On Thu, Jan 16, 2014 at 05:47:50PM +0100, Daniel Vetter wrote: > On Wed, Jan 15, 2014 at 04:57:28PM -0200, Rodrigo Vivi wrote: > > Hi Ville and Daniel, > > > > Ville, since this is heavily based on your kms_fbc_psr I'd appreciate > > your comments here please. > > Mainly regarding the page_flip+mmaps which I'm not sure how usefull > > they are on this psr case. > > Or my implementation is so wrong that they aren't being useful at all. > > > > Daniel, is this test what you had in mind? All comments, bikesheds, > > blames are appreciated! ;) > > Maybe I miss it, but Ville's fbc testcase not only checks that the crc > change, but that the new one perfectly matches the new frame. This is > useful for catching partial updates ... That's actually only true for the PAGE_FLIP test. The others (including the PAGE_FLIP_AND variants) just check that the crc doesn't match. But I suppose it should be possible to extend that to all the cases if we want to. As far as this PSR test case goes, the PAGE_FLIP_AND variant handling is pretty much busted. The only case where you actually do the page flip is TEST_PAGE_FLIP_AND_MMAP_GTT, and even then since the reference crc matches the original frame, the page flip itself would already cause the test to pass, so it's not actually testing the GTT write at all. The rest looks pretty much OK. It's mostly a copy of kms_fbc_crc though, so I wonder if we should try to unify them, just add some flag to indicate if we should be doing PSR or FBC tests. -- Ville Syrjälä Intel OTC _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/intel-gfx
