Bug#791599: xserver-xorg-video-intel: loss of hardware acceleration after resume
Just to paste some more info, when I start chromium after resume the
screen's black and dmesg output's as follows:
...
[38088.042961] [drm] stuck on render ring
[38088.043945] [drm] GPU HANG: ecode 0:0x00dfffff, reason: Ring hung,
action: reset
[38090.045530] [drm] Enabling RC6 states: RC6 on, RC6p off, RC6pp off
[42163.120279] [drm] stuck on render ring
[42163.121450] [drm] GPU HANG: ecode 0:0x00dfffff, in chromium [9505],
reason: Ring hung, action: reset
[42165.122279] [drm] Enabling RC6 states: RC6 on, RC6p off, RC6pp off
[42169.126284] [drm] stuck on render ring
[42169.127978] [drm] GPU HANG: ecode 0:0xf3cffffe, in chromium [9505],
reason: Ring hung, action: reset
[42171.128257] [drm] Enabling RC6 states: RC6 on, RC6p off, RC6pp off
...
On 06/07/15 21:22, Julien Cristau wrote:
> On Mon, Jul 6, 2015 at 16:44:37 +0100, olsen wrote:
>
>> Package: xserver-xorg-video-intel
>> Version: 2:2.99.917-1~bpo8+1
>> Severity: important
>>
>> Dear Maintainer,
>>
>> *** Reporter, please consider answering these questions, where appropriate ***
>>
>> * What led up to the situation?
>> The problem becomes apparent for example if I start chromium and try to move the whole browser window by holding down the alt key + the mouse, resulting in almost slow motion.
>> Similar effects with Iceweasel on various websites e.g. googlemaps. This is after a fresh reboot.
>> Furthermore I noticed after a couple of resumes I can't play videos with the default settings any more, e.g. in mplayer only with the -xv X11 option.
>> * What exactly did you do (or not do) that was effective (or
>> ineffective)?
>> I switched from testing to experimental (xserver-xorg-video-intel 2:2.99.917-1~bpo8+1) both with "uxa" acceleration in /etc/X11/xorg.conf.d/20-intel.conf as follows:
>> Section "Device"
>> Identifier "Intel Graphics"
>> Driver "intel"
>> Option "AccelMethod" "uxa"
>> EndSection
>> * What was the outcome of this action?
>> Slightly better performance but the issue still remains, especially the loss of the acceleration after resumes.
>> * What outcome did you expect instead?
>> No loss of the hardware acceleration
>>
> Does that also happen with a newer kernel and/or mesa?
>
> Cheers,
> Julien
>
Reply to: