On Tue, Aug 15, 2017 at 11:57:40PM +0100, Chris Wilson wrote: > For the set of execbuf flags who by definition are based on whether or > not the kernel supports that feature, the ultimate arbiter of whether or > not the kernel accepts the flag is the kernel. The negative tests were > second guessing the kernel and not checking behaviour. Indeed, the tests > failed quite spectacularly to spot a 5 year old bug in engine selection. > > The dubious invalid-flag negative test remains. It will always be > a catch-22, either the test is neutered before the kernel, or it will > lag behind the kernel and fail. It doesn't actually tell us if we fluked > out in failing the execbuf, but by testing all unknown flags it > functions better at the role it was meant for (that all as of yet unused > flags are not accepted). > > Signed-off-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@xxxxxxxxx> _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
