xref: /third_party/mesa3d/src/microsoft/vulkan/dzn_nir.c

/*
 * Copyright © Microsoft 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.
 */

#include "dzn_nir.h"

#include "spirv_to_dxil.h"
#include "nir_to_dxil.h"
#include "nir_builder.h"
#include "nir_vulkan.h"

static nir_ssa_def *
dzn_nir_create_bo_desc(nir_builder *b,
                       nir_variable_mode mode,
                       uint32_t desc_set,
                       uint32_t binding,
                       const char *name,
                       unsigned access)
{
   struct glsl_struct_field field = {
      .type = mode == nir_var_mem_ubo ?
              glsl_array_type(glsl_uint_type(), 4096, 4) :
              glsl_uint_type(),
      .name = "dummy_int",
   };
   const struct glsl_type *dummy_type =
      glsl_struct_type(&field, 1, "dummy_type", false);

   nir_variable *var =
      nir_variable_create(b->shader, mode, dummy_type, name);
   var->data.descriptor_set = desc_set;
   var->data.binding = binding;
   var->data.access = access;

   assert(mode == nir_var_mem_ubo || mode == nir_var_mem_ssbo);
   if (mode == nir_var_mem_ubo)
      b->shader->info.num_ubos++;
   else
      b->shader->info.num_ssbos++;

   VkDescriptorType desc_type =
      var->data.mode == nir_var_mem_ubo ?
      VK_DESCRIPTOR_TYPE_UNIFORM_BUFFER :
      VK_DESCRIPTOR_TYPE_STORAGE_BUFFER;
   nir_address_format addr_format = nir_address_format_32bit_index_offset;
   nir_ssa_def *index =
      nir_vulkan_resource_index(b,
                                nir_address_format_num_components(addr_format),
                                nir_address_format_bit_size(addr_format),
                                nir_imm_int(b, 0),
                                .desc_set = desc_set,
                                .binding = binding,
                                .desc_type = desc_type);

   nir_ssa_def *desc =
      nir_load_vulkan_descriptor(b,
                                 nir_address_format_num_components(addr_format),
                                 nir_address_format_bit_size(addr_format),
                                 index,
                                 .desc_type = desc_type);

   return nir_channel(b, desc, 0);
}

nir_shader *
dzn_nir_indirect_draw_shader(enum dzn_indirect_draw_type type)
{
   const char *type_str[] = {
      "draw",
      "draw_count",
      "indexed_draw",
      "indexed_draw_count",
      "draw_triangle_fan",
      "draw_count_triangle_fan",
      "indexed_draw_triangle_fan",
      "indexed_draw_count_triangle_fan",
      "indexed_draw_triangle_fan_prim_restart",
      "indexed_draw_count_triangle_fan_prim_restart",
   };

   assert(type < ARRAY_SIZE(type_str));

   bool indexed = type == DZN_INDIRECT_INDEXED_DRAW ||
                  type == DZN_INDIRECT_INDEXED_DRAW_COUNT ||
                  type == DZN_INDIRECT_INDEXED_DRAW_TRIANGLE_FAN ||
                  type == DZN_INDIRECT_INDEXED_DRAW_COUNT_TRIANGLE_FAN ||
                  type == DZN_INDIRECT_INDEXED_DRAW_TRIANGLE_FAN_PRIM_RESTART ||
                  type == DZN_INDIRECT_INDEXED_DRAW_COUNT_TRIANGLE_FAN_PRIM_RESTART;
   bool triangle_fan = type == DZN_INDIRECT_DRAW_TRIANGLE_FAN ||
                       type == DZN_INDIRECT_DRAW_COUNT_TRIANGLE_FAN ||
                       type == DZN_INDIRECT_INDEXED_DRAW_TRIANGLE_FAN ||
                       type == DZN_INDIRECT_INDEXED_DRAW_COUNT_TRIANGLE_FAN ||
                       type == DZN_INDIRECT_INDEXED_DRAW_TRIANGLE_FAN_PRIM_RESTART ||
                       type == DZN_INDIRECT_INDEXED_DRAW_COUNT_TRIANGLE_FAN_PRIM_RESTART;
   bool indirect_count = type == DZN_INDIRECT_DRAW_COUNT ||
                         type == DZN_INDIRECT_INDEXED_DRAW_COUNT ||
                         type == DZN_INDIRECT_DRAW_COUNT_TRIANGLE_FAN ||
                         type == DZN_INDIRECT_INDEXED_DRAW_COUNT_TRIANGLE_FAN ||
                         type == DZN_INDIRECT_INDEXED_DRAW_COUNT_TRIANGLE_FAN_PRIM_RESTART;
   bool prim_restart = type == DZN_INDIRECT_INDEXED_DRAW_TRIANGLE_FAN_PRIM_RESTART ||
                       type == DZN_INDIRECT_INDEXED_DRAW_COUNT_TRIANGLE_FAN_PRIM_RESTART;
   nir_builder b =
      nir_builder_init_simple_shader(MESA_SHADER_COMPUTE,
                                     dxil_get_nir_compiler_options(),
                                     "dzn_meta_indirect_%s()",
                                     type_str[type]);
   b.shader->info.internal = true;

   nir_ssa_def *params_desc =
      dzn_nir_create_bo_desc(&b, nir_var_mem_ubo, 0, 0, "params", 0);
   nir_ssa_def *draw_buf_desc =
      dzn_nir_create_bo_desc(&b, nir_var_mem_ssbo, 0, 1, "draw_buf", ACCESS_NON_WRITEABLE);
   nir_ssa_def *exec_buf_desc =
      dzn_nir_create_bo_desc(&b, nir_var_mem_ssbo, 0, 2, "exec_buf", ACCESS_NON_READABLE);

   unsigned params_size;
   if (triangle_fan)
      params_size = sizeof(struct dzn_indirect_draw_triangle_fan_rewrite_params);
   else
      params_size = sizeof(struct dzn_indirect_draw_rewrite_params);

   nir_ssa_def *params =
      nir_load_ubo(&b, params_size / 4, 32,
                   params_desc, nir_imm_int(&b, 0),
                   .align_mul = 4, .align_offset = 0, .range_base = 0, .range = ~0);

   nir_ssa_def *draw_stride = nir_channel(&b, params, 0);
   nir_ssa_def *exec_stride =
      triangle_fan ?
      nir_imm_int(&b, sizeof(struct dzn_indirect_triangle_fan_draw_exec_params)) :
      nir_imm_int(&b, sizeof(struct dzn_indirect_draw_exec_params));
   nir_ssa_def *index =
      nir_channel(&b, nir_load_global_invocation_id(&b, 32), 0);

   if (indirect_count) {
      nir_ssa_def *count_buf_desc =
         dzn_nir_create_bo_desc(&b, nir_var_mem_ssbo, 0, 3, "count_buf", ACCESS_NON_WRITEABLE);

      nir_ssa_def *draw_count =
         nir_load_ssbo(&b, 1, 32, count_buf_desc, nir_imm_int(&b, 0), .align_mul = 4);

      nir_push_if(&b, nir_ieq(&b, index, nir_imm_int(&b, 0)));
      nir_store_ssbo(&b, draw_count, exec_buf_desc, nir_imm_int(&b, 0),
                    .write_mask = 0x1, .access = ACCESS_NON_READABLE,
                    .align_mul = 16);
      nir_pop_if(&b, NULL);

      nir_push_if(&b, nir_ult(&b, index, draw_count));
   }

   nir_ssa_def *draw_offset = nir_imul(&b, draw_stride, index);

   /* The first entry contains the indirect count */
   nir_ssa_def *exec_offset =
      indirect_count ?
      nir_imul(&b, exec_stride, nir_iadd_imm(&b, index, 1)) :
      nir_imul(&b, exec_stride, index);

   nir_ssa_def *draw_info1 =
      nir_load_ssbo(&b, 4, 32, draw_buf_desc, draw_offset, .align_mul = 4);
   nir_ssa_def *draw_info2 =
      indexed ?
      nir_load_ssbo(&b, 1, 32, draw_buf_desc,
                    nir_iadd_imm(&b, draw_offset, 16), .align_mul = 4) :
      nir_imm_int(&b, 0);

   nir_ssa_def *first_vertex = nir_channel(&b, draw_info1, indexed ? 3 : 2);
   nir_ssa_def *base_instance =
      indexed ? draw_info2 : nir_channel(&b, draw_info1, 3);

   nir_ssa_def *exec_vals[8] = {
      first_vertex,
      base_instance,
      index,
   };

   if (triangle_fan) {
      /* Patch {vertex,index}_count and first_index */
      nir_ssa_def *triangle_count =
         nir_usub_sat(&b, nir_channel(&b, draw_info1, 0), nir_imm_int(&b, 2));
      exec_vals[3] = nir_imul_imm(&b, triangle_count, 3);
      exec_vals[4] = nir_channel(&b, draw_info1, 1);
      exec_vals[5] = nir_imm_int(&b, 0);
      exec_vals[6] = first_vertex;
      exec_vals[7] = base_instance;

      nir_ssa_def *triangle_fan_exec_buf_desc =
         dzn_nir_create_bo_desc(&b, nir_var_mem_ssbo, 0, 4,
                                "triangle_fan_exec_buf",
                                ACCESS_NON_READABLE);
      nir_ssa_def *triangle_fan_index_buf_stride = nir_channel(&b, params, 1);
      nir_ssa_def *triangle_fan_index_buf_addr_lo =
         nir_iadd(&b, nir_channel(&b, params, 2),
                  nir_imul(&b, triangle_fan_index_buf_stride, index));

      nir_ssa_def *triangle_fan_exec_vals[9] = { 0 };
      uint32_t triangle_fan_exec_param_count = 0;
      nir_ssa_def *addr_lo_overflow =
         nir_ult(&b, triangle_fan_index_buf_addr_lo, nir_channel(&b, params, 2));
      nir_ssa_def *triangle_fan_index_buf_addr_hi =
         nir_iadd(&b, nir_channel(&b, params, 3),
                  nir_bcsel(&b, addr_lo_overflow, nir_imm_int(&b, 1), nir_imm_int(&b, 0)));

      triangle_fan_exec_vals[triangle_fan_exec_param_count++] = triangle_fan_index_buf_addr_lo;
      triangle_fan_exec_vals[triangle_fan_exec_param_count++] = triangle_fan_index_buf_addr_hi;

      if (prim_restart) {
         triangle_fan_exec_vals[triangle_fan_exec_param_count++] = nir_channel(&b, draw_info1, 2);
         triangle_fan_exec_vals[triangle_fan_exec_param_count++] = nir_channel(&b, draw_info1, 0);
         uint32_t index_count_offset =
            offsetof(struct dzn_indirect_triangle_fan_draw_exec_params, indexed_draw.index_count);
         nir_ssa_def *exec_buf_start =
            nir_load_ubo(&b, 2, 32,
                         params_desc, nir_imm_int(&b, 16),
                         .align_mul = 4, .align_offset = 0, .range_base = 0, .range = ~0);
         nir_ssa_def *exec_buf_start_lo =
            nir_iadd(&b, nir_imm_int(&b, index_count_offset),
                     nir_iadd(&b, nir_channel(&b, exec_buf_start, 0),
                              nir_imul(&b, exec_stride, index)));
         addr_lo_overflow = nir_ult(&b, exec_buf_start_lo, nir_channel(&b, exec_buf_start, 0));
         nir_ssa_def *exec_buf_start_hi =
            nir_iadd(&b, nir_channel(&b, exec_buf_start, 0),
                     nir_bcsel(&b, addr_lo_overflow, nir_imm_int(&b, 1), nir_imm_int(&b, 0)));
         triangle_fan_exec_vals[triangle_fan_exec_param_count++] = exec_buf_start_lo;
         triangle_fan_exec_vals[triangle_fan_exec_param_count++] = exec_buf_start_hi;
         triangle_fan_exec_vals[triangle_fan_exec_param_count++] = nir_imm_int(&b, 1);
      } else {
         triangle_fan_exec_vals[triangle_fan_exec_param_count++] =
            indexed ? nir_channel(&b, draw_info1, 2) : nir_imm_int(&b, 0);
         triangle_fan_exec_vals[triangle_fan_exec_param_count++] =
            triangle_count;
      }
      triangle_fan_exec_vals[triangle_fan_exec_param_count++] = nir_imm_int(&b, 1);
      triangle_fan_exec_vals[triangle_fan_exec_param_count++] = nir_imm_int(&b, 1);

      unsigned rewrite_index_exec_params =
         prim_restart ?
         sizeof(struct dzn_indirect_triangle_fan_prim_restart_rewrite_index_exec_params) :
         sizeof(struct dzn_indirect_triangle_fan_rewrite_index_exec_params);
      nir_ssa_def *triangle_fan_exec_stride =
         nir_imm_int(&b, rewrite_index_exec_params);
      nir_ssa_def *triangle_fan_exec_offset =
         nir_imul(&b, triangle_fan_exec_stride, index);

      for (uint32_t i = 0; i < triangle_fan_exec_param_count; i += 4) {
         unsigned comps = MIN2(triangle_fan_exec_param_count - i, 4);
         uint32_t mask = (1 << comps) - 1;

         nir_store_ssbo(&b, nir_vec(&b, &triangle_fan_exec_vals[i], comps),
                        triangle_fan_exec_buf_desc,
                        nir_iadd_imm(&b, triangle_fan_exec_offset, i * 4),
                        .write_mask = mask, .access = ACCESS_NON_READABLE, .align_mul = 4);
      }

      nir_ssa_def *ibview_vals[] = {
         triangle_fan_index_buf_addr_lo,
         triangle_fan_index_buf_addr_hi,
         triangle_fan_index_buf_stride,
         nir_imm_int(&b, DXGI_FORMAT_R32_UINT),
      };

      nir_store_ssbo(&b, nir_vec(&b, ibview_vals, ARRAY_SIZE(ibview_vals)),
                     exec_buf_desc, exec_offset,
                     .write_mask = 0xf, .access = ACCESS_NON_READABLE, .align_mul = 16);
      exec_offset = nir_iadd_imm(&b, exec_offset, ARRAY_SIZE(ibview_vals) * 4);
   } else {
      exec_vals[3] = nir_channel(&b, draw_info1, 0);
      exec_vals[4] = nir_channel(&b, draw_info1, 1);
      exec_vals[5] = nir_channel(&b, draw_info1, 2);
      exec_vals[6] = nir_channel(&b, draw_info1, 3);
      exec_vals[7] = draw_info2;
   }

   nir_store_ssbo(&b, nir_vec(&b, exec_vals, 4),
                  exec_buf_desc, exec_offset,
                  .write_mask = 0xf, .access = ACCESS_NON_READABLE, .align_mul = 16);
   nir_store_ssbo(&b, nir_vec(&b, &exec_vals[4], 4),
                  exec_buf_desc, nir_iadd_imm(&b, exec_offset, 16),
                  .write_mask = 0xf, .access = ACCESS_NON_READABLE, .align_mul = 16);

   if (indirect_count)
      nir_pop_if(&b, NULL);

   return b.shader;
}

nir_shader *
dzn_nir_triangle_fan_prim_restart_rewrite_index_shader(uint8_t old_index_size)
{
   assert(old_index_size == 2 || old_index_size == 4);

   nir_builder b =
      nir_builder_init_simple_shader(MESA_SHADER_COMPUTE,
                                     dxil_get_nir_compiler_options(),
                                     "dzn_meta_triangle_prim_rewrite_index(old_index_size=%d)",
                                     old_index_size);
   b.shader->info.internal = true;

   nir_ssa_def *params_desc =
      dzn_nir_create_bo_desc(&b, nir_var_mem_ubo, 0, 0, "params", 0);
   nir_ssa_def *new_index_buf_desc =
      dzn_nir_create_bo_desc(&b, nir_var_mem_ssbo, 0, 1,
                             "new_index_buf", ACCESS_NON_READABLE);
   nir_ssa_def *old_index_buf_desc =
      dzn_nir_create_bo_desc(&b, nir_var_mem_ssbo, 0, 2,
                             "old_index_buf", ACCESS_NON_WRITEABLE);
   nir_ssa_def *new_index_count_ptr_desc =
      dzn_nir_create_bo_desc(&b, nir_var_mem_ssbo, 0, 3,
                             "new_index_count_ptr", ACCESS_NON_READABLE);

   nir_ssa_def *params =
      nir_load_ubo(&b, sizeof(struct dzn_triangle_fan_prim_restart_rewrite_index_params) / 4, 32,
                   params_desc, nir_imm_int(&b, 0),
                   .align_mul = 4, .align_offset = 0, .range_base = 0, .range = ~0);

   nir_ssa_def *prim_restart_val =
      nir_imm_int(&b, old_index_size == 2 ? 0xffff : 0xffffffff);
   nir_variable *old_index_ptr_var =
      nir_local_variable_create(b.impl, glsl_uint_type(), "old_index_ptr_var");
   nir_ssa_def *old_index_ptr = nir_channel(&b, params, 0);
   nir_store_var(&b, old_index_ptr_var, old_index_ptr, 1);
   nir_variable *new_index_ptr_var =
      nir_local_variable_create(b.impl, glsl_uint_type(), "new_index_ptr_var");
   nir_store_var(&b, new_index_ptr_var, nir_imm_int(&b, 0), 1);
   nir_ssa_def *old_index_count = nir_channel(&b, params, 1);
   nir_variable *index0_var =
      nir_local_variable_create(b.impl, glsl_uint_type(), "index0_var");
   nir_store_var(&b, index0_var, prim_restart_val, 1);

   /*
    * Filter out all primitive-restart magic values, and generate a triangle list
    * from the triangle fan definition.
    *
    * Basically:
    *
    * new_index_ptr = 0;
    * index0 = restart_prim_value; // 0xffff or 0xffffffff
    * for (old_index_ptr = firstIndex; old_index_ptr < indexCount;) {
    *    // If we have no starting-point we need at least 3 vertices,
    *    // otherwise we can do with two. If there's not enough vertices
    *    // to form a primitive, we just bail out.
    *    min_indices = index0 == restart_prim_value ? 3 : 2;
    *    if (old_index_ptr + min_indices > firstIndex + indexCount)
    *       break;
    *
    *    if (index0 == restart_prim_value) {
    *       // No starting point, skip all entries until we have a
    *       // non-primitive-restart value
    *       index0 = old_index_buf[old_index_ptr++];
    *       continue;
    *    }
    *
    *    // If at least one index contains the primitive-restart pattern,
         // ignore this triangle, and skip the unused entries
    *    if (old_index_buf[old_index_ptr + 1] == restart_prim_value) {
    *       old_index_ptr += 2;
    *       continue;
    *    }
    *    if (old_index_buf[old_index_ptr] == restart_prim_value) {
    *       old_index_ptr++;
    *       continue;
    *    }
    *
    *    // We have a valid primitive, queue it to the new index buffer
    *    new_index_buf[new_index_ptr++] = old_index_buf[old_index_ptr];
    *    new_index_buf[new_index_ptr++] = old_index_buf[old_index_ptr + 1];
    *    new_index_buf[new_index_ptr++] = index0;
    * }
    *
    * expressed in NIR, which admitedly is not super easy to grasp with.
    * TODO: Might be a good thing to use use the CL compiler we have and turn
    * those shaders into CL kernels.
    */
   nir_push_loop(&b);

   old_index_ptr = nir_load_var(&b, old_index_ptr_var);
   nir_ssa_def *index0 = nir_load_var(&b, index0_var);

   nir_ssa_def *read_index_count =
      nir_bcsel(&b, nir_ieq(&b, index0, prim_restart_val),
                nir_imm_int(&b, 3), nir_imm_int(&b, 2));
   nir_push_if(&b, nir_ult(&b, old_index_count, nir_iadd(&b, old_index_ptr, read_index_count)));
   nir_jump(&b, nir_jump_break);
   nir_pop_if(&b, NULL);

   nir_ssa_def *old_index_offset =
      nir_imul_imm(&b, old_index_ptr, old_index_size);

   nir_push_if(&b, nir_ieq(&b, index0, prim_restart_val));
   nir_ssa_def *index_val =
      nir_load_ssbo(&b, 1, 32, old_index_buf_desc,
                    old_index_size == 2 ? nir_iand_imm(&b, old_index_offset, ~3ULL) : old_index_offset,
                    .align_mul = 4);
   if (old_index_size == 2) {
     index_val = nir_bcsel(&b, nir_test_mask(&b, old_index_offset, 0x2),
                           nir_ushr_imm(&b, index_val, 16),
                           nir_iand_imm(&b, index_val, 0xffff));
   }

   nir_store_var(&b, index0_var, index_val, 1);
   nir_store_var(&b, old_index_ptr_var, nir_iadd_imm(&b, old_index_ptr, 1), 1);
   nir_jump(&b, nir_jump_continue);
   nir_pop_if(&b, NULL);

   nir_ssa_def *index12 =
      nir_load_ssbo(&b, 2, 32, old_index_buf_desc,
                    old_index_size == 2 ? nir_iand_imm(&b, old_index_offset, ~3ULL) : old_index_offset,
                    .align_mul = 4);
   if (old_index_size == 2) {
      nir_ssa_def *indices[] = {
         nir_iand_imm(&b, nir_channel(&b, index12, 0), 0xffff),
         nir_ushr_imm(&b, nir_channel(&b, index12, 0), 16),
         nir_iand_imm(&b, nir_channel(&b, index12, 1), 0xffff),
      };

      index12 = nir_bcsel(&b, nir_test_mask(&b, old_index_offset, 0x2),
                          nir_vec2(&b, indices[1], indices[2]),
                          nir_vec2(&b, indices[0], indices[1]));
   }

   nir_push_if(&b, nir_ieq(&b, nir_channel(&b, index12, 1), prim_restart_val));
   nir_store_var(&b, old_index_ptr_var, nir_iadd_imm(&b, old_index_ptr, 2), 1);
   nir_store_var(&b, index0_var, prim_restart_val, 1);
   nir_jump(&b, nir_jump_continue);
   nir_push_else(&b, NULL);
   nir_store_var(&b, old_index_ptr_var, nir_iadd_imm(&b, old_index_ptr, 1), 1);
   nir_push_if(&b, nir_ieq(&b, nir_channel(&b, index12, 0), prim_restart_val));
   nir_store_var(&b, index0_var, prim_restart_val, 1);
   nir_jump(&b, nir_jump_continue);
   nir_push_else(&b, NULL);
   nir_ssa_def *new_indices =
      nir_vec3(&b, nir_channel(&b, index12, 0), nir_channel(&b, index12, 1), index0);
   nir_ssa_def *new_index_ptr = nir_load_var(&b, new_index_ptr_var);
   nir_ssa_def *new_index_offset = nir_imul_imm(&b, new_index_ptr, sizeof(uint32_t));
   nir_store_ssbo(&b, new_indices, new_index_buf_desc,
                  new_index_offset,
                  .write_mask = 7, .access = ACCESS_NON_READABLE, .align_mul = 4);
   nir_store_var(&b, new_index_ptr_var, nir_iadd_imm(&b, new_index_ptr, 3), 1);
   nir_pop_if(&b, NULL);
   nir_pop_if(&b, NULL);
   nir_pop_loop(&b, NULL);

   nir_store_ssbo(&b, nir_load_var(&b, new_index_ptr_var),
                  new_index_count_ptr_desc, nir_imm_int(&b, 0),
                  .write_mask = 1, .access = ACCESS_NON_READABLE, .align_mul = 4);

   return b.shader;
}

nir_shader *
dzn_nir_triangle_fan_rewrite_index_shader(uint8_t old_index_size)
{
   assert(old_index_size == 0 || old_index_size == 2 || old_index_size == 4);

   nir_builder b =
      nir_builder_init_simple_shader(MESA_SHADER_COMPUTE,
                                     dxil_get_nir_compiler_options(),
                                     "dzn_meta_triangle_rewrite_index(old_index_size=%d)",
                                     old_index_size);
   b.shader->info.internal = true;

   nir_ssa_def *params_desc =
      dzn_nir_create_bo_desc(&b, nir_var_mem_ubo, 0, 0, "params", 0);
   nir_ssa_def *new_index_buf_desc =
      dzn_nir_create_bo_desc(&b, nir_var_mem_ssbo, 0, 1,
                             "new_index_buf", ACCESS_NON_READABLE);

   nir_ssa_def *old_index_buf_desc = NULL;
   if (old_index_size > 0) {
      old_index_buf_desc =
         dzn_nir_create_bo_desc(&b, nir_var_mem_ssbo, 0, 2,
                                "old_index_buf", ACCESS_NON_WRITEABLE);
   }

   nir_ssa_def *params =
      nir_load_ubo(&b, sizeof(struct dzn_triangle_fan_rewrite_index_params) / 4, 32,
                   params_desc, nir_imm_int(&b, 0),
                   .align_mul = 4, .align_offset = 0, .range_base = 0, .range = ~0);

   nir_ssa_def *triangle = nir_channel(&b, nir_load_global_invocation_id(&b, 32), 0);
   nir_ssa_def *new_indices;

   if (old_index_size > 0) {
      nir_ssa_def *old_first_index = nir_channel(&b, params, 0);
      nir_ssa_def *old_index0_offset =
         nir_imul_imm(&b, old_first_index, old_index_size);
      nir_ssa_def *old_index1_offset =
         nir_imul_imm(&b, nir_iadd(&b, nir_iadd_imm(&b, triangle, 1), old_first_index),
                      old_index_size);

      nir_ssa_def *old_index0 =
         nir_load_ssbo(&b, 1, 32, old_index_buf_desc,
                       old_index_size == 2 ? nir_iand_imm(&b, old_index0_offset, ~3ULL) : old_index0_offset,
                       .align_mul = 4);

      if (old_index_size == 2) {
        old_index0 = nir_bcsel(&b, nir_test_mask(&b, old_index0_offset, 0x2),
                               nir_ushr_imm(&b, old_index0, 16),
                               nir_iand_imm(&b, old_index0, 0xffff));
      }

      nir_ssa_def *old_index12 =
         nir_load_ssbo(&b, 2, 32, old_index_buf_desc,
                       old_index_size == 2 ? nir_iand_imm(&b, old_index1_offset, ~3ULL) : old_index1_offset,
                       .align_mul = 4);
      if (old_index_size == 2) {
         nir_ssa_def *indices[] = {
            nir_iand_imm(&b, nir_channel(&b, old_index12, 0), 0xffff),
            nir_ushr_imm(&b, nir_channel(&b, old_index12, 0), 16),
            nir_iand_imm(&b, nir_channel(&b, old_index12, 1), 0xffff),
         };

         old_index12 = nir_bcsel(&b, nir_test_mask(&b, old_index1_offset, 0x2),
                                 nir_vec2(&b, indices[1], indices[2]),
                                 nir_vec2(&b, indices[0], indices[1]));
      }

      /* TODO: VK_PROVOKING_VERTEX_MODE_LAST_VERTEX_EXT */
      new_indices =
         nir_vec3(&b, nir_channel(&b, old_index12, 0),
                  nir_channel(&b, old_index12, 1), old_index0);
   } else {
      new_indices =
         nir_vec3(&b,
                  nir_iadd_imm(&b, triangle, 1),
                  nir_iadd_imm(&b, triangle, 2),
                  nir_imm_int(&b, 0));
   }

   nir_ssa_def *new_index_offset =
      nir_imul_imm(&b, triangle, 4 * 3);

   nir_store_ssbo(&b, new_indices, new_index_buf_desc,
                  new_index_offset,
                  .write_mask = 7, .access = ACCESS_NON_READABLE, .align_mul = 4);

   return b.shader;
}

nir_shader *
dzn_nir_blit_vs(void)
{
   nir_builder b =
      nir_builder_init_simple_shader(MESA_SHADER_VERTEX,
                                     dxil_get_nir_compiler_options(),
                                     "dzn_meta_blit_vs()");
   b.shader->info.internal = true;

   nir_ssa_def *params_desc =
      dzn_nir_create_bo_desc(&b, nir_var_mem_ubo, 0, 0, "params", 0);

   nir_variable *out_pos =
      nir_variable_create(b.shader, nir_var_shader_out, glsl_vec4_type(),
                          "gl_Position");
   out_pos->data.location = VARYING_SLOT_POS;
   out_pos->data.driver_location = 0;

   nir_variable *out_coords =
      nir_variable_create(b.shader, nir_var_shader_out, glsl_vec_type(3),
                          "coords");
   out_coords->data.location = VARYING_SLOT_TEX0;
   out_coords->data.driver_location = 1;

   nir_ssa_def *vertex = nir_load_vertex_id(&b);
   nir_ssa_def *base = nir_imul_imm(&b, vertex, 4 * sizeof(float));
   nir_ssa_def *coords =
      nir_load_ubo(&b, 4, 32, params_desc, base,
                   .align_mul = 16, .align_offset = 0, .range_base = 0, .range = ~0);
   nir_ssa_def *pos =
      nir_vec4(&b, nir_channel(&b, coords, 0), nir_channel(&b, coords, 1),
               nir_imm_float(&b, 0.0), nir_imm_float(&b, 1.0));
   nir_ssa_def *z_coord =
      nir_load_ubo(&b, 1, 32, params_desc, nir_imm_int(&b, 4 * 4 * sizeof(float)),
                   .align_mul = 64, .align_offset = 0, .range_base = 0, .range = ~0);
   coords = nir_vec3(&b, nir_channel(&b, coords, 2), nir_channel(&b, coords, 3), z_coord);

   nir_store_var(&b, out_pos, pos, 0xf);
   nir_store_var(&b, out_coords, coords, 0x7);
   return b.shader;
}

nir_shader *
dzn_nir_blit_fs(const struct dzn_nir_blit_info *info)
{
   bool ms = info->src_samples > 1;
   nir_alu_type nir_out_type =
      nir_get_nir_type_for_glsl_base_type(info->out_type);
   uint32_t coord_comps =
      glsl_get_sampler_dim_coordinate_components(info->sampler_dim) +
      info->src_is_array;

   nir_builder b =
      nir_builder_init_simple_shader(MESA_SHADER_FRAGMENT,
                                     dxil_get_nir_compiler_options(),
                                     "dzn_meta_blit_fs()");
   b.shader->info.internal = true;

   const struct glsl_type *tex_type =
      glsl_texture_type(info->sampler_dim, info->src_is_array, info->out_type);
   nir_variable *tex_var =
      nir_variable_create(b.shader, nir_var_uniform, tex_type, "texture");
   nir_deref_instr *tex_deref = nir_build_deref_var(&b, tex_var);

   nir_variable *pos_var =
      nir_variable_create(b.shader, nir_var_shader_in,
                          glsl_vector_type(GLSL_TYPE_FLOAT, 4),
                          "gl_FragCoord");
   pos_var->data.location = VARYING_SLOT_POS;
   pos_var->data.driver_location = 0;

   nir_variable *coord_var =
      nir_variable_create(b.shader, nir_var_shader_in,
                          glsl_vector_type(GLSL_TYPE_FLOAT, 3),
                          "coord");
   coord_var->data.location = VARYING_SLOT_TEX0;
   coord_var->data.driver_location = 1;
   nir_ssa_def *coord =
      nir_channels(&b, nir_load_var(&b, coord_var), (1 << coord_comps) - 1);

   uint32_t out_comps =
      (info->loc == FRAG_RESULT_DEPTH || info->loc == FRAG_RESULT_STENCIL) ? 1 : 4;
   nir_variable *out =
      nir_variable_create(b.shader, nir_var_shader_out,
                          glsl_vector_type(info->out_type, out_comps),
                          "out");
   out->data.location = info->loc;

   nir_ssa_def *res = NULL;

   if (info->resolve) {
      /* When resolving a float type, we need to calculate the average of all
       * samples. For integer resolve, Vulkan says that one sample should be
       * chosen without telling which. Let's just pick the first one in that
       * case.
       */

      unsigned nsamples = info->out_type == GLSL_TYPE_FLOAT ?
                          info->src_samples : 1;
      for (unsigned s = 0; s < nsamples; s++) {
         nir_tex_instr *tex = nir_tex_instr_create(b.shader, 4);

         tex->op = nir_texop_txf_ms;
         tex->dest_type = nir_out_type;
         tex->texture_index = 0;
         tex->is_array = info->src_is_array;
         tex->sampler_dim = info->sampler_dim;

         tex->src[0].src_type = nir_tex_src_coord;
         tex->src[0].src = nir_src_for_ssa(nir_f2i32(&b, coord));
         tex->coord_components = coord_comps;

         tex->src[1].src_type = nir_tex_src_ms_index;
         tex->src[1].src = nir_src_for_ssa(nir_imm_int(&b, s));

         tex->src[2].src_type = nir_tex_src_lod;
         tex->src[2].src = nir_src_for_ssa(nir_imm_int(&b, 0));

         tex->src[3].src_type = nir_tex_src_texture_deref;
         tex->src[3].src = nir_src_for_ssa(&tex_deref->dest.ssa);

         nir_ssa_dest_init(&tex->instr, &tex->dest, 4, 32, NULL);

         nir_builder_instr_insert(&b, &tex->instr);
         res = res ? nir_fadd(&b, res, &tex->dest.ssa) : &tex->dest.ssa;
      }

      if (nsamples > 1) {
         unsigned type_sz = nir_alu_type_get_type_size(nir_out_type);
         res = nir_fmul(&b, res, nir_imm_floatN_t(&b, 1.0f / nsamples, type_sz));
      }
   } else {
      nir_tex_instr *tex =
         nir_tex_instr_create(b.shader, ms ? 4 : 3);

      tex->dest_type = nir_out_type;
      tex->is_array = info->src_is_array;
      tex->sampler_dim = info->sampler_dim;

      if (ms) {
         tex->op = nir_texop_txf_ms;

         tex->src[0].src_type = nir_tex_src_coord;
         tex->src[0].src = nir_src_for_ssa(nir_f2i32(&b, coord));
         tex->coord_components = coord_comps;

         tex->src[1].src_type = nir_tex_src_ms_index;
         tex->src[1].src = nir_src_for_ssa(nir_load_sample_id(&b));

         tex->src[2].src_type = nir_tex_src_lod;
         tex->src[2].src = nir_src_for_ssa(nir_imm_int(&b, 0));

         tex->src[3].src_type = nir_tex_src_texture_deref;
         tex->src[3].src = nir_src_for_ssa(&tex_deref->dest.ssa);
      } else {
         nir_variable *sampler_var =
            nir_variable_create(b.shader, nir_var_uniform, glsl_bare_sampler_type(), "sampler");
         nir_deref_instr *sampler_deref = nir_build_deref_var(&b, sampler_var);

         tex->op = nir_texop_tex;
         tex->sampler_index = 0;

         tex->src[0].src_type = nir_tex_src_coord;
         tex->src[0].src = nir_src_for_ssa(coord);
         tex->coord_components = coord_comps;

         tex->src[1].src_type = nir_tex_src_texture_deref;
         tex->src[1].src = nir_src_for_ssa(&tex_deref->dest.ssa);

         tex->src[2].src_type = nir_tex_src_sampler_deref;
         tex->src[2].src = nir_src_for_ssa(&sampler_deref->dest.ssa);
      }

      nir_ssa_dest_init(&tex->instr, &tex->dest, 4, 32, NULL);
      nir_builder_instr_insert(&b, &tex->instr);
      res = &tex->dest.ssa;
   }

   nir_store_var(&b, out, nir_channels(&b, res, (1 << out_comps) - 1), 0xf);

   return b.shader;
}