On 21 August 2017 at 09:28, Dave Airlie <airl...@gmail.com> wrote: > On 18 August 2017 at 04:31, Marek Olšák <mar...@gmail.com> wrote: >> From: Marek Olšák <marek.ol...@amd.com> >> > > This break r600 first. > >> - if (attrib) { >> + switch (type) { >> + case UTIL_BLITTER_ATTRIB_COLOR: >> memcpy(vb+4, attrib->f, sizeof(float)*4); >> memcpy(vb+12, attrib->f, sizeof(float)*4); >> memcpy(vb+20, attrib->f, sizeof(float)*4); >> + break; >> + case UTIL_BLITTER_ATTRIB_TEXCOORD: >> + vb[4] = attrib->f[0]; /* x1 */ >> + vb[5] = attrib->f[1]; /* y1 */ >> + vb[12] = attrib->f[0]; /* x1 */ >> + vb[13] = attrib->f[3]; /* y2 */ >> + vb[20] = attrib->f[2]; /* x2 */ >> + vb[21] = attrib->f[1]; /* y1 */ > > Something in here is backwards compared to the normal path. > > Making this (0,1) (2, 1), (2, 3) fixes evergreen for me, until the later > patch breaks it again. > > This is the same order that set_texcoords_in_vertices uses > before this patch.
Actually I can still get into a wierd place where simple tests that readback fail. I'll try and dig some more. Dave. _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev