On Wed, Feb 18, 2015 at 07:13:54PM +0000, Neil Roberts wrote: > Ben Widawsky <b...@bwidawsk.net> writes: > > > I promise to look at this again in more detail tomorrow when I am more > > awake, but meanwhile, I'd be very much in favor of just setting > > mt->qpith for all miptree layout types. > > I'm not sure what you mean. Do you mean that we should just always set > mt->qpitch to the actual value that we're going to upload? Yes, that > seems like a good idea. I was originally worried about that because I > had thought that other bits of code were looking at the qpitch to work > out some offsets but that is not the case. >
That was what I meant - it was very poorly stated. > The qpitch seems to also be broken for 3D textures and compressed > textures so I think it might be a good idea to squash this patch into a > general patch which fixes everything. I made a start on that as part of > fixing the copyteximage test for 3D textures here: > > https://github.com/bpeel/mesa/commit/fc95546c49f0e5c5805139712c51a0d13b2183c3 > > I don't want to post the patch to the mailing list yet because it > doesn't fix the test for the GL_RGBA32 format with 3D textures and I'd > like to work out why first. > > Regards, > - Neil Okay. I got distracted today trying to help Chris debug TS anyway. A squash sounds good to me, but I presume it's still worth reviewing these anyway - which I will do ASAP. _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
