anv: implement required PSS sync for Wa_18019816803

According to WA description, we need to track DS write state
and emit a PSS_STALL_SYNC whenever that state changes.

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18411>

diff --git a/src/intel/vulkan/anv_private.h b/src/intel/vulkan/anv_private.h
index 0e0eeb2..0348b52 100644 (file)
--- a/src/intel/vulkan/anv_private.h
+++ b/src/intel/vulkan/anv_private.h
@@ -2838,6 +2838,11 @@ struct anv_cmd_graphics_state {
    bool object_preemption;
    bool has_uint_rt;
 
+   /**
+    * DEPTH and STENCIL attachment write state for Wa_18019816803.
+    */
+   bool ds_write_state;
+
    uint32_t n_occlusion_queries;
 };
 

diff --git a/src/intel/vulkan/genX_blorp_exec.c b/src/intel/vulkan/genX_blorp_exec.c
index 84540f8..00cdcfc 100644 (file)
--- a/src/intel/vulkan/genX_blorp_exec.c
+++ b/src/intel/vulkan/genX_blorp_exec.c
@@ -294,6 +294,15 @@ blorp_exec_on_render(struct blorp_batch *batch,
    }
 #endif
 
+   /* Check if blorp ds state matches ours. */
+   if (intel_needs_workaround(cmd_buffer->device->info, 18019816803)) {
+      bool blorp_ds_state = params->depth.enabled || params->stencil.enabled;
+      if (cmd_buffer->state.gfx.ds_write_state != blorp_ds_state) {
+         batch->flags |= BLORP_BATCH_NEED_PSS_STALL_SYNC;
+         cmd_buffer->state.gfx.ds_write_state = blorp_ds_state;
+      }
+   }
+
    if (params->depth.enabled &&
        !(batch->flags & BLORP_BATCH_NO_EMIT_DEPTH_STENCIL))
       genX(cmd_buffer_emit_gfx12_depth_wa)(cmd_buffer, &params->depth.surf);

diff --git a/src/intel/vulkan/genX_cmd_buffer.c b/src/intel/vulkan/genX_cmd_buffer.c
index 86256f4..eafc20d 100644 (file)
--- a/src/intel/vulkan/genX_cmd_buffer.c
+++ b/src/intel/vulkan/genX_cmd_buffer.c
@@ -4101,6 +4101,7 @@ genX(CmdExecuteCommands)(
    primary->state.current_l3_config = NULL;
    primary->state.current_hash_scale = 0;
    primary->state.gfx.push_constant_stages = 0;
+   primary->state.gfx.ds_write_state = false;
    vk_dynamic_graphics_state_dirty_all(&primary->vk.dynamic_graphics_state);
 
    /* Each of the secondary command buffers will use its own state base

diff --git a/src/intel/vulkan/gfx8_cmd_buffer.c b/src/intel/vulkan/gfx8_cmd_buffer.c
index a0ad411..c4c4b32 100644 (file)
--- a/src/intel/vulkan/gfx8_cmd_buffer.c
+++ b/src/intel/vulkan/gfx8_cmd_buffer.c
@@ -651,6 +651,17 @@ genX(cmd_buffer_flush_dynamic_state)(struct anv_cmd_buffer *cmd_buffer)
          ds.BackfaceStencilTestFunction = genX(vk_to_intel_compare_op)[opt_ds.stencil.back.op.compare];
       }
 
+#if INTEL_NEEDS_WA_18019816803
+      if (intel_needs_workaround(cmd_buffer->device->info, 18019816803)) {
+         bool ds_write_state = opt_ds.depth.write_enable || opt_ds.stencil.write_enable;
+         if (cmd_buffer->state.gfx.ds_write_state != ds_write_state) {
+            genX(batch_emit_pipe_control)(&cmd_buffer->batch, cmd_buffer->device->info,
+                                          ANV_PIPE_PSS_STALL_SYNC_BIT);
+            cmd_buffer->state.gfx.ds_write_state = ds_write_state;
+         }
+      }
+#endif
+
       const bool pma = want_stencil_pma_fix(cmd_buffer, &opt_ds);
       genX(cmd_buffer_enable_pma_fix)(cmd_buffer, pma);
    }