On Sun, 10 Mar 2002, F. Heitkamp wrote:
> I'm still having the same problem with X on a
> dual G4 powermac, with Nvidia 2MX that I can't
> seem to trace down. I had a recently running
> nightly build of mozilla, some xterms, xclock,
> the icewm, and the newest CVS version of XFree86
> running and it stopped while I was surfing with
> mozilla.. I am able to move the mouse cursor around.
> I tried attaching to several of the programs running
> in the X server and this is what I get. X seems to
> be setting in 0x1027dc7c. How do I find out what
> that is?
[...]
>
> (gdb) bt
> #0 0x1027dc7c in ?? ()
> (gdb)
>
I'm not sure how to debug the XFree86 modules on PPC.
My guess is that something hosed the graphics engine and it's
stuck either in the RIVA_FIFO_FREE or RIVA_BUSY macros.
I don't know why it would do that under PPC and not X86.
Maybe there's a cornercase bug in XAA's BigEndian color expansion
code. You could check that 'w' and 'h' passed to
NVSubsequentScanlineCPUToScreenColorExpandFill is never zero.
I know the engine will choke on a zero width or height. XAA
shouldn't be passing such a thing, but maybe if the fonts were
broken by a BE bug in the renderer or if there was a BE bug in XAA...
Or maybe it's some weird bus thing - PCI retries and the like.
Mark.
_______________________________________________
Xpert mailing list
[EMAIL PROTECTED]
http://XFree86.Org/mailman/listinfo/xpert
