1 /*
2  * Copyright © 2019 Raspberry Pi
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 DEALINGS
21  * IN THE SOFTWARE.
22  */
23 
24 #include "nir_builder.h"
25 
26 /** @file nir_lower_point_size.c
27  *
28  * The OpenGL spec requires that implementations clamp gl_PointSize to an
29  * implementation-dependant point size range. The OpenGL ES 3.0 spec further
30  * requires that this range must match GL_ALIASED_POINT_SIZE_RANGE.
31  * Some hardware such as V3D don't clamp to a valid range automatically so
32  * the driver must clamp the point size written by the shader manually to a
33  * valid range.
34  */
35 
36 static void
lower_point_size_instr(nir_builder * b,nir_instr * psiz_instr,float min,float max)37 lower_point_size_instr(nir_builder *b, nir_instr *psiz_instr,
38                        float min, float max)
39 {
40    b->cursor = nir_before_instr(psiz_instr);
41 
42    nir_intrinsic_instr *instr = nir_instr_as_intrinsic(psiz_instr);
43 
44    assert(instr->src[1].is_ssa);
45    assert(instr->src[1].ssa->num_components == 1);
46    nir_ssa_def *psiz = instr->src[1].ssa;
47 
48    if (min > 0.0f)
49       psiz = nir_fmax(b, psiz, nir_imm_float(b, min));
50 
51    if (max > 0.0f)
52       psiz = nir_fmin(b, psiz, nir_imm_float(b, max));
53 
54    nir_instr_rewrite_src(&instr->instr, &instr->src[1], nir_src_for_ssa(psiz));
55 }
56 
57 static bool
instr_is_point_size(const nir_instr * instr)58 instr_is_point_size(const nir_instr *instr)
59 {
60    if (instr->type != nir_instr_type_intrinsic)
61       return false;
62 
63    nir_intrinsic_instr *intr = nir_instr_as_intrinsic(instr);
64    if (intr->intrinsic != nir_intrinsic_store_deref)
65       return false;
66 
67    nir_deref_instr *deref = nir_src_as_deref(intr->src[0]);
68    nir_variable *var = nir_deref_instr_get_variable(deref);
69    if (var->data.location != VARYING_SLOT_PSIZ)
70       return false;
71 
72    return true;
73 }
74 
75 /**
76  * Clamps gl_PointSize to the range [min, max]. If either min or max are not
77  * greater than 0 then no clamping is done for that side of the range.
78  */
79 bool
nir_lower_point_size(nir_shader * s,float min,float max)80 nir_lower_point_size(nir_shader *s, float min, float max)
81 {
82    assert(s->info.stage != MESA_SHADER_FRAGMENT &&
83           s->info.stage != MESA_SHADER_COMPUTE);
84 
85    assert(min > 0.0f || max > 0.0f);
86    assert(min <= 0.0f || max <= 0.0f || min <= max);
87 
88    bool progress = false;
89    nir_foreach_function(function, s) {
90       if (!function->impl)
91          continue;
92 
93       nir_builder b;
94       nir_builder_init(&b, function->impl);
95 
96       nir_foreach_block(block, function->impl) {
97          nir_foreach_instr_safe(instr, block) {
98             if (instr_is_point_size(instr)) {
99                lower_point_size_instr(&b, instr, min, max);
100                progress = true;
101             }
102          }
103       }
104 
105       if (progress) {
106          nir_metadata_preserve(function->impl,
107                                nir_metadata_block_index |
108                                nir_metadata_dominance);
109       }
110    }
111 
112    return progress;
113 }
114