radeonsi: switch to 3-spaces style

Generated automatically using clang-format and the following config:

AlignAfterOpenBracket: true
AlignConsecutiveMacros: true
AllowAllArgumentsOnNextLine: false
AllowShortCaseLabelsOnASingleLine: false
AllowShortFunctionsOnASingleLine: false
AlwaysBreakAfterReturnType: None
BasedOnStyle: LLVM
BraceWrapping:
  AfterControlStatement: false
  AfterEnum: true
  AfterFunction: true
  AfterStruct: false
  BeforeElse: false
  SplitEmptyFunction: true
BinPackArguments: true
BinPackParameters: true
BreakBeforeBraces: Custom
ColumnLimit: 100
ContinuationIndentWidth: 3
Cpp11BracedListStyle: false
Cpp11BracedListStyle: true
ForEachMacros:
  - LIST_FOR_EACH_ENTRY
  - LIST_FOR_EACH_ENTRY_SAFE
  - util_dynarray_foreach
  - nir_foreach_variable
  - nir_foreach_variable_safe
  - nir_foreach_register
  - nir_foreach_register_safe
  - nir_foreach_use
  - nir_foreach_use_safe
  - nir_foreach_if_use
  - nir_foreach_if_use_safe
  - nir_foreach_def
  - nir_foreach_def_safe
  - nir_foreach_phi_src
  - nir_foreach_phi_src_safe
  - nir_foreach_parallel_copy_entry
  - nir_foreach_instr
  - nir_foreach_instr_reverse
  - nir_foreach_instr_safe
  - nir_foreach_instr_reverse_safe
  - nir_foreach_function
  - nir_foreach_block
  - nir_foreach_block_safe
  - nir_foreach_block_reverse
  - nir_foreach_block_reverse_safe
  - nir_foreach_block_in_cf_node
IncludeBlocks: Regroup
IncludeCategories:
  - Regex:           '<[[:alnum:].]+>'
    Priority:        2
  - Regex:           '.*'
    Priority:        1
IndentWidth: 3
PenaltyBreakBeforeFirstCallParameter: 1
PenaltyExcessCharacter: 100
SpaceAfterCStyleCast: false
SpaceBeforeCpp11BracedList: false
SpaceBeforeCtorInitializerColon: false
SpacesInContainerLiterals: false

Reviewed-by: Marek Olšák <[email protected]>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4319>

1 files changed, 363 insertions, 392 deletions

diff --git a/src/gallium/drivers/radeonsi/si_state_streamout.c b/src/gallium/drivers/radeonsi/si_state_streamout.c
index 85ac4a119c5..2ce8de0ccde 100644
--- a/src/gallium/drivers/radeonsi/si_state_streamout.c
+++ b/src/gallium/drivers/radeonsi/si_state_streamout.c
@@ -23,395 +23,372 @@
  */
 
 #include "si_build_pm4.h"
-
 #include "util/u_memory.h"
 #include "util/u_suballoc.h"
 
 static void si_set_streamout_enable(struct si_context *sctx, bool enable);
 
 static inline void si_so_target_reference(struct si_streamout_target **dst,
-					  struct pipe_stream_output_target *src)
+                                          struct pipe_stream_output_target *src)
 {
-	pipe_so_target_reference((struct pipe_stream_output_target**)dst, src);
+   pipe_so_target_reference((struct pipe_stream_output_target **)dst, src);
 }
 
-static struct pipe_stream_output_target *
-si_create_so_target(struct pipe_context *ctx,
-		    struct pipe_resource *buffer,
-		    unsigned buffer_offset,
-		    unsigned buffer_size)
+static struct pipe_stream_output_target *si_create_so_target(struct pipe_context *ctx,
+                                                             struct pipe_resource *buffer,
+                                                             unsigned buffer_offset,
+                                                             unsigned buffer_size)
 {
-	struct si_context *sctx = (struct si_context *)ctx;
-	struct si_streamout_target *t;
-	struct si_resource *buf = si_resource(buffer);
-
-	t = CALLOC_STRUCT(si_streamout_target);
-	if (!t) {
-		return NULL;
-	}
-
-	unsigned buf_filled_size_size = sctx->screen->use_ngg_streamout ? 8 : 4;
-	u_suballocator_alloc(sctx->allocator_zeroed_memory, buf_filled_size_size, 4,
-			     &t->buf_filled_size_offset,
-			     (struct pipe_resource**)&t->buf_filled_size);
-	if (!t->buf_filled_size) {
-		FREE(t);
-		return NULL;
-	}
-
-	t->b.reference.count = 1;
-	t->b.context = ctx;
-	pipe_resource_reference(&t->b.buffer, buffer);
-	t->b.buffer_offset = buffer_offset;
-	t->b.buffer_size = buffer_size;
-
-	util_range_add(&buf->b.b, &buf->valid_buffer_range, buffer_offset,
-		       buffer_offset + buffer_size);
-	return &t->b;
+   struct si_context *sctx = (struct si_context *)ctx;
+   struct si_streamout_target *t;
+   struct si_resource *buf = si_resource(buffer);
+
+   t = CALLOC_STRUCT(si_streamout_target);
+   if (!t) {
+      return NULL;
+   }
+
+   unsigned buf_filled_size_size = sctx->screen->use_ngg_streamout ? 8 : 4;
+   u_suballocator_alloc(sctx->allocator_zeroed_memory, buf_filled_size_size, 4,
+                        &t->buf_filled_size_offset, (struct pipe_resource **)&t->buf_filled_size);
+   if (!t->buf_filled_size) {
+      FREE(t);
+      return NULL;
+   }
+
+   t->b.reference.count = 1;
+   t->b.context = ctx;
+   pipe_resource_reference(&t->b.buffer, buffer);
+   t->b.buffer_offset = buffer_offset;
+   t->b.buffer_size = buffer_size;
+
+   util_range_add(&buf->b.b, &buf->valid_buffer_range, buffer_offset, buffer_offset + buffer_size);
+   return &t->b;
 }
 
-static void si_so_target_destroy(struct pipe_context *ctx,
-				 struct pipe_stream_output_target *target)
+static void si_so_target_destroy(struct pipe_context *ctx, struct pipe_stream_output_target *target)
 {
-	struct si_streamout_target *t = (struct si_streamout_target*)target;
-	pipe_resource_reference(&t->b.buffer, NULL);
-	si_resource_reference(&t->buf_filled_size, NULL);
-	FREE(t);
+   struct si_streamout_target *t = (struct si_streamout_target *)target;
+   pipe_resource_reference(&t->b.buffer, NULL);
+   si_resource_reference(&t->buf_filled_size, NULL);
+   FREE(t);
 }
 
 void si_streamout_buffers_dirty(struct si_context *sctx)
 {
-	if (!sctx->streamout.enabled_mask)
-		return;
+   if (!sctx->streamout.enabled_mask)
+      return;
 
-	si_mark_atom_dirty(sctx, &sctx->atoms.s.streamout_begin);
-	si_set_streamout_enable(sctx, true);
+   si_mark_atom_dirty(sctx, &sctx->atoms.s.streamout_begin);
+   si_set_streamout_enable(sctx, true);
 }
 
-static void si_set_streamout_targets(struct pipe_context *ctx,
-				     unsigned num_targets,
-				     struct pipe_stream_output_target **targets,
-				     const unsigned *offsets)
+static void si_set_streamout_targets(struct pipe_context *ctx, unsigned num_targets,
+                                     struct pipe_stream_output_target **targets,
+                                     const unsigned *offsets)
 {
-	struct si_context *sctx = (struct si_context *)ctx;
-	unsigned old_num_targets = sctx->streamout.num_targets;
-	unsigned i;
-	bool wait_now = false;
-
-	/* We are going to unbind the buffers. Mark which caches need to be flushed. */
-	if (sctx->streamout.num_targets && sctx->streamout.begin_emitted) {
-		/* Since streamout uses vector writes which go through TC L2
-		 * and most other clients can use TC L2 as well, we don't need
-		 * to flush it.
-		 *
-		 * The only cases which requires flushing it is VGT DMA index
-		 * fetching (on <= GFX7) and indirect draw data, which are rare
-		 * cases. Thus, flag the TC L2 dirtiness in the resource and
-		 * handle it at draw call time.
-		 */
-		for (i = 0; i < sctx->streamout.num_targets; i++)
-			if (sctx->streamout.targets[i])
-				si_resource(sctx->streamout.targets[i]->b.buffer)->TC_L2_dirty = true;
-
-		/* Invalidate the scalar cache in case a streamout buffer is
-		 * going to be used as a constant buffer.
-		 *
-		 * Invalidate vL1, because streamout bypasses it (done by
-		 * setting GLC=1 in the store instruction), but vL1 in other
-		 * CUs can contain outdated data of streamout buffers.
-		 *
-		 * VS_PARTIAL_FLUSH is required if the buffers are going to be
-		 * used as an input immediately.
-		 */
-		sctx->flags |= SI_CONTEXT_INV_SCACHE |
-			       SI_CONTEXT_INV_VCACHE;
-
-		/* The BUFFER_FILLED_SIZE is written using a PS_DONE event. */
-		if (sctx->screen->use_ngg_streamout) {
-			sctx->flags |= SI_CONTEXT_PS_PARTIAL_FLUSH;
-
-			/* Wait now. This is needed to make sure that GDS is not
-			 * busy at the end of IBs.
-			 *
-			 * Also, the next streamout operation will overwrite GDS,
-			 * so we need to make sure that it's idle.
-			 */
-			wait_now = true;
-		} else {
-			sctx->flags |= SI_CONTEXT_VS_PARTIAL_FLUSH;
-		}
-	}
-
-	/* All readers of the streamout targets need to be finished before we can
-	 * start writing to the targets.
-	 */
-	if (num_targets) {
-		if (sctx->screen->use_ngg_streamout)
-			si_allocate_gds(sctx);
-
-		sctx->flags |= SI_CONTEXT_PS_PARTIAL_FLUSH |
-			       SI_CONTEXT_CS_PARTIAL_FLUSH;
-	}
-
-	/* Streamout buffers must be bound in 2 places:
-	 * 1) in VGT by setting the VGT_STRMOUT registers
-	 * 2) as shader resources
-	 */
-
-	/* Stop streamout. */
-	if (sctx->streamout.num_targets && sctx->streamout.begin_emitted)
-		si_emit_streamout_end(sctx);
-
-	/* Set the new targets. */
-	unsigned enabled_mask = 0, append_bitmask = 0;
-	for (i = 0; i < num_targets; i++) {
-		si_so_target_reference(&sctx->streamout.targets[i], targets[i]);
-		if (!targets[i])
-			continue;
-
-		si_context_add_resource_size(sctx, targets[i]->buffer);
-		enabled_mask |= 1 << i;
-
-		if (offsets[i] == ((unsigned)-1))
-			append_bitmask |= 1 << i;
-	}
-
-	for (; i < sctx->streamout.num_targets; i++)
-		si_so_target_reference(&sctx->streamout.targets[i], NULL);
-
-	sctx->streamout.enabled_mask = enabled_mask;
-	sctx->streamout.num_targets = num_targets;
-	sctx->streamout.append_bitmask = append_bitmask;
-
-	/* Update dirty state bits. */
-	if (num_targets) {
-		si_streamout_buffers_dirty(sctx);
-	} else {
-		si_set_atom_dirty(sctx, &sctx->atoms.s.streamout_begin, false);
-		si_set_streamout_enable(sctx, false);
-	}
-
-	/* Set the shader resources.*/
-	for (i = 0; i < num_targets; i++) {
-		if (targets[i]) {
-			struct pipe_shader_buffer sbuf;
-			sbuf.buffer = targets[i]->buffer;
-
-			if (sctx->screen->use_ngg_streamout) {
-				sbuf.buffer_offset = targets[i]->buffer_offset;
-				sbuf.buffer_size = targets[i]->buffer_size;
-			} else {
-				sbuf.buffer_offset = 0;
-				sbuf.buffer_size = targets[i]->buffer_offset +
-						   targets[i]->buffer_size;
-			}
-
-			si_set_rw_shader_buffer(sctx, SI_VS_STREAMOUT_BUF0 + i, &sbuf);
-			si_resource(targets[i]->buffer)->bind_history |= PIPE_BIND_STREAM_OUTPUT;
-		} else {
-			si_set_rw_shader_buffer(sctx, SI_VS_STREAMOUT_BUF0 + i, NULL);
-		}
-	}
-	for (; i < old_num_targets; i++)
-		si_set_rw_shader_buffer(sctx, SI_VS_STREAMOUT_BUF0 + i, NULL);
-
-	if (wait_now)
-		sctx->emit_cache_flush(sctx);
+   struct si_context *sctx = (struct si_context *)ctx;
+   unsigned old_num_targets = sctx->streamout.num_targets;
+   unsigned i;
+   bool wait_now = false;
+
+   /* We are going to unbind the buffers. Mark which caches need to be flushed. */
+   if (sctx->streamout.num_targets && sctx->streamout.begin_emitted) {
+      /* Since streamout uses vector writes which go through TC L2
+       * and most other clients can use TC L2 as well, we don't need
+       * to flush it.
+       *
+       * The only cases which requires flushing it is VGT DMA index
+       * fetching (on <= GFX7) and indirect draw data, which are rare
+       * cases. Thus, flag the TC L2 dirtiness in the resource and
+       * handle it at draw call time.
+       */
+      for (i = 0; i < sctx->streamout.num_targets; i++)
+         if (sctx->streamout.targets[i])
+            si_resource(sctx->streamout.targets[i]->b.buffer)->TC_L2_dirty = true;
+
+      /* Invalidate the scalar cache in case a streamout buffer is
+       * going to be used as a constant buffer.
+       *
+       * Invalidate vL1, because streamout bypasses it (done by
+       * setting GLC=1 in the store instruction), but vL1 in other
+       * CUs can contain outdated data of streamout buffers.
+       *
+       * VS_PARTIAL_FLUSH is required if the buffers are going to be
+       * used as an input immediately.
+       */
+      sctx->flags |= SI_CONTEXT_INV_SCACHE | SI_CONTEXT_INV_VCACHE;
+
+      /* The BUFFER_FILLED_SIZE is written using a PS_DONE event. */
+      if (sctx->screen->use_ngg_streamout) {
+         sctx->flags |= SI_CONTEXT_PS_PARTIAL_FLUSH;
+
+         /* Wait now. This is needed to make sure that GDS is not
+          * busy at the end of IBs.
+          *
+          * Also, the next streamout operation will overwrite GDS,
+          * so we need to make sure that it's idle.
+          */
+         wait_now = true;
+      } else {
+         sctx->flags |= SI_CONTEXT_VS_PARTIAL_FLUSH;
+      }
+   }
+
+   /* All readers of the streamout targets need to be finished before we can
+    * start writing to the targets.
+    */
+   if (num_targets) {
+      if (sctx->screen->use_ngg_streamout)
+         si_allocate_gds(sctx);
+
+      sctx->flags |= SI_CONTEXT_PS_PARTIAL_FLUSH | SI_CONTEXT_CS_PARTIAL_FLUSH;
+   }
+
+   /* Streamout buffers must be bound in 2 places:
+    * 1) in VGT by setting the VGT_STRMOUT registers
+    * 2) as shader resources
+    */
+
+   /* Stop streamout. */
+   if (sctx->streamout.num_targets && sctx->streamout.begin_emitted)
+      si_emit_streamout_end(sctx);
+
+   /* Set the new targets. */
+   unsigned enabled_mask = 0, append_bitmask = 0;
+   for (i = 0; i < num_targets; i++) {
+      si_so_target_reference(&sctx->streamout.targets[i], targets[i]);
+      if (!targets[i])
+         continue;
+
+      si_context_add_resource_size(sctx, targets[i]->buffer);
+      enabled_mask |= 1 << i;
+
+      if (offsets[i] == ((unsigned)-1))
+         append_bitmask |= 1 << i;
+   }
+
+   for (; i < sctx->streamout.num_targets; i++)
+      si_so_target_reference(&sctx->streamout.targets[i], NULL);
+
+   sctx->streamout.enabled_mask = enabled_mask;
+   sctx->streamout.num_targets = num_targets;
+   sctx->streamout.append_bitmask = append_bitmask;
+
+   /* Update dirty state bits. */
+   if (num_targets) {
+      si_streamout_buffers_dirty(sctx);
+   } else {
+      si_set_atom_dirty(sctx, &sctx->atoms.s.streamout_begin, false);
+      si_set_streamout_enable(sctx, false);
+   }
+
+   /* Set the shader resources.*/
+   for (i = 0; i < num_targets; i++) {
+      if (targets[i]) {
+         struct pipe_shader_buffer sbuf;
+         sbuf.buffer = targets[i]->buffer;
+
+         if (sctx->screen->use_ngg_streamout) {
+            sbuf.buffer_offset = targets[i]->buffer_offset;
+            sbuf.buffer_size = targets[i]->buffer_size;
+         } else {
+            sbuf.buffer_offset = 0;
+            sbuf.buffer_size = targets[i]->buffer_offset + targets[i]->buffer_size;
+         }
+
+         si_set_rw_shader_buffer(sctx, SI_VS_STREAMOUT_BUF0 + i, &sbuf);
+         si_resource(targets[i]->buffer)->bind_history |= PIPE_BIND_STREAM_OUTPUT;
+      } else {
+         si_set_rw_shader_buffer(sctx, SI_VS_STREAMOUT_BUF0 + i, NULL);
+      }
+   }
+   for (; i < old_num_targets; i++)
+      si_set_rw_shader_buffer(sctx, SI_VS_STREAMOUT_BUF0 + i, NULL);
+
+   if (wait_now)
+      sctx->emit_cache_flush(sctx);
 }
 
 static void gfx10_emit_streamout_begin(struct si_context *sctx)
 {
-	struct si_streamout_target **t = sctx->streamout.targets;
-	struct radeon_cmdbuf *cs = sctx->gfx_cs;
-	unsigned last_target = 0;
-
-	for (unsigned i = 0; i < sctx->streamout.num_targets; i++) {
-		if (t[i])
-			last_target = i;
-	}
-
-	for (unsigned i = 0; i < sctx->streamout.num_targets; i++) {
-		if (!t[i])
-			continue;
-
-		t[i]->stride_in_dw = sctx->streamout.stride_in_dw[i];
-
-		bool append = sctx->streamout.append_bitmask & (1 << i);
-		uint64_t va = 0;
-
-		if (append) {
-			radeon_add_to_buffer_list(sctx,  sctx->gfx_cs,
-						  t[i]->buf_filled_size,
-						  RADEON_USAGE_READ,
-						  RADEON_PRIO_SO_FILLED_SIZE);
-
-			va = t[i]->buf_filled_size->gpu_address +
-			     t[i]->buf_filled_size_offset;
-		}
-
-		radeon_emit(cs, PKT3(PKT3_DMA_DATA, 5, 0));
-		radeon_emit(cs, S_411_SRC_SEL(append ? V_411_SRC_ADDR_TC_L2 : V_411_DATA) |
-				S_411_DST_SEL(V_411_GDS) |
-				S_411_CP_SYNC(i == last_target));
-		radeon_emit(cs, va);
-		radeon_emit(cs, va >> 32);
-		radeon_emit(cs, 4 * i); /* destination in GDS */
-		radeon_emit(cs, 0);
-		radeon_emit(cs, S_414_BYTE_COUNT_GFX9(4) |
-				S_414_DISABLE_WR_CONFIRM_GFX9(i != last_target));
-	}
-
-	sctx->streamout.begin_emitted = true;
+   struct si_streamout_target **t = sctx->streamout.targets;
+   struct radeon_cmdbuf *cs = sctx->gfx_cs;
+   unsigned last_target = 0;
+
+   for (unsigned i = 0; i < sctx->streamout.num_targets; i++) {
+      if (t[i])
+         last_target = i;
+   }
+
+   for (unsigned i = 0; i < sctx->streamout.num_targets; i++) {
+      if (!t[i])
+         continue;
+
+      t[i]->stride_in_dw = sctx->streamout.stride_in_dw[i];
+
+      bool append = sctx->streamout.append_bitmask & (1 << i);
+      uint64_t va = 0;
+
+      if (append) {
+         radeon_add_to_buffer_list(sctx, sctx->gfx_cs, t[i]->buf_filled_size, RADEON_USAGE_READ,
+                                   RADEON_PRIO_SO_FILLED_SIZE);
+
+         va = t[i]->buf_filled_size->gpu_address + t[i]->buf_filled_size_offset;
+      }
+
+      radeon_emit(cs, PKT3(PKT3_DMA_DATA, 5, 0));
+      radeon_emit(cs, S_411_SRC_SEL(append ? V_411_SRC_ADDR_TC_L2 : V_411_DATA) |
+                         S_411_DST_SEL(V_411_GDS) | S_411_CP_SYNC(i == last_target));
+      radeon_emit(cs, va);
+      radeon_emit(cs, va >> 32);
+      radeon_emit(cs, 4 * i); /* destination in GDS */
+      radeon_emit(cs, 0);
+      radeon_emit(cs, S_414_BYTE_COUNT_GFX9(4) | S_414_DISABLE_WR_CONFIRM_GFX9(i != last_target));
+   }
+
+   sctx->streamout.begin_emitted = true;
 }
 
 static void gfx10_emit_streamout_end(struct si_context *sctx)
 {
-	struct si_streamout_target **t = sctx->streamout.targets;
+   struct si_streamout_target **t = sctx->streamout.targets;
 
-	for (unsigned i = 0; i < sctx->streamout.num_targets; i++) {
-		if (!t[i])
-			continue;
+   for (unsigned i = 0; i < sctx->streamout.num_targets; i++) {
+      if (!t[i])
+         continue;
 
-		uint64_t va = t[i]->buf_filled_size->gpu_address + t[i]->buf_filled_size_offset;
+      uint64_t va = t[i]->buf_filled_size->gpu_address + t[i]->buf_filled_size_offset;
 
-		si_cp_release_mem(sctx, sctx->gfx_cs, V_028A90_PS_DONE, 0,
-				  EOP_DST_SEL_TC_L2,
-				  EOP_INT_SEL_SEND_DATA_AFTER_WR_CONFIRM,
-				  EOP_DATA_SEL_GDS,
-				  t[i]->buf_filled_size, va,
-				  EOP_DATA_GDS(i, 1), 0);
+      si_cp_release_mem(sctx, sctx->gfx_cs, V_028A90_PS_DONE, 0, EOP_DST_SEL_TC_L2,
+                        EOP_INT_SEL_SEND_DATA_AFTER_WR_CONFIRM, EOP_DATA_SEL_GDS,
+                        t[i]->buf_filled_size, va, EOP_DATA_GDS(i, 1), 0);
 
-		t[i]->buf_filled_size_valid = true;
-	}
+      t[i]->buf_filled_size_valid = true;
+   }
 
-	sctx->streamout.begin_emitted = false;
+   sctx->streamout.begin_emitted = false;
 }
 
 static void si_flush_vgt_streamout(struct si_context *sctx)
 {
-	struct radeon_cmdbuf *cs = sctx->gfx_cs;
-	unsigned reg_strmout_cntl;
-
-	/* The register is at different places on different ASICs. */
-	if (sctx->chip_class >= GFX7) {
-		reg_strmout_cntl = R_0300FC_CP_STRMOUT_CNTL;
-		radeon_set_uconfig_reg(cs, reg_strmout_cntl, 0);
-	} else {
-		reg_strmout_cntl = R_0084FC_CP_STRMOUT_CNTL;
-		radeon_set_config_reg(cs, reg_strmout_cntl, 0);
-	}
-
-	radeon_emit(cs, PKT3(PKT3_EVENT_WRITE, 0, 0));
-	radeon_emit(cs, EVENT_TYPE(EVENT_TYPE_SO_VGTSTREAMOUT_FLUSH) | EVENT_INDEX(0));
-
-	radeon_emit(cs, PKT3(PKT3_WAIT_REG_MEM, 5, 0));
-	radeon_emit(cs, WAIT_REG_MEM_EQUAL); /* wait until the register is equal to the reference value */
-	radeon_emit(cs, reg_strmout_cntl >> 2);  /* register */
-	radeon_emit(cs, 0);
-	radeon_emit(cs, S_0084FC_OFFSET_UPDATE_DONE(1)); /* reference value */
-	radeon_emit(cs, S_0084FC_OFFSET_UPDATE_DONE(1)); /* mask */
-	radeon_emit(cs, 4); /* poll interval */
+   struct radeon_cmdbuf *cs = sctx->gfx_cs;
+   unsigned reg_strmout_cntl;
+
+   /* The register is at different places on different ASICs. */
+   if (sctx->chip_class >= GFX7) {
+      reg_strmout_cntl = R_0300FC_CP_STRMOUT_CNTL;
+      radeon_set_uconfig_reg(cs, reg_strmout_cntl, 0);
+   } else {
+      reg_strmout_cntl = R_0084FC_CP_STRMOUT_CNTL;
+      radeon_set_config_reg(cs, reg_strmout_cntl, 0);
+   }
+
+   radeon_emit(cs, PKT3(PKT3_EVENT_WRITE, 0, 0));
+   radeon_emit(cs, EVENT_TYPE(EVENT_TYPE_SO_VGTSTREAMOUT_FLUSH) | EVENT_INDEX(0));
+
+   radeon_emit(cs, PKT3(PKT3_WAIT_REG_MEM, 5, 0));
+   radeon_emit(cs,
+               WAIT_REG_MEM_EQUAL); /* wait until the register is equal to the reference value */
+   radeon_emit(cs, reg_strmout_cntl >> 2); /* register */
+   radeon_emit(cs, 0);
+   radeon_emit(cs, S_0084FC_OFFSET_UPDATE_DONE(1)); /* reference value */
+   radeon_emit(cs, S_0084FC_OFFSET_UPDATE_DONE(1)); /* mask */
+   radeon_emit(cs, 4);                              /* poll interval */
 }
 
 static void si_emit_streamout_begin(struct si_context *sctx)
 {
-	struct radeon_cmdbuf *cs = sctx->gfx_cs;
-	struct si_streamout_target **t = sctx->streamout.targets;
-	uint16_t *stride_in_dw = sctx->streamout.stride_in_dw;
-	unsigned i;
-
-	si_flush_vgt_streamout(sctx);
-
-	for (i = 0; i < sctx->streamout.num_targets; i++) {
-		if (!t[i])
-			continue;
-
-		t[i]->stride_in_dw = stride_in_dw[i];
-
-		/* AMD GCN binds streamout buffers as shader resources.
-		 * VGT only counts primitives and tells the shader
-		 * through SGPRs what to do. */
-		radeon_set_context_reg_seq(cs, R_028AD0_VGT_STRMOUT_BUFFER_SIZE_0 + 16*i, 2);
-		radeon_emit(cs, (t[i]->b.buffer_offset +
-				 t[i]->b.buffer_size) >> 2);	/* BUFFER_SIZE (in DW) */
-		radeon_emit(cs, stride_in_dw[i]);		/* VTX_STRIDE (in DW) */
-
-		if (sctx->streamout.append_bitmask & (1 << i) && t[i]->buf_filled_size_valid) {
-			uint64_t va = t[i]->buf_filled_size->gpu_address +
-				      t[i]->buf_filled_size_offset;
-
-			/* Append. */
-			radeon_emit(cs, PKT3(PKT3_STRMOUT_BUFFER_UPDATE, 4, 0));
-			radeon_emit(cs, STRMOUT_SELECT_BUFFER(i) |
-				    STRMOUT_OFFSET_SOURCE(STRMOUT_OFFSET_FROM_MEM)); /* control */
-			radeon_emit(cs, 0); /* unused */
-			radeon_emit(cs, 0); /* unused */
-			radeon_emit(cs, va); /* src address lo */
-			radeon_emit(cs, va >> 32); /* src address hi */
-
-			radeon_add_to_buffer_list(sctx,  sctx->gfx_cs,
-						  t[i]->buf_filled_size,
-						  RADEON_USAGE_READ,
-						  RADEON_PRIO_SO_FILLED_SIZE);
-		} else {
-			/* Start from the beginning. */
-			radeon_emit(cs, PKT3(PKT3_STRMOUT_BUFFER_UPDATE, 4, 0));
-			radeon_emit(cs, STRMOUT_SELECT_BUFFER(i) |
-				    STRMOUT_OFFSET_SOURCE(STRMOUT_OFFSET_FROM_PACKET)); /* control */
-			radeon_emit(cs, 0); /* unused */
-			radeon_emit(cs, 0); /* unused */
-			radeon_emit(cs, t[i]->b.buffer_offset >> 2); /* buffer offset in DW */
-			radeon_emit(cs, 0); /* unused */
-		}
-	}
-
-	sctx->streamout.begin_emitted = true;
+   struct radeon_cmdbuf *cs = sctx->gfx_cs;
+   struct si_streamout_target **t = sctx->streamout.targets;
+   uint16_t *stride_in_dw = sctx->streamout.stride_in_dw;
+   unsigned i;
+
+   si_flush_vgt_streamout(sctx);
+
+   for (i = 0; i < sctx->streamout.num_targets; i++) {
+      if (!t[i])
+         continue;
+
+      t[i]->stride_in_dw = stride_in_dw[i];
+
+      /* AMD GCN binds streamout buffers as shader resources.
+       * VGT only counts primitives and tells the shader
+       * through SGPRs what to do. */
+      radeon_set_context_reg_seq(cs, R_028AD0_VGT_STRMOUT_BUFFER_SIZE_0 + 16 * i, 2);
+      radeon_emit(cs, (t[i]->b.buffer_offset + t[i]->b.buffer_size) >> 2); /* BUFFER_SIZE (in DW) */
+      radeon_emit(cs, stride_in_dw[i]);                                    /* VTX_STRIDE (in DW) */
+
+      if (sctx->streamout.append_bitmask & (1 << i) && t[i]->buf_filled_size_valid) {
+         uint64_t va = t[i]->buf_filled_size->gpu_address + t[i]->buf_filled_size_offset;
+
+         /* Append. */
+         radeon_emit(cs, PKT3(PKT3_STRMOUT_BUFFER_UPDATE, 4, 0));
+         radeon_emit(cs, STRMOUT_SELECT_BUFFER(i) |
+                            STRMOUT_OFFSET_SOURCE(STRMOUT_OFFSET_FROM_MEM)); /* control */
+         radeon_emit(cs, 0);                                                 /* unused */
+         radeon_emit(cs, 0);                                                 /* unused */
+         radeon_emit(cs, va);                                                /* src address lo */
+         radeon_emit(cs, va >> 32);                                          /* src address hi */
+
+         radeon_add_to_buffer_list(sctx, sctx->gfx_cs, t[i]->buf_filled_size, RADEON_USAGE_READ,
+                                   RADEON_PRIO_SO_FILLED_SIZE);
+      } else {
+         /* Start from the beginning. */
+         radeon_emit(cs, PKT3(PKT3_STRMOUT_BUFFER_UPDATE, 4, 0));
+         radeon_emit(cs, STRMOUT_SELECT_BUFFER(i) |
+                            STRMOUT_OFFSET_SOURCE(STRMOUT_OFFSET_FROM_PACKET)); /* control */
+         radeon_emit(cs, 0);                                                    /* unused */
+         radeon_emit(cs, 0);                                                    /* unused */
+         radeon_emit(cs, t[i]->b.buffer_offset >> 2); /* buffer offset in DW */
+         radeon_emit(cs, 0);                          /* unused */
+      }
+   }
+
+   sctx->streamout.begin_emitted = true;
 }
 
 void si_emit_streamout_end(struct si_context *sctx)
 {
-	if (sctx->screen->use_ngg_streamout) {
-		gfx10_emit_streamout_end(sctx);
-		return;
-	}
-
-	struct radeon_cmdbuf *cs = sctx->gfx_cs;
-	struct si_streamout_target **t = sctx->streamout.targets;
-	unsigned i;
-	uint64_t va;
-
-	si_flush_vgt_streamout(sctx);
-
-	for (i = 0; i < sctx->streamout.num_targets; i++) {
-		if (!t[i])
-			continue;
-
-		va = t[i]->buf_filled_size->gpu_address + t[i]->buf_filled_size_offset;
-		radeon_emit(cs, PKT3(PKT3_STRMOUT_BUFFER_UPDATE, 4, 0));
-		radeon_emit(cs, STRMOUT_SELECT_BUFFER(i) |
-			    STRMOUT_OFFSET_SOURCE(STRMOUT_OFFSET_NONE) |
-			    STRMOUT_STORE_BUFFER_FILLED_SIZE); /* control */
-		radeon_emit(cs, va);     /* dst address lo */
-		radeon_emit(cs, va >> 32); /* dst address hi */
-		radeon_emit(cs, 0); /* unused */
-		radeon_emit(cs, 0); /* unused */
-
-		radeon_add_to_buffer_list(sctx,  sctx->gfx_cs,
-					  t[i]->buf_filled_size,
-					  RADEON_USAGE_WRITE,
-					  RADEON_PRIO_SO_FILLED_SIZE);
-
-		/* Zero the buffer size. The counters (primitives generated,
-		 * primitives emitted) may be enabled even if there is not
-		 * buffer bound. This ensures that the primitives-emitted query
-		 * won't increment. */
-		radeon_set_context_reg(cs, R_028AD0_VGT_STRMOUT_BUFFER_SIZE_0 + 16*i, 0);
-		sctx->context_roll = true;
-
-		t[i]->buf_filled_size_valid = true;
-	}
-
-	sctx->streamout.begin_emitted = false;
+   if (sctx->screen->use_ngg_streamout) {
+      gfx10_emit_streamout_end(sctx);
+      return;
+   }
+
+   struct radeon_cmdbuf *cs = sctx->gfx_cs;
+   struct si_streamout_target **t = sctx->streamout.targets;
+   unsigned i;
+   uint64_t va;
+
+   si_flush_vgt_streamout(sctx);
+
+   for (i = 0; i < sctx->streamout.num_targets; i++) {
+      if (!t[i])
+         continue;
+
+      va = t[i]->buf_filled_size->gpu_address + t[i]->buf_filled_size_offset;
+      radeon_emit(cs, PKT3(PKT3_STRMOUT_BUFFER_UPDATE, 4, 0));
+      radeon_emit(cs, STRMOUT_SELECT_BUFFER(i) | STRMOUT_OFFSET_SOURCE(STRMOUT_OFFSET_NONE) |
+                         STRMOUT_STORE_BUFFER_FILLED_SIZE); /* control */
+      radeon_emit(cs, va);                                  /* dst address lo */
+      radeon_emit(cs, va >> 32);                            /* dst address hi */
+      radeon_emit(cs, 0);                                   /* unused */
+      radeon_emit(cs, 0);                                   /* unused */
+
+      radeon_add_to_buffer_list(sctx, sctx->gfx_cs, t[i]->buf_filled_size, RADEON_USAGE_WRITE,
+                                RADEON_PRIO_SO_FILLED_SIZE);
+
+      /* Zero the buffer size. The counters (primitives generated,
+       * primitives emitted) may be enabled even if there is not
+       * buffer bound. This ensures that the primitives-emitted query
+       * won't increment. */
+      radeon_set_context_reg(cs, R_028AD0_VGT_STRMOUT_BUFFER_SIZE_0 + 16 * i, 0);
+      sctx->context_roll = true;
+
+      t[i]->buf_filled_size_valid = true;
+   }
+
+   sctx->streamout.begin_emitted = false;
 }
 
 /* STREAMOUT CONFIG DERIVED STATE
@@ -423,71 +400,65 @@ void si_emit_streamout_end(struct si_context *sctx)
 
 static void si_emit_streamout_enable(struct si_context *sctx)
 {
-	assert(!sctx->screen->use_ngg_streamout);
-
-	radeon_set_context_reg_seq(sctx->gfx_cs, R_028B94_VGT_STRMOUT_CONFIG, 2);
-	radeon_emit(sctx->gfx_cs,
-		    S_028B94_STREAMOUT_0_EN(si_get_strmout_en(sctx)) |
-		    S_028B94_RAST_STREAM(0) |
-		    S_028B94_STREAMOUT_1_EN(si_get_strmout_en(sctx)) |
-		    S_028B94_STREAMOUT_2_EN(si_get_strmout_en(sctx)) |
-		    S_028B94_STREAMOUT_3_EN(si_get_strmout_en(sctx)));
-	radeon_emit(sctx->gfx_cs,
-		    sctx->streamout.hw_enabled_mask &
-		    sctx->streamout.enabled_stream_buffers_mask);
+   assert(!sctx->screen->use_ngg_streamout);
+
+   radeon_set_context_reg_seq(sctx->gfx_cs, R_028B94_VGT_STRMOUT_CONFIG, 2);
+   radeon_emit(sctx->gfx_cs, S_028B94_STREAMOUT_0_EN(si_get_strmout_en(sctx)) |
+                                S_028B94_RAST_STREAM(0) |
+                                S_028B94_STREAMOUT_1_EN(si_get_strmout_en(sctx)) |
+                                S_028B94_STREAMOUT_2_EN(si_get_strmout_en(sctx)) |
+                                S_028B94_STREAMOUT_3_EN(si_get_strmout_en(sctx)));
+   radeon_emit(sctx->gfx_cs,
+               sctx->streamout.hw_enabled_mask & sctx->streamout.enabled_stream_buffers_mask);
 }
 
 static void si_set_streamout_enable(struct si_context *sctx, bool enable)
 {
-	bool old_strmout_en = si_get_strmout_en(sctx);
-	unsigned old_hw_enabled_mask = sctx->streamout.hw_enabled_mask;
+   bool old_strmout_en = si_get_strmout_en(sctx);
+   unsigned old_hw_enabled_mask = sctx->streamout.hw_enabled_mask;
 
-	sctx->streamout.streamout_enabled = enable;
+   sctx->streamout.streamout_enabled = enable;
 
-	sctx->streamout.hw_enabled_mask = sctx->streamout.enabled_mask |
-					  (sctx->streamout.enabled_mask << 4) |
-					  (sctx->streamout.enabled_mask << 8) |
-					  (sctx->streamout.enabled_mask << 12);
+   sctx->streamout.hw_enabled_mask =
+      sctx->streamout.enabled_mask | (sctx->streamout.enabled_mask << 4) |
+      (sctx->streamout.enabled_mask << 8) | (sctx->streamout.enabled_mask << 12);
 
-	if (!sctx->screen->use_ngg_streamout &&
-	    ((old_strmout_en != si_get_strmout_en(sctx)) ||
-	     (old_hw_enabled_mask != sctx->streamout.hw_enabled_mask)))
-		si_mark_atom_dirty(sctx, &sctx->atoms.s.streamout_enable);
+   if (!sctx->screen->use_ngg_streamout &&
+       ((old_strmout_en != si_get_strmout_en(sctx)) ||
+        (old_hw_enabled_mask != sctx->streamout.hw_enabled_mask)))
+      si_mark_atom_dirty(sctx, &sctx->atoms.s.streamout_enable);
 }
 
-void si_update_prims_generated_query_state(struct si_context *sctx,
-					   unsigned type, int diff)
+void si_update_prims_generated_query_state(struct si_context *sctx, unsigned type, int diff)
 {
-	if (!sctx->screen->use_ngg_streamout &&
-	    type == PIPE_QUERY_PRIMITIVES_GENERATED) {
-		bool old_strmout_en = si_get_strmout_en(sctx);
+   if (!sctx->screen->use_ngg_streamout && type == PIPE_QUERY_PRIMITIVES_GENERATED) {
+      bool old_strmout_en = si_get_strmout_en(sctx);
 
-		sctx->streamout.num_prims_gen_queries += diff;
-		assert(sctx->streamout.num_prims_gen_queries >= 0);
+      sctx->streamout.num_prims_gen_queries += diff;
+      assert(sctx->streamout.num_prims_gen_queries >= 0);
 
-		sctx->streamout.prims_gen_query_enabled =
-			sctx->streamout.num_prims_gen_queries != 0;
+      sctx->streamout.prims_gen_query_enabled = sctx->streamout.num_prims_gen_queries != 0;
 
-		if (old_strmout_en != si_get_strmout_en(sctx))
-			si_mark_atom_dirty(sctx, &sctx->atoms.s.streamout_enable);
+      if (old_strmout_en != si_get_strmout_en(sctx))
+         si_mark_atom_dirty(sctx, &sctx->atoms.s.streamout_enable);
 
-		if (si_update_ngg(sctx)) {
-			si_shader_change_notify(sctx);
-			sctx->do_update_shaders = true;
-		}
-	}
+      if (si_update_ngg(sctx)) {
+         si_shader_change_notify(sctx);
+         sctx->do_update_shaders = true;
+      }
+   }
 }
 
 void si_init_streamout_functions(struct si_context *sctx)
 {
-	sctx->b.create_stream_output_target = si_create_so_target;
-	sctx->b.stream_output_target_destroy = si_so_target_destroy;
-	sctx->b.set_stream_output_targets = si_set_streamout_targets;
-
-	if (sctx->screen->use_ngg_streamout) {
-		sctx->atoms.s.streamout_begin.emit = gfx10_emit_streamout_begin;
-	} else {
-		sctx->atoms.s.streamout_begin.emit = si_emit_streamout_begin;
-		sctx->atoms.s.streamout_enable.emit = si_emit_streamout_enable;
-	}
+   sctx->b.create_stream_output_target = si_create_so_target;
+   sctx->b.stream_output_target_destroy = si_so_target_destroy;
+   sctx->b.set_stream_output_targets = si_set_streamout_targets;
+
+   if (sctx->screen->use_ngg_streamout) {
+      sctx->atoms.s.streamout_begin.emit = gfx10_emit_streamout_begin;
+   } else {
+      sctx->atoms.s.streamout_begin.emit = si_emit_streamout_begin;
+      sctx->atoms.s.streamout_enable.emit = si_emit_streamout_enable;
+   }
 }