[Bug 101623] [IGT][BDW] kms_frontbuffer_tracking at fbc-*draw* has inconsistent results - FBC disabled

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=101623

--- Comment #26 from Marta Löfstedt <marta.lofstedt at intel.com> ---
(In reply to krisman from comment #25)
> (In reply to Elizabeth from comment #21)
> > Sorry for the spam. I mean the "FBC disabled" in KBL, but I need to retest
> > 100020 to be sure if it is still reproducible.
> > (In reply to Marta Löfstedt from comment #17)
> > > Note we are still seeing "FBC disabled" on KBL-shards:
> > > https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3016/shard-kbl2/
> > > igt at kms_frontbuffer_tracking@fbc-1p-offscren-pri-shrfb-draw-blt.html
> 
> 100020 is very specific to a workaround applied only to hsw and it is due
> the poor status reporting interface in fbc.  but I can see the FBC disabled
> happening on other platforms for unrelated reasons. It still triggers on
> HSW, btw.
> 
> This mentions an underrun detection as the cause of failure.  It is good to
> see if the underrun is the actual error or a side effect of the fbc status
> being overwritten after the actual failure. How hard is it to reproduce on
> any KBL? I can give it a try too.

I already have an IGT patch on the list that increase the wait time for the
tests. This was sort of Ack:ed, but I believe I would need to provide more
investigations as to why this issue was associated with specific draw domain
combinations. After that Chris suggested to drop caches, which also solved the
issue. At the time I was investigating this issue for BDW. However, just when I
was about to finalize a V3 based on dropping caches. The issue on BDW shifted
from hitting the "FBC disabled) to the WARN_ON(fbc->active) in bug 102473. So,
I started working on that. But unfortunately I have very little time over to do
this work. Anyways, if you have a KBL and spin all fbc related
frotnbuffer-tracking testcases I am sure that you will eventually reproduce
this issue.
/Marta

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20171019/a4bf7b67/attachment-0001.html>