[Bug 211033] [bisected][regression] amdgpu: *ERROR* Restoring old state failed with -12

bugzilla-daemon at bugzilla.kernel.org bugzilla-daemon at bugzilla.kernel.org
Sat Mar 6 13:33:24 UTC 2021


https://bugzilla.kernel.org/show_bug.cgi?id=211033

Jan Klos (honza.klos at gmail.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |honza.klos at gmail.com

--- Comment #17 from Jan Klos (honza.klos at gmail.com) ---
I don't think this is limited just to KVMs and such. On my Vega 64 +
5.11.3-arch1 (had the same problem with 5.11.2, 5.10 etc.), sometimes, when I
return to PC after a while and my 2 monitors are sleeping, only one monitor
wakes up, the other one remains in sleep mode. Dmesg shows this:

[bře 6 14:22] [drm] perform_link_training_with_retries: Link training attempt 1
of 4 failed
[bře 6 14:23] [drm] perform_link_training_with_retries: Link training attempt 2
of 4 failed
[  +0,473352] [drm] perform_link_training_with_retries: Link training attempt 3
of 4 failed
[  +0,437753] [drm] enabling link 0 failed: 15
[  +0,432276] [drm] perform_link_training_with_retries: Link training attempt 1
of 4 failed
[  +0,405827] [drm] perform_link_training_with_retries: Link training attempt 2
of 4 failed
[  +0,476172] [drm] perform_link_training_with_retries: Link training attempt 3
of 4 failed
[  +0,415466] [drm] enabling link 1 failed: 15

Switching to terminal and waiting a sec or two makes both monitors work,
switching back to X11 and everything is OK. It seems to me that there might be
some kind of bug where if the first link training attempt fails, the subsequent
ones ALWAYS fail as well, so there is actually only a single actual link
training attempt that has a chance to succeed.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.


More information about the dri-devel mailing list