2 * Copyright © 2010 Intel Corporation
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the "Software"),
6 * to deal in the Software without restriction, including without limitation
7 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8 * and/or sell copies of the Software, and to permit persons to whom the
9 * Software is furnished to do so, subject to the following conditions:
11 * The above copyright notice and this permission notice (including the next
12 * paragraph) shall be included in all copies or substantial portions of the
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
18 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
20 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
21 * DEALINGS IN THE SOFTWARE.
25 #include "util/half_float.h"
26 #include "compiler/glsl_types.h"
27 #include "glsl_parser_extras.h"
30 ir_rvalue::ir_rvalue(enum ir_node_type t
)
33 this->type
= glsl_type::error_type
;
36 bool ir_rvalue::is_zero() const
41 bool ir_rvalue::is_one() const
46 bool ir_rvalue::is_negative_one() const
52 * Modify the swizzle make to move one component to another
54 * \param m IR swizzle to be modified
55 * \param from Component in the RHS that is to be swizzled
56 * \param to Desired swizzle location of \c from
59 update_rhs_swizzle(ir_swizzle_mask
&m
, unsigned from
, unsigned to
)
62 case 0: m
.x
= from
; break;
63 case 1: m
.y
= from
; break;
64 case 2: m
.z
= from
; break;
65 case 3: m
.w
= from
; break;
66 default: assert(!"Should not get here.");
71 ir_assignment::set_lhs(ir_rvalue
*lhs
)
74 bool swizzled
= false;
77 ir_swizzle
*swiz
= lhs
->as_swizzle();
82 unsigned write_mask
= 0;
83 ir_swizzle_mask rhs_swiz
= { 0, 0, 0, 0, 0, 0 };
85 for (unsigned i
= 0; i
< swiz
->mask
.num_components
; i
++) {
89 case 0: c
= swiz
->mask
.x
; break;
90 case 1: c
= swiz
->mask
.y
; break;
91 case 2: c
= swiz
->mask
.z
; break;
92 case 3: c
= swiz
->mask
.w
; break;
93 default: assert(!"Should not get here.");
96 write_mask
|= (((this->write_mask
>> i
) & 1) << c
);
97 update_rhs_swizzle(rhs_swiz
, i
, c
);
98 rhs_swiz
.num_components
= swiz
->val
->type
->vector_elements
;
101 this->write_mask
= write_mask
;
104 this->rhs
= new(mem_ctx
) ir_swizzle(this->rhs
, rhs_swiz
);
109 /* Now, RHS channels line up with the LHS writemask. Collapse it
110 * to just the channels that will be written.
112 ir_swizzle_mask rhs_swiz
= { 0, 0, 0, 0, 0, 0 };
114 for (int i
= 0; i
< 4; i
++) {
115 if (write_mask
& (1 << i
))
116 update_rhs_swizzle(rhs_swiz
, i
, rhs_chan
++);
118 rhs_swiz
.num_components
= rhs_chan
;
119 this->rhs
= new(mem_ctx
) ir_swizzle(this->rhs
, rhs_swiz
);
122 assert((lhs
== NULL
) || lhs
->as_dereference());
124 this->lhs
= (ir_dereference
*) lhs
;
128 ir_assignment::whole_variable_written()
130 ir_variable
*v
= this->lhs
->whole_variable_referenced();
135 if (v
->type
->is_scalar())
138 if (v
->type
->is_vector()) {
139 const unsigned mask
= (1U << v
->type
->vector_elements
) - 1;
141 if (mask
!= this->write_mask
)
145 /* Either all the vector components are assigned or the variable is some
146 * composite type (and the whole thing is assigned.
151 ir_assignment::ir_assignment(ir_dereference
*lhs
, ir_rvalue
*rhs
,
152 ir_rvalue
*condition
, unsigned write_mask
)
153 : ir_instruction(ir_type_assignment
)
155 this->condition
= condition
;
158 this->write_mask
= write_mask
;
160 if (lhs
->type
->is_scalar() || lhs
->type
->is_vector()) {
161 int lhs_components
= 0;
162 for (int i
= 0; i
< 4; i
++) {
163 if (write_mask
& (1 << i
))
167 assert(lhs_components
== this->rhs
->type
->vector_elements
);
171 ir_assignment::ir_assignment(ir_rvalue
*lhs
, ir_rvalue
*rhs
,
172 ir_rvalue
*condition
)
173 : ir_instruction(ir_type_assignment
)
175 this->condition
= condition
;
178 /* If the RHS is a vector type, assume that all components of the vector
179 * type are being written to the LHS. The write mask comes from the RHS
180 * because we can have a case where the LHS is a vec4 and the RHS is a
181 * vec3. In that case, the assignment is:
183 * (assign (...) (xyz) (var_ref lhs) (var_ref rhs))
185 if (rhs
->type
->is_vector())
186 this->write_mask
= (1U << rhs
->type
->vector_elements
) - 1;
187 else if (rhs
->type
->is_scalar())
188 this->write_mask
= 1;
190 this->write_mask
= 0;
195 ir_expression::ir_expression(int op
, const struct glsl_type
*type
,
196 ir_rvalue
*op0
, ir_rvalue
*op1
,
197 ir_rvalue
*op2
, ir_rvalue
*op3
)
198 : ir_rvalue(ir_type_expression
)
201 this->operation
= ir_expression_operation(op
);
202 this->operands
[0] = op0
;
203 this->operands
[1] = op1
;
204 this->operands
[2] = op2
;
205 this->operands
[3] = op3
;
209 for (unsigned i
= num_operands
; i
< 4; i
++) {
210 assert(this->operands
[i
] == NULL
);
213 for (unsigned i
= 0; i
< num_operands
; i
++) {
214 assert(this->operands
[i
] != NULL
);
219 ir_expression::ir_expression(int op
, ir_rvalue
*op0
)
220 : ir_rvalue(ir_type_expression
)
222 this->operation
= ir_expression_operation(op
);
223 this->operands
[0] = op0
;
224 this->operands
[1] = NULL
;
225 this->operands
[2] = NULL
;
226 this->operands
[3] = NULL
;
228 assert(op
<= ir_last_unop
);
230 assert(num_operands
== 1);
231 assert(this->operands
[0]);
233 switch (this->operation
) {
234 case ir_unop_bit_not
:
235 case ir_unop_logic_not
:
250 case ir_unop_round_even
:
254 case ir_unop_dFdx_coarse
:
255 case ir_unop_dFdx_fine
:
257 case ir_unop_dFdy_coarse
:
258 case ir_unop_dFdy_fine
:
259 case ir_unop_bitfield_reverse
:
260 case ir_unop_interpolate_at_centroid
:
262 case ir_unop_saturate
:
264 this->type
= op0
->type
;
271 case ir_unop_bitcast_f2i
:
272 case ir_unop_bit_count
:
273 case ir_unop_find_msb
:
274 case ir_unop_find_lsb
:
275 case ir_unop_subroutine_to_int
:
278 this->type
= glsl_type::get_instance(GLSL_TYPE_INT
,
279 op0
->type
->vector_elements
, 1);
287 case ir_unop_bitcast_i2f
:
288 case ir_unop_bitcast_u2f
:
291 this->type
= glsl_type::get_instance(GLSL_TYPE_FLOAT
,
292 op0
->type
->vector_elements
, 1);
298 this->type
= glsl_type::get_instance(GLSL_TYPE_FLOAT16
,
299 op0
->type
->vector_elements
, 1);
307 this->type
= glsl_type::get_instance(GLSL_TYPE_BOOL
,
308 op0
->type
->vector_elements
, 1);
316 this->type
= glsl_type::get_instance(GLSL_TYPE_DOUBLE
,
317 op0
->type
->vector_elements
, 1);
323 case ir_unop_bitcast_f2u
:
326 this->type
= glsl_type::get_instance(GLSL_TYPE_UINT
,
327 op0
->type
->vector_elements
, 1);
335 case ir_unop_u642i64
:
336 this->type
= glsl_type::get_instance(GLSL_TYPE_INT64
,
337 op0
->type
->vector_elements
, 1);
344 case ir_unop_i642u64
:
345 this->type
= glsl_type::get_instance(GLSL_TYPE_UINT64
,
346 op0
->type
->vector_elements
, 1);
349 case ir_unop_unpack_double_2x32
:
350 case ir_unop_unpack_uint_2x32
:
351 this->type
= glsl_type::uvec2_type
;
354 case ir_unop_unpack_int_2x32
:
355 this->type
= glsl_type::ivec2_type
;
358 case ir_unop_pack_snorm_2x16
:
359 case ir_unop_pack_snorm_4x8
:
360 case ir_unop_pack_unorm_2x16
:
361 case ir_unop_pack_unorm_4x8
:
362 case ir_unop_pack_half_2x16
:
363 this->type
= glsl_type::uint_type
;
366 case ir_unop_pack_double_2x32
:
367 this->type
= glsl_type::double_type
;
370 case ir_unop_pack_int_2x32
:
371 this->type
= glsl_type::int64_t_type
;
374 case ir_unop_pack_uint_2x32
:
375 this->type
= glsl_type::uint64_t_type
;
378 case ir_unop_unpack_snorm_2x16
:
379 case ir_unop_unpack_unorm_2x16
:
380 case ir_unop_unpack_half_2x16
:
381 this->type
= glsl_type::vec2_type
;
384 case ir_unop_unpack_snorm_4x8
:
385 case ir_unop_unpack_unorm_4x8
:
386 this->type
= glsl_type::vec4_type
;
389 case ir_unop_unpack_sampler_2x32
:
390 case ir_unop_unpack_image_2x32
:
391 this->type
= glsl_type::uvec2_type
;
394 case ir_unop_pack_sampler_2x32
:
395 case ir_unop_pack_image_2x32
:
396 this->type
= op0
->type
;
399 case ir_unop_frexp_sig
:
400 this->type
= op0
->type
;
402 case ir_unop_frexp_exp
:
403 this->type
= glsl_type::get_instance(GLSL_TYPE_INT
,
404 op0
->type
->vector_elements
, 1);
407 case ir_unop_get_buffer_size
:
408 case ir_unop_ssbo_unsized_array_length
:
409 this->type
= glsl_type::int_type
;
412 case ir_unop_bitcast_i642d
:
413 case ir_unop_bitcast_u642d
:
414 this->type
= glsl_type::get_instance(GLSL_TYPE_DOUBLE
,
415 op0
->type
->vector_elements
, 1);
418 case ir_unop_bitcast_d2i64
:
419 this->type
= glsl_type::get_instance(GLSL_TYPE_INT64
,
420 op0
->type
->vector_elements
, 1);
422 case ir_unop_bitcast_d2u64
:
423 this->type
= glsl_type::get_instance(GLSL_TYPE_UINT64
,
424 op0
->type
->vector_elements
, 1);
428 assert(!"not reached: missing automatic type setup for ir_expression");
429 this->type
= op0
->type
;
434 ir_expression::ir_expression(int op
, ir_rvalue
*op0
, ir_rvalue
*op1
)
435 : ir_rvalue(ir_type_expression
)
437 this->operation
= ir_expression_operation(op
);
438 this->operands
[0] = op0
;
439 this->operands
[1] = op1
;
440 this->operands
[2] = NULL
;
441 this->operands
[3] = NULL
;
443 assert(op
> ir_last_unop
);
445 assert(num_operands
== 2);
446 for (unsigned i
= 0; i
< num_operands
; i
++) {
447 assert(this->operands
[i
] != NULL
);
450 switch (this->operation
) {
451 case ir_binop_all_equal
:
452 case ir_binop_any_nequal
:
453 this->type
= glsl_type::bool_type
;
465 if (op0
->type
->is_scalar()) {
466 this->type
= op1
->type
;
467 } else if (op1
->type
->is_scalar()) {
468 this->type
= op0
->type
;
470 if (this->operation
== ir_binop_mul
) {
471 this->type
= glsl_type::get_mul_type(op0
->type
, op1
->type
);
473 assert(op0
->type
== op1
->type
);
474 this->type
= op0
->type
;
479 case ir_binop_logic_and
:
480 case ir_binop_logic_xor
:
481 case ir_binop_logic_or
:
482 case ir_binop_bit_and
:
483 case ir_binop_bit_xor
:
484 case ir_binop_bit_or
:
485 assert(!op0
->type
->is_matrix());
486 assert(!op1
->type
->is_matrix());
487 if (op0
->type
->is_scalar()) {
488 this->type
= op1
->type
;
489 } else if (op1
->type
->is_scalar()) {
490 this->type
= op0
->type
;
492 assert(op0
->type
->vector_elements
== op1
->type
->vector_elements
);
493 this->type
= op0
->type
;
498 case ir_binop_nequal
:
499 case ir_binop_gequal
:
501 assert(op0
->type
== op1
->type
);
502 this->type
= glsl_type::get_instance(GLSL_TYPE_BOOL
,
503 op0
->type
->vector_elements
, 1);
507 this->type
= op0
->type
->get_base_type();
510 case ir_binop_imul_high
:
511 case ir_binop_mul_32x16
:
513 case ir_binop_borrow
:
514 case ir_binop_lshift
:
515 case ir_binop_rshift
:
517 case ir_binop_interpolate_at_offset
:
518 case ir_binop_interpolate_at_sample
:
519 this->type
= op0
->type
;
522 case ir_binop_add_sat
:
523 case ir_binop_sub_sat
:
525 case ir_binop_avg_round
:
526 assert(op0
->type
== op1
->type
);
527 this->type
= op0
->type
;
530 case ir_binop_abs_sub
: {
531 enum glsl_base_type base
;
533 assert(op0
->type
== op1
->type
);
535 switch (op0
->type
->base_type
) {
538 base
= GLSL_TYPE_UINT
;
540 case GLSL_TYPE_UINT8
:
542 base
= GLSL_TYPE_UINT8
;
544 case GLSL_TYPE_UINT16
:
545 case GLSL_TYPE_INT16
:
546 base
= GLSL_TYPE_UINT16
;
548 case GLSL_TYPE_UINT64
:
549 case GLSL_TYPE_INT64
:
550 base
= GLSL_TYPE_UINT64
;
553 unreachable(!"Invalid base type.");
556 this->type
= glsl_type::get_instance(base
, op0
->type
->vector_elements
, 1);
560 case ir_binop_vector_extract
:
561 this->type
= op0
->type
->get_scalar_type();
565 assert(!"not reached: missing automatic type setup for ir_expression");
566 this->type
= glsl_type::float_type
;
570 ir_expression::ir_expression(int op
, ir_rvalue
*op0
, ir_rvalue
*op1
,
572 : ir_rvalue(ir_type_expression
)
574 this->operation
= ir_expression_operation(op
);
575 this->operands
[0] = op0
;
576 this->operands
[1] = op1
;
577 this->operands
[2] = op2
;
578 this->operands
[3] = NULL
;
580 assert(op
> ir_last_binop
&& op
<= ir_last_triop
);
582 assert(num_operands
== 3);
583 for (unsigned i
= 0; i
< num_operands
; i
++) {
584 assert(this->operands
[i
] != NULL
);
587 switch (this->operation
) {
590 case ir_triop_bitfield_extract
:
591 case ir_triop_vector_insert
:
592 this->type
= op0
->type
;
596 this->type
= op1
->type
;
600 assert(!"not reached: missing automatic type setup for ir_expression");
601 this->type
= glsl_type::float_type
;
606 * This is only here for ir_reader to used for testing purposes. Please use
607 * the precomputed num_operands field if you need the number of operands.
610 ir_expression::get_num_operands(ir_expression_operation op
)
612 assert(op
<= ir_last_opcode
);
614 if (op
<= ir_last_unop
)
617 if (op
<= ir_last_binop
)
620 if (op
<= ir_last_triop
)
623 if (op
<= ir_last_quadop
)
626 unreachable("Could not calculate number of operands");
629 #include "ir_expression_operation_strings.h"
632 depth_layout_string(ir_depth_layout layout
)
635 case ir_depth_layout_none
: return "";
636 case ir_depth_layout_any
: return "depth_any";
637 case ir_depth_layout_greater
: return "depth_greater";
638 case ir_depth_layout_less
: return "depth_less";
639 case ir_depth_layout_unchanged
: return "depth_unchanged";
647 ir_expression_operation
648 ir_expression::get_operator(const char *str
)
650 for (int op
= 0; op
<= int(ir_last_opcode
); op
++) {
651 if (strcmp(str
, ir_expression_operation_strings
[op
]) == 0)
652 return (ir_expression_operation
) op
;
654 return (ir_expression_operation
) -1;
658 ir_expression::variable_referenced() const
661 case ir_binop_vector_extract
:
662 case ir_triop_vector_insert
:
663 /* We get these for things like a[0] where a is a vector type. In these
664 * cases we want variable_referenced() to return the actual vector
665 * variable this is wrapping.
667 return operands
[0]->variable_referenced();
669 return ir_rvalue::variable_referenced();
673 ir_constant::ir_constant()
674 : ir_rvalue(ir_type_constant
)
676 this->const_elements
= NULL
;
679 ir_constant::ir_constant(const struct glsl_type
*type
,
680 const ir_constant_data
*data
)
681 : ir_rvalue(ir_type_constant
)
683 this->const_elements
= NULL
;
685 assert((type
->base_type
>= GLSL_TYPE_UINT
)
686 && (type
->base_type
<= GLSL_TYPE_IMAGE
));
689 memcpy(& this->value
, data
, sizeof(this->value
));
692 ir_constant::ir_constant(float16_t f16
, unsigned vector_elements
)
693 : ir_rvalue(ir_type_constant
)
695 assert(vector_elements
<= 4);
696 this->type
= glsl_type::get_instance(GLSL_TYPE_FLOAT16
, vector_elements
, 1);
697 for (unsigned i
= 0; i
< vector_elements
; i
++) {
698 this->value
.f16
[i
] = f16
.bits
;
700 for (unsigned i
= vector_elements
; i
< 16; i
++) {
701 this->value
.f
[i
] = 0;
705 ir_constant::ir_constant(float f
, unsigned vector_elements
)
706 : ir_rvalue(ir_type_constant
)
708 assert(vector_elements
<= 4);
709 this->type
= glsl_type::get_instance(GLSL_TYPE_FLOAT
, vector_elements
, 1);
710 for (unsigned i
= 0; i
< vector_elements
; i
++) {
711 this->value
.f
[i
] = f
;
713 for (unsigned i
= vector_elements
; i
< 16; i
++) {
714 this->value
.f
[i
] = 0;
718 ir_constant::ir_constant(double d
, unsigned vector_elements
)
719 : ir_rvalue(ir_type_constant
)
721 assert(vector_elements
<= 4);
722 this->type
= glsl_type::get_instance(GLSL_TYPE_DOUBLE
, vector_elements
, 1);
723 for (unsigned i
= 0; i
< vector_elements
; i
++) {
724 this->value
.d
[i
] = d
;
726 for (unsigned i
= vector_elements
; i
< 16; i
++) {
727 this->value
.d
[i
] = 0.0;
731 ir_constant::ir_constant(unsigned int u
, unsigned vector_elements
)
732 : ir_rvalue(ir_type_constant
)
734 assert(vector_elements
<= 4);
735 this->type
= glsl_type::get_instance(GLSL_TYPE_UINT
, vector_elements
, 1);
736 for (unsigned i
= 0; i
< vector_elements
; i
++) {
737 this->value
.u
[i
] = u
;
739 for (unsigned i
= vector_elements
; i
< 16; i
++) {
740 this->value
.u
[i
] = 0;
744 ir_constant::ir_constant(int integer
, unsigned vector_elements
)
745 : ir_rvalue(ir_type_constant
)
747 assert(vector_elements
<= 4);
748 this->type
= glsl_type::get_instance(GLSL_TYPE_INT
, vector_elements
, 1);
749 for (unsigned i
= 0; i
< vector_elements
; i
++) {
750 this->value
.i
[i
] = integer
;
752 for (unsigned i
= vector_elements
; i
< 16; i
++) {
753 this->value
.i
[i
] = 0;
757 ir_constant::ir_constant(uint64_t u64
, unsigned vector_elements
)
758 : ir_rvalue(ir_type_constant
)
760 assert(vector_elements
<= 4);
761 this->type
= glsl_type::get_instance(GLSL_TYPE_UINT64
, vector_elements
, 1);
762 for (unsigned i
= 0; i
< vector_elements
; i
++) {
763 this->value
.u64
[i
] = u64
;
765 for (unsigned i
= vector_elements
; i
< 16; i
++) {
766 this->value
.u64
[i
] = 0;
770 ir_constant::ir_constant(int64_t int64
, unsigned vector_elements
)
771 : ir_rvalue(ir_type_constant
)
773 assert(vector_elements
<= 4);
774 this->type
= glsl_type::get_instance(GLSL_TYPE_INT64
, vector_elements
, 1);
775 for (unsigned i
= 0; i
< vector_elements
; i
++) {
776 this->value
.i64
[i
] = int64
;
778 for (unsigned i
= vector_elements
; i
< 16; i
++) {
779 this->value
.i64
[i
] = 0;
783 ir_constant::ir_constant(bool b
, unsigned vector_elements
)
784 : ir_rvalue(ir_type_constant
)
786 assert(vector_elements
<= 4);
787 this->type
= glsl_type::get_instance(GLSL_TYPE_BOOL
, vector_elements
, 1);
788 for (unsigned i
= 0; i
< vector_elements
; i
++) {
789 this->value
.b
[i
] = b
;
791 for (unsigned i
= vector_elements
; i
< 16; i
++) {
792 this->value
.b
[i
] = false;
796 ir_constant::ir_constant(const ir_constant
*c
, unsigned i
)
797 : ir_rvalue(ir_type_constant
)
799 this->const_elements
= NULL
;
800 this->type
= c
->type
->get_base_type();
802 switch (this->type
->base_type
) {
803 case GLSL_TYPE_UINT
: this->value
.u
[0] = c
->value
.u
[i
]; break;
804 case GLSL_TYPE_INT
: this->value
.i
[0] = c
->value
.i
[i
]; break;
805 case GLSL_TYPE_FLOAT
: this->value
.f
[0] = c
->value
.f
[i
]; break;
806 case GLSL_TYPE_FLOAT16
: this->value
.f16
[0] = c
->value
.f16
[i
]; break;
807 case GLSL_TYPE_BOOL
: this->value
.b
[0] = c
->value
.b
[i
]; break;
808 case GLSL_TYPE_DOUBLE
: this->value
.d
[0] = c
->value
.d
[i
]; break;
809 default: assert(!"Should not get here."); break;
813 ir_constant::ir_constant(const struct glsl_type
*type
, exec_list
*value_list
)
814 : ir_rvalue(ir_type_constant
)
816 this->const_elements
= NULL
;
819 assert(type
->is_scalar() || type
->is_vector() || type
->is_matrix()
820 || type
->is_struct() || type
->is_array());
822 /* If the constant is a record, the types of each of the entries in
823 * value_list must be a 1-for-1 match with the structure components. Each
824 * entry must also be a constant. Just move the nodes from the value_list
825 * to the list in the ir_constant.
827 if (type
->is_array() || type
->is_struct()) {
828 this->const_elements
= ralloc_array(this, ir_constant
*, type
->length
);
830 foreach_in_list(ir_constant
, value
, value_list
) {
831 assert(value
->as_constant() != NULL
);
833 this->const_elements
[i
++] = value
;
838 for (unsigned i
= 0; i
< 16; i
++) {
839 this->value
.u
[i
] = 0;
842 ir_constant
*value
= (ir_constant
*) (value_list
->get_head_raw());
844 /* Constructors with exactly one scalar argument are special for vectors
845 * and matrices. For vectors, the scalar value is replicated to fill all
846 * the components. For matrices, the scalar fills the components of the
847 * diagonal while the rest is filled with 0.
849 if (value
->type
->is_scalar() && value
->next
->is_tail_sentinel()) {
850 if (type
->is_matrix()) {
851 /* Matrix - fill diagonal (rest is already set to 0) */
852 for (unsigned i
= 0; i
< type
->matrix_columns
; i
++) {
853 switch (type
->base_type
) {
854 case GLSL_TYPE_FLOAT
:
855 this->value
.f
[i
* type
->vector_elements
+ i
] =
858 case GLSL_TYPE_DOUBLE
:
859 this->value
.d
[i
* type
->vector_elements
+ i
] =
862 case GLSL_TYPE_FLOAT16
:
863 this->value
.f16
[i
* type
->vector_elements
+ i
] =
867 assert(!"unexpected matrix base type");
871 /* Vector or scalar - fill all components */
872 switch (type
->base_type
) {
875 for (unsigned i
= 0; i
< type
->components(); i
++)
876 this->value
.u
[i
] = value
->value
.u
[0];
878 case GLSL_TYPE_FLOAT
:
879 for (unsigned i
= 0; i
< type
->components(); i
++)
880 this->value
.f
[i
] = value
->value
.f
[0];
882 case GLSL_TYPE_FLOAT16
:
883 for (unsigned i
= 0; i
< type
->components(); i
++)
884 this->value
.f16
[i
] = value
->value
.f16
[0];
886 case GLSL_TYPE_DOUBLE
:
887 for (unsigned i
= 0; i
< type
->components(); i
++)
888 this->value
.d
[i
] = value
->value
.d
[0];
890 case GLSL_TYPE_UINT64
:
891 case GLSL_TYPE_INT64
:
892 for (unsigned i
= 0; i
< type
->components(); i
++)
893 this->value
.u64
[i
] = value
->value
.u64
[0];
896 for (unsigned i
= 0; i
< type
->components(); i
++)
897 this->value
.b
[i
] = value
->value
.b
[0];
899 case GLSL_TYPE_SAMPLER
:
900 case GLSL_TYPE_IMAGE
:
901 this->value
.u64
[0] = value
->value
.u64
[0];
904 assert(!"Should not get here.");
911 if (type
->is_matrix() && value
->type
->is_matrix()) {
912 assert(value
->next
->is_tail_sentinel());
914 /* From section 5.4.2 of the GLSL 1.20 spec:
915 * "If a matrix is constructed from a matrix, then each component
916 * (column i, row j) in the result that has a corresponding component
917 * (column i, row j) in the argument will be initialized from there."
919 unsigned cols
= MIN2(type
->matrix_columns
, value
->type
->matrix_columns
);
920 unsigned rows
= MIN2(type
->vector_elements
, value
->type
->vector_elements
);
921 for (unsigned i
= 0; i
< cols
; i
++) {
922 for (unsigned j
= 0; j
< rows
; j
++) {
923 const unsigned src
= i
* value
->type
->vector_elements
+ j
;
924 const unsigned dst
= i
* type
->vector_elements
+ j
;
925 this->value
.f
[dst
] = value
->value
.f
[src
];
929 /* "All other components will be initialized to the identity matrix." */
930 for (unsigned i
= cols
; i
< type
->matrix_columns
; i
++)
931 this->value
.f
[i
* type
->vector_elements
+ i
] = 1.0;
936 /* Use each component from each entry in the value_list to initialize one
937 * component of the constant being constructed.
941 assert(value
->as_constant() != NULL
);
942 assert(!value
->is_tail_sentinel());
944 for (unsigned j
= 0; j
< value
->type
->components(); j
++) {
945 switch (type
->base_type
) {
947 this->value
.u
[i
] = value
->get_uint_component(j
);
950 this->value
.i
[i
] = value
->get_int_component(j
);
952 case GLSL_TYPE_FLOAT
:
953 this->value
.f
[i
] = value
->get_float_component(j
);
955 case GLSL_TYPE_FLOAT16
:
956 this->value
.f16
[i
] = value
->get_float16_component(j
);
959 this->value
.b
[i
] = value
->get_bool_component(j
);
961 case GLSL_TYPE_DOUBLE
:
962 this->value
.d
[i
] = value
->get_double_component(j
);
964 case GLSL_TYPE_UINT64
:
965 this->value
.u64
[i
] = value
->get_uint64_component(j
);
967 case GLSL_TYPE_INT64
:
968 this->value
.i64
[i
] = value
->get_int64_component(j
);
971 /* FINISHME: What to do? Exceptions are not the answer.
977 if (i
>= type
->components())
981 if (i
>= type
->components())
982 break; /* avoid downcasting a list sentinel */
983 value
= (ir_constant
*) value
->next
;
988 ir_constant::zero(void *mem_ctx
, const glsl_type
*type
)
990 assert(type
->is_scalar() || type
->is_vector() || type
->is_matrix()
991 || type
->is_struct() || type
->is_array());
993 ir_constant
*c
= new(mem_ctx
) ir_constant
;
995 memset(&c
->value
, 0, sizeof(c
->value
));
997 if (type
->is_array()) {
998 c
->const_elements
= ralloc_array(c
, ir_constant
*, type
->length
);
1000 for (unsigned i
= 0; i
< type
->length
; i
++)
1001 c
->const_elements
[i
] = ir_constant::zero(c
, type
->fields
.array
);
1004 if (type
->is_struct()) {
1005 c
->const_elements
= ralloc_array(c
, ir_constant
*, type
->length
);
1007 for (unsigned i
= 0; i
< type
->length
; i
++) {
1008 c
->const_elements
[i
] =
1009 ir_constant::zero(mem_ctx
, type
->fields
.structure
[i
].type
);
1017 ir_constant::get_bool_component(unsigned i
) const
1019 switch (this->type
->base_type
) {
1020 case GLSL_TYPE_UINT
: return this->value
.u
[i
] != 0;
1021 case GLSL_TYPE_INT
: return this->value
.i
[i
] != 0;
1022 case GLSL_TYPE_FLOAT
: return ((int)this->value
.f
[i
]) != 0;
1023 case GLSL_TYPE_FLOAT16
: return ((int)_mesa_half_to_float(this->value
.f16
[i
])) != 0;
1024 case GLSL_TYPE_BOOL
: return this->value
.b
[i
];
1025 case GLSL_TYPE_DOUBLE
: return this->value
.d
[i
] != 0.0;
1026 case GLSL_TYPE_SAMPLER
:
1027 case GLSL_TYPE_IMAGE
:
1028 case GLSL_TYPE_UINT64
: return this->value
.u64
[i
] != 0;
1029 case GLSL_TYPE_INT64
: return this->value
.i64
[i
] != 0;
1030 default: assert(!"Should not get here."); break;
1033 /* Must return something to make the compiler happy. This is clearly an
1040 ir_constant::get_float_component(unsigned i
) const
1042 switch (this->type
->base_type
) {
1043 case GLSL_TYPE_UINT
: return (float) this->value
.u
[i
];
1044 case GLSL_TYPE_INT
: return (float) this->value
.i
[i
];
1045 case GLSL_TYPE_FLOAT
: return this->value
.f
[i
];
1046 case GLSL_TYPE_FLOAT16
: return _mesa_half_to_float(this->value
.f16
[i
]);
1047 case GLSL_TYPE_BOOL
: return this->value
.b
[i
] ? 1.0f
: 0.0f
;
1048 case GLSL_TYPE_DOUBLE
: return (float) this->value
.d
[i
];
1049 case GLSL_TYPE_SAMPLER
:
1050 case GLSL_TYPE_IMAGE
:
1051 case GLSL_TYPE_UINT64
: return (float) this->value
.u64
[i
];
1052 case GLSL_TYPE_INT64
: return (float) this->value
.i64
[i
];
1053 default: assert(!"Should not get here."); break;
1056 /* Must return something to make the compiler happy. This is clearly an
1063 ir_constant::get_float16_component(unsigned i
) const
1065 if (this->type
->base_type
== GLSL_TYPE_FLOAT16
)
1066 return this->value
.f16
[i
];
1068 return _mesa_float_to_half(get_float_component(i
));
1072 ir_constant::get_double_component(unsigned i
) const
1074 switch (this->type
->base_type
) {
1075 case GLSL_TYPE_UINT
: return (double) this->value
.u
[i
];
1076 case GLSL_TYPE_INT
: return (double) this->value
.i
[i
];
1077 case GLSL_TYPE_FLOAT
: return (double) this->value
.f
[i
];
1078 case GLSL_TYPE_FLOAT16
: return (double) _mesa_half_to_float(this->value
.f16
[i
]);
1079 case GLSL_TYPE_BOOL
: return this->value
.b
[i
] ? 1.0 : 0.0;
1080 case GLSL_TYPE_DOUBLE
: return this->value
.d
[i
];
1081 case GLSL_TYPE_SAMPLER
:
1082 case GLSL_TYPE_IMAGE
:
1083 case GLSL_TYPE_UINT64
: return (double) this->value
.u64
[i
];
1084 case GLSL_TYPE_INT64
: return (double) this->value
.i64
[i
];
1085 default: assert(!"Should not get here."); break;
1088 /* Must return something to make the compiler happy. This is clearly an
1095 ir_constant::get_int_component(unsigned i
) const
1097 switch (this->type
->base_type
) {
1098 case GLSL_TYPE_UINT
: return this->value
.u
[i
];
1099 case GLSL_TYPE_INT
: return this->value
.i
[i
];
1100 case GLSL_TYPE_FLOAT
: return (int) this->value
.f
[i
];
1101 case GLSL_TYPE_FLOAT16
: return (int) _mesa_half_to_float(this->value
.f16
[i
]);
1102 case GLSL_TYPE_BOOL
: return this->value
.b
[i
] ? 1 : 0;
1103 case GLSL_TYPE_DOUBLE
: return (int) this->value
.d
[i
];
1104 case GLSL_TYPE_SAMPLER
:
1105 case GLSL_TYPE_IMAGE
:
1106 case GLSL_TYPE_UINT64
: return (int) this->value
.u64
[i
];
1107 case GLSL_TYPE_INT64
: return (int) this->value
.i64
[i
];
1108 default: assert(!"Should not get here."); break;
1111 /* Must return something to make the compiler happy. This is clearly an
1118 ir_constant::get_uint_component(unsigned i
) const
1120 switch (this->type
->base_type
) {
1121 case GLSL_TYPE_UINT
: return this->value
.u
[i
];
1122 case GLSL_TYPE_INT
: return this->value
.i
[i
];
1123 case GLSL_TYPE_FLOAT
: return (unsigned) this->value
.f
[i
];
1124 case GLSL_TYPE_FLOAT16
: return (unsigned) _mesa_half_to_float(this->value
.f16
[i
]);
1125 case GLSL_TYPE_BOOL
: return this->value
.b
[i
] ? 1 : 0;
1126 case GLSL_TYPE_DOUBLE
: return (unsigned) this->value
.d
[i
];
1127 case GLSL_TYPE_SAMPLER
:
1128 case GLSL_TYPE_IMAGE
:
1129 case GLSL_TYPE_UINT64
: return (unsigned) this->value
.u64
[i
];
1130 case GLSL_TYPE_INT64
: return (unsigned) this->value
.i64
[i
];
1131 default: assert(!"Should not get here."); break;
1134 /* Must return something to make the compiler happy. This is clearly an
1141 ir_constant::get_int64_component(unsigned i
) const
1143 switch (this->type
->base_type
) {
1144 case GLSL_TYPE_UINT
: return this->value
.u
[i
];
1145 case GLSL_TYPE_INT
: return this->value
.i
[i
];
1146 case GLSL_TYPE_FLOAT
: return (int64_t) this->value
.f
[i
];
1147 case GLSL_TYPE_FLOAT16
: return (int64_t) _mesa_half_to_float(this->value
.f16
[i
]);
1148 case GLSL_TYPE_BOOL
: return this->value
.b
[i
] ? 1 : 0;
1149 case GLSL_TYPE_DOUBLE
: return (int64_t) this->value
.d
[i
];
1150 case GLSL_TYPE_SAMPLER
:
1151 case GLSL_TYPE_IMAGE
:
1152 case GLSL_TYPE_UINT64
: return (int64_t) this->value
.u64
[i
];
1153 case GLSL_TYPE_INT64
: return this->value
.i64
[i
];
1154 default: assert(!"Should not get here."); break;
1157 /* Must return something to make the compiler happy. This is clearly an
1164 ir_constant::get_uint64_component(unsigned i
) const
1166 switch (this->type
->base_type
) {
1167 case GLSL_TYPE_UINT
: return this->value
.u
[i
];
1168 case GLSL_TYPE_INT
: return this->value
.i
[i
];
1169 case GLSL_TYPE_FLOAT
: return (uint64_t) this->value
.f
[i
];
1170 case GLSL_TYPE_FLOAT16
: return (uint64_t) _mesa_half_to_float(this->value
.f16
[i
]);
1171 case GLSL_TYPE_BOOL
: return this->value
.b
[i
] ? 1 : 0;
1172 case GLSL_TYPE_DOUBLE
: return (uint64_t) this->value
.d
[i
];
1173 case GLSL_TYPE_SAMPLER
:
1174 case GLSL_TYPE_IMAGE
:
1175 case GLSL_TYPE_UINT64
: return this->value
.u64
[i
];
1176 case GLSL_TYPE_INT64
: return (uint64_t) this->value
.i64
[i
];
1177 default: assert(!"Should not get here."); break;
1180 /* Must return something to make the compiler happy. This is clearly an
1187 ir_constant::get_array_element(unsigned i
) const
1189 assert(this->type
->is_array());
1191 /* From page 35 (page 41 of the PDF) of the GLSL 1.20 spec:
1193 * "Behavior is undefined if a shader subscripts an array with an index
1194 * less than 0 or greater than or equal to the size the array was
1197 * Most out-of-bounds accesses are removed before things could get this far.
1198 * There are cases where non-constant array index values can get constant
1203 else if (i
>= this->type
->length
)
1204 i
= this->type
->length
- 1;
1206 return const_elements
[i
];
1210 ir_constant::get_record_field(int idx
)
1212 assert(this->type
->is_struct());
1213 assert(idx
>= 0 && (unsigned) idx
< this->type
->length
);
1215 return const_elements
[idx
];
1219 ir_constant::copy_offset(ir_constant
*src
, int offset
)
1221 switch (this->type
->base_type
) {
1222 case GLSL_TYPE_UINT
:
1224 case GLSL_TYPE_FLOAT
:
1225 case GLSL_TYPE_FLOAT16
:
1226 case GLSL_TYPE_DOUBLE
:
1227 case GLSL_TYPE_SAMPLER
:
1228 case GLSL_TYPE_IMAGE
:
1229 case GLSL_TYPE_UINT64
:
1230 case GLSL_TYPE_INT64
:
1231 case GLSL_TYPE_BOOL
: {
1232 unsigned int size
= src
->type
->components();
1233 assert (size
<= this->type
->components() - offset
);
1234 for (unsigned int i
=0; i
<size
; i
++) {
1235 switch (this->type
->base_type
) {
1236 case GLSL_TYPE_UINT
:
1237 value
.u
[i
+offset
] = src
->get_uint_component(i
);
1240 value
.i
[i
+offset
] = src
->get_int_component(i
);
1242 case GLSL_TYPE_FLOAT
:
1243 value
.f
[i
+offset
] = src
->get_float_component(i
);
1245 case GLSL_TYPE_FLOAT16
:
1246 value
.f16
[i
+offset
] = src
->get_float16_component(i
);
1248 case GLSL_TYPE_BOOL
:
1249 value
.b
[i
+offset
] = src
->get_bool_component(i
);
1251 case GLSL_TYPE_DOUBLE
:
1252 value
.d
[i
+offset
] = src
->get_double_component(i
);
1254 case GLSL_TYPE_SAMPLER
:
1255 case GLSL_TYPE_IMAGE
:
1256 case GLSL_TYPE_UINT64
:
1257 value
.u64
[i
+offset
] = src
->get_uint64_component(i
);
1259 case GLSL_TYPE_INT64
:
1260 value
.i64
[i
+offset
] = src
->get_int64_component(i
);
1262 default: // Shut up the compiler
1269 case GLSL_TYPE_STRUCT
:
1270 case GLSL_TYPE_ARRAY
: {
1271 assert (src
->type
== this->type
);
1272 for (unsigned i
= 0; i
< this->type
->length
; i
++) {
1273 this->const_elements
[i
] = src
->const_elements
[i
]->clone(this, NULL
);
1279 assert(!"Should not get here.");
1285 ir_constant::copy_masked_offset(ir_constant
*src
, int offset
, unsigned int mask
)
1287 assert (!type
->is_array() && !type
->is_struct());
1289 if (!type
->is_vector() && !type
->is_matrix()) {
1295 for (int i
=0; i
<4; i
++) {
1296 if (mask
& (1 << i
)) {
1297 switch (this->type
->base_type
) {
1298 case GLSL_TYPE_UINT
:
1299 value
.u
[i
+offset
] = src
->get_uint_component(id
++);
1302 value
.i
[i
+offset
] = src
->get_int_component(id
++);
1304 case GLSL_TYPE_FLOAT
:
1305 value
.f
[i
+offset
] = src
->get_float_component(id
++);
1307 case GLSL_TYPE_FLOAT16
:
1308 value
.f16
[i
+offset
] = src
->get_float16_component(id
++);
1310 case GLSL_TYPE_BOOL
:
1311 value
.b
[i
+offset
] = src
->get_bool_component(id
++);
1313 case GLSL_TYPE_DOUBLE
:
1314 value
.d
[i
+offset
] = src
->get_double_component(id
++);
1316 case GLSL_TYPE_SAMPLER
:
1317 case GLSL_TYPE_IMAGE
:
1318 case GLSL_TYPE_UINT64
:
1319 value
.u64
[i
+offset
] = src
->get_uint64_component(id
++);
1321 case GLSL_TYPE_INT64
:
1322 value
.i64
[i
+offset
] = src
->get_int64_component(id
++);
1325 assert(!"Should not get here.");
1333 ir_constant::has_value(const ir_constant
*c
) const
1335 if (this->type
!= c
->type
)
1338 if (this->type
->is_array() || this->type
->is_struct()) {
1339 for (unsigned i
= 0; i
< this->type
->length
; i
++) {
1340 if (!this->const_elements
[i
]->has_value(c
->const_elements
[i
]))
1346 for (unsigned i
= 0; i
< this->type
->components(); i
++) {
1347 switch (this->type
->base_type
) {
1348 case GLSL_TYPE_UINT
:
1349 if (this->value
.u
[i
] != c
->value
.u
[i
])
1353 if (this->value
.i
[i
] != c
->value
.i
[i
])
1356 case GLSL_TYPE_FLOAT
:
1357 if (this->value
.f
[i
] != c
->value
.f
[i
])
1360 case GLSL_TYPE_FLOAT16
:
1361 /* Convert to float to make sure NaN and ±0.0 compares correctly */
1362 if (_mesa_half_to_float(this->value
.f16
[i
]) !=
1363 _mesa_half_to_float(c
->value
.f16
[i
]))
1366 case GLSL_TYPE_BOOL
:
1367 if (this->value
.b
[i
] != c
->value
.b
[i
])
1370 case GLSL_TYPE_DOUBLE
:
1371 if (this->value
.d
[i
] != c
->value
.d
[i
])
1374 case GLSL_TYPE_SAMPLER
:
1375 case GLSL_TYPE_IMAGE
:
1376 case GLSL_TYPE_UINT64
:
1377 if (this->value
.u64
[i
] != c
->value
.u64
[i
])
1380 case GLSL_TYPE_INT64
:
1381 if (this->value
.i64
[i
] != c
->value
.i64
[i
])
1385 assert(!"Should not get here.");
1394 ir_constant::is_value(float f
, int i
) const
1396 if (!this->type
->is_scalar() && !this->type
->is_vector())
1399 /* Only accept boolean values for 0/1. */
1400 if (int(bool(i
)) != i
&& this->type
->is_boolean())
1403 for (unsigned c
= 0; c
< this->type
->vector_elements
; c
++) {
1404 switch (this->type
->base_type
) {
1405 case GLSL_TYPE_FLOAT
:
1406 if (this->value
.f
[c
] != f
)
1409 case GLSL_TYPE_FLOAT16
:
1410 if (_mesa_half_to_float(this->value
.f16
[c
]) != f
)
1414 if (this->value
.i
[c
] != i
)
1417 case GLSL_TYPE_UINT
:
1418 if (this->value
.u
[c
] != unsigned(i
))
1421 case GLSL_TYPE_BOOL
:
1422 if (this->value
.b
[c
] != bool(i
))
1425 case GLSL_TYPE_DOUBLE
:
1426 if (this->value
.d
[c
] != double(f
))
1429 case GLSL_TYPE_SAMPLER
:
1430 case GLSL_TYPE_IMAGE
:
1431 case GLSL_TYPE_UINT64
:
1432 if (this->value
.u64
[c
] != uint64_t(i
))
1435 case GLSL_TYPE_INT64
:
1436 if (this->value
.i64
[c
] != i
)
1440 /* The only other base types are structures, arrays, and samplers.
1441 * Samplers cannot be constants, and the others should have been
1442 * filtered out above.
1444 assert(!"Should not get here.");
1453 ir_constant::is_zero() const
1455 return is_value(0.0, 0);
1459 ir_constant::is_one() const
1461 return is_value(1.0, 1);
1465 ir_constant::is_negative_one() const
1467 return is_value(-1.0, -1);
1471 ir_constant::is_uint16_constant() const
1473 if (!type
->is_integer_32())
1476 return value
.u
[0] < (1 << 16);
1480 : ir_instruction(ir_type_loop
)
1485 ir_dereference_variable::ir_dereference_variable(ir_variable
*var
)
1486 : ir_dereference(ir_type_dereference_variable
)
1488 assert(var
!= NULL
);
1491 this->type
= var
->type
;
1495 ir_dereference_array::ir_dereference_array(ir_rvalue
*value
,
1496 ir_rvalue
*array_index
)
1497 : ir_dereference(ir_type_dereference_array
)
1499 this->array_index
= array_index
;
1500 this->set_array(value
);
1504 ir_dereference_array::ir_dereference_array(ir_variable
*var
,
1505 ir_rvalue
*array_index
)
1506 : ir_dereference(ir_type_dereference_array
)
1508 void *ctx
= ralloc_parent(var
);
1510 this->array_index
= array_index
;
1511 this->set_array(new(ctx
) ir_dereference_variable(var
));
1516 ir_dereference_array::set_array(ir_rvalue
*value
)
1518 assert(value
!= NULL
);
1520 this->array
= value
;
1522 const glsl_type
*const vt
= this->array
->type
;
1524 if (vt
->is_array()) {
1525 type
= vt
->fields
.array
;
1526 } else if (vt
->is_matrix()) {
1527 type
= vt
->column_type();
1528 } else if (vt
->is_vector()) {
1529 type
= vt
->get_base_type();
1534 ir_dereference_record::ir_dereference_record(ir_rvalue
*value
,
1536 : ir_dereference(ir_type_dereference_record
)
1538 assert(value
!= NULL
);
1540 this->record
= value
;
1541 this->type
= this->record
->type
->field_type(field
);
1542 this->field_idx
= this->record
->type
->field_index(field
);
1546 ir_dereference_record::ir_dereference_record(ir_variable
*var
,
1548 : ir_dereference(ir_type_dereference_record
)
1550 void *ctx
= ralloc_parent(var
);
1552 this->record
= new(ctx
) ir_dereference_variable(var
);
1553 this->type
= this->record
->type
->field_type(field
);
1554 this->field_idx
= this->record
->type
->field_index(field
);
1558 ir_dereference::is_lvalue(const struct _mesa_glsl_parse_state
*state
) const
1560 ir_variable
*var
= this->variable_referenced();
1562 /* Every l-value derference chain eventually ends in a variable.
1564 if ((var
== NULL
) || var
->data
.read_only
)
1567 /* From section 4.1.7 of the ARB_bindless_texture spec:
1569 * "Samplers can be used as l-values, so can be assigned into and used as
1570 * "out" and "inout" function parameters."
1572 * From section 4.1.X of the ARB_bindless_texture spec:
1574 * "Images can be used as l-values, so can be assigned into and used as
1575 * "out" and "inout" function parameters."
1577 if ((!state
|| state
->has_bindless()) &&
1578 (this->type
->contains_sampler() || this->type
->contains_image()))
1581 /* From section 4.1.7 of the GLSL 4.40 spec:
1583 * "Opaque variables cannot be treated as l-values; hence cannot
1584 * be used as out or inout function parameters, nor can they be
1587 if (this->type
->contains_opaque())
1594 static const char * const tex_opcode_strs
[] = { "tex", "txb", "txl", "txd", "txf", "txf_ms", "txs", "lod", "tg4", "query_levels", "texture_samples", "samples_identical" };
1596 const char *ir_texture::opcode_string()
1598 assert((unsigned int) op
< ARRAY_SIZE(tex_opcode_strs
));
1599 return tex_opcode_strs
[op
];
1603 ir_texture::get_opcode(const char *str
)
1605 const int count
= sizeof(tex_opcode_strs
) / sizeof(tex_opcode_strs
[0]);
1606 for (int op
= 0; op
< count
; op
++) {
1607 if (strcmp(str
, tex_opcode_strs
[op
]) == 0)
1608 return (ir_texture_opcode
) op
;
1610 return (ir_texture_opcode
) -1;
1615 ir_texture::set_sampler(ir_dereference
*sampler
, const glsl_type
*type
)
1617 assert(sampler
!= NULL
);
1618 assert(type
!= NULL
);
1619 this->sampler
= sampler
;
1622 if (this->op
== ir_txs
|| this->op
== ir_query_levels
||
1623 this->op
== ir_texture_samples
) {
1624 assert(type
->base_type
== GLSL_TYPE_INT
);
1625 } else if (this->op
== ir_lod
) {
1626 assert(type
->vector_elements
== 2);
1627 assert(type
->is_float());
1628 } else if (this->op
== ir_samples_identical
) {
1629 assert(type
== glsl_type::bool_type
);
1630 assert(sampler
->type
->is_sampler());
1631 assert(sampler
->type
->sampler_dimensionality
== GLSL_SAMPLER_DIM_MS
);
1633 assert(sampler
->type
->sampled_type
== (int) type
->base_type
);
1634 if (sampler
->type
->sampler_shadow
)
1635 assert(type
->vector_elements
== 4 || type
->vector_elements
== 1);
1637 assert(type
->vector_elements
== 4);
1643 ir_swizzle::init_mask(const unsigned *comp
, unsigned count
)
1645 assert((count
>= 1) && (count
<= 4));
1647 memset(&this->mask
, 0, sizeof(this->mask
));
1648 this->mask
.num_components
= count
;
1650 unsigned dup_mask
= 0;
1653 assert(comp
[3] <= 3);
1654 dup_mask
|= (1U << comp
[3])
1655 & ((1U << comp
[0]) | (1U << comp
[1]) | (1U << comp
[2]));
1656 this->mask
.w
= comp
[3];
1659 assert(comp
[2] <= 3);
1660 dup_mask
|= (1U << comp
[2])
1661 & ((1U << comp
[0]) | (1U << comp
[1]));
1662 this->mask
.z
= comp
[2];
1665 assert(comp
[1] <= 3);
1666 dup_mask
|= (1U << comp
[1])
1667 & ((1U << comp
[0]));
1668 this->mask
.y
= comp
[1];
1671 assert(comp
[0] <= 3);
1672 this->mask
.x
= comp
[0];
1675 this->mask
.has_duplicates
= dup_mask
!= 0;
1677 /* Based on the number of elements in the swizzle and the base type
1678 * (i.e., float, int, unsigned, or bool) of the vector being swizzled,
1679 * generate the type of the resulting value.
1681 type
= glsl_type::get_instance(val
->type
->base_type
, mask
.num_components
, 1);
1684 ir_swizzle::ir_swizzle(ir_rvalue
*val
, unsigned x
, unsigned y
, unsigned z
,
1685 unsigned w
, unsigned count
)
1686 : ir_rvalue(ir_type_swizzle
), val(val
)
1688 const unsigned components
[4] = { x
, y
, z
, w
};
1689 this->init_mask(components
, count
);
1692 ir_swizzle::ir_swizzle(ir_rvalue
*val
, const unsigned *comp
,
1694 : ir_rvalue(ir_type_swizzle
), val(val
)
1696 this->init_mask(comp
, count
);
1699 ir_swizzle::ir_swizzle(ir_rvalue
*val
, ir_swizzle_mask mask
)
1700 : ir_rvalue(ir_type_swizzle
), val(val
), mask(mask
)
1702 this->type
= glsl_type::get_instance(val
->type
->base_type
,
1703 mask
.num_components
, 1);
1712 ir_swizzle::create(ir_rvalue
*val
, const char *str
, unsigned vector_length
)
1714 void *ctx
= ralloc_parent(val
);
1716 /* For each possible swizzle character, this table encodes the value in
1717 * \c idx_map that represents the 0th element of the vector. For invalid
1718 * swizzle characters (e.g., 'k'), a special value is used that will allow
1719 * detection of errors.
1721 static const unsigned char base_idx
[26] = {
1722 /* a b c d e f g h i j k l m */
1723 R
, R
, I
, I
, I
, I
, R
, I
, I
, I
, I
, I
, I
,
1724 /* n o p q r s t u v w x y z */
1725 I
, I
, S
, S
, R
, S
, S
, I
, I
, X
, X
, X
, X
1728 /* Each valid swizzle character has an entry in the previous table. This
1729 * table encodes the base index encoded in the previous table plus the actual
1730 * index of the swizzle character. When processing swizzles, the first
1731 * character in the string is indexed in the previous table. Each character
1732 * in the string is indexed in this table, and the value found there has the
1733 * value form the first table subtracted. The result must be on the range
1736 * For example, the string "wzyx" will get X from the first table. Each of
1737 * the charcaters will get X+3, X+2, X+1, and X+0 from this table. After
1738 * subtraction, the swizzle values are { 3, 2, 1, 0 }.
1740 * The string "wzrg" will get X from the first table. Each of the characters
1741 * will get X+3, X+2, R+0, and R+1 from this table. After subtraction, the
1742 * swizzle values are { 3, 2, 4, 5 }. Since 4 and 5 are outside the range
1743 * [0,3], the error is detected.
1745 static const unsigned char idx_map
[26] = {
1746 /* a b c d e f g h i j k l m */
1747 R
+3, R
+2, 0, 0, 0, 0, R
+1, 0, 0, 0, 0, 0, 0,
1748 /* n o p q r s t u v w x y z */
1749 0, 0, S
+2, S
+3, R
+0, S
+0, S
+1, 0, 0, X
+3, X
+0, X
+1, X
+2
1752 int swiz_idx
[4] = { 0, 0, 0, 0 };
1756 /* Validate the first character in the swizzle string and look up the base
1757 * index value as described above.
1759 if ((str
[0] < 'a') || (str
[0] > 'z'))
1762 const unsigned base
= base_idx
[str
[0] - 'a'];
1765 for (i
= 0; (i
< 4) && (str
[i
] != '\0'); i
++) {
1766 /* Validate the next character, and, as described above, convert it to a
1769 if ((str
[i
] < 'a') || (str
[i
] > 'z'))
1772 swiz_idx
[i
] = idx_map
[str
[i
] - 'a'] - base
;
1773 if ((swiz_idx
[i
] < 0) || (swiz_idx
[i
] >= (int) vector_length
))
1780 return new(ctx
) ir_swizzle(val
, swiz_idx
[0], swiz_idx
[1], swiz_idx
[2],
1790 ir_swizzle::variable_referenced() const
1792 return this->val
->variable_referenced();
1796 bool ir_variable::temporaries_allocate_names
= false;
1798 const char ir_variable::tmp_name
[] = "compiler_temp";
1800 ir_variable::ir_variable(const struct glsl_type
*type
, const char *name
,
1801 ir_variable_mode mode
)
1802 : ir_instruction(ir_type_variable
)
1806 if (mode
== ir_var_temporary
&& !ir_variable::temporaries_allocate_names
)
1809 /* The ir_variable clone method may call this constructor with name set to
1813 || mode
== ir_var_temporary
1814 || mode
== ir_var_function_in
1815 || mode
== ir_var_function_out
1816 || mode
== ir_var_function_inout
);
1817 assert(name
!= ir_variable::tmp_name
1818 || mode
== ir_var_temporary
);
1819 if (mode
== ir_var_temporary
1820 && (name
== NULL
|| name
== ir_variable::tmp_name
)) {
1821 this->name
= ir_variable::tmp_name
;
1822 } else if (name
== NULL
||
1823 strlen(name
) < ARRAY_SIZE(this->name_storage
)) {
1824 strcpy(this->name_storage
, name
? name
: "");
1825 this->name
= this->name_storage
;
1827 this->name
= ralloc_strdup(this, name
);
1830 this->u
.max_ifc_array_access
= NULL
;
1832 this->data
.explicit_location
= false;
1833 this->data
.explicit_index
= false;
1834 this->data
.explicit_binding
= false;
1835 this->data
.explicit_component
= false;
1836 this->data
.has_initializer
= false;
1837 this->data
.is_unmatched_generic_inout
= false;
1838 this->data
.is_xfb_only
= false;
1839 this->data
.explicit_xfb_buffer
= false;
1840 this->data
.explicit_xfb_offset
= false;
1841 this->data
.explicit_xfb_stride
= false;
1842 this->data
.location
= -1;
1843 this->data
.location_frac
= 0;
1844 this->data
.matrix_layout
= GLSL_MATRIX_LAYOUT_INHERITED
;
1845 this->data
.from_named_ifc_block
= false;
1846 this->data
.must_be_shader_input
= false;
1847 this->data
.index
= 0;
1848 this->data
.binding
= 0;
1849 this->data
.warn_extension_index
= 0;
1850 this->constant_value
= NULL
;
1851 this->constant_initializer
= NULL
;
1852 this->data
.depth_layout
= ir_depth_layout_none
;
1853 this->data
.used
= false;
1854 this->data
.assigned
= false;
1855 this->data
.always_active_io
= false;
1856 this->data
.read_only
= false;
1857 this->data
.centroid
= false;
1858 this->data
.sample
= false;
1859 this->data
.patch
= false;
1860 this->data
.explicit_invariant
= false;
1861 this->data
.invariant
= false;
1862 this->data
.precise
= false;
1863 this->data
.how_declared
= ir_var_declared_normally
;
1864 this->data
.mode
= mode
;
1865 this->data
.interpolation
= INTERP_MODE_NONE
;
1866 this->data
.max_array_access
= -1;
1867 this->data
.offset
= 0;
1868 this->data
.precision
= GLSL_PRECISION_NONE
;
1869 this->data
.memory_read_only
= false;
1870 this->data
.memory_write_only
= false;
1871 this->data
.memory_coherent
= false;
1872 this->data
.memory_volatile
= false;
1873 this->data
.memory_restrict
= false;
1874 this->data
.from_ssbo_unsized_array
= false;
1875 this->data
.implicit_sized_array
= false;
1876 this->data
.fb_fetch_output
= false;
1877 this->data
.bindless
= false;
1878 this->data
.bound
= false;
1879 this->data
.image_format
= PIPE_FORMAT_NONE
;
1880 this->data
._num_state_slots
= 0;
1881 this->data
.param_index
= 0;
1882 this->data
.stream
= 0;
1883 this->data
.xfb_buffer
= -1;
1884 this->data
.xfb_stride
= -1;
1886 this->interface_type
= NULL
;
1889 if (type
->is_interface())
1890 this->init_interface_type(type
);
1891 else if (type
->without_array()->is_interface())
1892 this->init_interface_type(type
->without_array());
1898 interpolation_string(unsigned interpolation
)
1900 switch (interpolation
) {
1901 case INTERP_MODE_NONE
: return "no";
1902 case INTERP_MODE_SMOOTH
: return "smooth";
1903 case INTERP_MODE_FLAT
: return "flat";
1904 case INTERP_MODE_NOPERSPECTIVE
: return "noperspective";
1907 assert(!"Should not get here.");
1911 const char *const ir_variable::warn_extension_table
[] = {
1913 "GL_ARB_shader_stencil_export",
1914 "GL_AMD_shader_stencil_export",
1918 ir_variable::enable_extension_warning(const char *extension
)
1920 for (unsigned i
= 0; i
< ARRAY_SIZE(warn_extension_table
); i
++) {
1921 if (strcmp(warn_extension_table
[i
], extension
) == 0) {
1922 this->data
.warn_extension_index
= i
;
1927 assert(!"Should not get here.");
1928 this->data
.warn_extension_index
= 0;
1932 ir_variable::get_extension_warning() const
1934 return this->data
.warn_extension_index
== 0
1935 ? NULL
: warn_extension_table
[this->data
.warn_extension_index
];
1938 ir_function_signature::ir_function_signature(const glsl_type
*return_type
,
1939 builtin_available_predicate b
)
1940 : ir_instruction(ir_type_function_signature
),
1941 return_type(return_type
), is_defined(false),
1942 return_precision(GLSL_PRECISION_NONE
),
1943 intrinsic_id(ir_intrinsic_invalid
), builtin_avail(b
), _function(NULL
)
1945 this->origin
= NULL
;
1950 ir_function_signature::is_builtin() const
1952 return builtin_avail
!= NULL
;
1957 ir_function_signature::is_builtin_available(const _mesa_glsl_parse_state
*state
) const
1959 /* We can't call the predicate without a state pointer, so just say that
1960 * the signature is available. At compile time, we need the filtering,
1961 * but also receive a valid state pointer. At link time, we're resolving
1962 * imported built-in prototypes to their definitions, which will always
1963 * be an exact match. So we can skip the filtering.
1968 assert(builtin_avail
!= NULL
);
1969 return builtin_avail(state
);
1974 modes_match(unsigned a
, unsigned b
)
1979 /* Accept "in" vs. "const in" */
1980 if ((a
== ir_var_const_in
&& b
== ir_var_function_in
) ||
1981 (b
== ir_var_const_in
&& a
== ir_var_function_in
))
1989 ir_function_signature::qualifiers_match(exec_list
*params
)
1991 /* check that the qualifiers match. */
1992 foreach_two_lists(a_node
, &this->parameters
, b_node
, params
) {
1993 ir_variable
*a
= (ir_variable
*) a_node
;
1994 ir_variable
*b
= (ir_variable
*) b_node
;
1996 if (a
->data
.read_only
!= b
->data
.read_only
||
1997 !modes_match(a
->data
.mode
, b
->data
.mode
) ||
1998 a
->data
.interpolation
!= b
->data
.interpolation
||
1999 a
->data
.centroid
!= b
->data
.centroid
||
2000 a
->data
.sample
!= b
->data
.sample
||
2001 a
->data
.patch
!= b
->data
.patch
||
2002 a
->data
.memory_read_only
!= b
->data
.memory_read_only
||
2003 a
->data
.memory_write_only
!= b
->data
.memory_write_only
||
2004 a
->data
.memory_coherent
!= b
->data
.memory_coherent
||
2005 a
->data
.memory_volatile
!= b
->data
.memory_volatile
||
2006 a
->data
.memory_restrict
!= b
->data
.memory_restrict
) {
2008 /* parameter a's qualifiers don't match */
2017 ir_function_signature::replace_parameters(exec_list
*new_params
)
2019 /* Destroy all of the previous parameter information. If the previous
2020 * parameter information comes from the function prototype, it may either
2021 * specify incorrect parameter names or not have names at all.
2023 new_params
->move_nodes_to(¶meters
);
2027 ir_function::ir_function(const char *name
)
2028 : ir_instruction(ir_type_function
)
2030 this->subroutine_index
= -1;
2031 this->name
= ralloc_strdup(this, name
);
2036 ir_function::has_user_signature()
2038 foreach_in_list(ir_function_signature
, sig
, &this->signatures
) {
2039 if (!sig
->is_builtin())
2047 ir_rvalue::error_value(void *mem_ctx
)
2049 ir_rvalue
*v
= new(mem_ctx
) ir_rvalue(ir_type_unset
);
2051 v
->type
= glsl_type::error_type
;
2057 visit_exec_list(exec_list
*list
, ir_visitor
*visitor
)
2059 foreach_in_list_safe(ir_instruction
, node
, list
) {
2060 node
->accept(visitor
);
2066 steal_memory(ir_instruction
*ir
, void *new_ctx
)
2068 ir_variable
*var
= ir
->as_variable();
2069 ir_function
*fn
= ir
->as_function();
2070 ir_constant
*constant
= ir
->as_constant();
2071 if (var
!= NULL
&& var
->constant_value
!= NULL
)
2072 steal_memory(var
->constant_value
, ir
);
2074 if (var
!= NULL
&& var
->constant_initializer
!= NULL
)
2075 steal_memory(var
->constant_initializer
, ir
);
2077 if (fn
!= NULL
&& fn
->subroutine_types
)
2078 ralloc_steal(new_ctx
, fn
->subroutine_types
);
2080 /* The components of aggregate constants are not visited by the normal
2081 * visitor, so steal their values by hand.
2083 if (constant
!= NULL
&&
2084 (constant
->type
->is_array() || constant
->type
->is_struct())) {
2085 for (unsigned int i
= 0; i
< constant
->type
->length
; i
++) {
2086 steal_memory(constant
->const_elements
[i
], ir
);
2090 ralloc_steal(new_ctx
, ir
);
2095 reparent_ir(exec_list
*list
, void *mem_ctx
)
2097 foreach_in_list(ir_instruction
, node
, list
) {
2098 visit_tree(node
, steal_memory
, mem_ctx
);
2104 try_min_one(ir_rvalue
*ir
)
2106 ir_expression
*expr
= ir
->as_expression();
2108 if (!expr
|| expr
->operation
!= ir_binop_min
)
2111 if (expr
->operands
[0]->is_one())
2112 return expr
->operands
[1];
2114 if (expr
->operands
[1]->is_one())
2115 return expr
->operands
[0];
2121 try_max_zero(ir_rvalue
*ir
)
2123 ir_expression
*expr
= ir
->as_expression();
2125 if (!expr
|| expr
->operation
!= ir_binop_max
)
2128 if (expr
->operands
[0]->is_zero())
2129 return expr
->operands
[1];
2131 if (expr
->operands
[1]->is_zero())
2132 return expr
->operands
[0];
2138 ir_rvalue::as_rvalue_to_saturate()
2140 ir_expression
*expr
= this->as_expression();
2145 ir_rvalue
*max_zero
= try_max_zero(expr
);
2147 return try_min_one(max_zero
);
2149 ir_rvalue
*min_one
= try_min_one(expr
);
2151 return try_max_zero(min_one
);
2160 vertices_per_prim(GLenum prim
)
2169 case GL_LINES_ADJACENCY
:
2171 case GL_TRIANGLES_ADJACENCY
:
2174 assert(!"Bad primitive");
2180 * Generate a string describing the mode of a variable
2183 mode_string(const ir_variable
*var
)
2185 switch (var
->data
.mode
) {
2187 return (var
->data
.read_only
) ? "global constant" : "global variable";
2189 case ir_var_uniform
:
2192 case ir_var_shader_storage
:
2195 case ir_var_shader_in
:
2196 return "shader input";
2198 case ir_var_shader_out
:
2199 return "shader output";
2201 case ir_var_function_in
:
2202 case ir_var_const_in
:
2203 return "function input";
2205 case ir_var_function_out
:
2206 return "function output";
2208 case ir_var_function_inout
:
2209 return "function inout";
2211 case ir_var_system_value
:
2212 return "shader input";
2214 case ir_var_temporary
:
2215 return "compiler temporary";
2217 case ir_var_mode_count
:
2221 assert(!"Should not get here.");
2222 return "invalid variable";