On Sat, 2003-11-22 at 14:23, Andreas Stenglein wrote:
> Am 2003.11.19 23:15:15 +0100 schrieb(en) Michel D=C3=A4nzer:
> > 'That' annoying hang? Have I missed something? *shrug*
>=20
> maybe..
> You will get something similar when running ut2003_demo in non-tcl mode,=20
> too.
You're saying there are UT2003 binaries for PPC? ;)
> [...]
> > > isnt it strange that RADEONWaitForIdleCP somehow calls fbBresDash32()=
?
> >=20
> > I suspect that you've been chasing red herrings, my bet for the
> > mysterious symbol is on drmCommandNone(), which gets resolved by the
> > server's dynamic loader (or doesn't, when the DRI is disabled). Build a
> > static server to get reliable debugging info, but I doubt it'll help fo=
r
> > the actual problem:
>=20
> yes, in the static build it points to drmCommandNone(),
> but the hang still occurs.
>=20
> how to trigger the bug:
> start glxgears, start quake3 in windowed mode,
You're saying... ;)
> go to the model choosing menu, fiddle around a bit,=20
> try different models, push the back/forw. buttons,
> choose another model... and somewhen it should happen:
> glxgears and quake3 exit with
> r200WaitIrq: drmRadeonIrqWait: -16
Another symptom of a lockup: the awaited interrupts never arrive.
I'm curious, does this only happen if the windows overlap, or
regardless?
> > > killall -KILL X -> System hangs, no ping
> >=20
> > Sounds like a good old chip lockup. :\
>=20
> But why does the system still work until
> after KILLing the Xserver?=20
I guess the chip lockup turns into a full system lockup when trying to
(re-)initialise the chip or something.
> Couldnt it be possible to somehow regenerate/reset the card?
Dunno, suspect not.
> What needs to be defined in host.def to get better debug-info in the=20
> static build? (XFree86 CVS HEAD)
What kind of information are you thinking of?
--=20
Earthling Michel D=C3=A4nzer | Debian (powerpc), X and DRI develop=
er
Software libre enthusiast | http://svcs.affero.net/rm.php?r=3Ddaenzer
|