return container_of(abi, ctx, abi);
}
-static LLVMValueRef get_sampler_desc(struct nir_to_llvm_context *ctx,
+static LLVMValueRef get_sampler_desc(struct ac_nir_context *ctx,
const nir_deref_var *deref,
- enum ac_descriptor_type desc_type);
+ enum ac_descriptor_type desc_type,
+ bool image, bool write);
static unsigned radeon_llvm_reg_index_soa(unsigned index, unsigned chan)
{
return sample_index;
}
-static LLVMValueRef get_image_coords(struct nir_to_llvm_context *ctx,
+static LLVMValueRef get_image_coords(struct ac_nir_context *ctx,
const nir_intrinsic_instr *instr)
{
const struct glsl_type *type = instr->variables[0]->var->type;
if(instr->variables[0]->deref.child)
type = instr->variables[0]->deref.child->type;
- LLVMValueRef src0 = get_src(ctx->nir, instr->src[0]);
+ LLVMValueRef src0 = get_src(ctx, instr->src[0]);
LLVMValueRef coords[4];
LLVMValueRef masks[] = {
- LLVMConstInt(ctx->i32, 0, false), LLVMConstInt(ctx->i32, 1, false),
- LLVMConstInt(ctx->i32, 2, false), LLVMConstInt(ctx->i32, 3, false),
+ LLVMConstInt(ctx->ac.i32, 0, false), LLVMConstInt(ctx->ac.i32, 1, false),
+ LLVMConstInt(ctx->ac.i32, 2, false), LLVMConstInt(ctx->ac.i32, 3, false),
};
LLVMValueRef res;
- LLVMValueRef sample_index = llvm_extract_elem(&ctx->ac, get_src(ctx->nir, instr->src[1]), 0);
+ LLVMValueRef sample_index = llvm_extract_elem(&ctx->ac, get_src(ctx, instr->src[1]), 0);
int count;
enum glsl_sampler_dim dim = glsl_get_sampler_dim(type);
LLVMValueRef fmask_load_address[3];
int chan;
- fmask_load_address[0] = LLVMBuildExtractElement(ctx->builder, src0, masks[0], "");
- fmask_load_address[1] = LLVMBuildExtractElement(ctx->builder, src0, masks[1], "");
+ fmask_load_address[0] = LLVMBuildExtractElement(ctx->ac.builder, src0, masks[0], "");
+ fmask_load_address[1] = LLVMBuildExtractElement(ctx->ac.builder, src0, masks[1], "");
if (glsl_sampler_type_is_array(type))
- fmask_load_address[2] = LLVMBuildExtractElement(ctx->builder, src0, masks[2], "");
+ fmask_load_address[2] = LLVMBuildExtractElement(ctx->ac.builder, src0, masks[2], "");
else
fmask_load_address[2] = NULL;
if (add_frag_pos) {
for (chan = 0; chan < 2; ++chan)
- fmask_load_address[chan] = LLVMBuildAdd(ctx->builder, fmask_load_address[chan], LLVMBuildFPToUI(ctx->builder, ctx->frag_pos[chan], ctx->i32, ""), "");
+ fmask_load_address[chan] = LLVMBuildAdd(ctx->ac.builder, fmask_load_address[chan], LLVMBuildFPToUI(ctx->ac.builder, ctx->nctx->frag_pos[chan], ctx->ac.i32, ""), "");
}
sample_index = adjust_sample_index_using_fmask(&ctx->ac,
fmask_load_address[0],
fmask_load_address[1],
fmask_load_address[2],
sample_index,
- get_sampler_desc(ctx, instr->variables[0], AC_DESC_FMASK));
+ get_sampler_desc(ctx, instr->variables[0], AC_DESC_FMASK, true, false));
}
if (count == 1) {
if (instr->src[0].ssa->num_components)
- res = LLVMBuildExtractElement(ctx->builder, src0, masks[0], "");
+ res = LLVMBuildExtractElement(ctx->ac.builder, src0, masks[0], "");
else
res = src0;
} else {
if (is_ms)
count--;
for (chan = 0; chan < count; ++chan) {
- coords[chan] = LLVMBuildExtractElement(ctx->builder, src0, masks[chan], "");
+ coords[chan] = LLVMBuildExtractElement(ctx->ac.builder, src0, masks[chan], "");
}
if (add_frag_pos) {
for (chan = 0; chan < count; ++chan)
- coords[chan] = LLVMBuildAdd(ctx->builder, coords[chan], LLVMBuildFPToUI(ctx->builder, ctx->frag_pos[chan], ctx->i32, ""), "");
+ coords[chan] = LLVMBuildAdd(ctx->ac.builder, coords[chan], LLVMBuildFPToUI(ctx->ac.builder, ctx->nctx->frag_pos[chan], ctx->ac.i32, ""), "");
}
if (is_ms) {
coords[count] = sample_index;
}
if (count == 3) {
- coords[3] = LLVMGetUndef(ctx->i32);
+ coords[3] = LLVMGetUndef(ctx->ac.i32);
count = 4;
}
res = ac_build_gather_values(&ctx->ac, coords, count);
return res;
}
-static LLVMValueRef visit_image_load(struct nir_to_llvm_context *ctx,
+static LLVMValueRef visit_image_load(struct ac_nir_context *ctx,
const nir_intrinsic_instr *instr)
{
LLVMValueRef params[7];
char intrinsic_name[64];
const nir_variable *var = instr->variables[0]->var;
const struct glsl_type *type = var->type;
+ LLVMValueRef i1false = LLVMConstInt(ctx->ac.i1, 0, false);
+ LLVMValueRef i1true = LLVMConstInt(ctx->ac.i1, 1, false);
+
if(instr->variables[0]->deref.child)
type = instr->variables[0]->deref.child->type;
type = glsl_without_array(type);
if (glsl_get_sampler_dim(type) == GLSL_SAMPLER_DIM_BUF) {
- params[0] = get_sampler_desc(ctx, instr->variables[0], AC_DESC_BUFFER);
- params[1] = LLVMBuildExtractElement(ctx->builder, get_src(ctx->nir, instr->src[0]),
- LLVMConstInt(ctx->i32, 0, false), ""); /* vindex */
- params[2] = LLVMConstInt(ctx->i32, 0, false); /* voffset */
- params[3] = ctx->i1false; /* glc */
- params[4] = ctx->i1false; /* slc */
- res = ac_build_intrinsic(&ctx->ac, "llvm.amdgcn.buffer.load.format.v4f32", ctx->v4f32,
+ params[0] = get_sampler_desc(ctx, instr->variables[0], AC_DESC_BUFFER, true, false);
+ params[1] = LLVMBuildExtractElement(ctx->ac.builder, get_src(ctx, instr->src[0]),
+ ctx->ac.i32_0, ""); /* vindex */
+ params[2] = ctx->ac.i32_0; /* voffset */
+ params[3] = i1false; /* glc */
+ params[4] = i1false; /* slc */
+ res = ac_build_intrinsic(&ctx->ac, "llvm.amdgcn.buffer.load.format.v4f32", ctx->ac.v4f32,
params, 5, 0);
res = trim_vector(&ctx->ac, res, instr->dest.ssa.num_components);
} else {
bool is_da = glsl_sampler_type_is_array(type) ||
glsl_get_sampler_dim(type) == GLSL_SAMPLER_DIM_CUBE;
- LLVMValueRef da = is_da ? ctx->i1true : ctx->i1false;
- LLVMValueRef glc = ctx->i1false;
- LLVMValueRef slc = ctx->i1false;
+ LLVMValueRef da = is_da ? i1true : i1false;
+ LLVMValueRef glc = i1false;
+ LLVMValueRef slc = i1false;
params[0] = get_image_coords(ctx, instr);
- params[1] = get_sampler_desc(ctx, instr->variables[0], AC_DESC_IMAGE);
- params[2] = LLVMConstInt(ctx->i32, 15, false); /* dmask */
+ params[1] = get_sampler_desc(ctx, instr->variables[0], AC_DESC_IMAGE, true, false);
+ params[2] = LLVMConstInt(ctx->ac.i32, 15, false); /* dmask */
if (HAVE_LLVM <= 0x0309) {
- params[3] = ctx->i1false; /* r128 */
+ params[3] = i1false; /* r128 */
params[4] = da;
params[5] = glc;
params[6] = slc;
} else {
- LLVMValueRef lwe = ctx->i1false;
+ LLVMValueRef lwe = i1false;
params[3] = glc;
params[4] = slc;
params[5] = lwe;
}
ac_get_image_intr_name("llvm.amdgcn.image.load",
- ctx->v4f32, /* vdata */
+ ctx->ac.v4f32, /* vdata */
LLVMTypeOf(params[0]), /* coords */
LLVMTypeOf(params[1]), /* rsrc */
intrinsic_name, sizeof(intrinsic_name));
- res = ac_build_intrinsic(&ctx->ac, intrinsic_name, ctx->v4f32,
+ res = ac_build_intrinsic(&ctx->ac, intrinsic_name, ctx->ac.v4f32,
params, 7, AC_FUNC_ATTR_READONLY);
}
return to_integer(&ctx->ac, res);
}
-static void visit_image_store(struct nir_to_llvm_context *ctx,
+static void visit_image_store(struct ac_nir_context *ctx,
nir_intrinsic_instr *instr)
{
LLVMValueRef params[8];
char intrinsic_name[64];
const nir_variable *var = instr->variables[0]->var;
const struct glsl_type *type = glsl_without_array(var->type);
- LLVMValueRef glc = ctx->i1false;
- bool force_glc = ctx->options->chip_class == SI;
+ LLVMValueRef i1false = LLVMConstInt(ctx->ac.i1, 0, false);
+ LLVMValueRef i1true = LLVMConstInt(ctx->ac.i1, 1, false);
+ LLVMValueRef glc = i1false;
+ bool force_glc = ctx->abi->chip_class == SI;
if (force_glc)
- glc = ctx->i1true;
- if (ctx->stage == MESA_SHADER_FRAGMENT)
- ctx->shader_info->fs.writes_memory = true;
+ glc = i1true;
if (glsl_get_sampler_dim(type) == GLSL_SAMPLER_DIM_BUF) {
- params[0] = to_float(&ctx->ac, get_src(ctx->nir, instr->src[2])); /* data */
- params[1] = get_sampler_desc(ctx, instr->variables[0], AC_DESC_BUFFER);
- params[2] = LLVMBuildExtractElement(ctx->builder, get_src(ctx->nir, instr->src[0]),
- LLVMConstInt(ctx->i32, 0, false), ""); /* vindex */
- params[3] = LLVMConstInt(ctx->i32, 0, false); /* voffset */
+ params[0] = to_float(&ctx->ac, get_src(ctx, instr->src[2])); /* data */
+ params[1] = get_sampler_desc(ctx, instr->variables[0], AC_DESC_BUFFER, true, true);
+ params[2] = LLVMBuildExtractElement(ctx->ac.builder, get_src(ctx, instr->src[0]),
+ ctx->ac.i32_0, ""); /* vindex */
+ params[3] = ctx->ac.i32_0; /* voffset */
params[4] = glc; /* glc */
- params[5] = ctx->i1false; /* slc */
- ac_build_intrinsic(&ctx->ac, "llvm.amdgcn.buffer.store.format.v4f32", ctx->voidt,
+ params[5] = i1false; /* slc */
+ ac_build_intrinsic(&ctx->ac, "llvm.amdgcn.buffer.store.format.v4f32", ctx->ac.voidt,
params, 6, 0);
} else {
bool is_da = glsl_sampler_type_is_array(type) ||
glsl_get_sampler_dim(type) == GLSL_SAMPLER_DIM_CUBE;
- LLVMValueRef da = is_da ? ctx->i1true : ctx->i1false;
- LLVMValueRef slc = ctx->i1false;
+ LLVMValueRef da = is_da ? i1true : i1false;
+ LLVMValueRef slc = i1false;
- params[0] = to_float(&ctx->ac, get_src(ctx->nir, instr->src[2]));
+ params[0] = to_float(&ctx->ac, get_src(ctx, instr->src[2]));
params[1] = get_image_coords(ctx, instr); /* coords */
- params[2] = get_sampler_desc(ctx, instr->variables[0], AC_DESC_IMAGE);
- params[3] = LLVMConstInt(ctx->i32, 15, false); /* dmask */
+ params[2] = get_sampler_desc(ctx, instr->variables[0], AC_DESC_IMAGE, true, true);
+ params[3] = LLVMConstInt(ctx->ac.i32, 15, false); /* dmask */
if (HAVE_LLVM <= 0x0309) {
- params[4] = ctx->i1false; /* r128 */
+ params[4] = i1false; /* r128 */
params[5] = da;
params[6] = glc;
params[7] = slc;
} else {
- LLVMValueRef lwe = ctx->i1false;
+ LLVMValueRef lwe = i1false;
params[4] = glc;
params[5] = slc;
params[6] = lwe;
LLVMTypeOf(params[2]), /* rsrc */
intrinsic_name, sizeof(intrinsic_name));
- ac_build_intrinsic(&ctx->ac, intrinsic_name, ctx->voidt,
+ ac_build_intrinsic(&ctx->ac, intrinsic_name, ctx->ac.voidt,
params, 8, 0);
}
}
-static LLVMValueRef visit_image_atomic(struct nir_to_llvm_context *ctx,
+static LLVMValueRef visit_image_atomic(struct ac_nir_context *ctx,
const nir_intrinsic_instr *instr)
{
LLVMValueRef params[6];
const char *atomic_name;
char intrinsic_name[41];
const struct glsl_type *type = glsl_without_array(var->type);
+ LLVMValueRef i1false = LLVMConstInt(ctx->ac.i1, 0, false);
+ LLVMValueRef i1true = LLVMConstInt(ctx->ac.i1, 1, false);
MAYBE_UNUSED int length;
- if (ctx->stage == MESA_SHADER_FRAGMENT)
- ctx->shader_info->fs.writes_memory = true;
-
switch (instr->intrinsic) {
case nir_intrinsic_image_atomic_add:
atomic_name = "add";
}
if (instr->intrinsic == nir_intrinsic_image_atomic_comp_swap)
- params[param_count++] = get_src(ctx->nir, instr->src[3]);
- params[param_count++] = get_src(ctx->nir, instr->src[2]);
+ params[param_count++] = get_src(ctx, instr->src[3]);
+ params[param_count++] = get_src(ctx, instr->src[2]);
if (glsl_get_sampler_dim(type) == GLSL_SAMPLER_DIM_BUF) {
- params[param_count++] = get_sampler_desc(ctx, instr->variables[0], AC_DESC_BUFFER);
- params[param_count++] = LLVMBuildExtractElement(ctx->builder, get_src(ctx->nir, instr->src[0]),
- LLVMConstInt(ctx->i32, 0, false), ""); /* vindex */
- params[param_count++] = ctx->i32zero; /* voffset */
- params[param_count++] = ctx->i1false; /* slc */
+ params[param_count++] = get_sampler_desc(ctx, instr->variables[0], AC_DESC_BUFFER,
+ true, true);
+ params[param_count++] = LLVMBuildExtractElement(ctx->ac.builder, get_src(ctx, instr->src[0]),
+ ctx->ac.i32_0, ""); /* vindex */
+ params[param_count++] = ctx->ac.i32_0; /* voffset */
+ params[param_count++] = i1false; /* slc */
length = snprintf(intrinsic_name, sizeof(intrinsic_name),
"llvm.amdgcn.buffer.atomic.%s", atomic_name);
glsl_get_sampler_dim(type) == GLSL_SAMPLER_DIM_CUBE;
LLVMValueRef coords = params[param_count++] = get_image_coords(ctx, instr);
- params[param_count++] = get_sampler_desc(ctx, instr->variables[0], AC_DESC_IMAGE);
- params[param_count++] = ctx->i1false; /* r128 */
- params[param_count++] = da ? ctx->i1true : ctx->i1false; /* da */
- params[param_count++] = ctx->i1false; /* slc */
+ params[param_count++] = get_sampler_desc(ctx, instr->variables[0], AC_DESC_IMAGE,
+ true, true);
+ params[param_count++] = i1false; /* r128 */
+ params[param_count++] = da ? i1true : i1false; /* da */
+ params[param_count++] = i1false; /* slc */
build_int_type_name(LLVMTypeOf(coords),
coords_type, sizeof(coords_type));
}
assert(length < sizeof(intrinsic_name));
- return ac_build_intrinsic(&ctx->ac, intrinsic_name, ctx->i32, params, param_count, 0);
+ return ac_build_intrinsic(&ctx->ac, intrinsic_name, ctx->ac.i32, params, param_count, 0);
}
static LLVMValueRef visit_image_size(struct ac_nir_context *ctx,
type = instr->variables[0]->deref.child->type;
if (glsl_get_sampler_dim(type) == GLSL_SAMPLER_DIM_BUF)
- return get_buffer_size(ctx, get_sampler_desc(ctx->nctx, instr->variables[0], AC_DESC_BUFFER), true);
+ return get_buffer_size(ctx,
+ get_sampler_desc(ctx, instr->variables[0],
+ AC_DESC_BUFFER, true, false), true);
struct ac_image_args args = { 0 };
args.da = da;
args.dmask = 0xf;
- args.resource = get_sampler_desc(ctx->nctx, instr->variables[0], AC_DESC_IMAGE);
+ args.resource = get_sampler_desc(ctx, instr->variables[0], AC_DESC_IMAGE, true, false);
args.opcode = ac_image_get_resinfo;
args.addr = ctx->ac.i32_0;
const nir_intrinsic_instr *instr)
{
LLVMValueRef cond;
- ctx->shader_info->fs.can_discard = true;
cond = LLVMBuildICmp(ctx->builder, LLVMIntNE,
get_src(ctx->nir, instr->src[0]),
break;
case nir_intrinsic_load_instance_id:
result = ctx->abi->instance_id;
- ctx->nctx->shader_info->vs.vgpr_comp_cnt = MAX2(3,
- ctx->nctx->shader_info->vs.vgpr_comp_cnt);
break;
case nir_intrinsic_load_num_work_groups:
result = ctx->nctx->num_work_groups;
visit_store_var(ctx, instr);
break;
case nir_intrinsic_image_load:
- result = visit_image_load(ctx->nctx, instr);
+ result = visit_image_load(ctx, instr);
break;
case nir_intrinsic_image_store:
- visit_image_store(ctx->nctx, instr);
+ visit_image_store(ctx, instr);
break;
case nir_intrinsic_image_atomic_add:
case nir_intrinsic_image_atomic_min:
case nir_intrinsic_image_atomic_xor:
case nir_intrinsic_image_atomic_exchange:
case nir_intrinsic_image_atomic_comp_swap:
- result = visit_image_atomic(ctx->nctx, instr);
+ result = visit_image_atomic(ctx, instr);
break;
case nir_intrinsic_image_size:
result = visit_image_size(ctx, instr);
break;
case nir_intrinsic_discard:
- ctx->nctx->shader_info->fs.can_discard = true;
ac_build_intrinsic(&ctx->ac, "llvm.AMDGPU.kilp",
LLVMVoidTypeInContext(ctx->ac.context),
NULL, 0, AC_FUNC_ATTR_LEGACY);
unsigned base_index,
unsigned constant_index,
LLVMValueRef index,
- enum ac_descriptor_type desc_type)
+ enum ac_descriptor_type desc_type,
+ bool image, bool write)
{
struct nir_to_llvm_context *ctx = nir_to_llvm_context_from_abi(abi);
LLVMValueRef list = ctx->descriptor_sets[descriptor_set];
assert(base_index < layout->binding_count);
+ if (write && ctx->stage == MESA_SHADER_FRAGMENT)
+ ctx->shader_info->fs.writes_memory = true;
+
switch (desc_type) {
case AC_DESC_IMAGE:
type = ctx->v8i32;
return ac_build_indexed_load_const(&ctx->ac, list, index);
}
-static LLVMValueRef get_sampler_desc(struct nir_to_llvm_context *ctx,
+static LLVMValueRef get_sampler_desc(struct ac_nir_context *ctx,
const nir_deref_var *deref,
- enum ac_descriptor_type desc_type)
+ enum ac_descriptor_type desc_type,
+ bool image, bool write)
{
LLVMValueRef index = NULL;
unsigned constant_index = 0;
+ const nir_deref *tail = &deref->deref;
+
+ while (tail->child) {
+ const nir_deref_array *child = nir_deref_as_array(tail->child);
+ unsigned array_size = glsl_get_aoa_size(tail->child->type);
- if (deref->deref.child) {
- const nir_deref_array *child =
- (const nir_deref_array *)deref->deref.child;
+ if (!array_size)
+ array_size = 1;
assert(child->deref_array_type != nir_deref_array_type_wildcard);
+
if (child->deref_array_type == nir_deref_array_type_indirect) {
- index = get_src(ctx->nir, child->indirect);
+ LLVMValueRef indirect = get_src(ctx, child->indirect);
+
+ indirect = LLVMBuildMul(ctx->ac.builder, indirect,
+ LLVMConstInt(ctx->ac.i32, array_size, false), "");
+
+ if (!index)
+ index = indirect;
+ else
+ index = LLVMBuildAdd(ctx->ac.builder, index, indirect, "");
}
- constant_index = child->base_offset;
+ constant_index += child->base_offset * array_size;
+
+ tail = &child->deref;
}
- return ctx->abi.load_sampler_desc(&ctx->abi,
+ return ctx->abi->load_sampler_desc(ctx->abi,
deref->var->data.descriptor_set,
deref->var->data.binding,
constant_index, index,
- desc_type);
+ desc_type, image, write);
}
static void set_tex_fetch_args(struct ac_llvm_context *ctx,
* VI:
* The ANISO_OVERRIDE sampler field enables this fix in TA.
*/
-static LLVMValueRef sici_fix_sampler_aniso(struct nir_to_llvm_context *ctx,
+static LLVMValueRef sici_fix_sampler_aniso(struct ac_nir_context *ctx,
LLVMValueRef res, LLVMValueRef samp)
{
- LLVMBuilderRef builder = ctx->builder;
+ LLVMBuilderRef builder = ctx->ac.builder;
LLVMValueRef img7, samp0;
- if (ctx->options->chip_class >= VI)
+ if (ctx->abi->chip_class >= VI)
return samp;
img7 = LLVMBuildExtractElement(builder, res,
- LLVMConstInt(ctx->i32, 7, 0), "");
+ LLVMConstInt(ctx->ac.i32, 7, 0), "");
samp0 = LLVMBuildExtractElement(builder, samp,
- LLVMConstInt(ctx->i32, 0, 0), "");
+ LLVMConstInt(ctx->ac.i32, 0, 0), "");
samp0 = LLVMBuildAnd(builder, samp0, img7, "");
return LLVMBuildInsertElement(builder, samp, samp0,
- LLVMConstInt(ctx->i32, 0, 0), "");
+ LLVMConstInt(ctx->ac.i32, 0, 0), "");
}
-static void tex_fetch_ptrs(struct nir_to_llvm_context *ctx,
+static void tex_fetch_ptrs(struct ac_nir_context *ctx,
nir_tex_instr *instr,
LLVMValueRef *res_ptr, LLVMValueRef *samp_ptr,
LLVMValueRef *fmask_ptr)
{
if (instr->sampler_dim == GLSL_SAMPLER_DIM_BUF)
- *res_ptr = get_sampler_desc(ctx, instr->texture, AC_DESC_BUFFER);
+ *res_ptr = get_sampler_desc(ctx, instr->texture, AC_DESC_BUFFER, false, false);
else
- *res_ptr = get_sampler_desc(ctx, instr->texture, AC_DESC_IMAGE);
+ *res_ptr = get_sampler_desc(ctx, instr->texture, AC_DESC_IMAGE, false, false);
if (samp_ptr) {
if (instr->sampler)
- *samp_ptr = get_sampler_desc(ctx, instr->sampler, AC_DESC_SAMPLER);
+ *samp_ptr = get_sampler_desc(ctx, instr->sampler, AC_DESC_SAMPLER, false, false);
else
- *samp_ptr = get_sampler_desc(ctx, instr->texture, AC_DESC_SAMPLER);
+ *samp_ptr = get_sampler_desc(ctx, instr->texture, AC_DESC_SAMPLER, false, false);
if (instr->sampler_dim < GLSL_SAMPLER_DIM_RECT)
*samp_ptr = sici_fix_sampler_aniso(ctx, *res_ptr, *samp_ptr);
}
if (fmask_ptr && !instr->sampler && (instr->op == nir_texop_txf_ms ||
instr->op == nir_texop_samples_identical))
- *fmask_ptr = get_sampler_desc(ctx, instr->texture, AC_DESC_FMASK);
+ *fmask_ptr = get_sampler_desc(ctx, instr->texture, AC_DESC_FMASK, false, false);
}
static LLVMValueRef apply_round_slice(struct ac_llvm_context *ctx,
unsigned const_src = 0, num_deriv_comp = 0;
bool lod_is_zero = false;
- tex_fetch_ptrs(ctx->nctx, instr, &res_ptr, &samp_ptr, &fmask_ptr);
+ tex_fetch_ptrs(ctx, instr, &res_ptr, &samp_ptr, &fmask_ptr);
for (unsigned i = 0; i < instr->num_srcs; i++) {
switch (instr->src[i].src_type) {
if (instr->op == nir_texop_query_levels)
result = LLVMBuildExtractElement(ctx->ac.builder, result, LLVMConstInt(ctx->ac.i32, 3, false), "");
- else if (instr->is_shadow && instr->op != nir_texop_txs && instr->op != nir_texop_lod && instr->op != nir_texop_tg4)
+ else if (instr->is_shadow && instr->is_new_style_shadow &&
+ instr->op != nir_texop_txs && instr->op != nir_texop_lod &&
+ instr->op != nir_texop_tg4)
result = LLVMBuildExtractElement(ctx->ac.builder, result, ctx->ac.i32_0, "");
else if (instr->op == nir_texop_txs &&
instr->sampler_dim == GLSL_SAMPLER_DIM_CUBE &&
memset(shader_info, 0, sizeof(*shader_info));
ac_nir_shader_info_pass(nir, options, &shader_info->info);
-
+
LLVMSetTarget(ctx.module, options->supports_spill ? "amdgcn-mesa-mesa3d" : "amdgcn--");
LLVMTargetDataRef data_layout = LLVMCreateTargetDataLayout(tm);
ctx.gs_max_out_vertices = nir->info.gs.vertices_out;
} else if (nir->stage == MESA_SHADER_TESS_EVAL) {
ctx.tes_primitive_mode = nir->info.tess.primitive_mode;
+ } else if (nir->stage == MESA_SHADER_VERTEX) {
+ if (shader_info->info.vs.needs_instance_id) {
+ ctx.shader_info->vs.vgpr_comp_cnt =
+ MAX2(3, ctx.shader_info->vs.vgpr_comp_cnt);
+ }
+ } else if (nir->stage == MESA_SHADER_FRAGMENT) {
+ shader_info->fs.can_discard = nir->info.fs.uses_discard;
}
ac_setup_rings(&ctx);