On Mon, May 3, 2010 at 6:07 PM, Marek Olšák <[email protected]> wrote: > When ColorTiling is enabled, macrotiled small pixmaps cannot be used as > textures because of MACRO_SWITCH. The attached patch fixes that and this > issue is well described in the commit message. Please review. > > This also fixes some piglit tests for r300g with ColorTiling enabled, e.g. > glean/readPixSanity and shaders/fp-fog among others. (The > texturing/texredefine test fails due to a flush missing somewhere, but the > driver now passes more tests than without ColorTiling.)
This patch looks good to me. > > This fix is only for r300. I am not sure whether there is the same issue on > non-r300 GPUs too. > I'm pretty sure r1xx/r2xx worked similarly. > Also can ColorTiling be enabled by default on KMS now? For r3xx-r5xx, sure. For r1xx, r2xx, we need to either implement span code in the 3D driver to deal with the tile patterns which are different from r3xx, or uses the 3D engine to blit to a linear surface before sw access. Alex _______________________________________________ xorg-driver-ati mailing list [email protected] http://lists.x.org/mailman/listinfo/xorg-driver-ati
