Hi Matt, On Fri, Dec 02, 2022 at 12:28:43PM +0000, Matthew Auld wrote: > Probably a good idea to do an igt_flush_test() at the end of each > subtest, just to be sure the previous work has been flushed and doesn't > somehow interfere with the current subtest. > > Signed-off-by: Matthew Auld <[email protected]> > Cc: Chris Wilson <[email protected]> > Cc: Andi Shyti <[email protected]> > Cc: Andrzej Hajda <[email protected]> > Cc: Nirmoy Das <[email protected]>
[...] > @@ -647,7 +661,7 @@ int intel_migrate_live_selftests(struct drm_i915_private > *i915) > if (!gt->migrate.context) > return 0; > > - return i915_subtests(tests, >->migrate); > + return intel_gt_live_subtests(tests, gt); yes... this calls the gt teardown function that waits for all the works to end and yields to userspace signals... makes sense. Reviewed-by: Andi Shyti <[email protected]> Thanks, Andi
