xorg 6.9.0 mem leak

[Jason Evans]

On Jan 30, 2006, at 2:05 PM, Randy Bush wrote:
>>> and the Xorg one just keeps growing and growing and growing.
>> That's a different matter though.
>
> and that's my point
>
>>   PID USERNAME    THR PRI NICE   SIZE    RES STATE    TIME   WCPU  
>> COMMAND
>>  2166 randy         4  20    0   166M   108M kserel   3:51  0.00%  
>> firefox-bin
>>  1343 randy         1  96    0   126M 83500K select   2:44  2.00%  
>> Xorg
>>  1394 randy         4  20    0   102M 51456K kserel   0:08  0.00%  
>> nautilus
>
> and now
>
>   PID USERNAME    THR PRI NICE   SIZE    RES STATE    TIME   WCPU  
> COMMAND
>  2166 randy         4  20    0   166M   110M kserel   4:26  0.00%  
> firefox-bin
>  1343 randy         1  96    0   142M 86032K select   3:14  2.39% Xorg
>  1394 randy         4  20    0   102M 51456K kserel   0:08  0.00%  
> nautilus
>
> notice the growth in xorg and only xorg.
>
> and, from x's pov, all i have been doing is typing in an emacs window,
> thought there are 42 other windows open.
>
> and it will just keep growing if i walk away for a few hours.

I've been running Xorg, firefox, and emacs, since last night (with  
jemalloc's redzone code enabled) on a two-day-old -current build, and  
after many invocations of firefox and emacs, with serious attempts to  
make firefox use lots of memory, Xorg's, memory usage has stabilized  
at ~75 MB mapped, and ~35 MB resident.  Neither number has increased  
this morning (other than fluctuations when launching/terminating  
firefox).

I see no evidence that jemalloc is doing anything wrong, or that  
Xorg's memory usage is excessive when utilizing jemalloc.  Perhaps  
gnome is causing issues for you.  In any case, I'm not going to look  
into this any further unless someone provides concrete, detailed  
evidence that jemalloc is behaving badly.

Thanks,
Jason