[Bug 91883] [drm:intel_pipe_update_end [i915]] *ERROR* Atomic update failure on pipe A

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=91883

--- Comment #33 from Ville Syrjala <ville.syrjala at linux.intel.com> ---
There are presumably two problems mixed up in this bug:

1) PSR/DMC problems, which result in bogus looking numbers in the debug output

[drm:intel_pipe_update_end [i915]] *ERROR* Atomic update failure on pipe A
(start=301707 end=301708) time 1 us, min 1073, max 1079, scanline start 1088,
end 1080

Fixing this requires avoiding PSR and/or DC5/6 while we need vblanks, and we
need to deal with the hardware resetting various counters on DC5/6 entry. There
have been some patches posted but nothing went in so far AFAICS.

2) Actual vblank evade problems, which result in numbers that make sense, just
we take way too long to do things

[ 4427.089764] [drm:intel_pipe_update_end [i915]] *ERROR* Atomic update failure
on pipe A (start=146309 end=146310) time 179 us, min 1073, max 1079, scanline
start 1071, end 1084

This is most likely a regression from the improved gamma table stuff since
we're now programming the gamma table(s) during the vblank evasion critical
section, which is pretty bogus since those registers aren't even double
buffered. I pointed this out but apparently no cared, and so the patches went
in anyway. The fix would involve moving the programming to a better spot.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
You are the QA Contact for the bug.
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20161108/98c45a85/attachment.html>