third_party_mesa3d/mesa_codegen.brg

/*
 * Copyright © 2010 Intel Corporation
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice (including the next
 * paragraph) shall be included in all copies or substantial portions of the
 * Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
 * IN THE SOFTWARE.
 *
 * Authors:
 *    Eric Anholt <eric@anholt.net>
 *
 */

/* DO NOT EDIT mesa_codegen.h.  It is a generated file produced
 * from mesa_codegen.brg and will be overwritten.
 */

#include <stdint.h>
#include <stdlib.h>
#include <string.h>
#include <assert.h>
#include <stdio.h>

/* Everything before the first %% is pasted at the start of the
 * mesa_codegen.h header file.
 */

#include "ir_to_mesa.h"

#define MBTREE_TYPE struct mbtree

%%
# The list of terminals is the set of things that ir_to_mesa.cpp will
# generate in its trees.
%term assign
%term reference_vec4
%term array_reference_vec4_vec4
%term exp_vec4
%term exp2_vec4
%term log_vec4
%term log2_vec4
%term sin_vec4
%term cos_vec4
%term add_vec4_vec4
%term sub_vec4_vec4
%term mul_vec4_vec4
%term div_vec4_vec4
%term slt_vec4_vec4
%term sgt_vec4_vec4
%term sle_vec4_vec4
%term sge_vec4_vec4
%term seq_vec4_vec4
%term sne_vec4_vec4
%term dp4_vec4_vec4
%term dp3_vec4_vec4
%term dp2_vec4_vec4
%term sqrt_vec4
%term rsq_vec4
%term swizzle_vec4
%term trunc_vec4

# Each tree will produce stmt.  Currently, the only production for
# stmt is from an assign rule -- every statement tree from
# ir_to_mesa.cpp assigns a result to a register.

%start stmt

# Now comes all the rules for code generation.  Each rule is of the
# general form
#
# produced: term(term, term) cost
# {
#	code_run_when_we_choose_this_rule();
# }
#
# where choosing this rule means we turn term(term, term) into
# produced at the cost of "cost".  We measure "cost" in approximate
# instruction count.  The BURG should then more or less minimize the
# number of instructions.

# A reference of a variable is just a vec4 register location,
# so it can be used as an argument for pretty much anything.
vec4: reference_vec4 0

# A reference of a variable is just a vec4 register location,
# so it can be used as an argument for pretty much anything.
vec4: array_reference_vec4_vec4(vec4, vec4) 1
{
	ir_to_mesa_dst_reg address_reg = {PROGRAM_ADDRESS, 0, WRITEMASK_X};

	ir_to_mesa_emit_op1_full(tree->v, tree->ir, OPCODE_ARL,
				 address_reg,
				 tree->right->src_reg);
	ir_to_mesa_emit_op1_full(tree->v, tree->ir, OPCODE_MOV,
				 tree->dst_reg,
				 tree->left->src_reg);
}

# Here's the rule everyone will hit: Moving the result of an
# expression into a variable-dereference register location.
#
# Note that this is likely a gratuitous move.  We could make variants
# of each of the following rules, e.g:
#
# vec4: add_vec4_vec4(vec4, vec4) 1
# {
#         emit(ADD, tree, tree->left, tree->right);
# }
#
# becoming
#
# vec4: assign(vec4_vec4, add_vec4_vec4(vec4, vec4) 1
# {
#         emit(ADD, tree->left, tree->right->left, tree->right->right);
# }
#
# But it seems like a lot of extra typing and duped code, when we
# probably want copy propagation and dead code after codegen anyway,
# which would clean these up.
stmt: assign(vec4, vec4) 1
{
	ir_to_mesa_emit_op1_full(tree->v, tree->ir, OPCODE_MOV,
				 tree->left->dst_reg,
				 tree->right->src_reg);
}

# Perform a swizzle by composing our swizzle with the swizzle
# required to get at the src reg.
vec4: swizzle_vec4(vec4) 1
{
	ir_to_mesa_src_reg reg = tree->left->src_reg;
	int swiz[4];
	int i;

	for (i = 0; i < 4; i++) {
		swiz[i] = GET_SWZ(tree->src_reg.swizzle, i);
		if (swiz[i] >= SWIZZLE_X && swiz[i] <= SWIZZLE_Y) {
			swiz[i] = GET_SWZ(tree->left->src_reg.swizzle, swiz[i]);
		}
	}
	reg.swizzle = MAKE_SWIZZLE4(swiz[0], swiz[1], swiz[2], swiz[3]);

	ir_to_mesa_emit_op1_full(tree->v, tree->ir, OPCODE_MOV,
				 tree->dst_reg,
				 reg);
}

vec4: sin_vec4(vec4) 1
{
	ir_to_mesa_emit_scalar_op1(tree, OPCODE_SIN,
				   tree->dst_reg,
				   tree->left->src_reg);
}

vec4: cos_vec4(vec4) 1
{
	ir_to_mesa_emit_scalar_op1(tree, OPCODE_COS,
				   tree->dst_reg,
				   tree->left->src_reg);
}

vec4: add_vec4_vec4(vec4, vec4) 1 { ir_to_mesa_emit_op2(tree, OPCODE_ADD); }
vec4: sub_vec4_vec4(vec4, vec4) 1 { ir_to_mesa_emit_op2(tree, OPCODE_SUB); }
vec4: mul_vec4_vec4(vec4, vec4) 1 { ir_to_mesa_emit_op2(tree, OPCODE_MUL); }

vec4: dp4_vec4_vec4(vec4, vec4) 1
{
	ir_to_mesa_emit_op2(tree, OPCODE_DP4);
	tree->src_reg.swizzle = SWIZZLE_XXXX;
}

vec4: dp3_vec4_vec4(vec4, vec4) 1
{
	ir_to_mesa_emit_op2(tree, OPCODE_DP3);
	tree->src_reg.swizzle = SWIZZLE_XXXX;
}


vec4: dp2_vec4_vec4(vec4, vec4) 1
{
	ir_to_mesa_emit_op2(tree, OPCODE_DP2);
	tree->src_reg.swizzle = SWIZZLE_XXXX;
}

vec4: div_vec4_vec4(vec4, vec4) 1
{
	ir_to_mesa_emit_scalar_op1(tree, OPCODE_RCP,
				   tree->dst_reg,
				   tree->right->src_reg);

	ir_to_mesa_emit_op2_full(tree->v, tree->ir, OPCODE_MUL,
				 tree->dst_reg,
				 tree->src_reg,
				 tree->left->src_reg);
}

vec4: slt_vec4_vec4(vec4, vec4) 1 { ir_to_mesa_emit_op2(tree, OPCODE_SLT); }
vec4: sgt_vec4_vec4(vec4, vec4) 1 { ir_to_mesa_emit_op2(tree, OPCODE_SGT); }
vec4: sle_vec4_vec4(vec4, vec4) 1 { ir_to_mesa_emit_op2(tree, OPCODE_SLE); }
vec4: sge_vec4_vec4(vec4, vec4) 1 { ir_to_mesa_emit_op2(tree, OPCODE_SGE); }
vec4: sne_vec4_vec4(vec4, vec4) 1 { ir_to_mesa_emit_op2(tree, OPCODE_SNE); }
vec4: seq_vec4_vec4(vec4, vec4) 1 { ir_to_mesa_emit_op2(tree, OPCODE_SEQ); }

vec4: sqrt_vec4(vec4) 1
{
	ir_to_mesa_emit_scalar_op1(tree, OPCODE_RSQ,
				   tree->dst_reg,
				   tree->left->src_reg);

	ir_to_mesa_emit_op1_full(tree->v, tree->ir, OPCODE_RCP,
				 tree->dst_reg,
				 tree->src_reg);
}

vec4: rsq_vec4(vec4) 1
{
	ir_to_mesa_emit_scalar_op1(tree, OPCODE_RSQ,
				   tree->dst_reg,
				   tree->left->src_reg);
}

vec4: exp_vec4(vec4) 1
{
	ir_to_mesa_emit_scalar_op1(tree, OPCODE_EXP,
				   tree->dst_reg,
				   tree->left->src_reg);
}

vec4: exp2_vec4(vec4) 1
{
	ir_to_mesa_emit_scalar_op1(tree, OPCODE_EX2,
				   tree->dst_reg,
				   tree->left->src_reg);
}

vec4: log_vec4(vec4) 1
{
	ir_to_mesa_emit_scalar_op1(tree, OPCODE_LOG,
				   tree->dst_reg,
				   tree->left->src_reg);
}

vec4: log2_vec4(vec4) 1
{
	ir_to_mesa_emit_scalar_op1(tree, OPCODE_LG2,
				   tree->dst_reg,
				   tree->left->src_reg);
}

vec4: trunc_vec4(vec4) 1 { ir_to_mesa_emit_op1(tree, OPCODE_TRUNC); }

%%