Mark Vojkovich wrote:

>On Sat, 6 Apr 2002, F. Heitkamp wrote:
>
>>I've posted some previous messages about the Xserver
>>problem I am having.  I have a G4 533 dualie Snakebit
>>I am running Debian unstable.  I am using ICEWM.
>>The machine came with the GForce 2MX.
>>
>>When I start up X it runs for several minutes fine.
>>At some point it appears to stop.  I can still move
>>the mouse cursor around.  I put a debug statement
>>in nv_xaa.c at the suggestion of Mark V.  The statement
>>checks if NVSebsequentScanlineCPUToScreenColorExpandFill
>>arguments w and h are ever zero.
>>
>   I don't know what would be causing this.  I thought maybe
>the BigEndian color expansion code in XAA had some bugs in
>it, but it doesn't sound like it.  I'm still under the impression
>that this is due to color expansion and not due to power management
>or something else.
>
>  Can you try disabling color exapansion with
>
>  Option "XaaNoScanlineCPUToScreenColorExpandFill"
>
This seemed to help a lot.  I left X running with xscreensaver for 
probably almost
two days without a crash.  Previous to that and after applying the 
option, I tried
to use Linux, with ICEWM like I would normally, (surfing the web with 
Mozilla,
reading news with pan, email etc.) and the system finally crashed after 
at least an
hour.  A little investigation, seemed to suggest that the crash happened 
in the
glibc select call.  (Mouse cursor would not move, display was  frozen.). 
 I had
to log in remotely to shutdown.

Another problem I have, but still investigating is when I launch X using 
'startx',
log out of X, and then run startx again, X doesn't start and I get a 
blank terminal.

I updated glibc to 2.2.5-4 and things seem better but I haven't banged 
on Linux
really hard or checked if 'startx' problem has gone away.  However, I 
did do
'normal' stuff without a crash, and shutdown Linux normally to boot to 
Mac OS X.

Anyway this is a vast improvement nonetheless.

Thanks Mark!

>
>
>  In the section device?  When you do that the driver
>should stop reporting
>
>      Indirect CPU to Screen color expansion
>
>in the XAA section of the log file.  If that makes it go away I
>think we may just have to disable color expansion on PPC.
>I'm assuming other people have seen this problem?  Can't say
>that I have, but I haven't logged much time on PPC.
>
>
>                       Mark.
>
>



_______________________________________________
Xpert mailing list
[EMAIL PROTECTED]
http://XFree86.Org/mailman/listinfo/xpert

Reply via email to