[Bug 61011] New: [ILK] system stalls with vaapi rendering

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=61011

          Priority: medium
            Bug ID: 61011
          Assignee: intel-gfx-bugs at lists.freedesktop.org
           Summary: [ILK] system stalls with vaapi rendering
        QA Contact: intel-gfx-bugs at lists.freedesktop.org
          Severity: normal
    Classification: Unclassified
                OS: Linux (All)
          Reporter: liquid.acid at gmx.net
          Hardware: x86-64 (AMD64)
            Status: NEW
           Version: DRI CVS
         Component: DRM/Intel
           Product: DRI

Hello,

first of all an overview of used system components:
xf86-video-intel git (43ba22ef4a4142f334e9ae2d926250988ecbe8bc)
libdrm git (36d18211b196cad4761ac70c4fd08aba323f5b0d)
libva-1.1.0
intel-driver git (7803fcae62c868c26f24425df04aee0405342563)
vanilla kernel 3.7.8

The DDX is using the SNA backend, however switching to UXA doesn't help.

The issue appears for me in combination with h264 playback through vaapi
(mplayer-vaapi -vo vaapi -va vaapi). I can't exclude that it also happens in
other situations.

THe issue: Playback stalls for several 100ms (very much noticeable, always
accompanied by sound dropout (audio backend is pulseaudio on a remote
machine)).

mplayer itself doesn't notice these stalls. If it would be high CPU usage
spikes, it would display the usual "system too slow, blabla" message. However
it doesn't, and monitoring CPU usage with top reveals that nothing is really
happening during these stalls. The XFCE CPU monitor graph even indicates that
CPU load drops to flat zero during these stalls.

I fiddled around with different versions of the DDX, libva's intel-driver and
SNA/UXA -- to no avail.
I could restore the original (stall-free) behaviour by going back to vanilla
3.4.31, so I presume that the issue was introduced by changes to the DRM code.

Any ideas on how to triage this (apart from bisecting of course)? E.g. I have
no idea how to find out what exactly is stalling playback -- which would be a
good starting point.

Greets,
Tobias

PS: Oh yes, nothing interesting in dmesg, Xorg.log or syslog in general.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20130217/f0bc4708/attachment.html>