| Age | Commit message (Collapse) | Author |
|---|
|
|
|
It is possible to bind texture images of an incomplete mipmapped texture.
Software fallbacks in this case incorrectly tried to mmap the entire texture.
Additionally, add span functions for 1555 and 4444 formats.
This fixes crashes in piglit's fbo-readpixels test; unfortunately, the test
itself still fails - this needs to be investigated.
Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
|
|
This updates some of the clear code from Intel gives a 5x clearspd perf
for me here. played openarena also, not sure if the viewport changes
broke anything,
|
|
it might still be misrendering not sure
|
|
fgl_glxgears -fbo runs, though the gears don't look right yet.
|
|
|
|
make sure to turn off when no texture is used in hw
|
|
|
|
|
|
Can be used for buffer swap as well.
|
|
texture bo setup isn't quite working yet
|
|
|
|
|
|
This moves a big chunk of the space checking code into libdrm so
it can be shared by the DDX.
|
|
Fixes fighting between GPU and software rendering with TTM.
|
|
half stealing the code without taking the intel regions
|
|
I suspect this might break TFP in some way but it makes firecube run here
|
|
Fixes #22181. R200 requires this since DP4 is used in hw tnl mode.
R300 prefers it (should be faster due to no instruction dependencies), but
both methods should be correct (when sw tcl is used though, MUL/MAD might
be faster). Probably doesn't make much difference for R100 since vertex progs
are executed in software anyway, but let's just keep it the same there too.
|
|
This reverts commit 0952645fe04a27968565ea4d913500c23b1b11e3.
Need to revisit where this is going wrong
|
|
This fixes an assertion in glReadPixels from the front buffer.
Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
|
|
This allows us to return the unused portion of the dma buffer
to the allocator instead of wasting nearly 16k a pop.
|
|
still always enable max, but the right values this time.
More work should probably be done for saner limits without mm, and/or
dri conf option allow_large_textures (which is ignored) removed.
3D limit on r100 is pretty arbitrary as still handled by swrast anyway.
Also fix r300 limits (except 3d I've no idea what the max is anyway so
keep using mesa default).
|
|
|
|
This was already correct in the GEM code
|
|
need to pass target parameter to radeon_teximage/radeon_subteximage functions
otherwise mipmap generation for cube maps can't work (assert/segfault in
_mesa_generate_mipmap)
|
|
drm cmd checker would refuse cube emits
also fix an issue in the cs path which would calculate the register
offset off by one dword.
Only same testing done as original code (none except compile tested).
|
|
Fixes #22181. R200 requires this since DP4 is used in hw tnl mode.
R300 prefers it (should be faster due to no instruction dependencies), but
both methods should be correct (when sw tcl is used though, MUL/MAD might
be faster). Probably doesn't make much difference for R100 since vertex progs
are executed in software anyway, but let's just keep it the same there too.
|
|
if its on the list its on the list don't go readding it.
multitexturing from the same texture could cause this.
|
|
We flush the command buffer so we don't emit mixed
state (with new and previous buffer size) command
buffer, this is especialy affecting zbuffer states.
|
|
|
|
|
|
Tested with glean/texture_srgb and wine/d3d9 tests on RV535
|
|
|
|
|
|
use the actual value set in the context
|
|
|
|
|
|
fixes last commit.
|
|
|
|
This gets the correct srclvl image map when uploading images to the new mipmap.
|
|
Display the chip family and PCI ID. This can be parsed easily,
and essentially all information that the driver has about the chip can be
deduced from it.
Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
To avoid locking bug we shouldn't not call firevertices from this path
as it's call from radeon get lock.
|
|
Revalidate can trigger flushing and dma buffer deallocation,
so retry allocation on such case.
|
|
The duplication of state data caused a crash due to double-free on destruction
of context, because a variable wasn't correctly null'ed out.
Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
|
|
For some unknown reasons the scratch reg value doesn't endup in the status
map at the scratch reg offset, this is a temporary work around until we
figure out why it doesn't work.
|
