[Intel-gfx] Proposal for rules for maintaining the 2.7 branch

Clemens Eisserer linuxhippy at gmail.com
Tue Mar 24 22:55:34 CET 2009


Well, after using Okular a lot (which seems to have a stupid pixmap
cache algorythm), I had:

 PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
 2686 root      20   0 2596m 1.8g 1.8g S 10.2 62.9   4:21.91 X

Of course, Okular should be improved (I was able to have one Okular
instance holding 500mb pixmaps), but on the other side I can't imaging
mapping that much is healthy.

I also experienced major troubles after some time, like firefox
refusing to display some large images or crashes when enabling kwin's
xrender composition manager - which only seem to happen after I've
used the system for some time.

- Clemens

2009/3/23 Adam Lantos <hege at playma.org>:
> Ok, just for the record I seem to experience similar issues like others:
>
> http://bugs.freedesktop.org/show_bug.cgi?id=20766
> http://bugs.freedesktop.org/show_bug.cgi?id=20704
>
>
> On Mon, Mar 23, 2009 at 4:55 PM, Adam Lantos <hege at playma.org> wrote:
>> Any idea how to investigate this further? Should I send lsof output,
>> register debug log or something? Is it a known issue or am I the only
>> one experiencing this? Are there any preliminary patches which might
>> help?
>>
>>
>> thanks,
>>  Adam
>>
>>
>>
>> On Fri, Mar 20, 2009 at 3:55 PM, Adam Lantos <hege at playma.org> wrote:
>>> On my i915 even swap fills up after several days, and it seems like
>>> mplayer+xv increases the memory leak.
>>>
>>> I use kernel 2.6.28.7 + intel-2008-q4 patches, mesa-intel-2008-q4,
>>> xf86-video-intel-2.6.3, xserver-1.6, uxa without kms
>>>
>>>
>>> -
>>>  Adam
>>>
>>>
>>> On Fri, Mar 20, 2009 at 2:08 PM, Clemens Eisserer <linuxhippy at gmail.com> wrote:
>>>> Hi Colin,
>>>>
>>>>> Could perhaps be related to:
>>>>> https://bugs.kde.org/show_bug.cgi?id=183646
>>>> I don't think its related to that - xrestop shows 40mb pixmaps in
>>>> total, but top already reports 1.1GB rss, and 8500 drm mm objects.
>>>>
>>>> I'll file a bug about that.
>>>>
>>>> - Clemens
>>>> _______________________________________________
>>>> Intel-gfx mailing list
>>>> Intel-gfx at lists.freedesktop.org
>>>> http://lists.freedesktop.org/mailman/listinfo/intel-gfx
>>>>
>>>
>>
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx
>



More information about the Intel-gfx mailing list