/* * 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. */ /** * \file ir_variable_refcount.cpp * * Provides a visitor which produces a list of variables referenced, * how many times they were referenced and assigned, and whether they * were defined in the scope. */ #include "ir.h" #include "ir_visitor.h" #include "ir_variable_refcount.h" #include "compiler/glsl_types.h" #include "util/hash_table.h" ir_variable_refcount_visitor::ir_variable_refcount_visitor() { this->mem_ctx = ralloc_context(NULL); this->ht = _mesa_hash_table_create(NULL, _mesa_hash_pointer, _mesa_key_pointer_equal); } static void free_entry(struct hash_entry *entry) { ir_variable_refcount_entry *ivre = (ir_variable_refcount_entry *) entry->data; /* Free assignment list */ exec_node *n; while ((n = ivre->assign_list.pop_head()) != NULL) { struct assignment_entry *assignment_entry = exec_node_data(struct assignment_entry, n, link); free(assignment_entry); } delete ivre; } ir_variable_refcount_visitor::~ir_variable_refcount_visitor() { ralloc_free(this->mem_ctx); _mesa_hash_table_destroy(this->ht, free_entry); } // constructor ir_variable_refcount_entry::ir_variable_refcount_entry(ir_variable *var) { this->var = var; assigned_count = 0; declaration = false; referenced_count = 0; } ir_variable_refcount_entry * ir_variable_refcount_visitor::get_variable_entry(ir_variable *var) { assert(var); struct hash_entry *e = _mesa_hash_table_search(this->ht, var); if (e) return (ir_variable_refcount_entry *)e->data; ir_variable_refcount_entry *entry = new ir_variable_refcount_entry(var); assert(entry->referenced_count == 0); _mesa_hash_table_insert(this->ht, var, entry); return entry; } ir_visitor_status ir_variable_refcount_visitor::visit(ir_variable *ir) { ir_variable_refcount_entry *entry = this->get_variable_entry(ir); if (entry) entry->declaration = true; return visit_continue; } ir_visitor_status ir_variable_refcount_visitor::visit(ir_dereference_variable *ir) { ir_variable *const var = ir->variable_referenced(); ir_variable_refcount_entry *entry = this->get_variable_entry(var); if (entry) entry->referenced_count++; return visit_continue; } ir_visitor_status ir_variable_refcount_visitor::visit_enter(ir_function_signature *ir) { /* We don't want to descend into the function parameters and * dead-code eliminate them, so just accept the body here. */ visit_list_elements(this, &ir->body); return visit_continue_with_parent; } ir_visitor_status ir_variable_refcount_visitor::visit_leave(ir_assignment *ir) { ir_variable_refcount_entry *entry; entry = this->get_variable_entry(ir->lhs->variable_referenced()); if (entry) { entry->assigned_count++; /* Build a list for dead code optimisation. Don't add assignment if it * was declared out of scope (outside the instruction stream). Also don't * bother adding any more to the list if there are more references than * assignments as this means the variable is used and won't be optimised * out. */ assert(entry->referenced_count >= entry->assigned_count); if (entry->referenced_count == entry->assigned_count) { struct assignment_entry *assignment_entry = (struct assignment_entry *)calloc(1, sizeof(*assignment_entry)); assignment_entry->assign = ir; entry->assign_list.push_head(&assignment_entry->link); } } return visit_continue; } href='#n35'>35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 
/*
 * Mesa 3-D graphics library
 *
 * Copyright (C) 2012-2013 LunarG, Inc.
 *
 * 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 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:
 *    Chia-I Wu <olv@lunarg.com>
 */

#include "util/u_transfer.h"

#include "ilo_cp.h"
#include "ilo_context.h"
#include "ilo_screen.h"
#include "ilo_resource.h"

static struct intel_bo *
alloc_buf_bo(const struct ilo_resource *res)
{
   struct ilo_screen *is = ilo_screen(res->base.screen);
   struct intel_bo *bo;
   const char *name;
   const unsigned size = res->bo_width;

   switch (res->base.bind) {
   case PIPE_BIND_VERTEX_BUFFER:
      name = "vertex buffer";
      break;
   case PIPE_BIND_INDEX_BUFFER:
      name = "index buffer";
      break;
   case PIPE_BIND_CONSTANT_BUFFER:
      name = "constant buffer";
      break;
   case PIPE_BIND_STREAM_OUTPUT:
      name = "stream output";
      break;
   default:
      name = "unknown buffer";
      break;
   }

   /* this is what a buffer supposed to be like */
   assert(res->bo_width * res->bo_height * res->bo_cpp == size);
   assert(res->tiling == INTEL_TILING_NONE);
   assert(res->bo_stride == 0);

   if (res->handle) {
      bo = is->winsys->import_handle(is->winsys, name,
            res->bo_width, res->bo_height, res->bo_cpp, res->handle);

      /* since the bo is shared to us, make sure it meets the expectations */
      if (bo) {
         assert(bo->get_size(res->bo) == size);
         assert(bo->get_tiling(res->bo) == res->tiling);
         assert(bo->get_pitch(res->bo) == res->bo_stride);
      }
   }
   else {
      bo = is->winsys->alloc_buffer(is->winsys, name, size, 0);
   }

   return bo;
}

static struct intel_bo *
alloc_tex_bo(const struct ilo_resource *res)
{
   struct ilo_screen *is = ilo_screen(res->base.screen);
   struct intel_bo *bo;
   const char *name;

   switch (res->base.target) {
   case PIPE_TEXTURE_1D:
      name = "1D texture";
      break;
   case PIPE_TEXTURE_2D:
      name = "2D texture";
      break;
   case PIPE_TEXTURE_3D:
      name = "3D texture";
      break;
   case PIPE_TEXTURE_CUBE:
      name = "cube texture";
      break;
   case PIPE_TEXTURE_RECT:
      name = "rectangle texture";
      break;
   case PIPE_TEXTURE_1D_ARRAY:
      name = "1D array texture";
      break;
   case PIPE_TEXTURE_2D_ARRAY:
      name = "2D array texture";
      break;
   case PIPE_TEXTURE_CUBE_ARRAY:
      name = "cube array texture";
      break;
   default:
      name ="unknown texture";
      break;
   }

   if (res->handle) {
      bo = is->winsys->import_handle(is->winsys, name,
            res->bo_width, res->bo_height, res->bo_cpp, res->handle);
   }
   else {
      const bool for_render =
         (res->base.bind & (PIPE_BIND_DEPTH_STENCIL |
                            PIPE_BIND_RENDER_TARGET));
      const unsigned long flags =
         (for_render) ? INTEL_ALLOC_FOR_RENDER : 0;

      bo = is->winsys->alloc(is->winsys, name,
            res->bo_width, res->bo_height, res->bo_cpp,
            res->tiling, flags);
   }

   return bo;
}

static bool
realloc_bo(struct ilo_resource *res)
{
   struct intel_bo *old_bo = res->bo;

   /* a shared bo cannot be reallocated */
   if (old_bo && res->handle)
      return false;

   if (res->base.target == PIPE_BUFFER)
      res->bo = alloc_buf_bo(res);
   else
      res->bo = alloc_tex_bo(res);

   if (!res->bo) {
      res->bo = old_bo;
      return false;
   }

   /* winsys may decide to use a different tiling */
   res->tiling = res->bo->get_tiling(res->bo);
   res->bo_stride = res->bo->get_pitch(res->bo);

   if (old_bo)
      old_bo->unreference(old_bo);

   return true;
}

static void
ilo_transfer_inline_write(struct pipe_context *pipe,
                          struct pipe_resource *r,
                          unsigned level,
                          unsigned usage,
                          const struct pipe_box *box,
                          const void *data,
                          unsigned stride,
                          unsigned layer_stride)
{
   struct ilo_context *ilo = ilo_context(pipe);
   struct ilo_resource *res = ilo_resource(r);
   int offset, size;
   bool will_be_busy;

   /*
    * Fall back to map(), memcpy(), and unmap().  We use this path for
    * unsynchronized write, as the buffer is likely to be busy and pwrite()
    * will stall.
    */
   if (unlikely(res->base.target != PIPE_BUFFER) ||
       (usage & PIPE_TRANSFER_UNSYNCHRONIZED)) {
      u_default_transfer_inline_write(pipe, r,
            level, usage, box, data, stride, layer_stride);

      return;
   }

   /*
    * XXX With hardware context support, the bo may be needed by GPU without
    * being referenced by ilo->cp->bo.  We have to flush unconditionally, and
    * that is bad.
    */
   if (ilo->cp->hw_ctx)
      ilo_cp_flush(ilo->cp);

   will_be_busy = ilo->cp->bo->references(ilo->cp->bo, res->bo);

   /* see if we can avoid stalling */
   if (will_be_busy || intel_bo_is_busy(res->bo)) {
      bool will_stall = true;

      if (usage & PIPE_TRANSFER_DISCARD_WHOLE_RESOURCE) {
         /* old data not needed so discard the old bo to avoid stalling */
         if (realloc_bo(res))
            will_stall = false;
      }
      else {
         /*
          * We could allocate a temporary bo to hold the data and emit
          * pipelined copy blit to move them to res->bo.  But for now, do
          * nothing.
          */
      }

      /* flush to make bo busy (so that pwrite() stalls as it should be) */
      if (will_stall && will_be_busy)
         ilo_cp_flush(ilo->cp);
   }

   /* they should specify just an offset and a size */
   assert(level == 0);
   assert(box->y == 0);
   assert(box->z == 0);
   assert(box->height == 1);
   assert(box->depth == 1);
   offset = box->x;
   size = box->width;

   res->bo->pwrite(res->bo, offset, size, data);
}

static void
ilo_transfer_unmap(struct pipe_context *pipe,
                   struct pipe_transfer *transfer)
{
   struct ilo_resource *res = ilo_resource(transfer->resource);

   res->bo->unmap(res->bo);

   pipe_resource_reference(&transfer->resource, NULL);
   FREE(transfer);
}

static void
ilo_transfer_flush_region(struct pipe_context *pipe,
                          struct pipe_transfer *transfer,
                          const struct pipe_box *box)
{
}

static bool
map_resource(struct ilo_context *ilo, struct ilo_resource *res,
             unsigned usage)
{
   struct ilo_screen *is = ilo_screen(res->base.screen);
   bool will_be_busy;
   int err;

   /* simply map unsynchronized */
   if (usage & PIPE_TRANSFER_UNSYNCHRONIZED) {
      err = res->bo->map_unsynchronized(res->bo);
      return !err;
   }

   /*
    * XXX With hardware context support, the bo may be needed by GPU without
    * being referenced by ilo->cp->bo.  We have to flush unconditionally, and
    * that is bad.
    */
   if (ilo->cp->hw_ctx)
      ilo_cp_flush(ilo->cp);

   will_be_busy = ilo->cp->bo->references(ilo->cp->bo, res->bo);

   /* see if we can avoid stalling */
   if (will_be_busy || intel_bo_is_busy(res->bo)) {
      bool will_stall = true;

      if (usage & PIPE_TRANSFER_DISCARD_WHOLE_RESOURCE) {
         /* discard old bo and allocate a new one for mapping */
         if (realloc_bo(res))
            will_stall = false;
      }
      else if (usage & PIPE_TRANSFER_MAP_DIRECTLY) {
         /* nothing we can do */
      }
      else if (usage & PIPE_TRANSFER_FLUSH_EXPLICIT) {
         /*
          * We could allocate and return a system buffer here.  When a region
          * of the buffer is explicitly flushed, we pwrite() the region to a
          * temporary bo and emit pipelined copy blit.
          *
          * For now, do nothing.
          */
      }
      else if (usage & PIPE_TRANSFER_DISCARD_RANGE) {
         /*
          * We could allocate a temporary bo for mapping, and emit pipelined
          * copy blit upon unmapping.
          *
          * For now, do nothing.
          */
      }

      if (will_stall) {
         if (usage & PIPE_TRANSFER_DONTBLOCK)
            return false;

         /* flush to make bo busy (so that map() stalls as it should be) */
         if (will_be_busy)
            ilo_cp_flush(ilo->cp);
      }
   }

   /* prefer map() when there is the last-level cache */
   if (res->tiling == INTEL_TILING_NONE &&
       (is->dev.has_llc || (usage & PIPE_TRANSFER_READ)))
      err = res->bo->map(res->bo, (usage & PIPE_TRANSFER_WRITE));
   else
      err = res->bo->map_gtt(res->bo);

   return !err;
}

static void *
ilo_transfer_map(struct pipe_context *pipe,
                 struct pipe_resource *r,
                 unsigned level,
                 unsigned usage,
                 const struct pipe_box *box,
                 struct pipe_transfer **transfer)
{
   struct ilo_context *ilo = ilo_context(pipe);
   struct ilo_resource *res = ilo_resource(r);
   struct pipe_transfer *xfer;
   void *ptr;
   int x, y;

   xfer = MALLOC_STRUCT(pipe_transfer);
   if (!xfer)
      return NULL;

   if (!map_resource(ilo, res, usage)) {
      FREE(xfer);
      return NULL;
   }

   /* init transfer */
   xfer->resource = NULL;
   pipe_resource_reference(&xfer->resource, &res->base);
   xfer->level = level;
   xfer->usage = usage;
   xfer->box = *box;
   /* stride for a block row, not a texel row */
   xfer->stride = res->bo_stride;

   /*
    * we can walk through layers when the resource is a texture array or
    * when this is the first level of a 3D texture being mapped
    */
   if (res->base.array_size > 1 ||
       (res->base.target == PIPE_TEXTURE_3D && level == 0)) {
      const unsigned qpitch =
         res->slice_offsets[level][1].y - res->slice_offsets[level][0].y;

      assert(qpitch % res->block_height == 0);
      xfer->layer_stride = (qpitch / res->block_height) * xfer->stride;
   }
   else {
      xfer->layer_stride = 0;
   }

   x = res->slice_offsets[level][box->z].x;
   y = res->slice_offsets[level][box->z].y;

   x += box->x;
   y += box->y;

   /* in blocks */
   assert(x % res->block_width == 0 && y % res->block_height == 0);
   x /= res->block_width;
   y /= res->block_height;

   ptr = res->bo->get_virtual(res->bo);
   ptr += y * res->bo_stride + x * res->bo_cpp;

   *transfer = xfer;

   return ptr;
}

static bool
alloc_slice_offsets(struct ilo_resource *res)
{
   int depth, lv;

   /* sum the depths of all levels */
   depth = 0;
   for (lv = 0; lv <= res->base.last_level; lv++)
      depth += u_minify(res->base.depth0, lv);

   /*
    * There are (depth * res->base.array_size) slices.  Either depth is one
    * (non-3D) or res->base.array_size is one (non-array), but it does not
    * matter.
    */
   res->slice_offsets[0] =
      CALLOC(depth * res->base.array_size, sizeof(res->slice_offsets[0][0]));
   if (!res->slice_offsets[0])
      return false;

   /* point to the respective positions in the buffer */
   for (lv = 1; lv <= res->base.last_level; lv++) {
      res->slice_offsets[lv] = res->slice_offsets[lv - 1] +
         u_minify(res->base.depth0, lv - 1) * res->base.array_size;
   }

   return true;
}

static void
free_slice_offsets(struct ilo_resource *res)
{
   int lv;

   FREE(res->slice_offsets[0]);
   for (lv = 0; lv <= res->base.last_level; lv++)
      res->slice_offsets[lv] = NULL;
}

struct layout_tex_info {
   bool compressed;
   int block_width, block_height;
   int align_i, align_j;
   bool array_spacing_full;
   int qpitch;

   struct {
      int w, h, d;
   } sizes[PIPE_MAX_TEXTURE_LEVELS];
};

/**
 * Prepare for texture layout.
 */
static void
layout_tex_init(const struct ilo_resource *res, struct layout_tex_info *info)
{
   struct ilo_screen *is = ilo_screen(res->base.screen);
   const enum intel_tiling_mode tiling = res->tiling;
   const struct pipe_resource *templ = &res->base;
   int last_level, lv;

   memset(info, 0, sizeof(*info));

   info->compressed = util_format_is_compressed(templ->format);
   info->block_width = util_format_get_blockwidth(templ->format);
   info->block_height = util_format_get_blockheight(templ->format);

   /*
    * From the Sandy Bridge PRM, volume 1 part 1, page 113:
    *
    *     "surface format           align_i     align_j
    *      YUV 4:2:2 formats        4           *see below
    *      BC1-5                    4           4
    *      FXT1                     8           4
    *      all other formats        4           *see below"
    *
    *     "- align_j = 4 for any depth buffer
    *      - align_j = 2 for separate stencil buffer
    *      - align_j = 4 for any render target surface is multisampled (4x)
    *      - align_j = 4 for any render target surface with Surface Vertical
    *        Alignment = VALIGN_4
    *      - align_j = 2 for any render target surface with Surface Vertical
    *        Alignment = VALIGN_2
    *      - align_j = 2 for all other render target surface
    *      - align_j = 2 for any sampling engine surface with Surface Vertical
    *        Alignment = VALIGN_2
    *      - align_j = 4 for any sampling engine surface with Surface Vertical
    *        Alignment = VALIGN_4"
    *
    * From the Sandy Bridge PRM, volume 4 part 1, page 86:
    *
    *     "This field (Surface Vertical Alignment) must be set to VALIGN_2 if
    *      the Surface Format is 96 bits per element (BPE)."
    *
    * They can be rephrased as
    *
    *                                  align_i        align_j
    *   compressed formats             block width    block height
    *   PIPE_FORMAT_S8_UINT            4              2
    *   other depth/stencil formats    4              4
    *   4x multisampled                4              4
    *   bpp 96                         4              2
    *   others                         4              2 or 4
    */

   /*
    * From the Ivy Bridge PRM, volume 1 part 1, page 110:
    *
    *     "surface defined by      surface format     align_i     align_j
    *      3DSTATE_DEPTH_BUFFER    D16_UNORM          8           4
    *                              not D16_UNORM      4           4
    *      3DSTATE_STENCIL_BUFFER  N/A                8           8
    *      SURFACE_STATE           BC*, ETC*, EAC*    4           4
    *                              FXT1               8           4
    *                              all others         (set by SURFACE_STATE)"
    *
    * From the Ivy Bridge PRM, volume 4 part 1, page 63:
    *
    *     "- This field (Surface Vertical Aligment) is intended to be set to
    *        VALIGN_4 if the surface was rendered as a depth buffer, for a
    *        multisampled (4x) render target, or for a multisampled (8x)
    *        render target, since these surfaces support only alignment of 4.
    *      - Use of VALIGN_4 for other surfaces is supported, but uses more
    *        memory.
    *      - This field must be set to VALIGN_4 for all tiled Y Render Target
    *        surfaces.
    *      - Value of 1 is not supported for format YCRCB_NORMAL (0x182),
    *        YCRCB_SWAPUVY (0x183), YCRCB_SWAPUV (0x18f), YCRCB_SWAPY (0x190)
    *      - If Number of Multisamples is not MULTISAMPLECOUNT_1, this field
    *        must be set to VALIGN_4."
    *      - VALIGN_4 is not supported for surface format R32G32B32_FLOAT."
    *
    *     "- This field (Surface Horizontal Aligment) is intended to be set to
    *        HALIGN_8 only if the surface was rendered as a depth buffer with
    *        Z16 format or a stencil buffer, since these surfaces support only
    *        alignment of 8.
    *      - Use of HALIGN_8 for other surfaces is supported, but uses more
    *        memory.
    *      - This field must be set to HALIGN_4 if the Surface Format is BC*.
    *      - This field must be set to HALIGN_8 if the Surface Format is
    *        FXT1."
    *
    * They can be rephrased as
    *
    *                                  align_i        align_j
    *  compressed formats              block width    block height
    *  PIPE_FORMAT_Z16_UNORM           8              4
    *  PIPE_FORMAT_S8_UINT             8              8
    *  other depth/stencil formats     4 or 8         4
    *  2x or 4x multisampled           4 or 8         4
    *  tiled Y                         4 or 8         4 (if rt)
    *  PIPE_FORMAT_R32G32B32_FLOAT     4 or 8         2
    *  others                          4 or 8         2 or 4
    */

   if (info->compressed) {
      /* this happens to be the case */
      info->align_i = info->block_width;
      info->align_j = info->block_height;
   }
   else if (util_format_is_depth_or_stencil(templ->format)) {
      if (is->dev.gen >= ILO_GEN(7)) {
         switch (templ->format) {
         case PIPE_FORMAT_Z16_UNORM:
            info->align_i = 8;
            info->align_j = 4;
            break;
         case PIPE_FORMAT_S8_UINT:
            info->align_i = 8;
            info->align_j = 8;
            break;
         default:
            /*
             * From the Ivy Bridge PRM, volume 2 part 1, page 319:
             *
             *     "The 3 LSBs of both offsets (Depth Coordinate Offset Y and
             *      Depth Coordinate Offset X) must be zero to ensure correct
             *      alignment"
             *
             * We will make use of them and setting align_i to 8 help us meet
             * the requirement.
             */
            info->align_i = (templ->last_level > 0) ? 8 : 4;
            info->align_j = 4;
            break;
         }
      }
      else {
         switch (templ->format) {
         case PIPE_FORMAT_S8_UINT:
            info->align_i = 4;
            info->align_j = 2;
            break;
         default:
            info->align_i = 4;
            info->align_j = 4;
            break;
         }
      }
   }
   else {
      const bool valign_4 = (templ->nr_samples > 1) ||
         (is->dev.gen >= ILO_GEN(7) &&
          (templ->bind & PIPE_BIND_RENDER_TARGET) &&
          tiling == INTEL_TILING_Y);

      if (valign_4)
         assert(util_format_get_blocksizebits(templ->format) != 96);

      info->align_i = 4;
      info->align_j = (valign_4) ? 4 : 2;
   }

   /*
    * the fact that align i and j are multiples of block width and height
    * respectively is what makes the size of the bo a multiple of the block
    * size, slices start at block boundaries, and many of the computations
    * work.
    */
   assert(info->align_i % info->block_width == 0);
   assert(info->align_j % info->block_height == 0);