[Intel-gfx] [PATCH] Error detection and reporting

Ben Gamari bgamari.foss at gmail.com
Sat Jul 11 22:48:02 CEST 2009


Hey all,

Here is the latest iteration of my/jbarnes' error detection and reporting
patchset. This set is at this point looking like it's in pretty good shape. The
lockups I was seeing earlier were actually a mistake in my hangcheck timer
(referring to ACTHD as opposed to ACTHD_I965, now fixed). The hangcheck timer
is able to detect gem_hang initiated hangs, although the chip reset logic
doesn't seem to do too much.

Anyways, I'd say much of this could be merged at this point. I reordered the
patches so that the safe stuff (error detection refactor, save/restore
refactor, and hangcheck timer) can be merged, while leaving the more
experimental parts (namely GPU reset logic) for later. Let me know what you
think. Thanks,

- Ben


P.S. I added jbarnes s-o-b to the patches which he was responsible for and had
previously signed off on in earlier versions. I hope this is kosher.




More information about the Intel-gfx mailing list