[Mesa-dev] [PATCH 4/4] program: remove _mesa_init_*_program wrappers

[Marek Olšák]

On Thu, Oct 8, 2015 at 3:49 PM, Emil Velikov <emil.l.velikov at gmail.com> wrote:
> On 8 October 2015 at 01:12, Marek Olšák <maraeo at gmail.com> wrote:
>> From: Marek Olšák <marek.olsak at amd.com>
>>
>> They didn't do anything useful.
>> ---
>>  src/mesa/drivers/dri/i915/i915_fragprog.c          |   7 +-
>>  src/mesa/drivers/dri/i965/brw_program.c            |  10 +-
>>  .../drivers/dri/i965/test_fs_cmod_propagation.cpp  |   2 +-
>>  .../dri/i965/test_fs_saturate_propagation.cpp      |   2 +-
>>  .../dri/i965/test_vec4_copy_propagation.cpp        |   2 +-
>>  .../dri/i965/test_vec4_register_coalesce.cpp       |   2 +-
>>  src/mesa/drivers/dri/r200/r200_vertprog.c          |   4 +-
>>  src/mesa/program/program.c                         | 133 +++------------------
>>  src/mesa/program/program.h                         |  29 +----
>>  src/mesa/state_tracker/st_cb_program.c             |  43 +++----
>>  10 files changed, 50 insertions(+), 184 deletions(-)
>>
>> diff --git a/src/mesa/drivers/dri/i915/i915_fragprog.c b/src/mesa/drivers/dri/i915/i915_fragprog.c
>> index 1a5943c..237d219 100644
>> --- a/src/mesa/drivers/dri/i915/i915_fragprog.c
>> +++ b/src/mesa/drivers/dri/i915/i915_fragprog.c
>> @@ -1316,8 +1316,8 @@ i915NewProgram(struct gl_context * ctx, GLenum target, GLuint id)
>>  {
>>     switch (target) {
>>     case GL_VERTEX_PROGRAM_ARB:
>> -      return _mesa_init_vertex_program(ctx, CALLOC_STRUCT(gl_vertex_program),
>> -                                       target, id);
>> +      return _mesa_init_gl_program(CALLOC_STRUCT(gl_vertex_program),
>> +                                   target, id);
>>
> Worth doing the memory allocation to a temporary variable, and feeding
> &prog->base to _mesa_init_gl_program ?

I tried that first, but it required more lines of code.

>
>> --- a/src/mesa/drivers/dri/r200/r200_vertprog.c
>> +++ b/src/mesa/drivers/dri/r200/r200_vertprog.c
>> @@ -1205,9 +1205,9 @@ r200NewProgram(struct gl_context *ctx, GLenum target, GLuint id)
>>     switch(target){
>>     case GL_VERTEX_PROGRAM_ARB:
>>        vp = CALLOC_STRUCT(r200_vertex_program);
>> -      return _mesa_init_vertex_program(ctx, &vp->mesa_program, target, id);
>> +      return _mesa_init_gl_program(&vp->mesa_program, target, id);
>>     case GL_FRAGMENT_PROGRAM_ARB:
>> -      return _mesa_init_fragment_program( ctx, CALLOC_STRUCT(gl_fragment_program), target, id );
>> +      return _mesa_init_gl_program(CALLOC_STRUCT(gl_fragment_program), target, id);
> Ditto.
>
>
>> --- a/src/mesa/program/program.c
>> +++ b/src/mesa/program/program.c
>
>> @@ -309,34 +217,29 @@ _mesa_new_program(struct gl_context *ctx, GLenum target, GLuint id)
>>     struct gl_program *prog;
>>     switch (target) {
>>     case GL_VERTEX_PROGRAM_ARB: /* == GL_VERTEX_PROGRAM_NV */
>> -      prog = _mesa_init_vertex_program(ctx, CALLOC_STRUCT(gl_vertex_program),
>> -                                       target, id );
>> +      prog = _mesa_init_gl_program(CALLOC_STRUCT(gl_vertex_program),
>> +                                   target, id);
> Ditto. + early return similar to st/mesa ?
>
>> --- a/src/mesa/program/program.h
>> +++ b/src/mesa/program/program.h
>
>> -extern struct gl_program *
>> -_mesa_init_compute_program(struct gl_context *ctx,
>> -                           struct gl_compute_program *prog,
>> -                           GLenum target, GLuint id);
>> +_mesa_init_gl_program(void *prog, GLenum target, GLuint id);
>>
> If you go for my suggestion you can keep prog as struct gl_program *.
>
>> --- a/src/mesa/state_tracker/st_cb_program.c
>> +++ b/src/mesa/state_tracker/st_cb_program.c
>
>> +   switch (target) {
>> +   case GL_VERTEX_PROGRAM_ARB:
>> +      prog = (struct gl_program*)ST_CALLOC_STRUCT(st_vertex_program);
>> +      break;
>> +   case GL_FRAGMENT_PROGRAM_ARB:
> Worth adding case GL_FRAGMENT_PROGRAM_NV, or perhaps nuking it from
> mesa/program/program.c above ?

I think GL_FRAGMENT_PROGRAM_NV is from an NV extension supported by
some classic drivers but not st/mesa.

> Can we keep the variables as is and avoid the casts ?

ST_CALLOC_STRUCT needs a cast.
Or &ST_CALLOC_STRUCT(...)->Base.Base if I fix the macro to allow that.

Marek