[Intel-gfx] Getting "rcs0: reset request timeout" on 4.19.16

[Marcin Owsiany]

Thank you for the quick response, Chris!

I still don't understand why an X server restart is not enough to make it
work again?
What does "intel_do_flush_locked failed: Input/output error" mean?
Is it just a fact of life these days that userspace can mess up the GPU to
a state where only a reboot helps?

śr., 22 maj 2019 o 10:05 Chris Wilson <chris at chris-wilson.co.uk> napisał(a):

> Quoting Marcin Owsiany (2019-05-22 08:54:50)
> > Hello,
> >
> > Every couple of weeks or so my laptop experiences some graphics-related
> > crash. Sometimes it's more frequent - happened twice yesterday.
> >
> > I'm wondering whether this could be a sign of a hardware failure? Or is
> this
> > some software lock-up?  Please do let me know whether I can do anything
> to help
> > debug this.
>
> It starts as a userspace hang that takes out the GPU. Update mesa to
> one of the 19.x stable releases.
>
> > As you can see from the log below, sometimes there is just a single
> > "Resetting rcs0 for hang on rcs0" from which it recovers, while at other
> > times it is followed by "reset request timeout", together with an X
> > server crash.
>
> File a regression report against debian, X should not crash and did not
> use to crash over a mere terminal GPU hang.
> -Chris
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-gfx/attachments/20190523/b481c829/attachment-0001.html>