The 3D destination shares the same cache so we don't have any trouble with
the later commands needing the writes flushed inside of the same batchbuffer.
Instructions such as RCP, RSQ, LOG must smear the result of the function
across the dest register's X, Y, Z and W channels (subject to write masking).
Before this change, only the X component was getting written.
Among other things, this fixes cube map texture sampling in GLSL shaders
(since cube lookups involve normalizing the texcoord).
The i965 hardware cannot do GL_CLAMP behavior on textures; an earlier
commit forced a software fallback if strict conformance was required
(i.e. the INTEL_STRICT_CONFORMANCE environment variable was set) and
2D textures were used, but it was somewhat flawed - it could trigger
the software fallback even if 2D textures weren't enabled, as long
as one texture unit was enabled.
This fixes that, and adds software fallback for GL_CLAMP behavior with
1D and 3D textures.
It also adds support for a particular setting of the INTEL_STRICT_CONFORMANCE
environment variable, which forces software fallbacks to be taken *all*
the time. This is helpful with debugging. The value is:
export INTEL_STRICT_CONFORMANCE=2
Move _mesa_init_all_x86_transform_asm() into x86.c so that common_x86.c
has no dependencies on the vertex transformation code.
Plus some comments and clean-ups.
Most of the time unfilled rendering requires a lot more thought than
just translating triangles to lines or points. But sometimes, you can
do exactly that, and it can be quite a bit quicker. Add code to do the
translation. The caller has to determine whether it's a legal thing
to do in the current state, in particular you'd need:
- culling disabled
- offset disabled
- same front and back fill modes
- possibly other stuff I can't think of.
This catches the linux-uclibc case and any others that were being set
prior to 98fcdf3f. Fixes bug 20345.
Signed-off-by: Dan Nicholson <dbn.lists@gmail.com>
