anv: implement gen9 post sync pipe control workaround

We've been missing this workaround for a while and since it's required
for Gen12, let's implement it for Gen9 first.

v2: Update comment for Gen9.

v3: Fix clearing of bits... (Lionel)

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: <mesa-stable@lists.freedesktop.org>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3405>

3 files changed, 39 insertions, 0 deletions

diff --git a/src/intel/vulkan/anv_private.h b/src/intel/vulkan/anv_private.h
index ad7612ca611..d794a9db5dd 100644
--- a/src/intel/vulkan/anv_private.h
+++ b/src/intel/vulkan/anv_private.h
@@ -2272,6 +2272,12 @@ enum anv_pipe_bits {
     * done by writing the AUX-TT register.
     */
    ANV_PIPE_AUX_TABLE_INVALIDATE_BIT         = (1 << 23),
+
+   /* This bit does not exist directly in PIPE_CONTROL. It means that a
+    * PIPE_CONTROL with a post-sync operation will follow. This is used to
+    * implement a workaround for Gen9.
+    */
+   ANV_PIPE_POST_SYNC_BIT                    = (1 << 24),
 };
 
 #define ANV_PIPE_FLUSH_BITS ( \
diff --git a/src/intel/vulkan/genX_cmd_buffer.c b/src/intel/vulkan/genX_cmd_buffer.c
index 608e2a06f6c..e2df904166c 100644
--- a/src/intel/vulkan/genX_cmd_buffer.c
+++ b/src/intel/vulkan/genX_cmd_buffer.c
@@ -2050,6 +2050,21 @@ genX(cmd_buffer_apply_pipe_flushes)(struct anv_cmd_buffer *cmd_buffer)
              sizeof(cmd_buffer->state.gfx.ib_dirty_range));
    }
 
+   /* Project: SKL / Argument: LRI Post Sync Operation [23]
+    *
+    * "PIPECONTROL command with “Command Streamer Stall Enable” must be
+    *  programmed prior to programming a PIPECONTROL command with "LRI
+    *  Post Sync Operation" in GPGPU mode of operation (i.e when
+    *  PIPELINE_SELECT command is set to GPGPU mode of operation)."
+    *
+    * The same text exists a few rows below for Post Sync Op.
+    */
+   if (bits & ANV_PIPE_POST_SYNC_BIT) {
+      if (GEN_GEN == 9 && cmd_buffer->state.current_pipeline == GPGPU)
+         bits |= ANV_PIPE_CS_STALL_BIT;
+      bits &= ~ANV_PIPE_POST_SYNC_BIT;
+   }
+
    if (bits & (ANV_PIPE_FLUSH_BITS | ANV_PIPE_CS_STALL_BIT)) {
       anv_batch_emit(&cmd_buffer->batch, GENX(PIPE_CONTROL), pipe) {
 #if GEN_GEN >= 12
@@ -4619,6 +4634,9 @@ cmd_buffer_emit_depth_stencil(struct anv_cmd_buffer *cmd_buffer)
    isl_emit_depth_stencil_hiz_s(&device->isl_dev, dw, &info);
 
    if (GEN_GEN >= 12) {
+      cmd_buffer->state.pending_pipe_bits |= ANV_PIPE_POST_SYNC_BIT;
+      genX(cmd_buffer_apply_pipe_flushes)(cmd_buffer);
+
       /* GEN:BUG:1408224581
        *
        * Workaround: Gen12LP Astep only An additional pipe control with
@@ -5570,6 +5588,9 @@ void genX(CmdSetEvent)(
    ANV_FROM_HANDLE(anv_cmd_buffer, cmd_buffer, commandBuffer);
    ANV_FROM_HANDLE(anv_event, event, _event);
 
+   cmd_buffer->state.pending_pipe_bits |= ANV_PIPE_POST_SYNC_BIT;
+   genX(cmd_buffer_apply_pipe_flushes)(cmd_buffer);
+
    anv_batch_emit(&cmd_buffer->batch, GENX(PIPE_CONTROL), pc) {
       if (stageMask & ANV_PIPELINE_STAGE_PIPELINED_BITS) {
          pc.StallAtPixelScoreboard = true;
@@ -5594,6 +5615,9 @@ void genX(CmdResetEvent)(
    ANV_FROM_HANDLE(anv_cmd_buffer, cmd_buffer, commandBuffer);
    ANV_FROM_HANDLE(anv_event, event, _event);
 
+   cmd_buffer->state.pending_pipe_bits |= ANV_PIPE_POST_SYNC_BIT;
+   genX(cmd_buffer_apply_pipe_flushes)(cmd_buffer);
+
    anv_batch_emit(&cmd_buffer->batch, GENX(PIPE_CONTROL), pc) {
       if (stageMask & ANV_PIPELINE_STAGE_PIPELINED_BITS) {
          pc.StallAtPixelScoreboard = true;
diff --git a/src/intel/vulkan/genX_query.c b/src/intel/vulkan/genX_query.c
index 77a04605867..82b48233c9a 100644
--- a/src/intel/vulkan/genX_query.c
+++ b/src/intel/vulkan/genX_query.c
@@ -424,6 +424,9 @@ static void
 emit_ps_depth_count(struct anv_cmd_buffer *cmd_buffer,
                     struct anv_address addr)
 {
+   cmd_buffer->state.pending_pipe_bits |= ANV_PIPE_POST_SYNC_BIT;
+   genX(cmd_buffer_apply_pipe_flushes)(cmd_buffer);
+
    anv_batch_emit(&cmd_buffer->batch, GENX(PIPE_CONTROL), pc) {
       pc.DestinationAddressType  = DAT_PPGTT;
       pc.PostSyncOperation       = WritePSDepthCount;
@@ -448,6 +451,9 @@ emit_query_pc_availability(struct anv_cmd_buffer *cmd_buffer,
                            struct anv_address addr,
                            bool available)
 {
+   cmd_buffer->state.pending_pipe_bits |= ANV_PIPE_POST_SYNC_BIT;
+   genX(cmd_buffer_apply_pipe_flushes)(cmd_buffer);
+
    anv_batch_emit(&cmd_buffer->batch, GENX(PIPE_CONTROL), pc) {
       pc.DestinationAddressType  = DAT_PPGTT;
       pc.PostSyncOperation       = WriteImmediateData;
@@ -832,6 +838,9 @@ void genX(CmdWriteTimestamp)(
 
    default:
       /* Everything else is bottom-of-pipe */
+      cmd_buffer->state.pending_pipe_bits |= ANV_PIPE_POST_SYNC_BIT;
+      genX(cmd_buffer_apply_pipe_flushes)(cmd_buffer);
+
       anv_batch_emit(&cmd_buffer->batch, GENX(PIPE_CONTROL), pc) {
          pc.DestinationAddressType  = DAT_PPGTT;
          pc.PostSyncOperation       = WriteTimestamp;