virtual void leave_record(const glsl_type *type, const char *,
bool row_major, const unsigned packing) {
assert(type->is_record());
if (this->buffer_block_index == -1)
return;
if (packing == GLSL_INTERFACE_PACKING_STD430)
this->ubo_byte_offset = glsl_align(
this->ubo_byte_offset, type->std430_base_alignment(row_major));
else
this->ubo_byte_offset = glsl_align(
this->ubo_byte_offset, type->std140_base_alignment(row_major));
}
virtual void visit_field(const glsl_type *type, const char *name,
bool row_major, const glsl_type *record_type)
{
assert(this->index < this->num_variables);
gl_uniform_buffer_variable *v = &this->variables[this->index++];
v->Name = ralloc_strdup(mem_ctx, name);
v->Type = type;
v->RowMajor = row_major;
if (this->is_array_instance) {
v->IndexName = ralloc_strdup(mem_ctx, name);
char *open_bracket = strchr(v->IndexName, '[');
assert(open_bracket != NULL);
char *close_bracket = strchr(open_bracket, ']');
assert(close_bracket != NULL);
/* Length of the tail without the ']' but with the NUL.
*/
unsigned len = strlen(close_bracket + 1) + 1;
memmove(open_bracket, close_bracket + 1, len);
} else {
v->IndexName = v->Name;
}
const unsigned alignment = record_type
? record_type->std140_base_alignment(!!v->RowMajor)
: type->std140_base_alignment(!!v->RowMajor);
unsigned size = type->std140_size(!!v->RowMajor);
this->offset = glsl_align(this->offset, alignment);
v->Offset = this->offset;
this->offset += size;
/* From the GL_ARB_uniform_buffer_object spec:
*
* "For uniform blocks laid out according to [std140] rules, the
* minimum buffer object size returned by the
* UNIFORM_BLOCK_DATA_SIZE query is derived by taking the offset of
* the last basic machine unit consumed by the last uniform of the
* uniform block (including any end-of-array or end-of-structure
* padding), adding one, and rounding up to the next multiple of
* the base alignment required for a vec4."
*/
this->buffer_size = glsl_align(this->offset, 16);
}
virtual void leave_record(const glsl_type *type, const char *,
bool row_major) {
assert(type->is_record());
if (this->ubo_block_index == -1)
return;
this->ubo_byte_offset = glsl_align(
this->ubo_byte_offset, type->std140_base_alignment(row_major));
}
virtual void visit_field(const glsl_struct_field *field)
{
/* FINISHME: When support for doubles (dvec4, etc.) is added to the
* FINISHME: compiler, this may be incorrect for a structure in a UBO
* FINISHME: like struct s { struct { float f } s1; dvec4 v; };.
*/
this->offset = glsl_align(this->offset,
field->type->std140_base_alignment(false));
}
virtual void visit_field(const glsl_type *type, const char *name,
bool row_major)
{
assert(!type->is_record());
assert(!(type->is_array() && type->fields.array->is_record()));
assert(!type->is_interface());
assert(!(type->is_array() && type->fields.array->is_interface()));
(void) row_major;
unsigned id;
bool found = this->map->get(id, name);
assert(found);
if (!found)
return;
/* If there is already storage associated with this uniform, it means
* that it was set while processing an earlier shader stage. For
* example, we may be processing the uniform in the fragment shader, but
* the uniform was already processed in the vertex shader.
*/
if (this->uniforms[id].storage != NULL) {
/* If the uniform already has storage set from another shader stage,
* mark the samplers used for this shader stage.
*/
if (type->contains_sampler()) {
const unsigned count = MAX2(1, this->uniforms[id].array_elements);
const unsigned shadow = (type->is_array())
? type->fields.array->sampler_shadow : type->sampler_shadow;
for (unsigned i = 0; i < count; i++) {
const unsigned s = this->uniforms[id].sampler + i;
this->shader_samplers_used |= 1U << s;
this->shader_shadow_samplers |= shadow << s;
}
}
return;
}
const glsl_type *base_type;
if (type->is_array()) {
this->uniforms[id].array_elements = type->length;
base_type = type->fields.array;
} else {
this->uniforms[id].array_elements = 0;
base_type = type;
}
if (base_type->is_sampler()) {
this->uniforms[id].sampler = this->next_sampler;
/* Increment the sampler by 1 for non-arrays and by the number of
* array elements for arrays.
*/
this->next_sampler += MAX2(1, this->uniforms[id].array_elements);
const gl_texture_index target = base_type->sampler_index();
const unsigned shadow = base_type->sampler_shadow;
for (unsigned i = this->uniforms[id].sampler
; i < MIN2(this->next_sampler, MAX_SAMPLERS)
; i++) {
this->targets[i] = target;
this->shader_samplers_used |= 1U << i;
this->shader_shadow_samplers |= shadow << i;
}
} else {
this->uniforms[id].sampler = ~0;
}
this->uniforms[id].name = ralloc_strdup(this->uniforms, name);
this->uniforms[id].type = base_type;
this->uniforms[id].initialized = 0;
this->uniforms[id].num_driver_storage = 0;
this->uniforms[id].driver_storage = NULL;
this->uniforms[id].storage = this->values;
if (this->ubo_block_index != -1) {
this->uniforms[id].block_index = this->ubo_block_index;
unsigned alignment = type->std140_base_alignment(ubo_row_major);
this->ubo_byte_offset = glsl_align(this->ubo_byte_offset, alignment);
this->uniforms[id].offset = this->ubo_byte_offset;
this->ubo_byte_offset += type->std140_size(ubo_row_major);
if (type->is_array()) {
this->uniforms[id].array_stride =
glsl_align(type->fields.array->std140_size(ubo_row_major), 16);
} else {
this->uniforms[id].array_stride = 0;
}
if (type->is_matrix() ||
(type->is_array() && type->fields.array->is_matrix())) {
this->uniforms[id].matrix_stride = 16;
this->uniforms[id].row_major = ubo_row_major;
} else {
this->uniforms[id].matrix_stride = 0;
this->uniforms[id].row_major = false;
}
} else {
this->uniforms[id].block_index = -1;
this->uniforms[id].offset = -1;
this->uniforms[id].array_stride = -1;
this->uniforms[id].matrix_stride = -1;
this->uniforms[id].row_major = false;
}
this->values += values_for_type(type);
}
virtual void visit_field(const glsl_type *type, const char *name,
bool row_major, const glsl_type *record_type,
bool last_field)
{
assert(!type->without_array()->is_record());
assert(!type->without_array()->is_interface());
unsigned id;
bool found = this->map->get(id, name);
assert(found);
if (!found)
return;
const glsl_type *base_type;
if (type->is_array()) {
this->uniforms[id].array_elements = type->length;
base_type = type->fields.array;
} else {
this->uniforms[id].array_elements = 0;
base_type = type;
}
/* This assigns uniform indices to sampler and image uniforms. */
handle_samplers(base_type, &this->uniforms[id]);
handle_images(base_type, &this->uniforms[id]);
/* If there is already storage associated with this uniform, it means
* that it was set while processing an earlier shader stage. For
* example, we may be processing the uniform in the fragment shader, but
* the uniform was already processed in the vertex shader.
*/
if (this->uniforms[id].storage != NULL) {
return;
}
/* Assign explicit locations. */
if (current_var->data.explicit_location) {
/* Set sequential locations for struct fields. */
if (record_type != NULL) {
const unsigned entries = MAX2(1, this->uniforms[id].array_elements);
this->uniforms[id].remap_location =
current_var->data.location + field_counter;
field_counter += entries;
} else {
this->uniforms[id].remap_location = current_var->data.location;
}
} else {
/* Initialize to to indicate that no location is set */
this->uniforms[id].remap_location = UNMAPPED_UNIFORM_LOC;
}
this->uniforms[id].name = ralloc_strdup(this->uniforms, name);
this->uniforms[id].type = base_type;
this->uniforms[id].initialized = 0;
this->uniforms[id].num_driver_storage = 0;
this->uniforms[id].driver_storage = NULL;
this->uniforms[id].storage = this->values;
this->uniforms[id].atomic_buffer_index = -1;
if (this->ubo_block_index != -1) {
this->uniforms[id].block_index = this->ubo_block_index;
const unsigned alignment = record_type
? record_type->std140_base_alignment(row_major)
: type->std140_base_alignment(row_major);
this->ubo_byte_offset = glsl_align(this->ubo_byte_offset, alignment);
this->uniforms[id].offset = this->ubo_byte_offset;
this->ubo_byte_offset += type->std140_size(row_major);
if (last_field)
this->ubo_byte_offset = glsl_align(this->ubo_byte_offset, 16);
if (type->is_array()) {
this->uniforms[id].array_stride =
glsl_align(type->fields.array->std140_size(row_major), 16);
} else {
this->uniforms[id].array_stride = 0;
}
if (type->without_array()->is_matrix()) {
this->uniforms[id].matrix_stride = 16;
this->uniforms[id].row_major = row_major;
} else {
this->uniforms[id].matrix_stride = 0;
this->uniforms[id].row_major = false;
}
} else {
this->uniforms[id].block_index = -1;
this->uniforms[id].offset = -1;
this->uniforms[id].array_stride = -1;
this->uniforms[id].matrix_stride = -1;
this->uniforms[id].row_major = false;
}
this->values += values_for_type(type);
}
unsigned
glsl_type::std140_size(bool row_major) const
{
/* (1) If the member is a scalar consuming <N> basic machine units, the
* base alignment is <N>.
*
* (2) If the member is a two- or four-component vector with components
* consuming <N> basic machine units, the base alignment is 2<N> or
* 4<N>, respectively.
*
* (3) If the member is a three-component vector with components consuming
* <N> basic machine units, the base alignment is 4<N>.
*/
if (this->is_scalar() || this->is_vector()) {
return this->vector_elements * 4;
}
/* (5) If the member is a column-major matrix with <C> columns and
* <R> rows, the matrix is stored identically to an array of
* <C> column vectors with <R> components each, according to
* rule (4).
*
* (6) If the member is an array of <S> column-major matrices with <C>
* columns and <R> rows, the matrix is stored identically to a row of
* <S>*<C> column vectors with <R> components each, according to rule
* (4).
*
* (7) If the member is a row-major matrix with <C> columns and <R>
* rows, the matrix is stored identically to an array of <R>
* row vectors with <C> components each, according to rule (4).
*
* (8) If the member is an array of <S> row-major matrices with <C> columns
* and <R> rows, the matrix is stored identically to a row of <S>*<R>
* row vectors with <C> components each, according to rule (4).
*/
if (this->is_matrix() || (this->is_array() &&
this->fields.array->is_matrix())) {
const struct glsl_type *element_type;
const struct glsl_type *vec_type;
unsigned int array_len;
if (this->is_array()) {
element_type = this->fields.array;
array_len = this->length;
} else {
element_type = this;
array_len = 1;
}
if (row_major) {
vec_type = get_instance(GLSL_TYPE_FLOAT,
element_type->matrix_columns, 1);
array_len *= element_type->vector_elements;
} else {
vec_type = get_instance(GLSL_TYPE_FLOAT,
element_type->vector_elements, 1);
array_len *= element_type->matrix_columns;
}
const glsl_type *array_type = glsl_type::get_array_instance(vec_type,
array_len);
return array_type->std140_size(false);
}
/* (4) If the member is an array of scalars or vectors, the base alignment
* and array stride are set to match the base alignment of a single
* array element, according to rules (1), (2), and (3), and rounded up
* to the base alignment of a vec4. The array may have padding at the
* end; the base offset of the member following the array is rounded up
* to the next multiple of the base alignment.
*
* (10) If the member is an array of <S> structures, the <S> elements of
* the array are laid out in order, according to rule (9).
*/
if (this->is_array()) {
if (this->fields.array->is_record()) {
return this->length * this->fields.array->std140_size(row_major);
} else {
unsigned element_base_align =
this->fields.array->std140_base_alignment(row_major);
return this->length * MAX2(element_base_align, 16);
}
}
/* (9) If the member is a structure, the base alignment of the
* structure is <N>, where <N> is the largest base alignment
* value of any of its members, and rounded up to the base
* alignment of a vec4. The individual members of this
* sub-structure are then assigned offsets by applying this set
* of rules recursively, where the base offset of the first
* member of the sub-structure is equal to the aligned offset
* of the structure. The structure may have padding at the end;
* the base offset of the member following the sub-structure is
* rounded up to the next multiple of the base alignment of the
* structure.
*/
if (this->is_record()) {
unsigned size = 0;
for (unsigned i = 0; i < this->length; i++) {
const struct glsl_type *field_type = this->fields.structure[i].type;
unsigned align = field_type->std140_base_alignment(row_major);
size = glsl_align(size, align);
size += field_type->std140_size(row_major);
}
size = glsl_align(size,
this->fields.structure[0].type->std140_base_alignment(row_major));
return size;
}
assert(!"not reached");
return -1;
}
virtual void visit_field(const glsl_type *type, const char *name,
bool row_major)
{
assert(!type->is_record());
assert(!(type->is_array() && type->fields.array->is_record()));
assert(!type->is_interface());
assert(!(type->is_array() && type->fields.array->is_interface()));
(void) row_major;
unsigned id;
bool found = this->map->get(id, name);
assert(found);
if (!found)
return;
const glsl_type *base_type;
if (type->is_array()) {
this->uniforms[id].array_elements = type->length;
base_type = type->fields.array;
} else {
this->uniforms[id].array_elements = 0;
base_type = type;
}
/* This assigns sampler uniforms to sampler units. */
handle_samplers(base_type, &this->uniforms[id]);
/* If there is already storage associated with this uniform, it means
* that it was set while processing an earlier shader stage. For
* example, we may be processing the uniform in the fragment shader, but
* the uniform was already processed in the vertex shader.
*/
if (this->uniforms[id].storage != NULL) {
return;
}
this->uniforms[id].name = ralloc_strdup(this->uniforms, name);
this->uniforms[id].type = base_type;
this->uniforms[id].initialized = 0;
this->uniforms[id].num_driver_storage = 0;
this->uniforms[id].driver_storage = NULL;
this->uniforms[id].storage = this->values;
if (this->ubo_block_index != -1) {
this->uniforms[id].block_index = this->ubo_block_index;
unsigned alignment = type->std140_base_alignment(ubo_row_major);
this->ubo_byte_offset = glsl_align(this->ubo_byte_offset, alignment);
this->uniforms[id].offset = this->ubo_byte_offset;
this->ubo_byte_offset += type->std140_size(ubo_row_major);
if (type->is_array()) {
this->uniforms[id].array_stride =
glsl_align(type->fields.array->std140_size(ubo_row_major), 16);
} else {
this->uniforms[id].array_stride = 0;
}
if (type->is_matrix() ||
(type->is_array() && type->fields.array->is_matrix())) {
this->uniforms[id].matrix_stride = 16;
this->uniforms[id].row_major = ubo_row_major;
} else {
this->uniforms[id].matrix_stride = 0;
this->uniforms[id].row_major = false;
}
} else {
this->uniforms[id].block_index = -1;
this->uniforms[id].offset = -1;
this->uniforms[id].array_stride = -1;
this->uniforms[id].matrix_stride = -1;
this->uniforms[id].row_major = false;
}
this->values += values_for_type(type);
}
virtual void visit_field(const glsl_type *type, const char *name,
bool row_major, const glsl_type *record_type,
bool /* last_field */)
{
assert(!type->without_array()->is_record());
assert(!type->without_array()->is_interface());
unsigned id;
bool found = this->map->get(id, name);
assert(found);
if (!found)
return;
const glsl_type *base_type;
if (type->is_array()) {
this->uniforms[id].array_elements = type->length;
base_type = type->fields.array;
} else {
this->uniforms[id].array_elements = 0;
base_type = type;
}
/* This assigns uniform indices to sampler and image uniforms. */
handle_samplers(base_type, &this->uniforms[id]);
handle_images(base_type, &this->uniforms[id]);
handle_subroutines(base_type, &this->uniforms[id]);
/* If there is already storage associated with this uniform or if the
* uniform is set as builtin, it means that it was set while processing
* an earlier shader stage. For example, we may be processing the
* uniform in the fragment shader, but the uniform was already processed
* in the vertex shader.
*/
if (this->uniforms[id].storage != NULL || this->uniforms[id].builtin) {
return;
}
/* Assign explicit locations. */
if (current_var->data.explicit_location) {
/* Set sequential locations for struct fields. */
if (record_type != NULL) {
const unsigned entries = MAX2(1, this->uniforms[id].array_elements);
this->uniforms[id].remap_location =
current_var->data.location + field_counter;
field_counter += entries;
} else {
this->uniforms[id].remap_location = current_var->data.location;
}
} else {
/* Initialize to to indicate that no location is set */
this->uniforms[id].remap_location = UNMAPPED_UNIFORM_LOC;
}
this->uniforms[id].name = ralloc_strdup(this->uniforms, name);
this->uniforms[id].type = base_type;
this->uniforms[id].initialized = 0;
this->uniforms[id].num_driver_storage = 0;
this->uniforms[id].driver_storage = NULL;
this->uniforms[id].atomic_buffer_index = -1;
this->uniforms[id].hidden =
current_var->data.how_declared == ir_var_hidden;
this->uniforms[id].builtin = is_gl_identifier(name);
/* Do not assign storage if the uniform is builtin */
if (!this->uniforms[id].builtin)
this->uniforms[id].storage = this->values;
if (this->ubo_block_index != -1) {
this->uniforms[id].block_index = this->ubo_block_index;
const unsigned alignment = type->std140_base_alignment(row_major);
this->ubo_byte_offset = glsl_align(this->ubo_byte_offset, alignment);
this->uniforms[id].offset = this->ubo_byte_offset;
this->ubo_byte_offset += type->std140_size(row_major);
if (type->is_array()) {
this->uniforms[id].array_stride =
glsl_align(type->fields.array->std140_size(row_major), 16);
} else {
this->uniforms[id].array_stride = 0;
}
if (type->without_array()->is_matrix()) {
const glsl_type *matrix = type->without_array();
const unsigned N = matrix->base_type == GLSL_TYPE_DOUBLE ? 8 : 4;
const unsigned items = row_major ? matrix->matrix_columns : matrix->vector_elements;
assert(items <= 4);
this->uniforms[id].matrix_stride = glsl_align(items * N, 16);
this->uniforms[id].row_major = row_major;
} else {
this->uniforms[id].matrix_stride = 0;
this->uniforms[id].row_major = false;
}
} else {
this->uniforms[id].block_index = -1;
this->uniforms[id].offset = -1;
this->uniforms[id].array_stride = -1;
this->uniforms[id].matrix_stride = -1;
this->uniforms[id].row_major = false;
}
this->values += values_for_type(type);
}
/**
* Takes a deref and recursively calls itself to break the deref down to the
* point that the reads or writes generated are contiguous scalars or vectors.
*/
void
lower_buffer_access::emit_access(void *mem_ctx,
bool is_write,
ir_dereference *deref,
ir_variable *base_offset,
unsigned int deref_offset,
bool row_major,
int matrix_columns,
unsigned int packing,
unsigned int write_mask)
{
if (deref->type->is_record()) {
unsigned int field_offset = 0;
for (unsigned i = 0; i < deref->type->length; i++) {
const struct glsl_struct_field *field =
&deref->type->fields.structure[i];
ir_dereference *field_deref =
new(mem_ctx) ir_dereference_record(deref->clone(mem_ctx, NULL),
field->name);
field_offset =
glsl_align(field_offset,
field->type->std140_base_alignment(row_major));
emit_access(mem_ctx, is_write, field_deref, base_offset,
deref_offset + field_offset,
row_major, 1, packing,
writemask_for_size(field_deref->type->vector_elements));
field_offset += field->type->std140_size(row_major);
}
return;
}
if (deref->type->is_array()) {
unsigned array_stride = packing == GLSL_INTERFACE_PACKING_STD430 ?
deref->type->fields.array->std430_array_stride(row_major) :
glsl_align(deref->type->fields.array->std140_size(row_major), 16);
for (unsigned i = 0; i < deref->type->length; i++) {
ir_constant *element = new(mem_ctx) ir_constant(i);
ir_dereference *element_deref =
new(mem_ctx) ir_dereference_array(deref->clone(mem_ctx, NULL),
element);
emit_access(mem_ctx, is_write, element_deref, base_offset,
deref_offset + i * array_stride,
row_major, 1, packing,
writemask_for_size(element_deref->type->vector_elements));
}
return;
}
if (deref->type->is_matrix()) {
for (unsigned i = 0; i < deref->type->matrix_columns; i++) {
ir_constant *col = new(mem_ctx) ir_constant(i);
ir_dereference *col_deref =
new(mem_ctx) ir_dereference_array(deref->clone(mem_ctx, NULL), col);
if (row_major) {
/* For a row-major matrix, the next column starts at the next
* element.
*/
int size_mul = deref->type->is_double() ? 8 : 4;
emit_access(mem_ctx, is_write, col_deref, base_offset,
deref_offset + i * size_mul,
row_major, deref->type->matrix_columns, packing,
writemask_for_size(col_deref->type->vector_elements));
} else {
int size_mul;
/* std430 doesn't round up vec2 size to a vec4 size */
if (packing == GLSL_INTERFACE_PACKING_STD430 &&
deref->type->vector_elements == 2 &&
!deref->type->is_double()) {
size_mul = 8;
} else {
/* std140 always rounds the stride of arrays (and matrices) to a
* vec4, so matrices are always 16 between columns/rows. With
* doubles, they will be 32 apart when there are more than 2 rows.
*
* For both std140 and std430, if the member is a
* three-'component vector with components consuming N basic
* machine units, the base alignment is 4N. For vec4, base
* alignment is 4N.
*/
size_mul = (deref->type->is_double() &&
deref->type->vector_elements > 2) ? 32 : 16;
}
emit_access(mem_ctx, is_write, col_deref, base_offset,
deref_offset + i * size_mul,
row_major, deref->type->matrix_columns, packing,
writemask_for_size(col_deref->type->vector_elements));
}
}
return;
}
assert(deref->type->is_scalar() || deref->type->is_vector());
if (!row_major) {
ir_rvalue *offset =
add(base_offset, new(mem_ctx) ir_constant(deref_offset));
unsigned mask =
is_write ? write_mask : (1 << deref->type->vector_elements) - 1;
insert_buffer_access(mem_ctx, deref, deref->type, offset, mask, -1);
} else {
unsigned N = deref->type->is_double() ? 8 : 4;
/* We're dereffing a column out of a row-major matrix, so we
* gather the vector from each stored row.
*/
assert(deref->type->base_type == GLSL_TYPE_FLOAT ||
deref->type->base_type == GLSL_TYPE_DOUBLE);
/* Matrices, row_major or not, are stored as if they were
* arrays of vectors of the appropriate size in std140.
* Arrays have their strides rounded up to a vec4, so the
* matrix stride is always 16. However a double matrix may either be 16
* or 32 depending on the number of columns.
*/
assert(matrix_columns <= 4);
unsigned matrix_stride = 0;
/* Matrix stride for std430 mat2xY matrices are not rounded up to
* vec4 size. From OpenGL 4.3 spec, section 7.6.2.2 "Standard Uniform
* Block Layout":
*
* "2. If the member is a two- or four-component vector with components
* consuming N basic machine units, the base alignment is 2N or 4N,
* respectively." [...]
* "4. If the member is an array of scalars or vectors, the base alignment
* and array stride are set to match the base alignment of a single array
* element, according to rules (1), (2), and (3), and rounded up to the
* base alignment of a vec4." [...]
* "7. If the member is a row-major matrix with C columns and R rows, the
* matrix is stored identically to an array of R row vectors with C
* components each, according to rule (4)." [...]
* "When using the std430 storage layout, shader storage blocks will be
* laid out in buffer storage identically to uniform and shader storage
* blocks using the std140 layout, except that the base alignment and
* stride of arrays of scalars and vectors in rule 4 and of structures in
* rule 9 are not rounded up a multiple of the base alignment of a vec4."
*/
if (packing == GLSL_INTERFACE_PACKING_STD430 && matrix_columns == 2)
matrix_stride = 2 * N;
else
matrix_stride = glsl_align(matrix_columns * N, 16);
const glsl_type *deref_type = deref->type->base_type == GLSL_TYPE_FLOAT ?
glsl_type::float_type : glsl_type::double_type;
for (unsigned i = 0; i < deref->type->vector_elements; i++) {
ir_rvalue *chan_offset =
add(base_offset,
new(mem_ctx) ir_constant(deref_offset + i * matrix_stride));
if (!is_write || ((1U << i) & write_mask))
insert_buffer_access(mem_ctx, deref, deref_type, chan_offset,
(1U << i), i);
}
}
}
/**
* This function initializes various values that will be used later by
* emit_access when actually emitting loads or stores.
*
* Note: const_offset is an input as well as an output, clients must
* initialize it to the offset of the variable in the underlying block, and
* this function will adjust it by adding the constant offset of the member
* being accessed into that variable.
*/
void
lower_buffer_access::setup_buffer_access(void *mem_ctx,
ir_variable *var,
ir_rvalue *deref,
ir_rvalue **offset,
unsigned *const_offset,
bool *row_major,
int *matrix_columns,
const glsl_struct_field **struct_field,
unsigned packing)
{
*offset = new(mem_ctx) ir_constant(0u);
*row_major = is_dereferenced_thing_row_major(deref);
*matrix_columns = 1;
/* Calculate the offset to the start of the region of the UBO
* dereferenced by *rvalue. This may be a variable offset if an
* array dereference has a variable index.
*/
while (deref) {
switch (deref->ir_type) {
case ir_type_dereference_variable: {
deref = NULL;
break;
}
case ir_type_dereference_array: {
ir_dereference_array *deref_array = (ir_dereference_array *) deref;
unsigned array_stride;
if (deref_array->array->type->is_vector()) {
/* We get this when storing or loading a component out of a vector
* with a non-constant index. This happens for v[i] = f where v is
* a vector (or m[i][j] = f where m is a matrix). If we don't
* lower that here, it gets turned into v = vector_insert(v, i,
* f), which loads the entire vector, modifies one component and
* then write the entire thing back. That breaks if another
* thread or SIMD channel is modifying the same vector.
*/
array_stride = 4;
if (deref_array->array->type->is_double())
array_stride *= 2;
} else if (deref_array->array->type->is_matrix() && *row_major) {
/* When loading a vector out of a row major matrix, the
* step between the columns (vectors) is the size of a
* float, while the step between the rows (elements of a
* vector) is handled below in emit_ubo_loads.
*/
array_stride = 4;
if (deref_array->array->type->is_double())
array_stride *= 2;
*matrix_columns = deref_array->array->type->matrix_columns;
} else if (deref_array->type->without_array()->is_interface()) {
/* We're processing an array dereference of an interface instance
* array. The thing being dereferenced *must* be a variable
* dereference because interfaces cannot be embedded in other
* types. In terms of calculating the offsets for the lowering
* pass, we don't care about the array index. All elements of an
* interface instance array will have the same offsets relative to
* the base of the block that backs them.
*/
deref = deref_array->array->as_dereference();
break;
} else {
/* Whether or not the field is row-major (because it might be a
* bvec2 or something) does not affect the array itself. We need
* to know whether an array element in its entirety is row-major.
*/
const bool array_row_major =
is_dereferenced_thing_row_major(deref_array);
/* The array type will give the correct interface packing
* information
*/
if (packing == GLSL_INTERFACE_PACKING_STD430) {
array_stride = deref_array->type->std430_array_stride(array_row_major);
} else {
array_stride = deref_array->type->std140_size(array_row_major);
array_stride = glsl_align(array_stride, 16);
}
}
ir_rvalue *array_index = deref_array->array_index;
if (array_index->type->base_type == GLSL_TYPE_INT)
array_index = i2u(array_index);
ir_constant *const_index =
array_index->constant_expression_value(NULL);
if (const_index) {
*const_offset += array_stride * const_index->value.u[0];
} else {
*offset = add(*offset,
mul(array_index,
new(mem_ctx) ir_constant(array_stride)));
}
deref = deref_array->array->as_dereference();
break;
}
case ir_type_dereference_record: {
ir_dereference_record *deref_record = (ir_dereference_record *) deref;
const glsl_type *struct_type = deref_record->record->type;
unsigned intra_struct_offset = 0;
for (unsigned int i = 0; i < struct_type->length; i++) {
const glsl_type *type = struct_type->fields.structure[i].type;
ir_dereference_record *field_deref = new(mem_ctx)
ir_dereference_record(deref_record->record,
struct_type->fields.structure[i].name);
const bool field_row_major =
is_dereferenced_thing_row_major(field_deref);
ralloc_free(field_deref);
unsigned field_align = 0;
if (packing == GLSL_INTERFACE_PACKING_STD430)
field_align = type->std430_base_alignment(field_row_major);
else
field_align = type->std140_base_alignment(field_row_major);
intra_struct_offset = glsl_align(intra_struct_offset, field_align);
if (strcmp(struct_type->fields.structure[i].name,
deref_record->field) == 0) {
if (struct_field)
*struct_field = &struct_type->fields.structure[i];
break;
}
if (packing == GLSL_INTERFACE_PACKING_STD430)
intra_struct_offset += type->std430_size(field_row_major);
else
intra_struct_offset += type->std140_size(field_row_major);
/* If the field just examined was itself a structure, apply rule
* #9:
*
* "The structure may have padding at the end; the base offset
* of the member following the sub-structure is rounded up to
* the next multiple of the base alignment of the structure."
*/
if (type->without_array()->is_record()) {
intra_struct_offset = glsl_align(intra_struct_offset,
field_align);
}
}
*const_offset += intra_struct_offset;
deref = deref_record->record->as_dereference();
break;
}
case ir_type_swizzle: {
ir_swizzle *deref_swizzle = (ir_swizzle *) deref;
assert(deref_swizzle->mask.num_components == 1);
*const_offset += deref_swizzle->mask.x * sizeof(int);
deref = deref_swizzle->val->as_dereference();
break;
}
default:
assert(!"not reached");
deref = NULL;
break;
}
}
}
virtual void visit_field(const glsl_struct_field *field)
{
this->offset = glsl_align(this->offset,
field->type->std140_base_alignment(false));
}
