anv: don't forget to add scratch buffer to BO list

We reference the scratch BO using a bindless index in the command
streamer instructions, but we forgot to add them to the BO list.

v2: Make use of pipeline reloc list (Jason)

v3: Don't add NULL BOs to the reloc list (Lionel)

v4: Don't add BOs twice to reloc list when dealing with addresses
    (Lionel)

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: eeeea5cb873b40 ("anv: Add support for scratch on XeHP")
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13544>
(cherry picked from commit 46c37c8600e0a48b8aa98f5635f7ae96db555fea)

2 files changed, 23 insertions, 7 deletions

diff --git a/.pick_status.json b/.pick_status.json
index db4ccb7a9bb..65ae497ff06 100644
--- a/.pick_status.json
+++ b/.pick_status.json
@@ -22,7 +22,7 @@
         "description": "anv: don't forget to add scratch buffer to BO list",
         "nominated": true,
         "nomination_type": 1,
-        "resolution": 0,
+        "resolution": 1,
         "main_sha": null,
         "because_sha": "eeeea5cb873b40418c738e25e4032fb1fd9b7ac9"
     },
diff --git a/src/intel/vulkan/genX_pipeline.c b/src/intel/vulkan/genX_pipeline.c
index 7675475e17f..60c43a69eb6 100644
--- a/src/intel/vulkan/genX_pipeline.c
+++ b/src/intel/vulkan/genX_pipeline.c
@@ -1743,8 +1743,18 @@ get_scratch_space(const struct anv_shader_bin *bin)
 
 static UNUSED uint32_t
 get_scratch_surf(struct anv_pipeline *pipeline,
+                 gl_shader_stage stage,
                  const struct anv_shader_bin *bin)
 {
+   if (bin->prog_data->total_scratch == 0)
+      return 0;
+
+   struct anv_bo *bo =
+      anv_scratch_pool_alloc(pipeline->device,
+                             &pipeline->device->scratch_pool,
+                             stage, bin->prog_data->total_scratch);
+   anv_reloc_list_add_bo(pipeline->batch.relocs,
+                         pipeline->batch.alloc, bo);
    return anv_scratch_pool_get_surf(pipeline->device,
                                     &pipeline->device->scratch_pool,
                                     bin->prog_data->total_scratch) >> 4;
@@ -1821,7 +1831,8 @@ emit_3dstate_vs(struct anv_graphics_pipeline *pipeline)
 #endif
 
 #if GFX_VERx10 >= 125
-      vs.ScratchSpaceBuffer = get_scratch_surf(&pipeline->base, vs_bin);
+      vs.ScratchSpaceBuffer =
+         get_scratch_surf(&pipeline->base, MESA_SHADER_VERTEX, vs_bin);
 #else
       vs.PerThreadScratchSpace   = get_scratch_space(vs_bin);
       vs.ScratchSpaceBasePointer =
@@ -1882,7 +1893,8 @@ emit_3dstate_hs_te_ds(struct anv_graphics_pipeline *pipeline,
 #endif
 
 #if GFX_VERx10 >= 125
-      hs.ScratchSpaceBuffer = get_scratch_surf(&pipeline->base, tcs_bin);
+      hs.ScratchSpaceBuffer =
+         get_scratch_surf(&pipeline->base, MESA_SHADER_TESS_CTRL, tcs_bin);
 #else
       hs.PerThreadScratchSpace = get_scratch_space(tcs_bin);
       hs.ScratchSpaceBasePointer =
@@ -1966,7 +1978,8 @@ emit_3dstate_hs_te_ds(struct anv_graphics_pipeline *pipeline,
 #endif
 
 #if GFX_VERx10 >= 125
-      ds.ScratchSpaceBuffer = get_scratch_surf(&pipeline->base, tes_bin);
+      ds.ScratchSpaceBuffer =
+         get_scratch_surf(&pipeline->base, MESA_SHADER_TESS_EVAL, tes_bin);
 #else
       ds.PerThreadScratchSpace = get_scratch_space(tes_bin);
       ds.ScratchSpaceBasePointer =
@@ -2037,7 +2050,8 @@ emit_3dstate_gs(struct anv_graphics_pipeline *pipeline)
 #endif
 
 #if GFX_VERx10 >= 125
-      gs.ScratchSpaceBuffer = get_scratch_surf(&pipeline->base, gs_bin);
+      gs.ScratchSpaceBuffer =
+         get_scratch_surf(&pipeline->base, MESA_SHADER_GEOMETRY, gs_bin);
 #else
       gs.PerThreadScratchSpace   = get_scratch_space(gs_bin);
       gs.ScratchSpaceBasePointer =
@@ -2309,7 +2323,8 @@ emit_3dstate_ps(struct anv_graphics_pipeline *pipeline,
          brw_wm_prog_data_dispatch_grf_start_reg(wm_prog_data, ps, 2);
 
 #if GFX_VERx10 >= 125
-      ps.ScratchSpaceBuffer = get_scratch_surf(&pipeline->base, fs_bin);
+      ps.ScratchSpaceBuffer =
+         get_scratch_surf(&pipeline->base, MESA_SHADER_FRAGMENT, fs_bin);
 #else
       ps.PerThreadScratchSpace   = get_scratch_space(fs_bin);
       ps.ScratchSpaceBasePointer =
@@ -2610,7 +2625,8 @@ emit_compute_state(struct anv_compute_pipeline *pipeline,
    anv_batch_emit(&pipeline->base.batch, GENX(CFE_STATE), cfe) {
       cfe.MaximumNumberofThreads =
          devinfo->max_cs_threads * devinfo->subslice_total - 1;
-      cfe.ScratchSpaceBuffer = get_scratch_surf(&pipeline->base, cs_bin);
+      cfe.ScratchSpaceBuffer =
+         get_scratch_surf(&pipeline->base, MESA_SHADER_COMPUTE, cs_bin);
    }
 }