On Tue, Jan 11, 2011 at 11:28 PM, Chris Wilson <[email protected]> wrote: > On Tue, 11 Jan 2011 11:17:16 +1000, Dave Airlie <[email protected]> wrote: >> This fixes a regression caused by the GTT rework that >> 450f2b3d51025a1749b694ee13f0e4e23ed58750 is part off. >> >> On my ILK laptop with DMAR enabled I was seeing the occasional >> DMAR warning. >> >> The ordering in the previous code was to rewrite the GTT table >> before unampping the pages and that makes sense to me. > > Dave you mentioned on IRC that this is not sufficient to quieten DMAR, > right? The unbind+unmap looks safer than unmap+unbind, so that seems to > be a good patch regardless. > -Chris >
This patch needs to go to stable as it fixes 2.6.37 DMAR problems. Separate problem is d-i-n, this code has changed around but the same idea still applies and it fixes this particular problem. However d-i-n has introduced another DMAR error that I posted in the dmesg earlier. Granted d-i-n also hangs and crashes GPU randomly. Dave. > -- > Chris Wilson, Intel Open Source Technology Centre > _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
