[Mesa-dev] [PATCH] radv: implement multisample image copies
Dave Airlie
airlied at gmail.com
Fri Mar 9 06:30:44 UTC 2018
From: Dave Airlie <airlied at redhat.com>
It appears its quite legal to do image copies on multisample
images, however due to a bug in our txf handling and incomplete
tests we never actually noticed we didn't do it properly in radv.
This patch implements a compute shader to copy multiple samples
of an image to another image. It implements the nir txf_ms_mcs
opcode for getting the fmask value, and then uses that value
to either copy sample 0 to all samples, or iterate across
the valid samples copying them.
The shader is inspired by one RE'd from AMDVLK.
Fixes:
dEQP-VK.api.copy_and_blit.core.resolve_image.whole_array_image*
Signed-off-by: Dave Airlie <airlied at redhat.com>
---
src/amd/common/ac_nir_to_llvm.c | 54 ++++++---
src/amd/vulkan/radv_meta_bufimage.c | 229 ++++++++++++++++++++++++++++++++++--
src/amd/vulkan/radv_meta_copy.c | 2 +-
src/amd/vulkan/radv_private.h | 2 +-
4 files changed, 258 insertions(+), 29 deletions(-)
diff --git a/src/amd/common/ac_nir_to_llvm.c b/src/amd/common/ac_nir_to_llvm.c
index 9b850698608..071f54a5b61 100644
--- a/src/amd/common/ac_nir_to_llvm.c
+++ b/src/amd/common/ac_nir_to_llvm.c
@@ -2277,6 +2277,7 @@ static LLVMValueRef build_tex_intrinsic(struct ac_nir_context *ctx,
switch (instr->op) {
case nir_texop_txf:
case nir_texop_txf_ms:
+ case nir_texop_txf_ms_mcs:
case nir_texop_samples_identical:
args->opcode = lod_is_zero ||
instr->sampler_dim == GLSL_SAMPLER_DIM_MS ?
@@ -3417,6 +3418,24 @@ glsl_is_array_image(const struct glsl_type *type)
}
+static LLVMValueRef get_fmask_desc_valid(struct ac_llvm_context *ctx,
+ LLVMValueRef fmask_desc_ptr)
+{
+ LLVMValueRef fmask_desc =
+ LLVMBuildBitCast(ctx->builder, fmask_desc_ptr,
+ ctx->v8i32, "");
+
+ LLVMValueRef fmask_word1 =
+ LLVMBuildExtractElement(ctx->builder, fmask_desc,
+ ctx->i32_1, "");
+
+ LLVMValueRef word1_is_nonzero =
+ LLVMBuildICmp(ctx->builder, LLVMIntNE,
+ fmask_word1, ctx->i32_0, "");
+
+ return word1_is_nonzero;
+}
+
/* Adjust the sample index according to FMASK.
*
* For uncompressed MSAA surfaces, FMASK should return 0x76543210,
@@ -3475,17 +3494,7 @@ static LLVMValueRef adjust_sample_index_using_fmask(struct ac_llvm_context *ctx,
/* Don't rewrite the sample index if WORD1.DATA_FORMAT of the FMASK
* resource descriptor is 0 (invalid),
*/
- LLVMValueRef fmask_desc =
- LLVMBuildBitCast(ctx->builder, fmask_desc_ptr,
- ctx->v8i32, "");
-
- LLVMValueRef fmask_word1 =
- LLVMBuildExtractElement(ctx->builder, fmask_desc,
- ctx->i32_1, "");
-
- LLVMValueRef word1_is_nonzero =
- LLVMBuildICmp(ctx->builder, LLVMIntNE,
- fmask_word1, ctx->i32_0, "");
+ LLVMValueRef word1_is_nonzero = get_fmask_desc_valid(ctx, fmask_desc_ptr);
/* Replace the MSAA sample index. */
sample_index =
@@ -3518,7 +3527,7 @@ static LLVMValueRef get_image_coords(struct ac_nir_context *ctx,
bool gfx9_1d = ctx->ac.chip_class >= GFX9 && dim == GLSL_SAMPLER_DIM_1D;
count = image_type_to_components_count(dim, is_array);
- if (is_ms) {
+ if (is_ms && instr->intrinsic == nir_intrinsic_image_load) {
LLVMValueRef fmask_load_address[3];
int chan;
@@ -4899,7 +4908,7 @@ static void tex_fetch_ptrs(struct ac_nir_context *ctx,
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 ||
+ if (fmask_ptr && !instr->sampler && (instr->op == nir_texop_txf_ms || instr->op == nir_texop_txf_ms_mcs ||
instr->op == nir_texop_samples_identical))
*fmask_ptr = get_sampler_desc(ctx, instr->texture, AC_DESC_FMASK, instr, false, false);
}
@@ -5150,7 +5159,7 @@ static void visit_tex(struct ac_nir_context *ctx, nir_tex_instr *instr)
/* Pack LOD */
if (lod && ((instr->op == nir_texop_txl || instr->op == nir_texop_txf) && !lod_is_zero)) {
address[count++] = lod;
- } else if (instr->op == nir_texop_txf_ms && sample_index) {
+ } else if ((instr->op == nir_texop_txf_ms || instr->op == nir_texop_txf) && sample_index) {
address[count++] = sample_index;
} else if(instr->op == nir_texop_txs) {
count = 0;
@@ -5165,7 +5174,8 @@ static void visit_tex(struct ac_nir_context *ctx, nir_tex_instr *instr)
address[chan], ctx->ac.i32, "");
}
- if (instr->op == nir_texop_samples_identical) {
+ if (instr->op == nir_texop_samples_identical ||
+ instr->op == nir_texop_txf_ms_mcs) {
LLVMValueRef txf_address[4];
struct ac_image_args txf_args = { 0 };
unsigned txf_count = count;
@@ -5182,12 +5192,22 @@ static void visit_tex(struct ac_nir_context *ctx, nir_tex_instr *instr)
result = build_tex_intrinsic(ctx, instr, false, &txf_args);
result = LLVMBuildExtractElement(ctx->ac.builder, result, ctx->ac.i32_0, "");
- result = emit_int_cmp(&ctx->ac, LLVMIntEQ, result, ctx->ac.i32_0);
+
+ /* Don't rewrite the sample index if WORD1.DATA_FORMAT of the FMASK
+ * resource descriptor is 0 (invalid),
+ */
+ LLVMValueRef word1_is_nonzero = get_fmask_desc_valid(&ctx->ac, fmask_ptr);
+ result = LLVMBuildSelect(ctx->ac.builder, word1_is_nonzero,
+ result, instr->op == nir_texop_samples_identical ? ctx->ac.i32_0 : LLVMConstInt(ctx->ac.i32, 0x76543210, false), "");
+
+ if (instr->op == nir_texop_samples_identical)
+ result = emit_int_cmp(&ctx->ac, LLVMIntEQ, result, ctx->ac.i32_0);
+
goto write_result;
}
if (instr->sampler_dim == GLSL_SAMPLER_DIM_MS &&
- instr->op != nir_texop_txs) {
+ instr->op == nir_texop_txf_ms) {
unsigned sample_chan = instr->is_array ? 3 : 2;
address[sample_chan] = adjust_sample_index_using_fmask(&ctx->ac,
address[0],
diff --git a/src/amd/vulkan/radv_meta_bufimage.c b/src/amd/vulkan/radv_meta_bufimage.c
index adf610a933e..534a39f032c 100644
--- a/src/amd/vulkan/radv_meta_bufimage.c
+++ b/src/amd/vulkan/radv_meta_bufimage.c
@@ -478,6 +478,192 @@ radv_device_finish_meta_btoi_state(struct radv_device *device)
state->btoi.pipeline_3d, &state->alloc);
}
+/*
+ * for multisample copies we need to copy per the fmask value in the src.
+ * - fetch the fmask value using txf_ms_mcs.
+ * - if it's 0 all samples are identical, write the 0 sample
+ * into all samples.
+ * - if it's a mask value have to iterate per sample
+ * and write the sample to the correct dest sample.
+ */
+static nir_shader *
+build_nir_itoi_compute_shader_ms(struct radv_device *dev)
+{
+ nir_builder b;
+ enum glsl_sampler_dim dim = GLSL_SAMPLER_DIM_MS;
+ const struct glsl_type *buf_type = glsl_sampler_type(dim,
+ false,
+ false,
+ GLSL_TYPE_FLOAT);
+ const struct glsl_type *img_type = glsl_sampler_type(dim,
+ false,
+ false,
+ GLSL_TYPE_FLOAT);
+ nir_builder_init_simple_shader(&b, NULL, MESA_SHADER_COMPUTE, NULL);
+ b.shader->info.name = ralloc_strdup(b.shader, "meta_itoi_cs_ms");
+
+ b.shader->info.cs.local_size[0] = 16;
+ b.shader->info.cs.local_size[1] = 16;
+ b.shader->info.cs.local_size[2] = 1;
+ nir_variable *input_img = nir_variable_create(b.shader, nir_var_uniform,
+ buf_type, "s_tex");
+ input_img->data.descriptor_set = 0;
+ input_img->data.binding = 0;
+
+ nir_variable *output_img = nir_variable_create(b.shader, nir_var_uniform,
+ img_type, "out_img");
+ output_img->data.descriptor_set = 0;
+ output_img->data.binding = 1;
+
+ nir_ssa_def *invoc_id = nir_load_system_value(&b, nir_intrinsic_load_local_invocation_id, 0);
+ nir_ssa_def *wg_id = nir_load_system_value(&b, nir_intrinsic_load_work_group_id, 0);
+ nir_ssa_def *block_size = nir_imm_ivec4(&b,
+ b.shader->info.cs.local_size[0],
+ b.shader->info.cs.local_size[1],
+ b.shader->info.cs.local_size[2], 0);
+
+ nir_ssa_def *global_id = nir_iadd(&b, nir_imul(&b, wg_id, block_size), invoc_id);
+
+ nir_intrinsic_instr *src_offset = nir_intrinsic_instr_create(b.shader, nir_intrinsic_load_push_constant);
+ nir_intrinsic_set_base(src_offset, 0);
+ nir_intrinsic_set_range(src_offset, 24);
+ src_offset->src[0] = nir_src_for_ssa(nir_imm_int(&b, 0));
+ src_offset->num_components = 3;
+ nir_ssa_dest_init(&src_offset->instr, &src_offset->dest, 3, 32, "src_offset");
+ nir_builder_instr_insert(&b, &src_offset->instr);
+
+ nir_intrinsic_instr *dst_offset = nir_intrinsic_instr_create(b.shader, nir_intrinsic_load_push_constant);
+ nir_intrinsic_set_base(dst_offset, 0);
+ nir_intrinsic_set_range(dst_offset, 24);
+ dst_offset->src[0] = nir_src_for_ssa(nir_imm_int(&b, 12));
+ dst_offset->num_components = 3;
+ nir_ssa_dest_init(&dst_offset->instr, &dst_offset->dest, 3, 32, "dst_offset");
+ nir_builder_instr_insert(&b, &dst_offset->instr);
+
+ nir_ssa_def *src_coord = nir_iadd(&b, global_id, &src_offset->dest.ssa);
+
+ nir_ssa_def *dst_coord = nir_iadd(&b, global_id, &dst_offset->dest.ssa);
+
+ nir_tex_instr *mcs_mask = nir_tex_instr_create(b.shader, 2);
+ mcs_mask->sampler_dim = dim;
+ mcs_mask->op = nir_texop_txf_ms_mcs;
+ mcs_mask->src[0].src_type = nir_tex_src_coord;
+ mcs_mask->src[0].src = nir_src_for_ssa(nir_channels(&b, src_coord, 0x3));
+ mcs_mask->src[1].src_type = nir_tex_src_ms_index;
+ mcs_mask->src[1].src = nir_src_for_ssa(nir_channel(&b, src_coord, 2));
+ mcs_mask->dest_type = nir_type_uint32;
+ mcs_mask->is_array = false;
+ mcs_mask->coord_components = 2;
+ mcs_mask->texture = nir_deref_var_create(mcs_mask, input_img);
+ mcs_mask->sampler = NULL;
+
+ nir_ssa_dest_init(&mcs_mask->instr, &mcs_mask->dest, 4, 32, "mcs");
+ nir_builder_instr_insert(&b, &mcs_mask->instr);
+
+ nir_ssa_def *mask_is_zero = nir_ieq(&b, nir_channel(&b, &mcs_mask->dest.ssa, 0), nir_imm_int(&b, 0));
+ nir_if *if_miz = nir_push_if(&b, mask_is_zero);
+
+ /* if all samples are 0 - just load once */
+ nir_tex_instr *tex = nir_tex_instr_create(b.shader, 2);
+ tex->sampler_dim = dim;
+ tex->op = nir_texop_txf_ms;
+ tex->src[0].src_type = nir_tex_src_coord;
+ tex->src[0].src = nir_src_for_ssa(nir_channels(&b, src_coord, 0x3));
+ tex->src[1].src_type = nir_tex_src_ms_index;
+ tex->src[1].src = nir_src_for_ssa(nir_imm_int(&b, 0));
+ tex->dest_type = nir_type_float;
+ tex->is_array = false;
+ tex->coord_components = 2;
+ tex->texture = nir_deref_var_create(tex, input_img);
+ tex->sampler = NULL;
+
+ nir_ssa_dest_init(&tex->instr, &tex->dest, 4, 32, "tex");
+ nir_builder_instr_insert(&b, &tex->instr);
+
+ /* src_offset[2] is num samples */
+ /* write out all the samples */
+ nir_variable *loop_count = nir_local_variable_create(b.impl, glsl_int_type(), "loop_count");
+ nir_variable *sample_idx = nir_local_variable_create(b.impl, glsl_uint_type(), "sample_idx");
+ nir_store_var(&b, loop_count, nir_channel(&b, &src_offset->dest.ssa, 2), 0x1);
+ nir_store_var(&b, sample_idx, nir_imm_int(&b, 0), 0x1);
+ nir_loop *loop1 = nir_push_loop(&b);
+
+ nir_ssa_def *current_sample_idx = nir_load_var(&b, sample_idx);
+ nir_ssa_def *outval = &tex->dest.ssa;
+ nir_intrinsic_instr *store = nir_intrinsic_instr_create(b.shader, nir_intrinsic_image_store);
+ store->src[0] = nir_src_for_ssa(dst_coord);
+ store->src[1] = nir_src_for_ssa(current_sample_idx);
+ store->src[2] = nir_src_for_ssa(outval);
+ store->variables[0] = nir_deref_var_create(store, output_img);
+ nir_builder_instr_insert(&b, &store->instr);
+ current_sample_idx = nir_iadd(&b, current_sample_idx, nir_imm_int(&b, 1));
+ nir_store_var(&b, sample_idx, current_sample_idx, 0x1);
+
+ /* decrement loop count */
+ nir_ssa_def *current_loop_count = nir_load_var(&b, loop_count);
+ current_loop_count = nir_isub(&b, current_loop_count, nir_imm_int(&b, 1));
+ nir_store_var(&b, loop_count, current_loop_count, 0x1);
+
+ nir_ssa_def *lcmp = nir_ieq(&b, current_loop_count, nir_imm_int(&b, 0));
+ nir_if *loop_over = nir_push_if(&b, lcmp);
+ nir_jump(&b, nir_jump_break);
+ nir_pop_if(&b, loop_over);
+
+ nir_pop_loop(&b, loop1);
+
+ nir_push_else(&b, if_miz);
+
+ nir_variable *sample_mask_var = nir_local_variable_create(b.impl, glsl_uint_type(), "sample_mask");
+ nir_store_var(&b, loop_count, nir_channel(&b, &src_offset->dest.ssa, 2), 0x1);
+ nir_store_var(&b, sample_mask_var, nir_channel(&b, &mcs_mask->dest.ssa, 0), 0x1);
+
+ nir_loop *loop2 = nir_push_loop(&b);
+ nir_ssa_def *sample_mask = nir_load_var(&b, sample_mask_var);
+ nir_ssa_def *sample_id = nir_iand(&b, sample_mask, nir_imm_int(&b, 0xf));
+ tex = nir_tex_instr_create(b.shader, 2);
+ tex->sampler_dim = dim;
+ tex->op = nir_texop_txf;
+ tex->src[0].src_type = nir_tex_src_coord;
+ tex->src[0].src = nir_src_for_ssa(nir_channels(&b, src_coord, 0x3));
+ tex->src[1].src_type = nir_tex_src_ms_index;
+ tex->src[1].src = nir_src_for_ssa(sample_id);
+ tex->dest_type = nir_type_float;
+ tex->is_array = false;
+ tex->coord_components = 2;
+ tex->texture = nir_deref_var_create(tex, input_img);
+ tex->sampler = NULL;
+ nir_ssa_dest_init(&tex->instr, &tex->dest, 4, 32, "tex");
+ nir_builder_instr_insert(&b, &tex->instr);
+
+ outval = &tex->dest.ssa;
+ store = nir_intrinsic_instr_create(b.shader, nir_intrinsic_image_store);
+ store->src[0] = nir_src_for_ssa(dst_coord);
+ store->src[1] = nir_src_for_ssa(sample_id);
+ store->src[2] = nir_src_for_ssa(outval);
+ store->variables[0] = nir_deref_var_create(store, output_img);
+
+ nir_builder_instr_insert(&b, &store->instr);
+
+ sample_mask = nir_ushr(&b, sample_mask, nir_imm_int(&b, 4));
+ nir_store_var(&b, sample_mask_var, sample_mask, 0x1);
+
+ /* decrement loop count */
+ current_loop_count = nir_load_var(&b, loop_count);
+ current_loop_count = nir_isub(&b, current_loop_count, nir_imm_int(&b, 1));
+ nir_store_var(&b, loop_count, current_loop_count, 0x1);
+
+ lcmp = nir_ieq(&b, current_loop_count, nir_imm_int(&b, 0));
+ loop_over = nir_push_if(&b, lcmp);
+ nir_jump(&b, nir_jump_break);
+ nir_pop_if(&b, loop_over);
+
+ nir_pop_loop(&b, loop2);
+
+ nir_pop_if(&b, if_miz);
+
+ return b.shader;
+}
+
static nir_shader *
build_nir_itoi_compute_shader(struct radv_device *dev, bool is_3d)
{
@@ -568,8 +754,10 @@ radv_device_init_meta_itoi_state(struct radv_device *device)
{
VkResult result;
struct radv_shader_module cs = { .nir = NULL };
+ struct radv_shader_module cs_ms = { .nir = NULL };
struct radv_shader_module cs_3d = { .nir = NULL };
cs.nir = build_nir_itoi_compute_shader(device, false);
+ cs_ms.nir = build_nir_itoi_compute_shader_ms(device);
if (device->physical_device->rad_info.chip_class >= GFX9)
cs_3d.nir = build_nir_itoi_compute_shader(device, true);
/*
@@ -631,17 +819,33 @@ radv_device_init_meta_itoi_state(struct radv_device *device)
.pSpecializationInfo = NULL,
};
- VkComputePipelineCreateInfo vk_pipeline_info = {
- .sType = VK_STRUCTURE_TYPE_COMPUTE_PIPELINE_CREATE_INFO,
- .stage = pipeline_shader_stage,
- .flags = 0,
- .layout = device->meta_state.itoi.img_p_layout,
+ VkPipelineShaderStageCreateInfo pipeline_shader_stage_ms = {
+ .sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO,
+ .stage = VK_SHADER_STAGE_COMPUTE_BIT,
+ .module = radv_shader_module_to_handle(&cs_ms),
+ .pName = "main",
+ .pSpecializationInfo = NULL,
+ };
+
+ VkComputePipelineCreateInfo vk_pipeline_info[2] = {
+ {
+ .sType = VK_STRUCTURE_TYPE_COMPUTE_PIPELINE_CREATE_INFO,
+ .stage = pipeline_shader_stage,
+ .flags = 0,
+ .layout = device->meta_state.itoi.img_p_layout,
+ },
+ {
+ .sType = VK_STRUCTURE_TYPE_COMPUTE_PIPELINE_CREATE_INFO,
+ .stage = pipeline_shader_stage_ms,
+ .flags = 0,
+ .layout = device->meta_state.itoi.img_p_layout,
+ }
};
result = radv_CreateComputePipelines(radv_device_to_handle(device),
radv_pipeline_cache_to_handle(&device->meta_state.cache),
- 1, &vk_pipeline_info, NULL,
- &device->meta_state.itoi.pipeline);
+ 2, vk_pipeline_info, NULL,
+ device->meta_state.itoi.pipeline);
if (result != VK_SUCCESS)
goto fail;
@@ -688,7 +892,9 @@ radv_device_finish_meta_itoi_state(struct radv_device *device)
state->itoi.img_ds_layout,
&state->alloc);
radv_DestroyPipeline(radv_device_to_handle(device),
- state->itoi.pipeline, &state->alloc);
+ state->itoi.pipeline[0], &state->alloc);
+ radv_DestroyPipeline(radv_device_to_handle(device),
+ state->itoi.pipeline[1], &state->alloc);
if (device->physical_device->rad_info.chip_class >= GFX9)
radv_DestroyPipeline(radv_device_to_handle(device),
state->itoi.pipeline_3d, &state->alloc);
@@ -1173,15 +1379,18 @@ radv_meta_image_to_image_cs(struct radv_cmd_buffer *cmd_buffer,
unsigned num_rects,
struct radv_meta_blit2d_rect *rects)
{
- VkPipeline pipeline = cmd_buffer->device->meta_state.itoi.pipeline;
+ VkPipeline pipeline = cmd_buffer->device->meta_state.itoi.pipeline[0];
struct radv_device *device = cmd_buffer->device;
struct radv_image_view src_view, dst_view;
+ int num_samples = dst->image->info.samples;
create_iview(cmd_buffer, src, &src_view);
create_iview(cmd_buffer, dst, &dst_view);
itoi_bind_descriptors(cmd_buffer, &src_view, &dst_view);
+ if (src->image->info.samples > 1)
+ pipeline = cmd_buffer->device->meta_state.itoi.pipeline[1];
if (device->physical_device->rad_info.chip_class >= GFX9 &&
src->image->type == VK_IMAGE_TYPE_3D)
pipeline = cmd_buffer->device->meta_state.itoi.pipeline_3d;
@@ -1192,7 +1401,7 @@ radv_meta_image_to_image_cs(struct radv_cmd_buffer *cmd_buffer,
unsigned push_constants[6] = {
rects[r].src_x,
rects[r].src_y,
- src->layer,
+ num_samples > 1 ? num_samples : src->layer,
rects[r].dst_x,
rects[r].dst_y,
dst->layer,
diff --git a/src/amd/vulkan/radv_meta_copy.c b/src/amd/vulkan/radv_meta_copy.c
index 2a3faa64f1c..d9a67659daf 100644
--- a/src/amd/vulkan/radv_meta_copy.c
+++ b/src/amd/vulkan/radv_meta_copy.c
@@ -338,7 +338,7 @@ meta_copy_image(struct radv_cmd_buffer *cmd_buffer,
uint32_t regionCount,
const VkImageCopy *pRegions)
{
- bool cs = cmd_buffer->queue_family_index == RADV_QUEUE_COMPUTE;
+ bool cs = cmd_buffer->queue_family_index == RADV_QUEUE_COMPUTE || dest_image->info.samples > 1;
struct radv_meta_saved_state saved_state;
/* From the Vulkan 1.0 spec:
diff --git a/src/amd/vulkan/radv_private.h b/src/amd/vulkan/radv_private.h
index 0f8ddb2e106..ab513e7dc9b 100644
--- a/src/amd/vulkan/radv_private.h
+++ b/src/amd/vulkan/radv_private.h
@@ -483,7 +483,7 @@ struct radv_meta_state {
struct {
VkPipelineLayout img_p_layout;
VkDescriptorSetLayout img_ds_layout;
- VkPipeline pipeline;
+ VkPipeline pipeline[2];
VkPipeline pipeline_3d;
} itoi;
struct {
--
2.14.3
More information about the mesa-dev
mailing list