From: Marek Olšák <marek.ol...@amd.com>
---
src/amd/common/ac_llvm_build.c | 9 ++++-----
src/amd/common/ac_llvm_build.h | 3 +--
src/amd/common/ac_nir_to_llvm.c | 6 +++---
src/gallium/drivers/radeonsi/si_shader.c | 25 ++++++++++++-------------
4 files changed, 20 insertions(+), 23 deletions(-)
diff --git a/src/amd/common/ac_llvm_build.c b/src/amd/common/ac_llvm_build.c
index cc1eaf1..08fedc7 100644
--- a/src/amd/common/ac_llvm_build.c
+++ b/src/amd/common/ac_llvm_build.c
@@ -540,53 +540,52 @@ ac_build_indexed_load_const(struct ac_llvm_context *ctx,
/* TBUFFER_STORE_FORMAT_{X,XY,XYZ,XYZW} <- the suffix is selected by
num_channels=1..4.
* The type of vdata must be one of i32 (num_channels=1), v2i32
(num_channels=2),
* or v4i32 (num_channels=3,4).
*/
void
ac_build_buffer_store_dword(struct ac_llvm_context *ctx,
LLVMValueRef rsrc,
LLVMValueRef vdata,
unsigned num_channels,
- LLVMValueRef vaddr,
+ LLVMValueRef voffset,
LLVMValueRef soffset,
unsigned inst_offset,
- bool offen,
bool glc,
bool slc)
{
static unsigned dfmt[] = {
V_008F0C_BUF_DATA_FORMAT_32,
V_008F0C_BUF_DATA_FORMAT_32_32,
V_008F0C_BUF_DATA_FORMAT_32_32_32,
V_008F0C_BUF_DATA_FORMAT_32_32_32_32
};
assert(num_channels >= 1 && num_channels <= 4);
LLVMValueRef args[] = {
rsrc,
vdata,
LLVMConstInt(ctx->i32, num_channels, 0),
- vaddr,
+ voffset ? voffset : LLVMGetUndef(ctx->i32),
soffset,
LLVMConstInt(ctx->i32, inst_offset, 0),
LLVMConstInt(ctx->i32, dfmt[num_channels - 1], 0),
LLVMConstInt(ctx->i32, V_008F0C_BUF_NUM_FORMAT_UINT, 0),
- LLVMConstInt(ctx->i32, offen, 0),
+ LLVMConstInt(ctx->i32, voffset != NULL, 0),
LLVMConstInt(ctx->i32, 0, 0), /* idxen */
LLVMConstInt(ctx->i32, glc, 0),
LLVMConstInt(ctx->i32, slc, 0),
LLVMConstInt(ctx->i32, 0, 0), /* tfe*/
};
/* The instruction offset field has 12 bits */
- assert(offen || inst_offset < (1 << 12));
+ assert(voffset || inst_offset < (1 << 12));
/* The intrinsic is overloaded, we need to add a type suffix for
overloading to work. */
unsigned func = CLAMP(num_channels, 1, 3) - 1;
const char *types[] = {"i32", "v2i32", "v4i32"};
char name[256];
snprintf(name, sizeof(name), "llvm.SI.tbuffer.store.%s", types[func]);
ac_emit_llvm_intrinsic(ctx, name, ctx->voidt,
args, ARRAY_SIZE(args),
AC_FUNC_ATTR_LEGACY);
diff --git a/src/amd/common/ac_llvm_build.h b/src/amd/common/ac_llvm_build.h
index 65a9a05..78df441 100644
--- a/src/amd/common/ac_llvm_build.h
+++ b/src/amd/common/ac_llvm_build.h
@@ -119,24 +119,23 @@ ac_build_indexed_load(struct ac_llvm_context *ctx,
LLVMValueRef
ac_build_indexed_load_const(struct ac_llvm_context *ctx,
LLVMValueRef base_ptr, LLVMValueRef index);
void
ac_build_buffer_store_dword(struct ac_llvm_context *ctx,
LLVMValueRef rsrc,
LLVMValueRef vdata,
unsigned num_channels,
- LLVMValueRef vaddr,
+ LLVMValueRef voffset,
LLVMValueRef soffset,
unsigned inst_offset,
- bool offen,
bool glc,
bool slc);
LLVMValueRef
ac_build_buffer_load(struct ac_llvm_context *ctx,
LLVMValueRef rsrc,
int num_channels,
LLVMValueRef vindex,
LLVMValueRef voffset,
LLVMValueRef soffset,
unsigned inst_offset,
diff --git a/src/amd/common/ac_nir_to_llvm.c b/src/amd/common/ac_nir_to_llvm.c
index 4143b3c..9a91e1a 100644
--- a/src/amd/common/ac_nir_to_llvm.c
+++ b/src/amd/common/ac_nir_to_llvm.c
@@ -3134,21 +3134,21 @@ visit_emit_vertex(struct nir_to_llvm_context *ctx,
out_ptr[j], "");
LLVMValueRef voffset = LLVMConstInt(ctx->i32, (slot * 4
+ j + start) * ctx->gs_max_out_vertices, false);
voffset = LLVMBuildAdd(ctx->builder, voffset,
gs_next_vertex, "");
voffset = LLVMBuildMul(ctx->builder, voffset,
LLVMConstInt(ctx->i32, 4, false), "");
out_val = LLVMBuildBitCast(ctx->builder, out_val,
ctx->i32, "");
ac_build_buffer_store_dword(&ctx->ac, ctx->gsvs_ring,
out_val, 1,
voffset, ctx->gs2vs_offset,
0,
- 1, 1, 1);
+ 1, 1);
}
idx += slot_inc;
}
gs_next_vertex = LLVMBuildAdd(ctx->builder, gs_next_vertex,
ctx->i32one, "");
LLVMBuildStore(ctx->builder, gs_next_vertex, ctx->gs_next_vertex);
ac_emit_sendmsg(&ctx->ac, AC_SENDMSG_GS_OP_EMIT | AC_SENDMSG_GS | (0 <<
8), ctx->gs_wave_id);
}
@@ -4632,23 +4632,23 @@ handle_es_outputs_post(struct nir_to_llvm_context *ctx)
if (param_index > max_output_written)
max_output_written = param_index;
for (j = 0; j < length; j++) {
LLVMValueRef out_val = LLVMBuildLoad(ctx->builder,
out_ptr[j], "");
out_val = LLVMBuildBitCast(ctx->builder, out_val,
ctx->i32, "");
ac_build_buffer_store_dword(&ctx->ac,
ctx->esgs_ring,
out_val, 1,
- LLVMGetUndef(ctx->i32),
ctx->es2gs_offset,
+ NULL, ctx->es2gs_offset,
(4 * param_index + j + start) *
4,
- 0, 1, 1);
+ 1, 1);
}
}
ctx->shader_info->vs.esgs_itemsize = (max_output_written + 1) * 16;
}
static void
si_export_mrt_color(struct nir_to_llvm_context *ctx,
LLVMValueRef *color, unsigned param, bool is_last)
{
LLVMValueRef args[9];
diff --git a/src/gallium/drivers/radeonsi/si_shader.c
b/src/gallium/drivers/radeonsi/si_shader.c
index 71b5b7a..dd5bdf6 100644
--- a/src/gallium/drivers/radeonsi/si_shader.c
+++ b/src/gallium/drivers/radeonsi/si_shader.c
@@ -1042,29 +1042,29 @@ static void store_output_tcs(struct
lp_build_tgsi_context *bld_base,
/* Skip LDS stores if there is no LDS read of this output. */
if (!skip_lds_store)
lds_store(bld_base, chan_index, dw_addr, value);
value = LLVMBuildBitCast(gallivm->builder, value, ctx->i32, "");
values[chan_index] = value;
if (inst->Dst[0].Register.WriteMask != 0xF && !is_tess_factor) {
ac_build_buffer_store_dword(&ctx->ac, buffer, value, 1,
buf_addr, base,
- 4 * chan_index, 1, 1, 0);
+ 4 * chan_index, 1, 0);
}
}
if (inst->Dst[0].Register.WriteMask == 0xF && !is_tess_factor) {
LLVMValueRef value =
lp_build_gather_values(bld_base->base.gallivm,
values, 4);
ac_build_buffer_store_dword(&ctx->ac, buffer, value, 4,
buf_addr,
- base, 0, 1, 1, 0);
+ base, 0, 1, 0);
}
}
static LLVMValueRef fetch_input_gs(
struct lp_build_tgsi_context *bld_base,
const struct tgsi_full_src_register *reg,
enum tgsi_opcode_type type,
unsigned swizzle)
{
struct lp_build_context *base = &bld_base->base;
@@ -2080,21 +2080,21 @@ static void emit_streamout_output(struct
si_shader_context *ctx,
vdata = LLVMBuildInsertElement(builder, vdata, out[j],
LLVMConstInt(ctx->i32,
j, 0), "");
}
break;
}
ac_build_buffer_store_dword(&ctx->ac, so_buffers[buf_idx],
vdata, num_comps,
so_write_offsets[buf_idx],
LLVMConstInt(ctx->i32, 0, 0),
- stream_out->dst_offset * 4, 1, 1, 1);
+ stream_out->dst_offset * 4, 1, 1);
}
/**
* Write streamout data to buffers for vertex stream @p stream (different
* vertex streams can occur for GS copy shaders).
*/
static void si_llvm_emit_streamout(struct si_shader_context *ctx,
struct si_shader_output_values *outputs,
unsigned noutput, unsigned stream)
{
@@ -2405,21 +2405,21 @@ static void si_copy_tcs_inputs(struct
lp_build_tgsi_context *bld_base)
LLVMValueRef buffer_addr = get_tcs_tes_buffer_address(ctx,
get_rel_patch_id(ctx),
invocation_id,
lp_build_const_int32(gallivm, i));
LLVMValueRef value = lds_load(bld_base, TGSI_TYPE_SIGNED, ~0,
lds_ptr);
ac_build_buffer_store_dword(&ctx->ac, buffer, value, 4,
buffer_addr,
- buffer_offset, 0, 1, 1, 0);
+ buffer_offset, 0, 1, 0);
}
}
static void si_write_tess_factors(struct lp_build_tgsi_context *bld_base,
LLVMValueRef rel_patch_id,
LLVMValueRef invocation_id,
LLVMValueRef
tcs_out_current_patch_data_offset)
{
struct si_shader_context *ctx = si_shader_context(bld_base);
struct gallivm_state *gallivm = bld_base->base.gallivm;
@@ -2520,32 +2520,32 @@ static void si_write_tess_factors(struct
lp_build_tgsi_context *bld_base,
lp_build_const_int32(gallivm, 4 * stride),
"");
lp_build_if(&inner_if_ctx, gallivm,
LLVMBuildICmp(gallivm->builder, LLVMIntEQ,
rel_patch_id, bld_base->uint_bld.zero, ""));
/* Store the dynamic HS control word. */
ac_build_buffer_store_dword(&ctx->ac, buffer,
lp_build_const_int32(gallivm, 0x80000000),
1, lp_build_const_int32(gallivm, 0),
tf_base,
- 0, 1, 1, 0);
+ 0, 1, 0);
lp_build_endif(&inner_if_ctx);
/* Store the tessellation factors. */
ac_build_buffer_store_dword(&ctx->ac, buffer, vec0,
MIN2(stride, 4), byteoffset, tf_base,
- 4, 1, 1, 0);
+ 4, 1, 0);
if (vec1)
ac_build_buffer_store_dword(&ctx->ac, buffer, vec1,
stride - 4, byteoffset, tf_base,
- 20, 1, 1, 0);
+ 20, 1, 0);
/* Store the tess factors into the offchip buffer if TES reads them. */
if (shader->key.part.tcs.epilog.tes_reads_tess_factors) {
LLVMValueRef buf, base, inner_vec, outer_vec, tf_outer_offset;
LLVMValueRef tf_inner_offset;
unsigned param_outer, param_inner;
buf = ac_build_indexed_load_const(&ctx->ac, rw_buffers,
LLVMConstInt(ctx->i32, SI_HS_RING_TESS_OFFCHIP,
0));
base = LLVMGetParam(ctx->main_fn, ctx->param_oc_lds);
@@ -2553,32 +2553,32 @@ static void si_write_tess_factors(struct
lp_build_tgsi_context *bld_base,
param_outer = si_shader_io_get_unique_index(
TGSI_SEMANTIC_TESSOUTER, 0);
tf_outer_offset = get_tcs_tes_buffer_address(ctx, rel_patch_id,
NULL,
LLVMConstInt(ctx->i32, param_outer, 0));
outer_vec = lp_build_gather_values(gallivm, outer,
util_next_power_of_two(outer_comps));
ac_build_buffer_store_dword(&ctx->ac, buf, outer_vec,
outer_comps, tf_outer_offset,
- base, 0, 1, 1, 0);
+ base, 0, 1, 0);
if (inner_comps) {
param_inner = si_shader_io_get_unique_index(
TGSI_SEMANTIC_TESSINNER, 0);
tf_inner_offset = get_tcs_tes_buffer_address(ctx,
rel_patch_id, NULL,
LLVMConstInt(ctx->i32, param_inner, 0));
inner_vec = inner_comps == 1 ? inner[0] :
lp_build_gather_values(gallivm, inner,
inner_comps);
ac_build_buffer_store_dword(&ctx->ac, buf, inner_vec,
inner_comps,
tf_inner_offset,
- base, 0, 1, 1, 0);
+ base, 0, 1, 0);
}
}
lp_build_endif(&if_ctx);
}
/* This only writes the tessellation factor levels. */
static void si_llvm_emit_tcs_epilogue(struct lp_build_tgsi_context *bld_base)
{
struct si_shader_context *ctx = si_shader_context(bld_base);
@@ -2686,24 +2686,23 @@ static void si_llvm_emit_es_epilogue(struct
lp_build_tgsi_context *bld_base)
param_index =
si_shader_io_get_unique_index(info->output_semantic_name[i],
info->output_semantic_index[i]);
for (chan = 0; chan < 4; chan++) {
LLVMValueRef out_val = LLVMBuildLoad(gallivm->builder,
out_ptr[chan], "");
out_val = LLVMBuildBitCast(gallivm->builder, out_val,
ctx->i32, "");
ac_build_buffer_store_dword(&ctx->ac,
ctx->esgs_ring,
- out_val, 1,
- LLVMGetUndef(ctx->i32),
soffset,
+ out_val, 1, NULL, soffset,
(4 * param_index + chan) *
4,
- 0, 1, 1);
+ 1, 1);
}
}
}
static void si_llvm_emit_gs_epilogue(struct lp_build_tgsi_context *bld_base)
{
struct si_shader_context *ctx = si_shader_context(bld_base);
ac_emit_sendmsg(&ctx->ac, AC_SENDMSG_GS_OP_NOP | AC_SENDMSG_GS_DONE,
LLVMGetParam(ctx->main_fn, SI_PARAM_GS_WAVE_ID));
@@ -5054,21 +5053,21 @@ static void si_llvm_emit_vertex(
voffset = lp_build_add(uint, voffset, gs_next_vertex);
voffset = lp_build_mul_imm(uint, voffset, 4);
out_val = LLVMBuildBitCast(gallivm->builder, out_val,
ctx->i32, "");
ac_build_buffer_store_dword(&ctx->ac,
ctx->gsvs_ring[stream],
out_val, 1,
voffset, soffset, 0,
- 1, 1, 1);
+ 1, 1);
}
}
gs_next_vertex = lp_build_add(uint, gs_next_vertex,
lp_build_const_int32(gallivm, 1));
LLVMBuildStore(gallivm->builder, gs_next_vertex,
ctx->gs_next_vertex[stream]);
/* Signal vertex emission */
ac_emit_sendmsg(&ctx->ac, AC_SENDMSG_GS_OP_EMIT | AC_SENDMSG_GS |
(stream << 8),
--
2.7.4
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev
