Quoting Chris Wilson (2017-10-08 14:43:22) > As we emit and retire requests, we track how many are still in flight. > As we have this knowledge, we can make use of that inside > i915_gem_evict_something() as a better test to see if the device is > idle. > > Signed-off-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> I've just confirmed this was the bug that triggered ENOSPC with (failure to do) context eviction. With this fixed, the test times out (because we are now forced to wait for the fence to be signaled) and so preemption/scheduling is fubar. I guess we just pencil that in as yet-another-pi-bug. -Chris _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
