[Bug 276985] crash in LinuxKPI/drm

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=276985

--- Comment #26 from Olivier Certner <olce@FreeBSD.org> ---
(In reply to Tomasz "CeDeROM" CEDRO from comment #24)

Do you prefer "not able to test"?  But to me, your own reactions so far,
besides what you expressly wrote, border on "not willing to test".

> There are various problems with AMDGPU / DRM / LinuxKPI that are reported by various people

Yes, but...

> so I am not alone

Well, judging by your stack traces and those reported by other users, it
currently looks like you're alone, yes.

> and those problems are disqualifying FreeBSD 14+ from a production use in that setup

Please stop spreading this FUD.

Lots of people use FreeBSD 14 in production with DRM.  I personally have
several machines with AMD and Intel GPUs using stacks with 14 (and DRM 5.10,
5.15 and 6.1) and am really happy with them.  And, at the risk of sounding
repetitive, I also have a RX 580 like you, but contrary to you never saw a
single crash on 14 (last ones were around 2 years ago... on 13-STABLE).

There are some problems, more or less annoying, which I (and others) have
reported in some bugs, but none are critical and more importantly they all can
be worked around.

Some people are also reporting crashes, part of which seem to happen in very
specific situations (unlike what you're reporting).  I have not seen any proof
so far that these behaviors are regressions compared to an older stack (e.g.,
involving 13) or that the bug isn't present in a comparable Linux stack (same
DRM version, and ideally same libraries).  To my current knowledge, you're
absolutely the only one claiming a regression from 13.

If you don't have the time or will to test newer stacks, it's perfectly fine,
and in any case you're responsible for your own strategy going forward.  But
you're also:
- Spreading FUD by over-generalizing your own case.
- Mixing problems by dumping your reports, traces and related advice in some
bugs without having any evidence that these are really related (besides "my RX
580 crashes with amdgpu and 14").  Actually, a simple examination of stack
traces in the various bugs could have lead you to see that on your own.  When
in doubt, create a new bug anyway, we can always mark it as a duplicate
afterwards.
This attitude is actively hampering, or even harming, problem reporting and
resolution for *everyone*.  So please stop it.

-- 
You are receiving this mail because:
You are the assignee for the bug.