1/*
2 * Copyright © 2014 Intel Corporation
3 *
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:
10 *
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
13 * Software.
14 *
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.
22 */
23
24/**
25 * \file lower_const_arrays_to_uniforms.cpp
26 *
27 * Lower constant arrays to uniform arrays.
28 *
29 * Some driver backends (such as i965 and nouveau) don't handle constant arrays
30 * gracefully, instead treating them as ordinary writable temporary arrays.
31 * Since arrays can be large, this often means spilling them to scratch memory,
32 * which usually involves a large number of instructions.
33 *
34 * This must be called prior to link_set_uniform_initializers(); we need the
35 * linker to process our new uniform's constant initializer.
36 *
37 * This should be called after optimizations, since those can result in
38 * splitting and removing arrays that are indexed by constant expressions.
39 */
40#include "ir.h"
41#include "ir_visitor.h"
42#include "ir_rvalue_visitor.h"
43#include "compiler/glsl_types.h"
44
45namespace {
46class lower_const_array_visitor : public ir_rvalue_visitor {
47public:
48   lower_const_array_visitor(exec_list *insts, unsigned s)
49   {
50      instructions = insts;
51      stage = s;
52      const_count = 0;
53      progress = false;
54   }
55
56   bool run()
57   {
58      visit_list_elements(this, instructions);
59      return progress;
60   }
61
62   ir_visitor_status visit_enter(ir_texture *);
63   void handle_rvalue(ir_rvalue **rvalue);
64
65private:
66   exec_list *instructions;
67   unsigned stage;
68   unsigned const_count;
69   bool progress;
70};
71
72ir_visitor_status
73lower_const_array_visitor::visit_enter(ir_texture *)
74{
75   return visit_continue_with_parent;
76}
77
78void
79lower_const_array_visitor::handle_rvalue(ir_rvalue **rvalue)
80{
81   if (!*rvalue)
82      return;
83
84   ir_constant *con = (*rvalue)->as_constant();
85   if (!con || !con->type->is_array())
86      return;
87
88   void *mem_ctx = ralloc_parent(con);
89
90   /* In the very unlikely event of 4294967295 constant arrays in a single
91    * shader, don't promote this to a uniform.
92    */
93   unsigned limit = ~0;
94   if (const_count == limit)
95      return;
96
97   char *uniform_name = ralloc_asprintf(mem_ctx, "constarray_%x_%u",
98                                        const_count, stage);
99   const_count++;
100
101   ir_variable *uni =
102      new(mem_ctx) ir_variable(con->type, uniform_name, ir_var_uniform);
103   uni->constant_initializer = con;
104   uni->constant_value = con;
105   uni->data.has_initializer = true;
106   uni->data.how_declared = ir_var_hidden;
107   uni->data.read_only = true;
108   /* Assume the whole thing is accessed. */
109   uni->data.max_array_access = uni->type->length - 1;
110   instructions->push_head(uni);
111
112   *rvalue = new(mem_ctx) ir_dereference_variable(uni);
113
114   progress = true;
115}
116
117} /* anonymous namespace */
118
119bool
120lower_const_arrays_to_uniforms(exec_list *instructions, unsigned stage)
121{
122   lower_const_array_visitor v(instructions, stage);
123   return v.run();
124}
125