Quoting Chris Wilson (2018-11-08 12:11:05) > Quoting Mika Kuoppala (2018-11-08 12:00:39) > > Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> writes: > > > > > Ensure that the writes into the context image are completed prior to the > > > register mmio to trigger execution. Although previously we were assured > > > by the SDM that all writes are flushed before an uncached memory > > > transaction (our mmio write to submit the context to HW for execution), > > > we have empirical evidence to believe that this is not actually the > > > case. > > > > > > Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=108656 > > > > I would have marked this also as References. > > I'm confident in my local results indicating some success here, albeit > in not exactly the same quick death, but still out-of-bounds execution. For the record, bsw just hung after close to a day, with the same out of bounds execution (BBADDR after the batch). /o\ -Chris _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
