src/gallium/drivers/freedreno/a3xx/fd3_format.c

   1 /*
   2  * Copyright (C) 2013 Rob Clark <robclark@freedesktop.org>
   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
  25 #include "pipe/p_defines.h"
  26 #include "util/u_format.h"
  27
  28 #include "fd3_format.h"
  29
  30 /* Specifies the table of all the formats and their features. Also supplies
  31  * the helpers that look up various data in those tables.
  32  */
  33
  34 struct fd3_format {
  35         enum a3xx_vtx_fmt vtx;
  36         enum a3xx_tex_fmt tex;
  37         enum a3xx_color_fmt rb;
  38         enum a3xx_color_swap swap;
  39         boolean present;
  40 };
  41
  42 #define RB_NONE ~0
  43
  44 /* vertex + texture */
  45 #define VT(pipe, fmt, rbfmt, swapfmt) \
  46         [PIPE_FORMAT_ ## pipe] = { \
  47                 .present = 1, \
  48                 .vtx = VFMT_ ## fmt, \
  49                 .tex = TFMT_ ## fmt, \
  50                 .rb = RB_ ## rbfmt, \
  51                 .swap = swapfmt \
  52         }
  53
  54 /* texture-only */
  55 #define _T(pipe, fmt, rbfmt, swapfmt) \
  56         [PIPE_FORMAT_ ## pipe] = { \
  57                 .present = 1, \
  58                 .vtx = ~0, \
  59                 .tex = TFMT_ ## fmt, \
  60                 .rb = RB_ ## rbfmt, \
  61                 .swap = swapfmt \
  62         }
  63
  64 /* vertex-only */
  65 #define V_(pipe, fmt, rbfmt, swapfmt) \
  66         [PIPE_FORMAT_ ## pipe] = { \
  67                 .present = 1, \
  68                 .vtx = VFMT_ ## fmt, \
  69                 .tex = ~0, \
  70                 .rb = RB_ ## rbfmt, \
  71                 .swap = swapfmt \
  72         }
  73
  74 static struct fd3_format formats[PIPE_FORMAT_COUNT] = {
  75         /* 8-bit */
  76         VT(R8_UNORM,   8_UNORM, R8_UNORM, WZYX),
  77         VT(R8_SNORM,   8_SNORM, NONE,     WZYX),
  78         VT(R8_UINT,    8_UINT,  R8_UINT,  WZYX),
  79         VT(R8_SINT,    8_SINT,  R8_SINT,  WZYX),
  80         V_(R8_USCALED, 8_UINT,  NONE,     WZYX),
  81         V_(R8_SSCALED, 8_UINT,  NONE,     WZYX),
  82
  83         _T(A8_UNORM,   8_UNORM, A8_UNORM, WZYX),
  84         _T(L8_UNORM,   8_UNORM, R8_UNORM, WZYX),
  85         _T(I8_UNORM,   8_UNORM, NONE,     WZYX),
  86
  87         _T(A8_UINT,    8_UINT,  NONE,     WZYX),
  88         _T(A8_SINT,    8_SINT,  NONE,     WZYX),
  89         _T(L8_UINT,    8_UINT,  NONE,     WZYX),
  90         _T(L8_SINT,    8_SINT,  NONE,     WZYX),
  91         _T(I8_UINT,    8_UINT,  NONE,     WZYX),
  92         _T(I8_SINT,    8_SINT,  NONE,     WZYX),
  93
  94         _T(S8_UINT,    8_UINT,  R8_UNORM, WZYX),
  95
  96         /* 16-bit */
  97         VT(R16_UNORM,   16_UNORM, NONE,     WZYX),
  98         VT(R16_SNORM,   16_SNORM, NONE,     WZYX),
  99         VT(R16_UINT,    16_UINT,  R16_UINT, WZYX),
 100         VT(R16_SINT,    16_SINT,  R16_SINT, WZYX),
 101         V_(R16_USCALED, 16_UINT,  NONE,     WZYX),
 102         V_(R16_SSCALED, 16_UINT,  NONE,     WZYX),
 103         VT(R16_FLOAT,   16_FLOAT, R16_FLOAT,WZYX),
 104
 105         _T(A16_UINT,    16_UINT,  NONE,     WZYX),
 106         _T(A16_SINT,    16_SINT,  NONE,     WZYX),
 107         _T(L16_UINT,    16_UINT,  NONE,     WZYX),
 108         _T(L16_SINT,    16_SINT,  NONE,     WZYX),
 109         _T(I16_UINT,    16_UINT,  NONE,     WZYX),
 110         _T(I16_SINT,    16_SINT,  NONE,     WZYX),
 111
 112         VT(R8G8_UNORM,   8_8_UNORM, R8G8_UNORM, WZYX),
 113         VT(R8G8_SNORM,   8_8_SNORM, R8G8_SNORM, WZYX),
 114         VT(R8G8_UINT,    8_8_UINT,  NONE,       WZYX),
 115         VT(R8G8_SINT,    8_8_SINT,  NONE,       WZYX),
 116         V_(R8G8_USCALED, 8_8_UINT,  NONE,       WZYX),
 117         V_(R8G8_SSCALED, 8_8_SINT,  NONE,       WZYX),
 118
 119         _T(L8A8_UINT,    8_8_UINT,  NONE,       WZYX),
 120         _T(L8A8_SINT,    8_8_SINT,  NONE,       WZYX),
 121
 122         _T(Z16_UNORM,      Z16_UNORM,     R8G8_UNORM,     WZYX),
 123         _T(B5G6R5_UNORM,   5_6_5_UNORM,   R5G6B5_UNORM,   WXYZ),
 124         _T(B5G5R5A1_UNORM, 5_5_5_1_UNORM, R5G5B5A1_UNORM, WXYZ),
 125         _T(B5G5R5X1_UNORM, 5_5_5_1_UNORM, R5G5B5A1_UNORM, WXYZ),
 126         _T(B4G4R4A4_UNORM, 4_4_4_4_UNORM, R4G4B4A4_UNORM, WXYZ),
 127
 128         /* 24-bit */
 129         V_(R8G8B8_UNORM,   8_8_8_UNORM, NONE, WZYX),
 130         V_(R8G8B8_SNORM,   8_8_8_SNORM, NONE, WZYX),
 131         V_(R8G8B8_UINT,    8_8_8_UINT,  NONE, WZYX),
 132         V_(R8G8B8_SINT,    8_8_8_SINT,  NONE, WZYX),
 133         V_(R8G8B8_USCALED, 8_8_8_UINT,  NONE, WZYX),
 134         V_(R8G8B8_SSCALED, 8_8_8_SINT,  NONE, WZYX),
 135
 136         /* 32-bit */
 137         VT(R32_UINT,    32_UINT,  R32_UINT, WZYX),
 138         VT(R32_SINT,    32_SINT,  R32_SINT, WZYX),
 139         V_(R32_USCALED, 32_UINT,  NONE,     WZYX),
 140         V_(R32_SSCALED, 32_UINT,  NONE,     WZYX),
 141         VT(R32_FLOAT,   32_FLOAT, R32_FLOAT,WZYX),
 142         V_(R32_FIXED,   32_FIXED, NONE,     WZYX),
 143
 144         _T(A32_UINT,    32_UINT,  NONE,     WZYX),
 145         _T(A32_SINT,    32_SINT,  NONE,     WZYX),
 146         _T(L32_UINT,    32_UINT,  NONE,     WZYX),
 147         _T(L32_SINT,    32_SINT,  NONE,     WZYX),
 148         _T(I32_UINT,    32_UINT,  NONE,     WZYX),
 149         _T(I32_SINT,    32_SINT,  NONE,     WZYX),
 150
 151         VT(R16G16_UNORM,   16_16_UNORM, NONE,        WZYX),
 152         VT(R16G16_SNORM,   16_16_SNORM, NONE,        WZYX),
 153         VT(R16G16_UINT,    16_16_UINT,  R16G16_UINT, WZYX),
 154         VT(R16G16_SINT,    16_16_SINT,  R16G16_SINT, WZYX),
 155         V_(R16G16_USCALED, 16_16_UINT,  NONE,        WZYX),
 156         V_(R16G16_SSCALED, 16_16_SINT,  NONE,        WZYX),
 157         VT(R16G16_FLOAT,   16_16_FLOAT, R16G16_FLOAT,WZYX),
 158
 159         _T(L16A16_UINT,    16_16_UINT,  NONE,        WZYX),
 160         _T(L16A16_SINT,    16_16_SINT,  NONE,        WZYX),
 161
 162         VT(R8G8B8A8_UNORM,   8_8_8_8_UNORM, R8G8B8A8_UNORM, WZYX),
 163         _T(R8G8B8X8_UNORM,   8_8_8_8_UNORM, R8G8B8A8_UNORM, WZYX),
 164         _T(R8G8B8A8_SRGB,    8_8_8_8_UNORM, R8G8B8A8_UNORM, WZYX),
 165         _T(R8G8B8X8_SRGB,    8_8_8_8_UNORM, R8G8B8A8_UNORM, WZYX),
 166         VT(R8G8B8A8_SNORM,   8_8_8_8_SNORM, R8G8B8A8_SNORM, WZYX),
 167         VT(R8G8B8A8_UINT,    8_8_8_8_UINT,  R8G8B8A8_UINT,  WZYX),
 168         VT(R8G8B8A8_SINT,    8_8_8_8_SINT,  R8G8B8A8_SINT,  WZYX),
 169         V_(R8G8B8A8_USCALED, 8_8_8_8_UINT,  NONE,           WZYX),
 170         V_(R8G8B8A8_SSCALED, 8_8_8_8_SINT,  NONE,           WZYX),
 171
 172         VT(B8G8R8A8_UNORM,   8_8_8_8_UNORM, R8G8B8A8_UNORM, WXYZ),
 173         _T(B8G8R8X8_UNORM,   8_8_8_8_UNORM, R8G8B8A8_UNORM, WXYZ),
 174         VT(B8G8R8A8_SRGB,    8_8_8_8_UNORM, R8G8B8A8_UNORM, WXYZ),
 175         _T(B8G8R8X8_SRGB,    8_8_8_8_UNORM, R8G8B8A8_UNORM, WXYZ),
 176
 177         VT(A8B8G8R8_UNORM,   8_8_8_8_UNORM, R8G8B8A8_UNORM, XYZW),
 178         _T(X8B8G8R8_UNORM,   8_8_8_8_UNORM, R8G8B8A8_UNORM, XYZW),
 179         _T(A8B8G8R8_SRGB,    8_8_8_8_UNORM, R8G8B8A8_UNORM, XYZW),
 180         _T(X8B8G8R8_SRGB,    8_8_8_8_UNORM, R8G8B8A8_UNORM, XYZW),
 181
 182         VT(A8R8G8B8_UNORM,   8_8_8_8_UNORM, R8G8B8A8_UNORM, ZYXW),
 183         _T(X8R8G8B8_UNORM,   8_8_8_8_UNORM, R8G8B8A8_UNORM, ZYXW),
 184         _T(A8R8G8B8_SRGB,    8_8_8_8_UNORM, R8G8B8A8_UNORM, ZYXW),
 185         _T(X8R8G8B8_SRGB,    8_8_8_8_UNORM, R8G8B8A8_UNORM, ZYXW),
 186
 187         VT(R10G10B10A2_UNORM,   10_10_10_2_UNORM, R10G10B10A2_UNORM, WZYX),
 188         VT(B10G10R10A2_UNORM,   10_10_10_2_UNORM, R10G10B10A2_UNORM, WXYZ),
 189         _T(B10G10R10X2_UNORM,   10_10_10_2_UNORM, R10G10B10A2_UNORM, WXYZ),
 190         V_(R10G10B10A2_SNORM,   10_10_10_2_SNORM, NONE,              WZYX),
 191         V_(R10G10B10A2_UINT,    10_10_10_2_UINT,  NONE,              WZYX),
 192         V_(R10G10B10A2_USCALED, 10_10_10_2_UINT,  NONE,              WZYX),
 193         V_(R10G10B10A2_SSCALED, 10_10_10_2_SINT,  NONE,              WZYX),
 194
 195         _T(R11G11B10_FLOAT, 11_11_10_FLOAT, R11G11B10_FLOAT, WZYX),
 196         _T(R9G9B9E5_FLOAT,  9_9_9_E5_FLOAT, NONE,            WZYX),
 197
 198         _T(Z24X8_UNORM,       X8Z24_UNORM, R8G8B8A8_UNORM, WZYX),
 199         _T(Z24_UNORM_S8_UINT, X8Z24_UNORM, R8G8B8A8_UNORM, WZYX),
 200         _T(Z32_FLOAT,         Z32_FLOAT,   R8G8B8A8_UNORM, WZYX),
 201         _T(Z32_FLOAT_S8X24_UINT, Z32_FLOAT,R8G8B8A8_UNORM, WZYX),
 202
 203         /* 48-bit */
 204         V_(R16G16B16_UNORM,   16_16_16_UNORM, NONE, WZYX),
 205         V_(R16G16B16_SNORM,   16_16_16_SNORM, NONE, WZYX),
 206         V_(R16G16B16_UINT,    16_16_16_UINT,  NONE, WZYX),
 207         V_(R16G16B16_SINT,    16_16_16_SINT,  NONE, WZYX),
 208         V_(R16G16B16_USCALED, 16_16_16_UINT,  NONE, WZYX),
 209         V_(R16G16B16_SSCALED, 16_16_16_SINT,  NONE, WZYX),
 210         V_(R16G16B16_FLOAT,   16_16_16_FLOAT, NONE, WZYX),
 211
 212         /* 64-bit */
 213         VT(R16G16B16A16_UNORM,   16_16_16_16_UNORM, NONE,               WZYX),
 214         VT(R16G16B16A16_SNORM,   16_16_16_16_SNORM, NONE,               WZYX),
 215         VT(R16G16B16A16_UINT,    16_16_16_16_UINT,  R16G16B16A16_UINT,  WZYX),
 216         _T(R16G16B16X16_UINT,    16_16_16_16_UINT,  R16G16B16A16_UINT,  WZYX),
 217         VT(R16G16B16A16_SINT,    16_16_16_16_SINT,  R16G16B16A16_SINT,  WZYX),
 218         _T(R16G16B16X16_SINT,    16_16_16_16_SINT,  R16G16B16A16_SINT,  WZYX),
 219         V_(R16G16B16A16_USCALED, 16_16_16_16_UINT,  NONE,               WZYX),
 220         V_(R16G16B16A16_SSCALED, 16_16_16_16_SINT,  NONE,               WZYX),
 221         VT(R16G16B16A16_FLOAT,   16_16_16_16_FLOAT, R16G16B16A16_FLOAT, WZYX),
 222         _T(R16G16B16X16_FLOAT,   16_16_16_16_FLOAT, R16G16B16A16_FLOAT, WZYX),
 223
 224         VT(R32G32_UINT,    32_32_UINT,  R32G32_UINT, WZYX),
 225         VT(R32G32_SINT,    32_32_SINT,  R32G32_SINT, WZYX),
 226         V_(R32G32_USCALED, 32_32_UINT,  NONE,        WZYX),
 227         V_(R32G32_SSCALED, 32_32_SINT,  NONE,        WZYX),
 228         VT(R32G32_FLOAT,   32_32_FLOAT, R32G32_FLOAT,WZYX),
 229         V_(R32G32_FIXED,   32_32_FIXED, NONE,        WZYX),
 230
 231         _T(L32A32_UINT,    32_32_UINT,  NONE,        WZYX),
 232         _T(L32A32_SINT,    32_32_SINT,  NONE,        WZYX),
 233
 234         /* 96-bit */
 235         V_(R32G32B32_UINT,    32_32_32_UINT,  NONE, WZYX),
 236         V_(R32G32B32_SINT,    32_32_32_SINT,  NONE, WZYX),
 237         V_(R32G32B32_USCALED, 32_32_32_UINT,  NONE, WZYX),
 238         V_(R32G32B32_SSCALED, 32_32_32_SINT,  NONE, WZYX),
 239         V_(R32G32B32_FLOAT,   32_32_32_FLOAT, NONE, WZYX),
 240         V_(R32G32B32_FIXED,   32_32_32_FIXED, NONE, WZYX),
 241
 242         /* 128-bit */
 243         VT(R32G32B32A32_UINT,    32_32_32_32_UINT,  R32G32B32A32_UINT,  WZYX),
 244         _T(R32G32B32X32_UINT,    32_32_32_32_UINT,  R32G32B32A32_UINT,  WZYX),
 245         VT(R32G32B32A32_SINT,    32_32_32_32_SINT,  R32G32B32A32_SINT,  WZYX),
 246         _T(R32G32B32X32_SINT,    32_32_32_32_SINT,  R32G32B32A32_SINT,  WZYX),
 247         V_(R32G32B32A32_USCALED, 32_32_32_32_UINT,  NONE,               WZYX),
 248         V_(R32G32B32A32_SSCALED, 32_32_32_32_SINT,  NONE,               WZYX),
 249         VT(R32G32B32A32_FLOAT,   32_32_32_32_FLOAT, R32G32B32A32_FLOAT, WZYX),
 250         _T(R32G32B32X32_FLOAT,   32_32_32_32_FLOAT, R32G32B32A32_FLOAT, WZYX),
 251         V_(R32G32B32A32_FIXED,   32_32_32_32_FIXED, NONE,               WZYX),
 252
 253         /* compressed */
 254         _T(ETC1_RGB8, ETC1, NONE, WZYX),
 255         _T(ETC2_RGB8, ETC2_RGB8, NONE, WZYX),
 256         _T(ETC2_SRGB8, ETC2_RGB8, NONE, WZYX),
 257         _T(ETC2_RGB8A1, ETC2_RGB8A1, NONE, WZYX),
 258         _T(ETC2_SRGB8A1, ETC2_RGB8A1, NONE, WZYX),
 259         _T(ETC2_RGBA8, ETC2_RGBA8, NONE, WZYX),
 260         _T(ETC2_SRGBA8, ETC2_RGBA8, NONE, WZYX),
 261         _T(ETC2_R11_UNORM, ETC2_R11_UNORM, NONE, WZYX),
 262         _T(ETC2_R11_SNORM, ETC2_R11_SNORM, NONE, WZYX),
 263         _T(ETC2_RG11_UNORM, ETC2_RG11_UNORM, NONE, WZYX),
 264         _T(ETC2_RG11_SNORM, ETC2_RG11_SNORM, NONE, WZYX),
 265 };
 266
 267 enum a3xx_vtx_fmt
 268 fd3_pipe2vtx(enum pipe_format format)
 269 {
 270         if (!formats[format].present)
 271                 return ~0;
 272         return formats[format].vtx;
 273 }
 274
 275 enum a3xx_tex_fmt
 276 fd3_pipe2tex(enum pipe_format format)
 277 {
 278         if (!formats[format].present)
 279                 return ~0;
 280         return formats[format].tex;
 281 }
 282
 283 enum a3xx_color_fmt
 284 fd3_pipe2color(enum pipe_format format)
 285 {
 286         if (!formats[format].present)
 287                 return ~0;
 288         return formats[format].rb;
 289 }
 290
 291 enum a3xx_color_swap
 292 fd3_pipe2swap(enum pipe_format format)
 293 {
 294         if (!formats[format].present)
 295                 return WZYX;
 296         return formats[format].swap;
 297 }
 298
 299 enum a3xx_tex_fetchsize
 300 fd3_pipe2fetchsize(enum pipe_format format)
 301 {
 302         if (format == PIPE_FORMAT_Z32_FLOAT_S8X24_UINT)
 303                 format = PIPE_FORMAT_Z32_FLOAT;
 304         switch (util_format_get_blocksizebits(format)) {
 305         case 8: return TFETCH_1_BYTE;
 306         case 16: return TFETCH_2_BYTE;
 307         case 32: return TFETCH_4_BYTE;
 308         case 64: return TFETCH_8_BYTE;
 309         case 128: return TFETCH_16_BYTE;
 310         default:
 311                 debug_printf("Unknown block size for format %s: %d\n",
 312                                          util_format_name(format),
 313                                          util_format_get_blocksizebits(format));
 314                 return TFETCH_DISABLE;
 315         }
 316 }
 317
 318 /* we need to special case a bit the depth/stencil restore, because we are
 319  * using the texture sampler to blit into the depth/stencil buffer, *not*
 320  * into a color buffer.  Otherwise fd3_tex_swiz() will do the wrong thing,
 321  * as it is assuming that you are sampling into normal render target..
 322  */
 323 enum pipe_format
 324 fd3_gmem_restore_format(enum pipe_format format)
 325 {
 326         switch (format) {
 327         case PIPE_FORMAT_Z24X8_UNORM:
 328         case PIPE_FORMAT_Z24_UNORM_S8_UINT:
 329                 return PIPE_FORMAT_R8G8B8A8_UNORM;
 330         case PIPE_FORMAT_Z16_UNORM:
 331                 return PIPE_FORMAT_R8G8_UNORM;
 332         case PIPE_FORMAT_S8_UINT:
 333                 return PIPE_FORMAT_R8_UNORM;
 334         default:
 335                 return format;
 336         }
 337 }
 338
 339 enum a3xx_color_fmt
 340 fd3_fs_output_format(enum pipe_format format)
 341 {
 342         if (util_format_is_srgb(format))
 343                 return RB_R16G16B16A16_FLOAT;
 344         switch (format) {
 345         case PIPE_FORMAT_R16_FLOAT:
 346         case PIPE_FORMAT_R16G16_FLOAT:
 347         case PIPE_FORMAT_R11G11B10_FLOAT:
 348                 return RB_R16G16B16A16_FLOAT;
 349         default:
 350                 return fd3_pipe2color(format);
 351         }
 352 }
 353
 354 static inline enum a3xx_tex_swiz
 355 tex_swiz(unsigned swiz)
 356 {
 357         switch (swiz) {
 358         default:
 359         case PIPE_SWIZZLE_RED:   return A3XX_TEX_X;
 360         case PIPE_SWIZZLE_GREEN: return A3XX_TEX_Y;
 361         case PIPE_SWIZZLE_BLUE:  return A3XX_TEX_Z;
 362         case PIPE_SWIZZLE_ALPHA: return A3XX_TEX_W;
 363         case PIPE_SWIZZLE_ZERO:  return A3XX_TEX_ZERO;
 364         case PIPE_SWIZZLE_ONE:   return A3XX_TEX_ONE;
 365         }
 366 }
 367
 368 uint32_t
 369 fd3_tex_swiz(enum pipe_format format, unsigned swizzle_r, unsigned swizzle_g,
 370                 unsigned swizzle_b, unsigned swizzle_a)
 371 {
 372         const struct util_format_description *desc =
 373                         util_format_description(format);
 374         unsigned char swiz[4] = {
 375                         swizzle_r, swizzle_g, swizzle_b, swizzle_a,
 376         }, rswiz[4];
 377
 378         util_format_compose_swizzles(desc->swizzle, swiz, rswiz);
 379
 380         return A3XX_TEX_CONST_0_SWIZ_X(tex_swiz(rswiz[0])) |
 381                         A3XX_TEX_CONST_0_SWIZ_Y(tex_swiz(rswiz[1])) |
 382                         A3XX_TEX_CONST_0_SWIZ_Z(tex_swiz(rswiz[2])) |
 383                         A3XX_TEX_CONST_0_SWIZ_W(tex_swiz(rswiz[3]));
 384 }