On Sun, Jun 15, 2025 at 5:51 AM SyntheticBird <syntheticbird...@protonmail.com> wrote: > > Dear Alex, > > The situation is more than concerning and my frustration, as well as others, > is growing. > > https://gitlab.freedesktop.org/drm/amd/-/issues/3787 > https://gitlab.freedesktop.org/drm/amd/-/issues/2950 > > There has been a hundred of reports on gitlab, I can find more on reddit, > tech1level forum, phoronix forum, I got confirmation by some folks on Matrix > that are staying in 6.1 and are starting to mock the amdgpu driver. I've seen > features after features being merged in amdgpu yet people are laughing in > forum, at how bugs are not fixed. On reddit, some users decided to go back on > Windows because of it. > > There has been 4 bisection attempts so far, that led to no result, and (I'm > in part at fault here for having used an experimental filesystem with major > version changes). There are people that managed to grasp debug infos before > the timeout. > > In two months, this significant issue will be a year old. > > I do not understand, how you guys are not hitting it/trying to bisect it. Are > you not keeping your machine up for more than 30 sec. > With the immense spectrum of hardware concerned, I cannot believe that you > guys are immune to this. And I cannot believe that you guys are letting this > rot. > > Right now, only a minority is managing to recover their session after a few > minutes. Some people are able to get away with it by enabling the debug flag, > but this only work for some people and there are no guarantees that it will > last long. > > Please, > It's time to forget trying to pass the hotball and get a maintainer > investigate and fix this before it gets too late. If we let it rot another > year, no one will be able to bisect it because distros and hardware will make > use of recent versions only.
Unfortunately, not every issue is readily reproducible and the issues you've noted are pretty generic. There can be a lot of root causes for these kinds of issues. If you can't bisect your issue, I would suggest trying a newer kernel as we've fixed quite a number of issues that manifest in a similar way over the last few months. Alex
