| Age | Commit message (Collapse) | Author |
|---|
|
As osiris pointed out, glxgears slowly gets slower for some reason
when it's enabled, and it's not helping at the moment, so just turn
it off.
|
|
|
|
|
|
If you really want to see it again, check the history.
|
|
We *must* recalculate something in vformat every rebind; let's see
if we can't narrow it down a bit.
|
|
|
|
I don't even know if texkill works right now.
|
|
Uf. Lots of files touched. Would people with working vega, xorg, dri1, etc.
please make sure you are not broken, and fix yourself up if you are.
There were only two or three places where the code did not have painful
fallbacks, so I would advise st maintainers to find less painful workarounds,
or consider overhauling util_surface_copy and util_surface_fill.
Per ymanton, darktama, and Dr_Jakob's suggestions, clear has been left as-is.
I will not add PIPE_CAP_BLITTER unless it is deemed necessary.
|
|
Won't ever be supported.
|
|
This is just split out from the next commit, that's all.
|
|
Eliminates part of the glxgears corruption here. Need to clean up PSC more,
to get rid of the rest of it.
|
|
|
|
Need to move rs_block to this, too.
Also, I'm getting massive amounts of flicker for some reason; I bet we've gotta
re-re-examine PSC and friends. :C
|
|
multitexarray works on my r300, but texrect doesn't.
|
|
It may seem pointless, but this avoids a fair amount of predicted CSO pain.
|
|
Need to get it into its own atom instead of piggybacking on DSA.
|
|
|
|
This should be handled in the emit fine
|
|
I missed this, thanks to Corbin for pointing it out.
|
|
Still missing the frag uses kill support, hopefully nha can point that out.
|
|
Add support for begin/end in each CS so we don't get any other
processes rendering in between.
TODO:
blame other parts of driver for this not working like Z.
|
|
This means we don't emit in the begin query but when we have
to flush. Similiar to classic.
TODO:
make query object actually work.
|
|
1: add rv530 support
- num z pipes cap
- add proper start/finish query options for rv530
2: convert to use linked list properly.
3: add flushing required check.
4: initial Z top disabling support.
TODO:
make it actually work on my rv530.
|
|
This attempts to make r300g do proper bo space checking as opposed
to whatever it was doing now.
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
to reproduce, start texrect, disable 0 texture in menu.
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
|
|
It gets really annoying watching r300g tell me how it's filling surfaces.
Or falling back during filling surfaces.
|
|
Still not sure why st keeps handing down things we can't render to.
|
|
|
|
|
|
Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
|
|
this makes the default state same as r300
|
|
|
|
Conflicts:
src/gallium/drivers/r300/r300_tgsi_to_rc.c
Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
|
|
So I didn't touch r300compiler, but r300g now compiles after having
declarations and code untangled. As nha so gently points out, we shouldn't
have to do this just to comply with MSVC compilers.
|
|
We now have 48 GLX visuals. Pretty soon, we'll have 90+ visuals,
only five of which ever get tested. :3
|
|
Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
|
|
Note that control flow instruction support isn't actually fully functional yet.
Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
|
|
|
|
Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
|
|
Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
|
|
This fixes the glxgears bug, among other things.
Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
|
|
Always test for PIPE_TRANSFER_READ/WRITE using the bit-wise and operator, and
add a pipe_transfer_buffer_flags() helper for getting the buffer usage flags
corresponding to them.
|
|
why there are two input position semantic tags is that ureg doesn't
set vs input semantic due to commit: 6d8dbd3d1ec888
so use vs input index instead of semantic name.
|
|
|
|
It could be generated by vertex shader.
|
|
Not sure why we are getting a shader with two inputs with position semantic, but we don't know how to handle it correctly so it's better to stop the app than lock the machine.
|
|
|
|
|
|
|
