nir: Pass a type_size() function pointer into nir_lower_io().

Previously, there were four type_size() functions in play - the i965
compiler backend defined scalar and vec4 type_size() functions, and
nir_lower_io contained its own similar functions.

In fact, the i965 driver used nir_lower_io() and then looped over the
components using its own type_size - meaning both were in play.  The
two are /basically/ the same, but not exactly in obscure cases like
subroutines and images.

This patch removes nir_lower_io's functions, and instead makes the
driver supply a function pointer.  This gives the driver ultimate
flexibility in deciding how it wants to count things, reduces code
duplication, and improves consistency.

v2 (Jason Ekstrand):
 - One side-effect of passing in a function pointer is that nir_lower_io is
   now aware of and properly allocates space for image uniforms, allowing
   us to drop hacks in the backend

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
v2 Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
This commit is contained in:
Kenneth Graunke
2015-08-12 14:29:25 -07:00
committed by Jason Ekstrand
parent a23f82053d
commit 6c33d6bbf9
4 changed files with 31 additions and 124 deletions

View File

@@ -37,99 +37,12 @@
struct lower_io_state {
nir_builder builder;
void *mem_ctx;
bool is_scalar;
int (*type_size)(const struct glsl_type *type);
};
static int
type_size_vec4(const struct glsl_type *type)
{
unsigned int i;
int size;
switch (glsl_get_base_type(type)) {
case GLSL_TYPE_UINT:
case GLSL_TYPE_INT:
case GLSL_TYPE_FLOAT:
case GLSL_TYPE_BOOL:
if (glsl_type_is_matrix(type)) {
return glsl_get_matrix_columns(type);
} else {
return 1;
}
case GLSL_TYPE_ARRAY:
return type_size_vec4(glsl_get_array_element(type)) * glsl_get_length(type);
case GLSL_TYPE_STRUCT:
size = 0;
for (i = 0; i < glsl_get_length(type); i++) {
size += type_size_vec4(glsl_get_struct_field(type, i));
}
return size;
case GLSL_TYPE_SUBROUTINE:
return 1;
case GLSL_TYPE_SAMPLER:
return 0;
case GLSL_TYPE_ATOMIC_UINT:
return 0;
case GLSL_TYPE_IMAGE:
case GLSL_TYPE_VOID:
case GLSL_TYPE_DOUBLE:
case GLSL_TYPE_ERROR:
case GLSL_TYPE_INTERFACE:
unreachable("not reached");
}
return 0;
}
static unsigned
type_size_scalar(const struct glsl_type *type)
{
unsigned int size, i;
switch (glsl_get_base_type(type)) {
case GLSL_TYPE_UINT:
case GLSL_TYPE_INT:
case GLSL_TYPE_FLOAT:
case GLSL_TYPE_BOOL:
return glsl_get_components(type);
case GLSL_TYPE_ARRAY:
return type_size_scalar(glsl_get_array_element(type)) * glsl_get_length(type);
case GLSL_TYPE_STRUCT:
size = 0;
for (i = 0; i < glsl_get_length(type); i++) {
size += type_size_scalar(glsl_get_struct_field(type, i));
}
return size;
case GLSL_TYPE_SUBROUTINE:
return 1;
case GLSL_TYPE_SAMPLER:
return 0;
case GLSL_TYPE_ATOMIC_UINT:
return 0;
case GLSL_TYPE_INTERFACE:
return 0;
case GLSL_TYPE_IMAGE:
return 0;
case GLSL_TYPE_VOID:
case GLSL_TYPE_ERROR:
case GLSL_TYPE_DOUBLE:
unreachable("not reached");
}
return 0;
}
static unsigned
type_size(const struct glsl_type *type, bool is_scalar)
{
if (is_scalar)
return type_size_scalar(type);
else
return type_size_vec4(type);
}
void
nir_assign_var_locations(struct exec_list *var_list, unsigned *size, bool is_scalar)
nir_assign_var_locations(struct exec_list *var_list, unsigned *size,
int (*type_size)(const struct glsl_type *))
{
unsigned location = 0;
@@ -143,7 +56,7 @@ nir_assign_var_locations(struct exec_list *var_list, unsigned *size, bool is_sca
continue;
var->data.driver_location = location;
location += type_size(var->type, is_scalar);
location += type_size(var->type);
}
*size = location;
@@ -193,7 +106,7 @@ nir_assign_var_locations_direct_first(nir_shader *shader,
struct exec_list *var_list,
unsigned *direct_size,
unsigned *size,
bool is_scalar)
int (*type_size)(const struct glsl_type *))
{
struct set *indirect_set = _mesa_set_create(NULL, _mesa_hash_pointer,
_mesa_key_pointer_equal);
@@ -215,7 +128,7 @@ nir_assign_var_locations_direct_first(nir_shader *shader,
continue;
var->data.driver_location = location;
location += type_size(var->type, is_scalar);
location += type_size(var->type);
}
*direct_size = location;
@@ -229,7 +142,7 @@ nir_assign_var_locations_direct_first(nir_shader *shader,
continue;
var->data.driver_location = location;
location += type_size(var->type, is_scalar);
location += type_size(var->type);
}
*size = location;
@@ -254,7 +167,7 @@ get_io_offset(nir_deref_var *deref, nir_instr *instr, nir_src *indirect,
if (tail->deref_type == nir_deref_type_array) {
nir_deref_array *deref_array = nir_deref_as_array(tail);
unsigned size = type_size(tail->type, state->is_scalar);
unsigned size = state->type_size(tail->type);
base_offset += size * deref_array->base_offset;
@@ -275,9 +188,10 @@ get_io_offset(nir_deref_var *deref, nir_instr *instr, nir_src *indirect,
} else if (tail->deref_type == nir_deref_type_struct) {
nir_deref_struct *deref_struct = nir_deref_as_struct(tail);
for (unsigned i = 0; i < deref_struct->index; i++)
base_offset += type_size(glsl_get_struct_field(parent_type, i),
state->is_scalar);
for (unsigned i = 0; i < deref_struct->index; i++) {
base_offset +=
state->type_size(glsl_get_struct_field(parent_type, i));
}
}
}
@@ -395,13 +309,13 @@ nir_lower_io_block(nir_block *block, void *void_state)
}
static void
nir_lower_io_impl(nir_function_impl *impl, bool is_scalar)
nir_lower_io_impl(nir_function_impl *impl, int(*type_size)(const struct glsl_type *))
{
struct lower_io_state state;
nir_builder_init(&state.builder, impl);
state.mem_ctx = ralloc_parent(impl);
state.is_scalar = is_scalar;
state.type_size = type_size;
nir_foreach_block(impl, nir_lower_io_block, &state);
@@ -410,10 +324,10 @@ nir_lower_io_impl(nir_function_impl *impl, bool is_scalar)
}
void
nir_lower_io(nir_shader *shader, bool is_scalar)
nir_lower_io(nir_shader *shader, int(*type_size)(const struct glsl_type *))
{
nir_foreach_overload(shader, overload) {
if (overload->impl)
nir_lower_io_impl(overload->impl, is_scalar);
nir_lower_io_impl(overload->impl, type_size);
}
}