xref: /third_party/mesa3d/src/amd/common/ac_binary.c (revision bf215546)
1/*
2 * Copyright 2014 Advanced Micro Devices, Inc.
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 FROM,
20 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21 * SOFTWARE.
22 */
23
24#include "ac_binary.h"
25
26#include "ac_gpu_info.h"
27#include "util/u_math.h"
28#include "util/u_memory.h"
29
30#ifndef _WIN32
31#include <gelf.h>
32#include <libelf.h>
33#endif
34#include <sid.h>
35#include <stdio.h>
36
37#define SPILLED_SGPRS 0x4
38#define SPILLED_VGPRS 0x8
39
40/* Parse configuration data in .AMDGPU.config section format. */
41void ac_parse_shader_binary_config(const char *data, size_t nbytes, unsigned wave_size,
42                                   const struct radeon_info *info, struct ac_shader_config *conf)
43{
44   for (size_t i = 0; i < nbytes; i += 8) {
45      unsigned reg = util_le32_to_cpu(*(uint32_t *)(data + i));
46      unsigned value = util_le32_to_cpu(*(uint32_t *)(data + i + 4));
47      switch (reg) {
48      case R_00B028_SPI_SHADER_PGM_RSRC1_PS:
49      case R_00B128_SPI_SHADER_PGM_RSRC1_VS:
50      case R_00B228_SPI_SHADER_PGM_RSRC1_GS:
51      case R_00B848_COMPUTE_PGM_RSRC1:
52      case R_00B428_SPI_SHADER_PGM_RSRC1_HS:
53         if (wave_size == 32 || info->wave64_vgpr_alloc_granularity == 8)
54            conf->num_vgprs = MAX2(conf->num_vgprs, (G_00B028_VGPRS(value) + 1) * 8);
55         else
56            conf->num_vgprs = MAX2(conf->num_vgprs, (G_00B028_VGPRS(value) + 1) * 4);
57
58         conf->num_sgprs = MAX2(conf->num_sgprs, (G_00B028_SGPRS(value) + 1) * 8);
59         /* TODO: LLVM doesn't set FLOAT_MODE for non-compute shaders */
60         conf->float_mode = G_00B028_FLOAT_MODE(value);
61         conf->rsrc1 = value;
62         break;
63      case R_00B02C_SPI_SHADER_PGM_RSRC2_PS:
64         conf->lds_size = MAX2(conf->lds_size, G_00B02C_EXTRA_LDS_SIZE(value));
65         /* TODO: LLVM doesn't set SHARED_VGPR_CNT for all shader types */
66         conf->num_shared_vgprs = G_00B02C_SHARED_VGPR_CNT(value);
67         conf->rsrc2 = value;
68         break;
69      case R_00B12C_SPI_SHADER_PGM_RSRC2_VS:
70         conf->num_shared_vgprs = G_00B12C_SHARED_VGPR_CNT(value);
71         conf->rsrc2 = value;
72         break;
73      case R_00B22C_SPI_SHADER_PGM_RSRC2_GS:
74         conf->num_shared_vgprs = G_00B22C_SHARED_VGPR_CNT(value);
75         conf->rsrc2 = value;
76         break;
77      case R_00B42C_SPI_SHADER_PGM_RSRC2_HS:
78         conf->num_shared_vgprs = G_00B42C_SHARED_VGPR_CNT(value);
79         conf->rsrc2 = value;
80         break;
81      case R_00B84C_COMPUTE_PGM_RSRC2:
82         conf->lds_size = MAX2(conf->lds_size, G_00B84C_LDS_SIZE(value));
83         conf->rsrc2 = value;
84         break;
85      case R_00B8A0_COMPUTE_PGM_RSRC3:
86         conf->num_shared_vgprs = G_00B8A0_SHARED_VGPR_CNT(value);
87         conf->rsrc3 = value;
88         break;
89      case R_0286CC_SPI_PS_INPUT_ENA:
90         conf->spi_ps_input_ena = value;
91         break;
92      case R_0286D0_SPI_PS_INPUT_ADDR:
93         conf->spi_ps_input_addr = value;
94         break;
95      case R_0286E8_SPI_TMPRING_SIZE:
96      case R_00B860_COMPUTE_TMPRING_SIZE:
97         if (info->gfx_level >= GFX11)
98            conf->scratch_bytes_per_wave = G_00B860_WAVESIZE(value) * 256;
99         else
100            conf->scratch_bytes_per_wave = G_00B860_WAVESIZE(value) * 1024;
101         break;
102      case SPILLED_SGPRS:
103         conf->spilled_sgprs = value;
104         break;
105      case SPILLED_VGPRS:
106         conf->spilled_vgprs = value;
107         break;
108      default: {
109         static bool printed;
110
111         if (!printed) {
112            fprintf(stderr,
113                    "Warning: LLVM emitted unknown "
114                    "config register: 0x%x\n",
115                    reg);
116            printed = true;
117         }
118      } break;
119      }
120   }
121
122   if (!conf->spi_ps_input_addr)
123      conf->spi_ps_input_addr = conf->spi_ps_input_ena;
124
125   /* GFX 10.3 internally:
126    * - aligns VGPRS to 16 for Wave32 and 8 for Wave64
127    * - aligns LDS to 1024
128    *
129    * For shader-db stats, set num_vgprs that the hw actually uses.
130    */
131   if (info->gfx_level == GFX10_3) {
132      conf->num_vgprs = align(conf->num_vgprs, wave_size == 32 ? 16 : 8);
133   }
134
135   /* Enable 64-bit and 16-bit denormals, because there is no performance
136    * cost.
137    *
138    * Don't enable denormals for 32-bit floats, because:
139    * - denormals disable output modifiers
140    * - denormals break v_mad_f32
141    * - GFX6 & GFX7 would be very slow
142    */
143   conf->float_mode &= ~V_00B028_FP_32_DENORMS;
144   conf->float_mode |= V_00B028_FP_16_64_DENORMS;
145}
146