no idea if whatever the problem might be was fixed already by someone. If it was as obvious to others as it seemed to me then it should very likely be fixed for 7.2.1.