blorp: Handle 2D MSAA array image copies on compute shader

We are passing number of layers as inline parameter register, so figure
out z_pos and write to 2D MSAA array images in compute
shader. We already get component X, Y and sample index, all we needed
was the number of layers.

Ken:
- Use load/store var instead of derefs

Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33905>

src/intel/blorp/blorp_blit.c

@@ -110,7 +110,7 @@ blorp_blit_get_cs_dst_coords(nir_builder *b,
       coord = nir_isub(b, coord, nir_load_var(b, v->v_dst_offset));
 
    assert(!key->persample_msaa_dispatch);
-   return nir_trim_vector(b, coord, 2);
+   return nir_trim_vector(b, coord, key->dst_samples > 1 ? 3 : 2);
 }
 
 /**
@@ -1183,6 +1183,11 @@ blorp_build_nir_shader(struct blorp_context *blorp,
                            const struct blorp_blit_prog_key *key)
 {
    const struct intel_device_info *devinfo = blorp->isl_dev->info;
+
+   /* Compute MSAA is only available on Gfx30+ */
+   if (key->base.shader_pipeline == BLORP_SHADER_PIPELINE_COMPUTE)
+      assert(key->dst_samples == 1 || devinfo->ver >= 30);
+
    nir_def *src_pos, *dst_pos, *color;
 
    /* Sanity checks */
@@ -1479,12 +1484,30 @@ blorp_build_nir_shader(struct blorp_context *blorp,
 
    if (compute) {
       nir_def *store_pos = nir_load_global_invocation_id(&b, 32);
+
+      /* Load sample index for MSAA image store */
+      nir_def *sample_idx = nir_imm_int(&b, 0);
+
+      if (key->dst_samples > 1) {
+         nir_def *num_layers_data =
+            nir_load_inline_data_intel(&b, 1, 32,
+               .base = BLORP_INLINE_PARAM_THREAD_GROUP_ID_Z_DIMENSION);
+
+         nir_def *z_pos = nir_umod(&b, nir_channel(&b, store_pos, 2),
+                                   num_layers_data);
+         sample_idx = nir_idiv(&b, nir_channel(&b, store_pos, 2),
+                               num_layers_data);
+
+         store_pos = nir_vector_insert_imm(&b, store_pos, z_pos, 2);
+      }
       nir_image_store(&b, nir_imm_int(&b, 0),
                       nir_pad_vector_imm_int(&b, store_pos, 0, 4),
-                      nir_imm_int(&b, 0),
+                      sample_idx,
                       nir_pad_vector_imm_int(&b, color, 0, 4),
                       nir_imm_int(&b, 0),
-                      .image_dim = GLSL_SAMPLER_DIM_2D,
+                      .image_dim = key->dst_samples > 1 ?
+                                   GLSL_SAMPLER_DIM_MS:
+                                   GLSL_SAMPLER_DIM_2D,
                       .image_array = true,
                       .access = ACCESS_NON_READABLE);
    } else if (key->dst_usage == ISL_SURF_USAGE_RENDER_TARGET_BIT) {
@@ -1566,7 +1589,7 @@ blorp_get_blit_kernel_cs(struct blorp_batch *batch,
    nir->info.name = ralloc_strdup(nir, "BLORP-gpgpu-blit");
    blorp_set_cs_dims(nir, prog_key->local_y);
 
-   assert(prog_key->rt_samples == 1);
+   assert(batch->blorp->isl_dev->info->ver >= 30 || prog_key->rt_samples == 1);
 
    const struct blorp_program p =
       blorp_compile_cs(blorp, mem_ctx, nir);
@@ -2436,10 +2459,8 @@ blorp_blit_supports_compute(struct blorp_context *blorp,
                             const struct isl_surf *dst_surf,
                             enum isl_aux_usage dst_aux_usage)
 {
-   /* Our compiler doesn't currently support typed image writes with MSAA.
+   /* Platforms < Xe3 doesn't support typed image writes with MSAA. */
-    * Also, our BLORP compute shaders don't handle multisampling cases.
+   if (blorp->isl_dev->info->ver < 30 && dst_surf->samples > 1)
-    */
-   if (dst_surf->samples > 1 || src_surf->samples > 1)
       return false;
 
    if (blorp->isl_dev->info->ver >= 12) {

src/intel/blorp/blorp_clear.c

@@ -169,12 +169,27 @@ blorp_params_get_clear_kernel_cs(struct blorp_batch *batch,
 
    nir_push_if(&b, in_bounds);
 
+   nir_def *sample_idx = nir_imm_int(&b, 0);
+
+   /* Strip sample index from the coord since we are going to send it
+    * separately and include that later in coord as 4th component in
+    * lower_image_sample_index_in_coord lowering pass.
+    */
+   if (blorp->isl_dev->info->ver >= 30 && params->num_samples > 1) {
+      sample_idx = nir_channel(&b, dst_pos, 2);
+      dst_pos = nir_vec3(&b, nir_channel(&b, dst_pos, 0),
+                         nir_channel(&b, dst_pos, 1),
+                         nir_imm_int(&b, 0));
+   }
+
    nir_image_store(&b, nir_imm_int(&b, 0),
                    nir_pad_vector_imm_int(&b, dst_pos, 0, 4),
-                   nir_imm_int(&b, 0),
+                   sample_idx,
                    nir_pad_vector_imm_int(&b, color, 0, 4),
                    nir_imm_int(&b, 0),
-                   .image_dim = GLSL_SAMPLER_DIM_2D,
+                   .image_dim = params->num_samples > 1 ?
+                                GLSL_SAMPLER_DIM_MS :
+                                GLSL_SAMPLER_DIM_2D,
                    .image_array = true,
                    .access = ACCESS_NON_READABLE);
 

src/intel/blorp/blorp_genX_exec_brw.h

@@ -1727,7 +1727,7 @@ blorp_exec_compute(struct blorp_batch *batch, const struct blorp_params *params)
    uint32_t group_x1 = DIV_ROUND_UP(params->x1, cs_prog_data->local_size[0]);
    uint32_t group_y1 = DIV_ROUND_UP(params->y1, cs_prog_data->local_size[1]);
    assert(params->num_layers >= 1);
-   uint32_t group_z1 = params->dst.z_offset + params->num_layers;
+   uint32_t group_z1 = params->num_samples * params->num_layers;
    assert(cs_prog_data->local_size[2] == 1);
 
 #if GFX_VERx10 >= 125
@@ -1791,6 +1791,14 @@ blorp_exec_compute(struct blorp_batch *batch, const struct blorp_params *params)
       .IndirectDataStartAddress       = push_const_offset,
       .IndirectDataLength             = push_const_size,
 
+      /* Send number of layers as inline register parameter to copy 2D MSAA
+       * array image/texture properly.
+       */
+      .EmitInlineParameter            = true,
+      .InlineData                     = {
+         [BLORP_INLINE_PARAM_THREAD_GROUP_ID_Z_DIMENSION / 4 + 0] = params->num_layers,
+      },
+
 #if GFX_VERx10 >= 125
       .GenerateLocalID                = cs_prog_data->generate_local_id != 0,
       .EmitLocal                      = cs_prog_data->generate_local_id,

src/intel/blorp/blorp_priv.h

@@ -35,6 +35,9 @@
 extern "C" {
 #endif
 
+
+#define BLORP_INLINE_PARAM_THREAD_GROUP_ID_Z_DIMENSION (0)
+
 void blorp_init(struct blorp_context *blorp, void *driver_ctx,
                 struct isl_device *isl_dev, const struct blorp_config *config);