[Mesa-dev] [PATCH 19/37] i965/gen6/gs: Handle the case where a geometry shader emits no output.
Iago Toral Quiroga
itoral at igalia.com
Thu Aug 14 04:11:51 PDT 2014
In gen6 we need to end the thread differently depending on whether we have
emitted at least one vertex or not. In case we did, the EOT message must
always include the COMPLETE flag or else the GPU hangs. If we have not
produced any output, however, we can't use the COMPLETE flag.
This would lead us to end the program with an ENDIF opcode, which we want
to avoid (and actually is not permitted since it hits an assertion), so
instead what we do is that we always request a new VUE handle every time we do
an URB WRITE, even for the last vertex we emit. With this we make sure that
whether we have emitted at least one vertex or none at all we have to finish the
thread without writing to the URB, which works for both cases by setting the
COMPLETE and UNUSED flags in the EOT message.
---
src/mesa/drivers/dri/i965/gen6_gs_visitor.cpp | 237 +++++++++++++-------------
src/mesa/drivers/dri/i965/gen6_gs_visitor.h | 3 +-
2 files changed, 118 insertions(+), 122 deletions(-)
diff --git a/src/mesa/drivers/dri/i965/gen6_gs_visitor.cpp b/src/mesa/drivers/dri/i965/gen6_gs_visitor.cpp
index 252e585..4a440eb 100644
--- a/src/mesa/drivers/dri/i965/gen6_gs_visitor.cpp
+++ b/src/mesa/drivers/dri/i965/gen6_gs_visitor.cpp
@@ -166,7 +166,7 @@ gen6_gs_visitor::visit(ir_end_primitive *)
/* Otheriwse we know that the last vertex we have processed was the last
* vertex in the primitive and we need to set its PrimEnd flag, so do this
- * unless we haven't emitted that vertex at all.
+ * unless we haven't emitted that vertex at all (vertex_count != 0).
*
* Notice that we have already incremented vertex_count when we processed
* the last emit_vertex, so we need to take that into account in the
@@ -176,6 +176,10 @@ gen6_gs_visitor::visit(ir_end_primitive *)
unsigned num_output_vertices = c->gp->program.VerticesOut;
emit(CMP(dst_null_d(), this->vertex_count, src_reg(num_output_vertices + 1),
BRW_CONDITIONAL_L));
+ vec4_instruction *inst = emit(CMP(dst_null_d(),
+ this->vertex_count, 0u,
+ BRW_CONDITIONAL_NEQ));
+ inst->predicate = BRW_PREDICATE_NORMAL;
emit(IF(BRW_PREDICATE_NORMAL));
{
/* vertex_output_offset is already pointing at the first entry of the
@@ -224,47 +228,40 @@ gen6_gs_visitor::emit_urb_write_header(int mrf)
}
void
-gen6_gs_visitor::emit_urb_write_opcode(bool complete, src_reg vertex,
- int base_mrf, int mlen, int urb_offset)
+gen6_gs_visitor::emit_urb_write_opcode(bool complete, int base_mrf,
+ int last_mrf, int urb_offset)
{
vec4_instruction *inst = NULL;
- /* If the vertex is not complete we don't have to do anything special */
if (!complete) {
+ /* If the vertex is not complete we don't have to do anything special */
inst = emit(GS_OPCODE_URB_WRITE);
inst->urb_write_flags = BRW_URB_WRITE_NO_FLAGS;
- inst->base_mrf = base_mrf;
- inst->mlen = mlen;
- inst->offset = urb_offset;
- return;
- }
-
- /* Otherwise, if this is not the last vertex we are going to write,
- * we have to request a new VUE handle for the next vertex.
- *
- * Notice that the vertex parameter has been pre-incremented in
- * emit_thread_end() to make this comparison easier.
- */
- emit(CMP(dst_null_d(), vertex, this->vertex_count, BRW_CONDITIONAL_L));
- emit(IF(BRW_PREDICATE_NORMAL));
- {
+ } else {
+ /* Otherwise we always request to allocate a new VUE handle. If this is
+ * the last write before the EOT message and the new handle never gets
+ * used it will be dereferenced when we send the EOT message. This is
+ * necessary to avoid different setups for the EOT message (one for the
+ * case when there is no output and another for the case when there is)
+ * which would require to end the program with an IF/ELSE/ENDIF block,
+ * something we do not want.
+ */
inst = emit(GS_OPCODE_URB_WRITE_ALLOCATE);
inst->urb_write_flags = BRW_URB_WRITE_COMPLETE;
- inst->base_mrf = base_mrf;
- inst->mlen = mlen;
- inst->offset = urb_offset;
inst->dst = dst_reg(MRF, base_mrf);
inst->src[0] = this->temp;
}
- emit(BRW_OPCODE_ELSE);
- {
- inst = emit(GS_OPCODE_URB_WRITE);
- inst->urb_write_flags = BRW_URB_WRITE_COMPLETE;
- inst->base_mrf = base_mrf;
- inst->mlen = mlen;
- inst->offset = urb_offset;
- }
- emit(BRW_OPCODE_ENDIF);
+
+ inst->base_mrf = base_mrf;
+ /* URB data written (does not include the message header reg) must
+ * be a multiple of 256 bits, or 2 VS registers. See vol5c.5,
+ * section 5.4.3.2.2: URB_INTERLEAVED.
+ */
+ int mlen = last_mrf - base_mrf;
+ if ((mlen % 2) != 1)
+ mlen++;
+ inst->mlen = mlen;
+ inst->offset = urb_offset;
}
void
@@ -303,112 +300,112 @@ gen6_gs_visitor::emit_thread_end()
int max_usable_mrf = 13;
/* Issue the FF_SYNC message and obtain the initial VUE handle. */
- this->current_annotation = "gen6 thread end: ff_sync";
- emit(GS_OPCODE_FF_SYNC,
- dst_reg(MRF, base_mrf), this->temp, this->prim_count);
-
- /* Loop over all buffered vertices and emit URB write messages */
- this->current_annotation = "gen6 thread end: urb writes init";
- src_reg vertex(this, glsl_type::uint_type);
- emit(MOV(dst_reg(vertex), 0u));
- emit(MOV(dst_reg(this->vertex_output_offset), 0u));
-
- this->current_annotation = "gen6 thread end: urb writes";
- emit(BRW_OPCODE_DO);
+ emit(CMP(dst_null_d(), this->vertex_count, 0u, BRW_CONDITIONAL_G));
+ emit(IF(BRW_PREDICATE_NORMAL));
{
- emit(CMP(dst_null_d(), vertex, this->vertex_count, BRW_CONDITIONAL_GE));
- vec4_instruction *inst = emit(BRW_OPCODE_BREAK);
- inst->predicate = BRW_PREDICATE_NORMAL;
-
- /* First we prepare the message header */
- emit_urb_write_header(base_mrf);
-
- /* Then add vertex data to the message in interleaved fashion */
- int slot = 0;
- bool complete = false;
- do {
- int mrf = base_mrf + 1;
-
- /* URB offset is in URB row increments, and each of our MRFs is half
- * of one of those, since we're doing interleaved writes.
- */
- int urb_offset = slot / 2;
-
- for (; slot < prog_data->vue_map.num_slots; ++slot) {
- int varying = prog_data->vue_map.slot_to_varying[slot];
- current_annotation = output_reg_annotation[varying];
+ this->current_annotation = "gen6 thread end: ff_sync";
+ emit(GS_OPCODE_FF_SYNC,
+ dst_reg(MRF, base_mrf), this->temp, this->prim_count);
+
+ /* Loop over all buffered vertices and emit URB write messages */
+ this->current_annotation = "gen6 thread end: urb writes init";
+ src_reg vertex(this, glsl_type::uint_type);
+ emit(MOV(dst_reg(vertex), 0u));
+ emit(MOV(dst_reg(this->vertex_output_offset), 0u));
+
+ this->current_annotation = "gen6 thread end: urb writes";
+ emit(BRW_OPCODE_DO);
+ {
+ emit(CMP(dst_null_d(), vertex, this->vertex_count, BRW_CONDITIONAL_GE));
+ vec4_instruction *inst = emit(BRW_OPCODE_BREAK);
+ inst->predicate = BRW_PREDICATE_NORMAL;
- /* Compute offset of this slot for the current vertex
- * in vertex_output
- */
- src_reg data(this->vertex_output);
- data.reladdr = ralloc(mem_ctx, src_reg);
- memcpy(data.reladdr, &this->vertex_output_offset, sizeof(src_reg));
-
- if (varying == VARYING_SLOT_PSIZ) {
- /* We did not buffer PSIZ, emit it directly here */
- emit_urb_slot(dst_reg(MRF, mrf), varying);
- } else {
- /* Copy this slot to the appropriate message register */
- dst_reg reg = dst_reg(MRF, mrf);
- reg.type = output_reg[varying].type;
- data.type = reg.type;
- vec4_instruction *inst = emit(MOV(reg, data));
- inst->force_writemask_all = true;
- }
+ /* First we prepare the message header */
+ emit_urb_write_header(base_mrf);
- mrf++;
- emit(ADD(dst_reg(this->vertex_output_offset),
- this->vertex_output_offset, 1u));
+ /* Then add vertex data to the message in interleaved fashion */
+ int slot = 0;
+ bool complete = false;
+ do {
+ int mrf = base_mrf + 1;
- /* If this was max_usable_mrf, we can't fit anything more into this
- * URB WRITE.
+ /* URB offset is in URB row increments, and each of our MRFs is half
+ * of one of those, since we're doing interleaved writes.
*/
- if (mrf > max_usable_mrf) {
- slot++;
- break;
+ int urb_offset = slot / 2;
+
+ for (; slot < prog_data->vue_map.num_slots; ++slot) {
+ int varying = prog_data->vue_map.slot_to_varying[slot];
+ current_annotation = output_reg_annotation[varying];
+
+ /* Compute offset of this slot for the current vertex
+ * in vertex_output
+ */
+ src_reg data(this->vertex_output);
+ data.reladdr = ralloc(mem_ctx, src_reg);
+ memcpy(data.reladdr, &this->vertex_output_offset,
+ sizeof(src_reg));
+
+ if (varying == VARYING_SLOT_PSIZ) {
+ /* We did not buffer PSIZ, emit it directly here */
+ emit_urb_slot(dst_reg(MRF, mrf), varying);
+ } else {
+ /* Copy this slot to the appropriate message register */
+ dst_reg reg = dst_reg(MRF, mrf);
+ reg.type = output_reg[varying].type;
+ data.type = reg.type;
+ vec4_instruction *inst = emit(MOV(reg, data));
+ inst->force_writemask_all = true;
+ }
+
+ mrf++;
+ emit(ADD(dst_reg(this->vertex_output_offset),
+ this->vertex_output_offset, 1u));
+
+ /* If this was max_usable_mrf, we can't fit anything more into
+ * this URB WRITE.
+ */
+ if (mrf > max_usable_mrf) {
+ slot++;
+ break;
+ }
}
- }
- complete = slot >= prog_data->vue_map.num_slots;
+ complete = slot >= prog_data->vue_map.num_slots;
+ emit_urb_write_opcode(complete, base_mrf, mrf, urb_offset);
+ } while (!complete);
- /* When we emit the URB_WRITE below we need to do different things
- * depending on whether this is the last vertex we are going to
- * write. That means that we will need to check if
- * vertex >= vertex_count - 1. However, by increasing vertex early
- * we transform that comparison into vertex >= vertex_count, which
- * is more convenient.
+ /* Skip over the flags data item so that vertex_output_offset points
+ * to the first data item of the next vertex, so that we can start
+ * writing the next vertex.
*/
- if (complete)
- emit(ADD(dst_reg(vertex), vertex, 1u));
+ emit(ADD(dst_reg(this->vertex_output_offset),
+ this->vertex_output_offset, 1u));
- /* URB data written (does not include the message header reg) must
- * be a multiple of 256 bits, or 2 VS registers. See vol5c.5,
- * section 5.4.3.2.2: URB_INTERLEAVED.
- */
- int mlen = mrf - base_mrf;
- if ((mlen % 2) != 1)
- mlen++;
- emit_urb_write_opcode(complete, vertex, base_mrf, mlen, urb_offset);
- } while (!complete);
-
- /* Skip over the flags data item so that vertex_output_offset points to
- * the first data item of the next vertex, so that we can start writing
- * the next vertex.
- */
- emit(ADD(dst_reg(this->vertex_output_offset),
- this->vertex_output_offset, 1u));
+ emit(ADD(dst_reg(vertex), vertex, 1u));
+ }
+ emit(BRW_OPCODE_WHILE);
}
- emit(BRW_OPCODE_WHILE);
+ emit(BRW_OPCODE_ENDIF);
/* Finally, emit EOT message.
*
- * In gen6 it looks like we have to set the complete flag too, otherwise
- * the GPU hangs.
+ * In gen6 we need to end the thread differently depending on whether we have
+ * emitted at least one vertex or not. In case we did, the EOT message must
+ * always include the COMPLETE flag or else the GPU hangs. If we have not
+ * produced any output we can't use the COMPLETE flag.
+ *
+ * However, this would lead us to end the program with an ENDIF opcode,
+ * which we want to avoid, so what we do is that we always request a new
+ * VUE handle every time we do a URB WRITE, even for the last vertex we emit.
+ * With this we make sure that whether we have emitted at least one vertex
+ * or none at all, we have to finish the thread without writing to the URB,
+ * which works for both cases by setting the COMPLETE and UNUSED flags in
+ * the EOT message.
*/
this->current_annotation = "gen6 thread end: EOT";
vec4_instruction *inst = emit(GS_OPCODE_THREAD_END);
- inst->urb_write_flags = BRW_URB_WRITE_COMPLETE;
+ inst->urb_write_flags = BRW_URB_WRITE_COMPLETE | BRW_URB_WRITE_UNUSED;
inst->base_mrf = base_mrf;
inst->mlen = 1;
}
diff --git a/src/mesa/drivers/dri/i965/gen6_gs_visitor.h b/src/mesa/drivers/dri/i965/gen6_gs_visitor.h
index 68fe88d..7af6405 100644
--- a/src/mesa/drivers/dri/i965/gen6_gs_visitor.h
+++ b/src/mesa/drivers/dri/i965/gen6_gs_visitor.h
@@ -49,9 +49,8 @@ protected:
virtual void visit(ir_end_primitive *);
virtual void emit_urb_write_header(int mrf);
virtual void emit_urb_write_opcode(bool complete,
- src_reg vertex,
int base_mrf,
- int mlen,
+ int last_mrf,
int urb_offset);
private:
--
1.9.1
More information about the mesa-dev
mailing list