On Mon, Nov 4, 2013 at 6:26 PM, Mateo Lozano, Oscar <oscar.mateo@xxxxxxxxx> wrote: >> > IMHO, gem_quiescent_gpu() does need the drop cache call, otherwise it >> doesn´t really do what it advertises. However, calling gem_quiescent_gpu() >> inside get_object_count() is probably overkill, and could potentially mask >> something going wrong (not so much when called at the beginning of a >> subtest, but definitely when called at the end!). >> >> Well i-g-t testcase run under the assumption that nothing else does, so most >> often the drop_cache call would be a noop. Save when there's indeed >> something left to retire, in which case it matters actually. >> I'll add it if you don't feel like doing it yourself. > > It´s ok, I´ll add it. But... which call do you want?: > > A) igt_drop_caches_set() inside gem_quiescent_gpu() > B) gem_quiescent_gpu() inside get_object_count() > C) Both of them I'd go with drop_caches(RETIRE) in get_object_count + in gem_quiescent_gpu. -Daniel -- Daniel Vetter Software Engineer, Intel Corporation +41 (0) 79 365 57 48 - http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/intel-gfx
