Dieter Nützel wrote:
I saw problems here without irqs. Try 'R200_NO_IRQS=t' - I bet you get the same results.Am Freitag, 25. Oktober 2002 23:36 schrieb Dieter Nützel:Uh, no. Big mistake.Am Freitag, 25. Oktober 2002 22:22 schrieb Brian Paul:Charl P. Botha wrote:Dear list,
This bug has so far been reproduced on my M7 (mobile 7500) laptop with
DRI CVS as well as on a TiBook (i.e. PPC) with 7500 (thanks Wouter).
Download http://cpbotha.net/thingies/glthreads.c and compile it with
something like:
gcc -I../include -O3 -ansi -pedantic -fPIC -ffast-math -D_SVID_SOURCE \
-D_BSD_SOURCE -DUSE_XSHM -DPTHREADS -I/usr/X11R6/include glthreads.c \
-L../lib -lglut -lGLU -lGL -lm -o glthreads
This is an example from the mesademos I've modified to show the bug. You have to run with RADEON_TCL_FORCE_DISABLE=1 (if you have TCL
hardware) else you will get the VTXFMT assert bug or your X will lock
up. This is the case with the stock glthreads too. If you have an
R200, try with R200_NO_TCL=1
Run glthreads with something like: glthreads -n 5
Focus one of the windows and press any key. This will terminate the
thread, wait for it and then kill the glxcontext and window (my
modification). At this point, if you're using a radeon, you should
see: drmCmdBuffer: -22 and the app will terminate.
Does anyone have any ideas on this?
I found the problem. The call to R200_FIREVERTICES() was in the wrong place (as someone suggested). It's supposed to be near the top. I'll check in the fix for both the radeon and r200 driver soon. glthreads seems to work fine now (with or without h/w tcl) on the r200.This is true got it running with n=100,150,200 ;-) Felix fix cured it.
I have confused "manywin" which runs fine for n=100,150,200 with "glthread". The later lockup (only the X server, STRG+Alt+BS works) sometimes with n=2, then it runs OK, then lockup wit n=10 (after ~6 threads).
With R200_NO_VTXFMT it is a little bit better.
n=
2 OK
10 OK
12 OK
13 OK
13 - all windows pop up, then hang
14 - ditto
15 - ditto
25 - ~16 windows pop up, then hang
Could it be scheduling, IRQs, waiting, again?
Keith
-------------------------------------------------------
This sf.net email is sponsored by: Influence the future
of Java(TM) technology. Join the Java Community
Process(SM) (JCP(SM)) program now.
http://ads.sourceforge.net/cgi-bin/redirect.pl?sunm0004en
_______________________________________________
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel
