Hi savage users,
those of you who experience more or less reproducible lockups, could you
try the attached patch with the latest Mesa trunk? There were
substantial changes in the hardware state management over the weekend.
Therefore comparisons of the latest patched Mesa trunk with an older
unpatched Mesa trunk are basically useless. Please proceed as follows:
- "cvs update" your mesa sources
- "make" and "make install" DRI
- try to reproduce a lockup
- patch your mesa sources with the attached patch
- "make" and "make install" DRI
- try to reproduce the same lockup
Please let me know if the patch makes any difference.
Best regards,
Felix
Index: src/mesa/drivers/dri/savage/savagestate.c
===================================================================
RCS file: /cvs/mesa/Mesa/src/mesa/drivers/dri/savage/savagestate.c,v
retrieving revision 1.3
diff -u -r1.3 savagestate.c
--- a/src/mesa/drivers/dri/savage/savagestate.c 29 Feb 2004 20:42:22 -0000 1.3
+++ b/src/mesa/drivers/dri/savage/savagestate.c 1 Mar 2004 14:53:07 -0000
@@ -1507,7 +1507,9 @@
WRITE_CMD (pBCIBase, WAIT_3D_IDLE, GLuint);
savageDMACommit (imesa, pBCIBase);
}
- if (imesa->lostContext)
+ /* Try to always emit all registers. May prevent lockups on certain
+ * hardware. */
+ if (1 || imesa->lostContext)
savageEmitContiguousRegs (imesa, 0x1e, 0x39);
else
savageEmitChangedRegs (imesa, 0x1e, 0x39);
