On Fri, Nov 27, 2015 at 11:03:05AM +0100, Neil Roberts wrote: > "Pohjolainen, Topi" <topi.pohjolai...@intel.com> writes: > > >> glBindFramebuffer(GL_FRAMEBUFFER, piglit_winsys_fbo); > >> piglit_draw_rect(offset * 16 * 2.0f / piglit_width - 1.0f, > > > > This is a question regarding the existing logic. Earlier the test > > calls "glBindFramebuffer(GL_FRAMEBUFFER, fbo)" and clears the > > framebuffer desigbated by "fbo". Then just above the test sets the > > target framebuffer to "piglit_winsys_fbo", and blits into > > "piglit_winsys_fbo" using piglit_draw_rect(). Please bare with me, but > > I understand the idea being that the cleared values from "fbo" are > > blit to "piglit_winsys_fbo". But "glBindFramebuffer(GL_FRAMEBUFFER, > > piglit_winsys_fbo)" sets "piglit_winsys_fbo" both as source and > > destination, doesn't it? > > That is correct, but the read buffer is only used for a blit, ie when > calling glBlitFramebuffer. This is not actually doing a blit but is > instead drawing a regular rectangle while using the texture from the > test framebuffer as a texture source. That means only the draw > framebuffer is actually used. The idea of the test is to test sampling > from the surface so that we can be sure the clear color programmed in > the texture surface state works correctly, so I think it makes sense > here to explicitly use the surface as a texture source rather than a > blit.
Thanks for the explanation, it makes sense now. The patch is: Reviewed-by: Topi Pohjolainen <topi.pohjolai...@intel.com> _______________________________________________ Piglit mailing list Piglit@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/piglit