On Wed, Jun 20, 2018 at 10:01 AM, Ilia Mirkin <imir...@alum.mit.edu> wrote: > On Wed, Jun 20, 2018 at 9:56 AM, Rob Clark <robdcl...@gmail.com> wrote: >> On Tue, Jun 19, 2018 at 9:02 PM, Jonathan Marek <jonat...@marek.ca> wrote: >>> blend can be NULL, so check for that >> >> hmm, which case are you hitting that? From a quick look I think >> a3xx-a5xx would have the same issue. Maybe we should just init >> ctx->blend to a dummy state obj when the ctx is created? > > This generally happens with clears. A long time ago, st/mesa got > rejiggered slightly, and a glClear() could happen before a blend state > was bound. Many drivers implement clear by doing some amount of > validation, so they have to watch out. I'm sure a3xx+ were already > fixed for this, but not surprising that a2xx still has issues. >
ahh, ok.. that makes more sense.. a5xx uses hw blitter for clears, and a3xx-a4xx use u_blitter.. a2xx is doing something a bit more custom since (iirc) I didn't know how to write depth from frag shader.. BR, -R _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
