Should be easier to read and work with than the older ways of emitting
TGSI tokens.
Also, emit simpler TGSI than previously:
- translate away source and dest extended modifiers
- translate away the SWZ opcode
Previously, the FOGC attribute contained the fragment fog coord, front/back-
face flag and the gl_PointCoord.xy values. Now each of those things are
separate fragment program attributes. This simplifies quite a few things in
Mesa and gallium.
Need to test i965 driver and fix up point coord handling in the gallium/draw
module...
mesa allocates both frontface and pointcoord registers within the fog
coordinate register, by using swizzling. to make it cleaner and easier
for drivers we want each of them in its own register. so when doing
compilation from the mesa IR to tgsi allocate new registers for both
and add new semantics to the respective declarations.
It's legal for ARB_vertex_program programs to not write to result.position.
The results are undefined in that case. This assertion was causing us to
abort/exit though.
The recent increase ST_MAX_SHADER_TOKENS to 8K causes stack overflows on
windows.
Failure to allocate is not being propagated to the caller. This is not
a regression since the previous _mesa_malloc result wasn't being
checked as well. Unfortunately it is not easy to fix, as the callers of
these functions do not have failure propagation mechanism either, and
so on. So leaving a just fixme note for now.
s/FRAG_RESULT_DEPR/FRAG_RESULT_DEPTH/
s/FRAG_RESULT_COLR/FRAG_RESULT/COLOR/
Remove FRAG_RESULT_COLH (NV half-precision) output since we never used it.
Next, we might merge the COLOR and DATA outputs (COLOR0, COLOR1, etc).
In some cases, the vertex program output's semantic info didn't match up
with the fragment program's input semantic info. This info is now explicitly
passed into the st_translate_fragment_program() function.
Use the code in cso_context.c rather than st_cache.c.
Basically, binding of state objects now goes through the CSO module.
But Vertex/fragment shaders go through pipe->bind_fs/vs_state() since they're
not cached by the CSO module at this time.
Also, update softpipe driver to handle NULL state objects in various places.
This happens during context destruction. May need to update other drivers...
Update the Makefiles and includes for the new paths.
Note that there hasn't been no separation of the Makefiles yet, and make is
jumping all over the place. That will be taken care shortly. But for now, make
should work. It was tested with linux and linux-dri. Linux-cell and linux-llvm
might require some minor tweaks.
The CSO returned by pipe->create_vs_state() can't be passed to the
private draw module. That was causing glRasterPos to blow up.
Add a 'draw_shader' field to st_vertex_program for use with the private
draw module.
Change st_context->state.vs type from cso_vertex_shader to st_vertex_program.
We were doing this for the sake of softpipe and the tgsi intergrepter since
we always need the fragment position and W-coordinate information in order
to compute fragment interpolants.
But that's not appropriate for hardware drivers.
The tgsi interpreter now get x,y,w information from a separate tgsi_exec_vector
variable setup by softpipe.
The new pipe_shader_state->input_map[] defines how vert shader outputs map
to frag shader inputs. It may go away though, since one can also examine
the semantic label on frag shader input[0] to figure things out.
Move the CPU vertex shader execution code to the draw
module, remove traces of LLVM from the state tracker,
abstract execution engine for the purposes of the draw module.
