vkd3d: Ignore unsupported execute indirect calls.

Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>

include/vkd3d.h

@@ -82,6 +82,7 @@ enum vkd3d_config_flags
     VKD3D_CONFIG_FLAG_PIPELINE_LIBRARY_SANITIZE_SPIRV = 0x00080000,
     VKD3D_CONFIG_FLAG_PIPELINE_LIBRARY_LOG = 0x00100000,
     VKD3D_CONFIG_FLAG_PIPELINE_LIBRARY_IGNORE_SPIRV = 0x00200000,
+    VKD3D_CONFIG_FLAG_FORCE_RAW_VA_CBV = 0x00400000,
 };
 
 typedef HRESULT (*PFN_vkd3d_signal_event)(HANDLE event);

include/vkd3d_shader.h

@@ -241,6 +241,7 @@ struct vkd3d_shader_root_constant
 struct vkd3d_shader_root_descriptor
 {
     struct vkd3d_shader_resource_binding *binding;
+    uint32_t raw_va_root_descriptor_index;
 };
 
 struct vkd3d_shader_root_parameter

libs/vkd3d/device.c

@@ -114,6 +114,7 @@ static const struct vkd3d_optional_extension_info optional_device_extensions[] =
     VK_EXTENSION(NVX_IMAGE_VIEW_HANDLE, NVX_image_view_handle),
     VK_EXTENSION(NV_FRAGMENT_SHADER_BARYCENTRIC, NV_fragment_shader_barycentric),
     VK_EXTENSION(NV_COMPUTE_SHADER_DERIVATIVES, NV_compute_shader_derivatives),
+    VK_EXTENSION(NV_DEVICE_GENERATED_COMMANDS, NV_device_generated_commands),
     /* VALVE extensions */
     VK_EXTENSION(VALVE_MUTABLE_DESCRIPTOR_TYPE, VALVE_mutable_descriptor_type),
 };
@@ -609,6 +610,7 @@ static const struct vkd3d_debug_option vkd3d_config_options[] =
     {"pipeline_library_sanitize_spirv", VKD3D_CONFIG_FLAG_PIPELINE_LIBRARY_SANITIZE_SPIRV},
     {"pipeline_library_log", VKD3D_CONFIG_FLAG_PIPELINE_LIBRARY_LOG},
     {"pipeline_library_ignore_spirv", VKD3D_CONFIG_FLAG_PIPELINE_LIBRARY_IGNORE_SPIRV},
+    {"force_raw_va_cbv", VKD3D_CONFIG_FLAG_FORCE_RAW_VA_CBV},
 };
 
 static void vkd3d_config_flags_init_once(void)
@@ -1336,6 +1338,16 @@ static void vkd3d_physical_device_info_init(struct vkd3d_physical_device_info *i
         vk_prepend_struct(&info->features2, &info->compute_shader_derivatives_features_nv);
     }
 
+    if (vulkan_info->NV_device_generated_commands)
+    {
+        info->device_generated_commands_features_nv.sType =
+                VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_DEVICE_GENERATED_COMMANDS_FEATURES_NV;
+        info->device_generated_commands_properties_nv.sType =
+                VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_DEVICE_GENERATED_COMMANDS_PROPERTIES_NV;
+        vk_prepend_struct(&info->features2, &info->device_generated_commands_features_nv);
+        vk_prepend_struct(&info->properties2, &info->device_generated_commands_properties_nv);
+    }
+
     if (vulkan_info->KHR_shader_atomic_int64)
     {
         info->shader_atomic_int64_features.sType =
@@ -2400,22 +2412,50 @@ static void d3d12_remove_device_singleton(LUID luid)
     }
 }
 
-static HRESULT d3d12_device_create_scratch_buffer(struct d3d12_device *device, VkDeviceSize size, struct vkd3d_scratch_buffer *scratch)
+static HRESULT d3d12_device_create_scratch_buffer(struct d3d12_device *device, enum vkd3d_scratch_pool_kind kind,
+        VkDeviceSize size, uint32_t memory_types, struct vkd3d_scratch_buffer *scratch)
 {
-    struct vkd3d_allocate_heap_memory_info alloc_info;
     HRESULT hr;
 
     TRACE("device %p, size %llu, scratch %p.\n", device, size, scratch);
 
-    memset(&alloc_info, 0, sizeof(alloc_info));
-    alloc_info.heap_desc.Properties.Type = D3D12_HEAP_TYPE_DEFAULT;
-    alloc_info.heap_desc.SizeInBytes = size;
-    alloc_info.heap_desc.Alignment = D3D12_DEFAULT_RESOURCE_PLACEMENT_ALIGNMENT;
-    alloc_info.heap_desc.Flags = D3D12_HEAP_FLAG_ALLOW_ONLY_BUFFERS;
+    if (kind == VKD3D_SCRATCH_POOL_KIND_DEVICE_STORAGE)
+    {
+        struct vkd3d_allocate_heap_memory_info alloc_info;
 
-    if (FAILED(hr = vkd3d_allocate_heap_memory(device, &device->memory_allocator,
-            &alloc_info, &scratch->allocation)))
-        return hr;
+        /* We only care about memory types for INDIRECT_PREPROCESS. */
+        assert(memory_types == ~0u);
+
+        memset(&alloc_info, 0, sizeof(alloc_info));
+        alloc_info.heap_desc.Properties.Type = D3D12_HEAP_TYPE_DEFAULT;
+        alloc_info.heap_desc.SizeInBytes = size;
+        alloc_info.heap_desc.Alignment = D3D12_DEFAULT_RESOURCE_PLACEMENT_ALIGNMENT;
+        alloc_info.heap_desc.Flags = D3D12_HEAP_FLAG_ALLOW_ONLY_BUFFERS | D3D12_HEAP_FLAG_CREATE_NOT_ZEROED;
+
+        if (FAILED(hr = vkd3d_allocate_heap_memory(device, &device->memory_allocator,
+                &alloc_info, &scratch->allocation)))
+            return hr;
+    }
+    else if (kind == VKD3D_SCRATCH_POOL_KIND_INDIRECT_PREPROCESS)
+    {
+        struct vkd3d_allocate_memory_info alloc_info;
+        memset(&alloc_info, 0, sizeof(alloc_info));
+
+        alloc_info.heap_properties.Type = D3D12_HEAP_TYPE_DEFAULT;
+        alloc_info.memory_requirements.size = size;
+        alloc_info.memory_requirements.memoryTypeBits = memory_types;
+        alloc_info.heap_flags = D3D12_HEAP_FLAG_ALLOW_ONLY_BUFFERS | D3D12_HEAP_FLAG_CREATE_NOT_ZEROED;
+        alloc_info.optional_memory_properties = VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT;
+        alloc_info.flags = VKD3D_ALLOCATION_FLAG_GLOBAL_BUFFER;
+
+        if (FAILED(hr = vkd3d_allocate_memory(device, &device->memory_allocator,
+                &alloc_info, &scratch->allocation)))
+            return hr;
+    }
+    else
+    {
+        return E_INVALIDARG;
+    }
 
     scratch->offset = 0;
     return S_OK;
@@ -2428,35 +2468,47 @@ static void d3d12_device_destroy_scratch_buffer(struct d3d12_device *device, con
     vkd3d_free_memory(device, &device->memory_allocator, &scratch->allocation);
 }
 
-HRESULT d3d12_device_get_scratch_buffer(struct d3d12_device *device, VkDeviceSize min_size, struct vkd3d_scratch_buffer *scratch)
+HRESULT d3d12_device_get_scratch_buffer(struct d3d12_device *device, enum vkd3d_scratch_pool_kind kind,
+        VkDeviceSize min_size, uint32_t memory_types, struct vkd3d_scratch_buffer *scratch)
 {
+    struct d3d12_device_scratch_pool *pool = &device->scratch_pools[kind];
+    struct vkd3d_scratch_buffer *candidate;
+    size_t i;
+
     if (min_size > VKD3D_SCRATCH_BUFFER_SIZE)
-        return d3d12_device_create_scratch_buffer(device, min_size, scratch);
+        return d3d12_device_create_scratch_buffer(device, kind, min_size, memory_types, scratch);
 
     pthread_mutex_lock(&device->mutex);
 
-    if (device->scratch_buffer_count)
+    for (i = pool->scratch_buffer_count; i; i--)
     {
-        *scratch = device->scratch_buffers[--device->scratch_buffer_count];
-        scratch->offset = 0;
-        pthread_mutex_unlock(&device->mutex);
-        return S_OK;
-    }
-    else
-    {
-        pthread_mutex_unlock(&device->mutex);
-        return d3d12_device_create_scratch_buffer(device, VKD3D_SCRATCH_BUFFER_SIZE, scratch);
+        candidate = &pool->scratch_buffers[i - 1];
+
+        /* Extremely unlikely to fail since we have separate lists per pool kind, but to be 100% correct ... */
+        if (memory_types & (1u << candidate->allocation.device_allocation.vk_memory_type))
+        {
+            *scratch = *candidate;
+            scratch->offset = 0;
+            pool->scratch_buffers[i - 1] = pool->scratch_buffers[--pool->scratch_buffer_count];
+            pthread_mutex_unlock(&device->mutex);
+            return S_OK;
+        }
     }
+
+    pthread_mutex_unlock(&device->mutex);
+    return d3d12_device_create_scratch_buffer(device, kind, VKD3D_SCRATCH_BUFFER_SIZE, memory_types, scratch);
 }
 
-void d3d12_device_return_scratch_buffer(struct d3d12_device *device, const struct vkd3d_scratch_buffer *scratch)
+void d3d12_device_return_scratch_buffer(struct d3d12_device *device, enum vkd3d_scratch_pool_kind kind,
+        const struct vkd3d_scratch_buffer *scratch)
 {
+    struct d3d12_device_scratch_pool *pool = &device->scratch_pools[kind];
     pthread_mutex_lock(&device->mutex);
 
     if (scratch->allocation.resource.size == VKD3D_SCRATCH_BUFFER_SIZE &&
-            device->scratch_buffer_count < VKD3D_SCRATCH_BUFFER_COUNT)
+            pool->scratch_buffer_count < VKD3D_SCRATCH_BUFFER_COUNT)
     {
-        device->scratch_buffers[device->scratch_buffer_count++] = *scratch;
+        pool->scratch_buffers[pool->scratch_buffer_count++] = *scratch;
         pthread_mutex_unlock(&device->mutex);
     }
     else
@@ -2707,10 +2759,11 @@ static void d3d12_device_global_pipeline_cache_cleanup(struct d3d12_device *devi
 static void d3d12_device_destroy(struct d3d12_device *device)
 {
     const struct vkd3d_vk_device_procs *vk_procs = &device->vk_procs;
-    size_t i;
+    size_t i, j;
 
-    for (i = 0; i < device->scratch_buffer_count; i++)
-        d3d12_device_destroy_scratch_buffer(device, &device->scratch_buffers[i]);
+    for (i = 0; i < VKD3D_SCRATCH_POOL_KIND_COUNT; i++)
+        for (j = 0; j < device->scratch_pools[i].scratch_buffer_count; j++)
+            d3d12_device_destroy_scratch_buffer(device, &device->scratch_pools[i].scratch_buffers[j]);
 
     for (i = 0; i < device->query_pool_count; i++)
         d3d12_device_destroy_query_pool(device, &device->query_pools[i]);
@@ -4297,9 +4350,10 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_SetStablePowerState(d3d12_device_i
 }
 
 static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommandSignature(d3d12_device_iface *iface,
-        const D3D12_COMMAND_SIGNATURE_DESC *desc, ID3D12RootSignature *root_signature,
+        const D3D12_COMMAND_SIGNATURE_DESC *desc, ID3D12RootSignature *root_signature_iface,
         REFIID iid, void **command_signature)
 {
+    struct d3d12_root_signature *root_signature = impl_from_ID3D12RootSignature(root_signature_iface);
     struct d3d12_device *device = impl_from_ID3D12Device(iface);
     struct d3d12_command_signature *object;
     HRESULT hr;
@@ -4307,7 +4361,7 @@ static HRESULT STDMETHODCALLTYPE d3d12_device_CreateCommandSignature(d3d12_devic
     TRACE("iface %p, desc %p, root_signature %p, iid %s, command_signature %p.\n",
             iface, desc, root_signature, debugstr_guid(iid), command_signature);
 
-    if (FAILED(hr = d3d12_command_signature_create(device, desc, &object)))
+    if (FAILED(hr = d3d12_command_signature_create(device, root_signature, desc, &object)))
         return hr;
 
     return return_interface(&object->ID3D12CommandSignature_iface,

libs/vkd3d/meson.build

@@ -27,6 +27,7 @@ vkd3d_shaders =[
 
   'shaders/vs_swapchain_fullscreen.vert',
   'shaders/fs_swapchain_fullscreen.frag',
+  'shaders/cs_execute_indirect_patch.comp',
 ]
 
 vkd3d_src = [

libs/vkd3d/meta.c

@@ -1361,6 +1361,106 @@ void vkd3d_meta_get_predicate_pipeline(struct vkd3d_meta_ops *meta_ops,
     info->data_size = predicate_ops->data_sizes[command_type];
 }
 
+HRESULT vkd3d_execute_indirect_ops_init(struct vkd3d_execute_indirect_ops *meta_indirect_ops,
+        struct d3d12_device *device)
+{
+    VkPushConstantRange push_constant_range;
+    VkResult vr;
+    int rc;
+
+    if ((rc = pthread_mutex_init(&meta_indirect_ops->mutex, NULL)))
+        return hresult_from_errno(rc);
+
+    push_constant_range.offset = 0;
+    push_constant_range.size = sizeof(struct vkd3d_execute_indirect_args);
+    push_constant_range.stageFlags = VK_SHADER_STAGE_COMPUTE_BIT;
+
+    if ((vr = vkd3d_meta_create_pipeline_layout(device, 0, NULL, 1,
+            &push_constant_range, &meta_indirect_ops->vk_pipeline_layout)) < 0)
+    {
+        pthread_mutex_destroy(&meta_indirect_ops->mutex);
+        return hresult_from_vk_result(vr);
+    }
+
+    meta_indirect_ops->pipelines_count = 0;
+    meta_indirect_ops->pipelines_size = 0;
+    meta_indirect_ops->pipelines = NULL;
+    return S_OK;
+}
+
+HRESULT vkd3d_meta_get_execute_indirect_pipeline(struct vkd3d_meta_ops *meta_ops,
+        uint32_t patch_command_count, struct vkd3d_execute_indirect_info *info)
+{
+    struct vkd3d_execute_indirect_ops *meta_indirect_ops = &meta_ops->execute_indirect;
+    VkSpecializationMapEntry map_entry;
+    VkSpecializationInfo spec;
+    HRESULT hr = S_OK;
+    VkResult vr;
+    size_t i;
+    int rc;
+
+    if ((rc = pthread_mutex_lock(&meta_indirect_ops->mutex)))
+    {
+        ERR("Failed to lock mutex, error %d.\n", rc);
+        return hresult_from_errno(rc);
+    }
+
+    for (i = 0; i < meta_indirect_ops->pipelines_count; i++)
+    {
+        if (meta_indirect_ops->pipelines[i].workgroup_size_x == patch_command_count)
+        {
+            info->vk_pipeline_layout = meta_indirect_ops->vk_pipeline_layout;
+            info->vk_pipeline = meta_indirect_ops->pipelines[i].vk_pipeline;
+            goto out;
+        }
+    }
+
+    map_entry.constantID = 0;
+    map_entry.offset = 0;
+    map_entry.size = sizeof(patch_command_count);
+
+    spec.pMapEntries = &map_entry;
+    spec.pData = &patch_command_count;
+    spec.mapEntryCount = 1;
+    spec.dataSize = sizeof(patch_command_count);
+
+    vkd3d_array_reserve((void**)&meta_indirect_ops->pipelines, &meta_indirect_ops->pipelines_size,
+            meta_indirect_ops->pipelines_count + 1, sizeof(*meta_indirect_ops->pipelines));
+
+    meta_indirect_ops->pipelines[meta_indirect_ops->pipelines_count].workgroup_size_x = patch_command_count;
+
+    vr = vkd3d_meta_create_compute_pipeline(meta_ops->device,
+            sizeof(cs_execute_indirect_patch), cs_execute_indirect_patch,
+            meta_indirect_ops->vk_pipeline_layout, &spec,
+            &meta_indirect_ops->pipelines[meta_indirect_ops->pipelines_count].vk_pipeline);
+
+    if (vr)
+    {
+        hr = hresult_from_vk_result(vr);
+        goto out;
+    }
+
+    info->vk_pipeline_layout = meta_indirect_ops->vk_pipeline_layout;
+    info->vk_pipeline = meta_indirect_ops->pipelines[meta_indirect_ops->pipelines_count].vk_pipeline;
+    meta_indirect_ops->pipelines_count++;
+
+out:
+    pthread_mutex_unlock(&meta_indirect_ops->mutex);
+    return hr;
+}
+
+void vkd3d_execute_indirect_ops_cleanup(struct vkd3d_execute_indirect_ops *meta_indirect_ops,
+        struct d3d12_device *device)
+{
+    const struct vkd3d_vk_device_procs *vk_procs = &device->vk_procs;
+    size_t i;
+
+    for (i = 0; i < meta_indirect_ops->pipelines_count; i++)
+        VK_CALL(vkDestroyPipeline(device->vk_device, meta_indirect_ops->pipelines[i].vk_pipeline, NULL));
+    VK_CALL(vkDestroyPipelineLayout(device->vk_device, meta_indirect_ops->vk_pipeline_layout, NULL));
+    pthread_mutex_destroy(&meta_indirect_ops->mutex);
+}
+
 HRESULT vkd3d_meta_ops_init(struct vkd3d_meta_ops *meta_ops, struct d3d12_device *device)
 {
     HRESULT hr;
@@ -1386,8 +1486,13 @@ HRESULT vkd3d_meta_ops_init(struct vkd3d_meta_ops *meta_ops, struct d3d12_device
     if (FAILED(hr = vkd3d_predicate_ops_init(&meta_ops->predicate, device)))
         goto fail_predicate_ops;
 
+    if (FAILED(hr = vkd3d_execute_indirect_ops_init(&meta_ops->execute_indirect, device)))
+        goto fail_execute_indirect_ops;
+
     return S_OK;
 
+fail_execute_indirect_ops:
+    vkd3d_predicate_ops_cleanup(&meta_ops->predicate, device);
 fail_predicate_ops:
     vkd3d_query_ops_cleanup(&meta_ops->query, device);
 fail_query_ops:
@@ -1404,6 +1509,7 @@ fail_common:
 
 HRESULT vkd3d_meta_ops_cleanup(struct vkd3d_meta_ops *meta_ops, struct d3d12_device *device)
 {
+    vkd3d_execute_indirect_ops_cleanup(&meta_ops->execute_indirect, device);
     vkd3d_predicate_ops_cleanup(&meta_ops->predicate, device);
     vkd3d_query_ops_cleanup(&meta_ops->query, device);
     vkd3d_swapchain_ops_cleanup(&meta_ops->swapchain, device);

libs/vkd3d/shaders/cs_execute_indirect_patch.comp

@@ -0,0 +1,76 @@
+#version 450
+#extension GL_EXT_buffer_reference : require
+#extension GL_EXT_buffer_reference_uvec2 : require
+
+layout(local_size_x_id = 0) in;
+
+struct Command
+{
+	uint type;
+	uint src_offset;
+	uint dst_offset;
+};
+
+const int COMMAND_TYPE_COPY_U32 = 0;
+const int COMMAND_TYPE_COPY_INDEX_TYPE = 1;
+const int DXGI_FORMAT_R32_UINT = 0x2a;
+const int VK_INDEX_TYPE_UINT16 = 0;
+const int VK_INDEX_TYPE_UINT32 = 1;
+
+layout(buffer_reference, std430, buffer_reference_align = 4) readonly buffer Commands
+{
+	Command commands[];
+};
+
+layout(buffer_reference, std430, buffer_reference_align = 4) readonly buffer SrcBuffer {
+	uint values[];
+};
+
+layout(buffer_reference, std430, buffer_reference_align = 4) writeonly buffer DstBuffer {
+	uint values[];
+};
+
+layout(buffer_reference, std430, buffer_reference_align = 4) readonly buffer IndirectCount {
+	uint count;
+};
+
+layout(buffer_reference, std430, buffer_reference_align = 4) writeonly buffer IndirectCountWrite {
+	uint count;
+};
+
+layout(push_constant) uniform Registers
+{
+	Commands commands_va;
+	SrcBuffer src_buffer_va;
+	DstBuffer dst_buffer_va;
+	uvec2 indirect_count_va;
+	IndirectCountWrite dst_indirect_count_va;
+	uint src_stride;
+	uint dst_stride;
+};
+
+void main()
+{
+	Command cmd = commands_va.commands[gl_LocalInvocationIndex];
+
+	uint draw_id = gl_WorkGroupID.x;
+	uint max_draws = ~0u;
+	if (any(notEqual(indirect_count_va, uvec2(0))))
+	{
+		max_draws = IndirectCount(indirect_count_va).count;
+		if (gl_WorkGroupID.x == 0u)
+			dst_indirect_count_va.count = max_draws;
+	}
+
+	if (draw_id < max_draws)
+	{
+		uint src_offset = src_stride * draw_id + cmd.src_offset;
+		uint dst_offset = dst_stride * draw_id + cmd.dst_offset;
+
+		uint src_value = src_buffer_va.values[src_offset];
+		if (cmd.type == COMMAND_TYPE_COPY_INDEX_TYPE)
+			src_value = src_value == DXGI_FORMAT_R32_UINT ? VK_INDEX_TYPE_UINT32 : VK_INDEX_TYPE_UINT16;
+
+		dst_buffer_va.values[dst_offset] = src_value;
+	}
+}

libs/vkd3d/state.c

@@ -923,6 +923,7 @@ static HRESULT d3d12_root_signature_init_root_descriptors(struct d3d12_root_sign
     struct vkd3d_shader_resource_binding *binding;
     VkDescriptorSetLayoutCreateFlags vk_flags;
     struct vkd3d_shader_root_parameter *param;
+    uint32_t raw_va_root_descriptor_count = 0;
     unsigned int hoisted_parameter_index;
     const D3D12_DESCRIPTOR_RANGE1 *range;
     unsigned int i, j, k;
@@ -1039,10 +1040,13 @@ static HRESULT d3d12_root_signature_init_root_descriptors(struct d3d12_root_sign
         param = &root_signature->parameters[i];
         param->parameter_type = p->ParameterType;
         param->descriptor.binding = binding;
+        param->descriptor.raw_va_root_descriptor_index = raw_va_root_descriptor_count;
 
         context->binding_index += 1;
 
-        if (!raw_va)
+        if (raw_va)
+            raw_va_root_descriptor_count += 1;
+        else
             context->vk_binding += 1;
     }
 
@@ -4709,7 +4713,8 @@ static uint32_t vkd3d_bindless_state_get_bindless_flags(struct d3d12_device *dev
          * The difference in performance is profound (~15% in some cases).
          * On ACO, BDA with NonWritable can be promoted directly to scalar loads,
          * which is great. */
-        if (device_info->properties2.properties.vendorID != VKD3D_VENDOR_ID_NVIDIA)
+        if ((vkd3d_config_flags & VKD3D_CONFIG_FLAG_FORCE_RAW_VA_CBV) ||
+                device_info->properties2.properties.vendorID != VKD3D_VENDOR_ID_NVIDIA)
             flags |= VKD3D_RAW_VA_ROOT_DESCRIPTOR_CBV;
     }
 

libs/vkd3d/vkd3d_private.h

@@ -162,6 +162,7 @@ struct vkd3d_vulkan_info
     bool NVX_image_view_handle;
     bool NV_fragment_shader_barycentric;
     bool NV_compute_shader_derivatives;
+    bool NV_device_generated_commands;
     /* VALVE extensions */
     bool VALVE_mutable_descriptor_type;
 
@@ -1703,6 +1704,20 @@ struct vkd3d_query_pool
     uint32_t next_index;
 };
 
+struct d3d12_command_allocator_scratch_pool
+{
+    struct vkd3d_scratch_buffer *scratch_buffers;
+    size_t scratch_buffers_size;
+    size_t scratch_buffer_count;
+};
+
+enum vkd3d_scratch_pool_kind
+{
+    VKD3D_SCRATCH_POOL_KIND_DEVICE_STORAGE = 0,
+    VKD3D_SCRATCH_POOL_KIND_INDIRECT_PREPROCESS,
+    VKD3D_SCRATCH_POOL_KIND_COUNT
+};
+
 /* ID3D12CommandAllocator */
 struct d3d12_command_allocator
 {
@@ -1736,9 +1751,7 @@ struct d3d12_command_allocator
     size_t command_buffers_size;
     size_t command_buffer_count;
 
-    struct vkd3d_scratch_buffer *scratch_buffers;
-    size_t scratch_buffers_size;
-    size_t scratch_buffer_count;
+    struct d3d12_command_allocator_scratch_pool scratch_pools[VKD3D_SCRATCH_POOL_KIND_COUNT];
 
     struct vkd3d_query_pool *query_pools;
     size_t query_pools_size;
@@ -1916,12 +1929,26 @@ struct d3d12_command_list
     bool is_valid;
     bool debug_capture;
     bool has_replaced_shaders;
-    bool has_valid_index_buffer;
+
+    struct
+    {
+        VkBuffer buffer;
+        VkDeviceSize offset;
+        DXGI_FORMAT dxgi_format;
+        VkIndexType vk_type;
+        bool is_non_null;
+        bool is_dirty;
+    } index_buffer;
+
+    struct
+    {
+        bool has_observed_transition_to_indirect;
+        bool has_emitted_indirect_to_compute_barrier;
+    } execute_indirect;
+
     VkCommandBuffer vk_command_buffer;
     VkCommandBuffer vk_init_commands;
 
-    DXGI_FORMAT index_buffer_format;
-
     struct d3d12_rtv_desc rtvs[D3D12_SIMULTANEOUS_RENDER_TARGET_COUNT];
     struct d3d12_rtv_desc dsv;
     uint32_t rtv_nonnull_mask;
@@ -2218,6 +2245,12 @@ HRESULT d3d12_command_queue_create(struct d3d12_device *device,
         const D3D12_COMMAND_QUEUE_DESC *desc, struct d3d12_command_queue **queue);
 void d3d12_command_queue_submit_stop(struct d3d12_command_queue *queue);
 
+struct vkd3d_execute_indirect_info
+{
+    VkPipelineLayout vk_pipeline_layout;
+    VkPipeline vk_pipeline;
+};
+
 /* ID3D12CommandSignature */
 struct d3d12_command_signature
 {
@@ -2225,13 +2258,27 @@ struct d3d12_command_signature
     LONG refcount;
 
     D3D12_COMMAND_SIGNATURE_DESC desc;
+    uint32_t argument_buffer_offset;
+
+    /* Complex command signatures require some work to stamp out device generated commands. */
+    struct
+    {
+        VkBuffer buffer;
+        VkDeviceAddress buffer_va;
+        struct vkd3d_device_memory_allocation memory;
+        VkIndirectCommandsLayoutNV layout;
+        uint32_t stride;
+        struct vkd3d_execute_indirect_info pipeline;
+    } state_template;
+    bool requires_state_template;
 
     struct d3d12_device *device;
 
     struct vkd3d_private_store private_store;
 };
 
-HRESULT d3d12_command_signature_create(struct d3d12_device *device, const D3D12_COMMAND_SIGNATURE_DESC *desc,
+HRESULT d3d12_command_signature_create(struct d3d12_device *device, struct d3d12_root_signature *root_signature,
+        const D3D12_COMMAND_SIGNATURE_DESC *desc,
         struct d3d12_command_signature **signature);
 
 static inline struct d3d12_command_signature *impl_from_ID3D12CommandSignature(ID3D12CommandSignature *iface)
@@ -2646,6 +2693,37 @@ HRESULT vkd3d_predicate_ops_init(struct vkd3d_predicate_ops *meta_predicate_ops,
 void vkd3d_predicate_ops_cleanup(struct vkd3d_predicate_ops *meta_predicate_ops,
         struct d3d12_device *device);
 
+struct vkd3d_execute_indirect_args
+{
+    VkDeviceAddress template_va;
+    VkDeviceAddress api_buffer_va;
+    VkDeviceAddress device_generated_commands_va;
+    VkDeviceAddress indirect_count_va;
+    VkDeviceAddress dst_indirect_count_va;
+    uint32_t api_buffer_word_stride;
+    uint32_t device_generated_commands_word_stride;
+};
+
+struct vkd3d_execute_indirect_pipeline
+{
+    VkPipeline vk_pipeline;
+    uint32_t workgroup_size_x;
+};
+
+struct vkd3d_execute_indirect_ops
+{
+    VkPipelineLayout vk_pipeline_layout;
+    struct vkd3d_execute_indirect_pipeline *pipelines;
+    size_t pipelines_count;
+    size_t pipelines_size;
+    pthread_mutex_t mutex;
+};
+
+HRESULT vkd3d_execute_indirect_ops_init(struct vkd3d_execute_indirect_ops *meta_indirect_ops,
+        struct d3d12_device *device);
+void vkd3d_execute_indirect_ops_cleanup(struct vkd3d_execute_indirect_ops *meta_indirect_ops,
+        struct d3d12_device *device);
+
 struct vkd3d_meta_ops_common
 {
     VkShaderModule vk_module_fullscreen_vs;
@@ -2661,6 +2739,7 @@ struct vkd3d_meta_ops
     struct vkd3d_swapchain_ops swapchain;
     struct vkd3d_query_ops query;
     struct vkd3d_predicate_ops predicate;
+    struct vkd3d_execute_indirect_ops execute_indirect;
 };
 
 HRESULT vkd3d_meta_ops_init(struct vkd3d_meta_ops *meta_ops, struct d3d12_device *device);
@@ -2693,6 +2772,9 @@ bool vkd3d_meta_get_query_gather_pipeline(struct vkd3d_meta_ops *meta_ops,
 void vkd3d_meta_get_predicate_pipeline(struct vkd3d_meta_ops *meta_ops,
         enum vkd3d_predicate_command_type command_type, struct vkd3d_predicate_command_info *info);
 
+HRESULT vkd3d_meta_get_execute_indirect_pipeline(struct vkd3d_meta_ops *meta_ops,
+        uint32_t patch_command_count, struct vkd3d_execute_indirect_info *info);
+
 enum vkd3d_time_domain_flag
 {
     VKD3D_TIME_DOMAIN_DEVICE = 0x00000001u,
@@ -2725,6 +2807,7 @@ struct vkd3d_physical_device_info
     VkPhysicalDeviceFragmentShadingRatePropertiesKHR fragment_shading_rate_properties;
     VkPhysicalDeviceConservativeRasterizationPropertiesEXT conservative_rasterization_properties;
     VkPhysicalDeviceShaderIntegerDotProductPropertiesKHR shader_integer_dot_product_properties;
+    VkPhysicalDeviceDeviceGeneratedCommandsPropertiesNV device_generated_commands_properties_nv;
 
     VkPhysicalDeviceProperties2KHR properties2;
 
@@ -2761,6 +2844,7 @@ struct vkd3d_physical_device_info
     VkPhysicalDeviceShaderImageAtomicInt64FeaturesEXT shader_image_atomic_int64_features;
     VkPhysicalDeviceScalarBlockLayoutFeaturesEXT scalar_block_layout_features;
     VkPhysicalDeviceImageViewMinLodFeaturesEXT image_view_min_lod_features;
+    VkPhysicalDeviceDeviceGeneratedCommandsFeaturesNV device_generated_commands_features_nv;
 
     VkPhysicalDeviceFeatures2 features2;
 
@@ -2819,6 +2903,12 @@ struct vkd3d_descriptor_qa_heap_buffer_data;
 /* ID3D12DeviceExt */
 typedef ID3D12DeviceExt d3d12_device_vkd3d_ext_iface;
 
+struct d3d12_device_scratch_pool
+{
+    struct vkd3d_scratch_buffer scratch_buffers[VKD3D_SCRATCH_BUFFER_COUNT];
+    size_t scratch_buffer_count;
+};
+
 struct d3d12_device
 {
     d3d12_device_iface ID3D12Device_iface;
@@ -2854,8 +2944,7 @@ struct d3d12_device
 
     struct vkd3d_memory_allocator memory_allocator;
 
-    struct vkd3d_scratch_buffer scratch_buffers[VKD3D_SCRATCH_BUFFER_COUNT];
-    size_t scratch_buffer_count;
+    struct d3d12_device_scratch_pool scratch_pools[VKD3D_SCRATCH_POOL_KIND_COUNT];
 
     struct vkd3d_query_pool query_pools[VKD3D_VIRTUAL_QUERY_POOL_COUNT];
     size_t query_pool_count;
@@ -2917,8 +3006,10 @@ static inline struct d3d12_device *impl_from_ID3D12Device(d3d12_device_iface *if
 
 bool d3d12_device_validate_shader_meta(struct d3d12_device *device, const struct vkd3d_shader_meta *meta);
 
-HRESULT d3d12_device_get_scratch_buffer(struct d3d12_device *device, VkDeviceSize min_size, struct vkd3d_scratch_buffer *scratch);
-void d3d12_device_return_scratch_buffer(struct d3d12_device *device, const struct vkd3d_scratch_buffer *scratch);
+HRESULT d3d12_device_get_scratch_buffer(struct d3d12_device *device, enum vkd3d_scratch_pool_kind kind,
+        VkDeviceSize min_size, uint32_t memory_types, struct vkd3d_scratch_buffer *scratch);
+void d3d12_device_return_scratch_buffer(struct d3d12_device *device, enum vkd3d_scratch_pool_kind kind,
+        const struct vkd3d_scratch_buffer *scratch);
 
 HRESULT d3d12_device_get_query_pool(struct d3d12_device *device, uint32_t type_index, struct vkd3d_query_pool *pool);
 void d3d12_device_return_query_pool(struct d3d12_device *device, const struct vkd3d_query_pool *pool);

libs/vkd3d/vkd3d_shaders.h

@@ -45,6 +45,7 @@ enum vkd3d_meta_copy_mode
 #include <cs_resolve_binary_queries.h>
 #include <cs_resolve_predicate.h>
 #include <cs_resolve_query.h>
+#include <cs_execute_indirect_patch.h>
 #include <vs_fullscreen_layer.h>
 #include <vs_fullscreen.h>
 #include <gs_fullscreen.h>

libs/vkd3d/vulkan_procs.h

@@ -293,6 +293,12 @@ VK_DEVICE_EXT_PFN(vkCmdCuLaunchKernelNVX)
 VK_DEVICE_EXT_PFN(vkGetImageViewHandleNVX)
 VK_DEVICE_EXT_PFN(vkGetImageViewAddressNVX)
 
+/* VK_NV_device_generated_commands */
+VK_DEVICE_EXT_PFN(vkCreateIndirectCommandsLayoutNV)
+VK_DEVICE_EXT_PFN(vkDestroyIndirectCommandsLayoutNV)
+VK_DEVICE_EXT_PFN(vkGetGeneratedCommandsMemoryRequirementsNV)
+VK_DEVICE_EXT_PFN(vkCmdExecuteGeneratedCommandsNV)
+
 #undef VK_INSTANCE_PFN
 #undef VK_INSTANCE_EXT_PFN
 #undef VK_DEVICE_PFN

tests/d3d12_command.c

@@ -1449,6 +1449,697 @@ void test_vbv_stride_edge_cases(void)
     destroy_test_context(&context);
 }
 
+void test_execute_indirect_state(void)
+{
+    static const struct vec4 values = { 1000.0f, 2000.0f, 3000.0f, 4000.0f };
+    D3D12_INDIRECT_ARGUMENT_DESC indirect_argument_descs[2];
+    D3D12_COMMAND_SIGNATURE_DESC command_signature_desc;
+    D3D12_ROOT_SIGNATURE_DESC root_signature_desc;
+    D3D12_GRAPHICS_PIPELINE_STATE_DESC pso_desc;
+    ID3D12CommandSignature *command_signature;
+    D3D12_SO_DECLARATION_ENTRY so_entries[1];
+    ID3D12GraphicsCommandList *command_list;
+    D3D12_ROOT_PARAMETER root_parameters[4];
+    ID3D12RootSignature *root_signatures[2];
+    ID3D12Resource *argument_buffer_late;
+    D3D12_STREAM_OUTPUT_BUFFER_VIEW sov;
+    ID3D12Resource *streamout_buffer;
+    D3D12_VERTEX_BUFFER_VIEW vbvs[2];
+    ID3D12Resource *argument_buffer;
+    struct test_context_desc desc;
+    ID3D12Resource *count_buffer;
+    ID3D12PipelineState *psos[2];
+    struct test_context context;
+    struct resource_readback rb;
+    D3D12_INDEX_BUFFER_VIEW ibv;
+    ID3D12CommandQueue *queue;
+    const UINT so_stride = 16;
+    ID3D12Resource *vbo[3];
+    ID3D12Resource *ibo[2];
+    unsigned int i, j, k;
+    ID3D12Resource *cbv;
+    ID3D12Resource *srv;
+    ID3D12Resource *uav;
+    HRESULT hr;
+
+    static const D3D12_INPUT_ELEMENT_DESC layout_desc[] =
+    {
+        {"COLOR", 0, DXGI_FORMAT_R32_FLOAT, 0, 0, D3D12_INPUT_CLASSIFICATION_PER_VERTEX_DATA, 0},
+        {"COLOR", 1, DXGI_FORMAT_R32_FLOAT, 1, 0, D3D12_INPUT_CLASSIFICATION_PER_VERTEX_DATA, 0},
+    };
+
+    struct test
+    {
+        const D3D12_INDIRECT_ARGUMENT_DESC *indirect_arguments;
+        uint32_t indirect_argument_count;
+        const void *argument_buffer_data;
+        size_t argument_buffer_size;
+        uint32_t api_max_count;
+        const struct vec4 *expected_output;
+        uint32_t expected_output_count;
+        uint32_t stride;
+        uint32_t pso_index;
+        bool needs_root_sig;
+    };
+
+    /* Modify root parameters. */
+    struct root_constant_data
+    {
+        float constants[2];
+        D3D12_DRAW_INDEXED_ARGUMENTS indexed;
+    };
+
+    static const D3D12_INDIRECT_ARGUMENT_DESC root_constant_sig[2] =
+    {
+        { .Type = D3D12_INDIRECT_ARGUMENT_TYPE_CONSTANT, .Constant = {
+            .RootParameterIndex = 0, .DestOffsetIn32BitValues = 1, .Num32BitValuesToSet = 2 }},
+        { .Type = D3D12_INDIRECT_ARGUMENT_TYPE_DRAW_INDEXED }
+    };
+
+    static const struct root_constant_data root_constant_data[] =
+    {
+        {
+            .constants = { 100.0f, 500.0f },
+            .indexed = { .IndexCountPerInstance = 2, .InstanceCount = 1 }
+        },
+        {
+            .constants = { 200.0f, 800.0f },
+            .indexed = { .IndexCountPerInstance = 1, .InstanceCount = 2,
+                         .StartIndexLocation = 1, .StartInstanceLocation = 100, }
+        },
+    };
+
+    static const struct vec4 root_constant_expected[] =
+    {
+        { 1000.0f, 64.0f + 100.0f, 500.0f, 4000.0f },
+        { 1001.0f, 65.0f + 100.0f, 500.0f, 4000.0f },
+        { 1001.0f, 65.0f + 200.0f, 800.0f, 4000.0f },
+        { 1001.0f, 65.0f + 200.0f, 800.0f, 4001.0f },
+    };
+
+    /* Modify root parameters, but very large root signature to test boundary conditions. */
+    static const D3D12_INDIRECT_ARGUMENT_DESC root_constant_spill_sig[2] =
+    {
+        { .Type = D3D12_INDIRECT_ARGUMENT_TYPE_CONSTANT, .Constant = {
+                .RootParameterIndex = 0, .DestOffsetIn32BitValues = 44 + 1, .Num32BitValuesToSet = 2 }},
+        { .Type = D3D12_INDIRECT_ARGUMENT_TYPE_DRAW_INDEXED }
+    };
+
+    static const struct root_constant_data root_constant_spill_data[] =
+    {
+        {
+            .constants = { 100.0f, 500.0f },
+            .indexed = { .IndexCountPerInstance = 2, .InstanceCount = 1 }
+        },
+        {
+            .constants = { 200.0f, 800.0f },
+            .indexed = { .IndexCountPerInstance = 1, .InstanceCount = 2,
+                    .StartIndexLocation = 1, .StartInstanceLocation = 100, }
+        },
+    };
+
+    static const struct vec4 root_constant_spill_expected[] =
+    {
+        { 1000.0f, 64.0f + 100.0f, 500.0f, 4000.0f },
+        { 1001.0f, 65.0f + 100.0f, 500.0f, 4000.0f },
+        { 1001.0f, 65.0f + 200.0f, 800.0f, 4000.0f },
+        { 1001.0f, 65.0f + 200.0f, 800.0f, 4001.0f },
+    };
+
+    /* Modify VBOs. */
+    struct indirect_vbo_data
+    {
+        D3D12_VERTEX_BUFFER_VIEW view[2];
+        D3D12_DRAW_INDEXED_ARGUMENTS indexed;
+    };
+
+    static const D3D12_INDIRECT_ARGUMENT_DESC indirect_vbo_sig[3] =
+    {
+        { .Type = D3D12_INDIRECT_ARGUMENT_TYPE_VERTEX_BUFFER_VIEW, .VertexBuffer = { .Slot = 0 }},
+        { .Type = D3D12_INDIRECT_ARGUMENT_TYPE_VERTEX_BUFFER_VIEW, .VertexBuffer = { .Slot = 1 }},
+        { .Type = D3D12_INDIRECT_ARGUMENT_TYPE_DRAW_INDEXED },
+    };
+
+    /* Fill buffer locations later. */
+    struct indirect_vbo_data indirect_vbo_data[] =
+    {
+        {
+            .view = { { 0, 64, 8 }, { 0, 64, 16 } },
+            .indexed = { .IndexCountPerInstance = 2, .InstanceCount = 2 }
+        },
+        {
+            /* Test indirectly binding NULL descriptor and 0 stride. */
+            .view = { { 0, 0, 0 }, { 0, 64, 0 } },
+            .indexed = { .IndexCountPerInstance = 2, .InstanceCount = 1 }
+        }
+    };
+
+    static const struct vec4 indirect_vbo_expected[] =
+    {
+        { 1064.0f, 2128.0f, 3000.0f, 4000.0f },
+        { 1066.0f, 2132.0f, 3000.0f, 4000.0f },
+        { 1064.0f, 2128.0f, 3000.0f, 4001.0f },
+        { 1066.0f, 2132.0f, 3000.0f, 4001.0f },
+        { 1000.0f, 2016.0f, 3000.0f, 4000.0f }, /* This is buggy on WARP and AMD. We seem to get null descriptor instead. */
+        { 1000.0f, 2016.0f, 3000.0f, 4000.0f }, /* This is buggy on WARP and AMD. */
+    };
+
+    /* Modify just one VBO. */
+    struct indirect_vbo_one_data
+    {
+        D3D12_VERTEX_BUFFER_VIEW view;
+        D3D12_DRAW_INDEXED_ARGUMENTS indexed;
+    };
+
+    static const D3D12_INDIRECT_ARGUMENT_DESC indirect_vbo_one_sig[2] =
+    {
+        { .Type = D3D12_INDIRECT_ARGUMENT_TYPE_VERTEX_BUFFER_VIEW, .VertexBuffer = { .Slot = 0 }},
+        { .Type = D3D12_INDIRECT_ARGUMENT_TYPE_DRAW_INDEXED },
+    };
+
+    /* Fill buffer locations later. */
+    struct indirect_vbo_one_data indirect_vbo_one_data[] =
+    {
+        {
+            .view = { 0, 64, 8 },
+            .indexed = { .IndexCountPerInstance = 2, .InstanceCount = 1 }
+        },
+        {
+            .indexed = { .IndexCountPerInstance = 1, .InstanceCount = 1 }
+        }
+    };
+
+    static const struct vec4 indirect_vbo_one_expected[] =
+    {
+        { 1128.0f, 2064.0f, 3000.0f, 4000.0f },
+        { 1130.0f, 2065.0f, 3000.0f, 4000.0f },
+        { 1000.0f, 2064.0f, 3000.0f, 4000.0f },
+    };
+
+    /* Indirect IBO */
+    struct indirect_ibo_data
+    {
+        D3D12_INDEX_BUFFER_VIEW view;
+        D3D12_DRAW_INDEXED_ARGUMENTS indexed;
+    };
+
+    static const D3D12_INDIRECT_ARGUMENT_DESC indirect_ibo_sig[2] =
+    {
+        { .Type = D3D12_INDIRECT_ARGUMENT_TYPE_INDEX_BUFFER_VIEW },
+        { .Type = D3D12_INDIRECT_ARGUMENT_TYPE_DRAW_INDEXED },
+    };
+
+    struct indirect_ibo_data indirect_ibo_data[] =
+    {
+        {
+            .view = { 0, 0, DXGI_FORMAT_R32_UINT },
+            .indexed = { .IndexCountPerInstance = 2, .InstanceCount = 1 }
+        },
+        {
+            .view = { 0, 64, DXGI_FORMAT_R16_UINT },
+            .indexed = { .IndexCountPerInstance = 4, .InstanceCount = 1 }
+        },
+    };
+
+    static const struct vec4 indirect_ibo_expected[] =
+    {
+        { 1000.0f, 2064.0f, 3000.0f, 4000.0f },
+        { 1000.0f, 2064.0f, 3000.0f, 4000.0f },
+        { 1016.0f, 2080.0f, 3000.0f, 4000.0f },
+        { 1000.0f, 2064.0f, 3000.0f, 4000.0f },
+        { 1017.0f, 2081.0f, 3000.0f, 4000.0f },
+        { 1000.0f, 2064.0f, 3000.0f, 4000.0f },
+    };
+
+    /* Indirect root arguments */
+    struct indirect_root_descriptor_data
+    {
+        D3D12_GPU_VIRTUAL_ADDRESS cbv;
+        D3D12_GPU_VIRTUAL_ADDRESS srv;
+        D3D12_GPU_VIRTUAL_ADDRESS uav;
+        D3D12_DRAW_ARGUMENTS array;
+    };
+
+    static const D3D12_INDIRECT_ARGUMENT_DESC indirect_root_descriptor_sig[4] =
+    {
+        { .Type = D3D12_INDIRECT_ARGUMENT_TYPE_CONSTANT_BUFFER_VIEW, .ConstantBufferView = { .RootParameterIndex = 1 } },
+        { .Type = D3D12_INDIRECT_ARGUMENT_TYPE_SHADER_RESOURCE_VIEW, .ShaderResourceView = { .RootParameterIndex = 2 } },
+        { .Type = D3D12_INDIRECT_ARGUMENT_TYPE_UNORDERED_ACCESS_VIEW, .UnorderedAccessView = { .RootParameterIndex = 3 } },
+        { .Type = D3D12_INDIRECT_ARGUMENT_TYPE_DRAW },
+    };
+
+    struct indirect_root_descriptor_data indirect_root_descriptor_data[] =
+    {
+        { .array = { .VertexCountPerInstance = 1, .InstanceCount = 1 } },
+        { .array = { .VertexCountPerInstance = 1, .InstanceCount = 1 } },
+    };
+
+    static const struct vec4 indirect_root_descriptor_expected[] =
+    {
+        { 1000.0f, 2064.0f, 3000.0f + 64.0f, 4000.0f + 2.0f },
+        { 1000.0f, 2064.0f, 3000.0f + 128.0f, 4000.0f + 3.0f },
+    };
+
+    /* Test packing rules.
+     * 64-bit aligned values are tightly packed with 32-bit alignment when they are in indirect command buffers. */
+    struct indirect_alignment_data
+    {
+        float value;
+        uint32_t cbv_va[2];
+        D3D12_DRAW_ARGUMENTS arrays;
+    };
+    static const D3D12_INDIRECT_ARGUMENT_DESC indirect_alignment_sig[3] =
+    {
+        { .Type = D3D12_INDIRECT_ARGUMENT_TYPE_CONSTANT, .Constant = {
+                .RootParameterIndex = 0, .DestOffsetIn32BitValues = 1, .Num32BitValuesToSet = 1 }},
+        { .Type = D3D12_INDIRECT_ARGUMENT_TYPE_CONSTANT_BUFFER_VIEW, .ConstantBufferView = { .RootParameterIndex = 1 }},
+        { .Type = D3D12_INDIRECT_ARGUMENT_TYPE_DRAW },
+    };
+
+    struct indirect_alignment_data indirect_alignment_data[] =
+    {
+        {
+            .value = 5.0f,
+            .arrays = { .VertexCountPerInstance = 1, .InstanceCount = 1 }
+        },
+        {
+            .value = 6.0f,
+            .arrays = { .VertexCountPerInstance = 1, .InstanceCount = 1 }
+        },
+    };
+
+    static const struct vec4 indirect_alignment_expected[] =
+    {
+        { 1000.0f, 69.0f, 3064.0f, 4000.0f },
+        { 1000.0f, 70.0f, 3128.0f, 4000.0f },
+    };
+
+#define DECL_TEST(t, pso_index, needs_root_sig) { t##_sig, ARRAY_SIZE(t##_sig), t##_data, sizeof(t##_data), ARRAY_SIZE(t##_data), \
+        t##_expected, ARRAY_SIZE(t##_expected), sizeof(*(t##_data)), pso_index, needs_root_sig }
+    const struct test tests[] =
+    {
+        DECL_TEST(root_constant, 0, true),
+        DECL_TEST(indirect_vbo, 0, false),
+        DECL_TEST(indirect_vbo_one, 0, false),
+        DECL_TEST(indirect_ibo, 0, false),
+        DECL_TEST(indirect_root_descriptor, 0, true),
+        DECL_TEST(indirect_alignment, 0, true),
+        DECL_TEST(root_constant_spill, 1, true),
+        DECL_TEST(indirect_root_descriptor, 1, true),
+    };
+#undef DECL_TEST
+
+    uint32_t ibo_data[ARRAY_SIZE(ibo)][64];
+    float vbo_data[ARRAY_SIZE(vbo)][64];
+    float generic_data[4096];
+
+    static const DWORD vs_code_small_cbv[] =
+    {
+#if 0
+    cbuffer RootCBV : register(b0)
+    {
+            float a;
+    };
+
+    StructuredBuffer<float> RootSRV : register(t0);
+
+    cbuffer RootConstants : register(b0, space1)
+    {
+            float4 root;
+    };
+
+    float4 main(float c0 : COLOR0, float c1 : COLOR1, uint iid : SV_InstanceID) : SV_Position
+    {
+            return float4(c0, c1, a, RootSRV[0] + float(iid)) + root;
+    }
+#endif
+        0x43425844, 0x33b7b302, 0x34259b9b, 0x3e8568d9, 0x5a5e0c3e, 0x00000001, 0x00000268, 0x00000003,
+        0x0000002c, 0x00000098, 0x000000cc, 0x4e475349, 0x00000064, 0x00000003, 0x00000008, 0x00000050,
+        0x00000000, 0x00000000, 0x00000003, 0x00000000, 0x00000101, 0x00000050, 0x00000001, 0x00000000,
+        0x00000003, 0x00000001, 0x00000101, 0x00000056, 0x00000000, 0x00000008, 0x00000001, 0x00000002,
+        0x00000101, 0x4f4c4f43, 0x56530052, 0x736e495f, 0x636e6174, 0x00444965, 0x4e47534f, 0x0000002c,
+        0x00000001, 0x00000008, 0x00000020, 0x00000000, 0x00000001, 0x00000003, 0x00000000, 0x0000000f,
+        0x505f5653, 0x7469736f, 0x006e6f69, 0x58454853, 0x00000194, 0x00010051, 0x00000065, 0x0100086a,
+        0x07000059, 0x00308e46, 0x00000000, 0x00000000, 0x00000000, 0x00000001, 0x00000000, 0x07000059,
+        0x00308e46, 0x00000001, 0x00000000, 0x00000000, 0x00000001, 0x00000001, 0x070000a2, 0x00307e46,
+        0x00000000, 0x00000000, 0x00000000, 0x00000004, 0x00000000, 0x0300005f, 0x00101012, 0x00000000,
+        0x0300005f, 0x00101012, 0x00000001, 0x04000060, 0x00101012, 0x00000002, 0x00000008, 0x04000067,
+        0x001020f2, 0x00000000, 0x00000001, 0x02000068, 0x00000001, 0x0a0000a7, 0x00100012, 0x00000000,
+        0x00004001, 0x00000000, 0x00004001, 0x00000000, 0x00207006, 0x00000000, 0x00000000, 0x05000056,
+        0x00100022, 0x00000000, 0x0010100a, 0x00000002, 0x07000000, 0x00100012, 0x00000000, 0x0010001a,
+        0x00000000, 0x0010000a, 0x00000000, 0x09000000, 0x00102012, 0x00000000, 0x0010100a, 0x00000000,
+        0x0030800a, 0x00000001, 0x00000000, 0x00000000, 0x09000000, 0x00102022, 0x00000000, 0x0010100a,
+        0x00000001, 0x0030801a, 0x00000001, 0x00000000, 0x00000000, 0x0b000000, 0x00102042, 0x00000000,
+        0x0030800a, 0x00000000, 0x00000000, 0x00000000, 0x0030802a, 0x00000001, 0x00000000, 0x00000000,
+        0x09000000, 0x00102082, 0x00000000, 0x0010000a, 0x00000000, 0x0030803a, 0x00000001, 0x00000000,
+        0x00000000, 0x0100003e,
+    };
+
+    static const DWORD vs_code_large_cbv[] =
+    {
+#if 0
+    cbuffer RootCBV : register(b0)
+    {
+        float a;
+    };
+
+    StructuredBuffer<float> RootSRV : register(t0);
+
+    cbuffer RootConstants : register(b0, space1)
+    {
+        // Cannot use arrays for root constants in D3D12.
+        float4 pad0, pad1, pad2, pad3, pad4, pad5, pad6, pad7, pad8, pad9, pad10;
+        float4 root;
+    };
+
+    float4 main(float c0 : COLOR0, float c1 : COLOR1, uint iid : SV_InstanceID) : SV_Position
+    {
+        return float4(c0, c1, a, RootSRV[0] + float(iid)) + root;
+    }
+#endif
+        0x43425844, 0x99a057e8, 0x20344569, 0x434f8a7a, 0xf9171e08, 0x00000001, 0x00000268, 0x00000003,
+        0x0000002c, 0x00000098, 0x000000cc, 0x4e475349, 0x00000064, 0x00000003, 0x00000008, 0x00000050,
+        0x00000000, 0x00000000, 0x00000003, 0x00000000, 0x00000101, 0x00000050, 0x00000001, 0x00000000,
+        0x00000003, 0x00000001, 0x00000101, 0x00000056, 0x00000000, 0x00000008, 0x00000001, 0x00000002,
+        0x00000101, 0x4f4c4f43, 0x56530052, 0x736e495f, 0x636e6174, 0x00444965, 0x4e47534f, 0x0000002c,
+        0x00000001, 0x00000008, 0x00000020, 0x00000000, 0x00000001, 0x00000003, 0x00000000, 0x0000000f,
+        0x505f5653, 0x7469736f, 0x006e6f69, 0x58454853, 0x00000194, 0x00010051, 0x00000065, 0x0100086a,
+        0x07000059, 0x00308e46, 0x00000000, 0x00000000, 0x00000000, 0x00000001, 0x00000000, 0x07000059,
+        0x00308e46, 0x00000001, 0x00000000, 0x00000000, 0x0000000c, 0x00000001, 0x070000a2, 0x00307e46,
+        0x00000000, 0x00000000, 0x00000000, 0x00000004, 0x00000000, 0x0300005f, 0x00101012, 0x00000000,
+        0x0300005f, 0x00101012, 0x00000001, 0x04000060, 0x00101012, 0x00000002, 0x00000008, 0x04000067,
+        0x001020f2, 0x00000000, 0x00000001, 0x02000068, 0x00000001, 0x0a0000a7, 0x00100012, 0x00000000,
+        0x00004001, 0x00000000, 0x00004001, 0x00000000, 0x00207006, 0x00000000, 0x00000000, 0x05000056,
+        0x00100022, 0x00000000, 0x0010100a, 0x00000002, 0x07000000, 0x00100012, 0x00000000, 0x0010001a,
+        0x00000000, 0x0010000a, 0x00000000, 0x09000000, 0x00102012, 0x00000000, 0x0010100a, 0x00000000,
+        0x0030800a, 0x00000001, 0x00000000, 0x0000000b, 0x09000000, 0x00102022, 0x00000000, 0x0010100a,
+        0x00000001, 0x0030801a, 0x00000001, 0x00000000, 0x0000000b, 0x0b000000, 0x00102042, 0x00000000,
+        0x0030800a, 0x00000000, 0x00000000, 0x00000000, 0x0030802a, 0x00000001, 0x00000000, 0x0000000b,
+        0x09000000, 0x00102082, 0x00000000, 0x0010000a, 0x00000000, 0x0030803a, 0x00000001, 0x00000000,
+        0x0000000b, 0x0100003e,
+    };
+
+    memset(&desc, 0, sizeof(desc));
+    desc.no_root_signature = true;
+    desc.no_pipeline = true;
+    if (!init_test_context(&context, &desc))
+        return;
+    command_list = context.list;
+    queue = context.queue;
+
+    for (j = 0; j < ARRAY_SIZE(ibo); j++)
+        for (i = 0; i < ARRAY_SIZE(ibo_data[j]); i++)
+            ibo_data[j][i] = j * 16 + i;
+
+    for (j = 0; j < ARRAY_SIZE(vbo); j++)
+        for (i = 0; i < ARRAY_SIZE(vbo_data[j]); i++)
+            vbo_data[j][i] = (float)(j * ARRAY_SIZE(vbo_data[j]) + i);
+
+    for (i = 0; i < ARRAY_SIZE(generic_data); i++)
+        generic_data[i] = (float)i;
+
+    for (i = 0; i < ARRAY_SIZE(ibo); i++)
+        ibo[i] = create_upload_buffer(context.device, sizeof(ibo_data[i]), ibo_data[i]);
+    for (i = 0; i < ARRAY_SIZE(vbo); i++)
+        vbo[i] = create_upload_buffer(context.device, sizeof(vbo_data[i]), vbo_data[i]);
+    cbv = create_upload_buffer(context.device, sizeof(generic_data), generic_data);
+    srv = create_upload_buffer(context.device, sizeof(generic_data), generic_data);
+    uav = create_default_buffer(context.device, sizeof(generic_data),
+            D3D12_RESOURCE_FLAG_ALLOW_UNORDERED_ACCESS,
+            D3D12_RESOURCE_STATE_UNORDERED_ACCESS);
+
+    indirect_vbo_data[0].view[0].BufferLocation = ID3D12Resource_GetGPUVirtualAddress(vbo[1]);
+    indirect_vbo_data[0].view[1].BufferLocation = ID3D12Resource_GetGPUVirtualAddress(vbo[2]);
+    indirect_vbo_data[1].view[0].BufferLocation = 0;
+    indirect_vbo_data[1].view[1].BufferLocation = ID3D12Resource_GetGPUVirtualAddress(vbo[0]) + 64;
+
+    indirect_vbo_one_data[0].view.BufferLocation = ID3D12Resource_GetGPUVirtualAddress(vbo[2]);
+    indirect_vbo_one_data[1].view.BufferLocation = 0;
+
+    indirect_ibo_data[1].view.BufferLocation = ID3D12Resource_GetGPUVirtualAddress(ibo[1]);
+
+    indirect_root_descriptor_data[0].cbv = ID3D12Resource_GetGPUVirtualAddress(cbv) + 256;
+    indirect_root_descriptor_data[0].srv = ID3D12Resource_GetGPUVirtualAddress(srv) + 8;
+    indirect_root_descriptor_data[0].uav = ID3D12Resource_GetGPUVirtualAddress(uav) + 4;
+    indirect_root_descriptor_data[1].cbv = ID3D12Resource_GetGPUVirtualAddress(cbv) + 512;
+    indirect_root_descriptor_data[1].srv = ID3D12Resource_GetGPUVirtualAddress(srv) + 12;
+    indirect_root_descriptor_data[1].uav = ID3D12Resource_GetGPUVirtualAddress(uav) + 8;
+
+    memcpy(indirect_alignment_data[0].cbv_va, &indirect_root_descriptor_data[0].cbv, sizeof(D3D12_GPU_VIRTUAL_ADDRESS));
+    memcpy(indirect_alignment_data[1].cbv_va, &indirect_root_descriptor_data[1].cbv, sizeof(D3D12_GPU_VIRTUAL_ADDRESS));
+
+    memset(&root_signature_desc, 0, sizeof(root_signature_desc));
+    root_signature_desc.Flags = D3D12_ROOT_SIGNATURE_FLAG_ALLOW_INPUT_ASSEMBLER_INPUT_LAYOUT |
+            D3D12_ROOT_SIGNATURE_FLAG_ALLOW_STREAM_OUTPUT;
+
+    memset(root_parameters, 0, sizeof(root_parameters));
+    root_signature_desc.pParameters = root_parameters;
+    root_signature_desc.NumParameters = ARRAY_SIZE(root_parameters);
+    root_parameters[0].ShaderVisibility = D3D12_SHADER_VISIBILITY_ALL;
+    root_parameters[0].ParameterType = D3D12_ROOT_PARAMETER_TYPE_32BIT_CONSTANTS;
+    root_parameters[0].Constants.RegisterSpace = 1;
+    root_parameters[0].Constants.Num32BitValues = 4;
+    root_parameters[1].ShaderVisibility = D3D12_SHADER_VISIBILITY_ALL;
+    root_parameters[1].ParameterType = D3D12_ROOT_PARAMETER_TYPE_CBV;
+    root_parameters[2].ShaderVisibility = D3D12_SHADER_VISIBILITY_ALL;
+    root_parameters[2].ParameterType = D3D12_ROOT_PARAMETER_TYPE_SRV;
+    root_parameters[3].ShaderVisibility = D3D12_SHADER_VISIBILITY_ALL;
+    root_parameters[3].ParameterType = D3D12_ROOT_PARAMETER_TYPE_UAV;
+    hr = create_root_signature(context.device, &root_signature_desc, &root_signatures[0]);
+    ok(SUCCEEDED(hr), "Failed to create root signature, hr #%x.\n", hr);
+    root_parameters[0].Constants.Num32BitValues = 48;
+    hr = create_root_signature(context.device, &root_signature_desc, &root_signatures[1]);
+    ok(SUCCEEDED(hr), "Failed to create root signature, hr #%x.\n", hr);
+
+    memset(so_entries, 0, sizeof(so_entries));
+    so_entries[0].ComponentCount = 4;
+    so_entries[0].SemanticName = "SV_Position";
+
+    memset(&pso_desc, 0, sizeof(pso_desc));
+    pso_desc.VS.pShaderBytecode = vs_code_small_cbv;
+    pso_desc.VS.BytecodeLength = sizeof(vs_code_small_cbv);
+    pso_desc.StreamOutput.NumStrides = 1;
+    pso_desc.StreamOutput.pBufferStrides = &so_stride;
+    pso_desc.StreamOutput.pSODeclaration = so_entries;
+    pso_desc.StreamOutput.NumEntries = ARRAY_SIZE(so_entries);
+    pso_desc.StreamOutput.RasterizedStream = D3D12_SO_NO_RASTERIZED_STREAM;
+    pso_desc.pRootSignature = root_signatures[0];
+    pso_desc.SampleDesc.Count = 1;
+    pso_desc.PrimitiveTopologyType = D3D12_PRIMITIVE_TOPOLOGY_TYPE_POINT;
+    pso_desc.RasterizerState.FillMode = D3D12_FILL_MODE_SOLID;
+    pso_desc.RasterizerState.CullMode = D3D12_CULL_MODE_NONE;
+    pso_desc.InputLayout.NumElements = ARRAY_SIZE(layout_desc);
+    pso_desc.InputLayout.pInputElementDescs = layout_desc;
+    hr = ID3D12Device_CreateGraphicsPipelineState(context.device, &pso_desc, &IID_ID3D12PipelineState, (void**)&psos[0]);
+    ok(SUCCEEDED(hr), "Failed to create PSO, hr #%x.\n", hr);
+    pso_desc.VS.pShaderBytecode = vs_code_large_cbv;
+    pso_desc.VS.BytecodeLength = sizeof(vs_code_large_cbv);
+    pso_desc.pRootSignature = root_signatures[1];
+    hr = ID3D12Device_CreateGraphicsPipelineState(context.device, &pso_desc, &IID_ID3D12PipelineState, (void**)&psos[1]);
+    ok(SUCCEEDED(hr), "Failed to create PSO, hr #%x.\n", hr);
+
+    /* Verify sanity checks.
+     * As per validation layers, there must be exactly one command in the signature.
+     * It must come last. Verify that we check for this. */
+    memset(&command_signature_desc, 0, sizeof(command_signature_desc));
+    command_signature_desc.NumArgumentDescs = 1;
+    command_signature_desc.pArgumentDescs = indirect_argument_descs;
+    command_signature_desc.ByteStride = sizeof(D3D12_VERTEX_BUFFER_VIEW);
+    indirect_argument_descs[0].Type = D3D12_INDIRECT_ARGUMENT_TYPE_VERTEX_BUFFER_VIEW;
+    hr = ID3D12Device_CreateCommandSignature(context.device, &command_signature_desc, NULL,
+            &IID_ID3D12CommandSignature, (void**)&command_signature);
+    ok(hr == E_INVALIDARG, "Unexpected hr #%x.\n", hr);
+
+    command_signature_desc.NumArgumentDescs = 2;
+    command_signature_desc.pArgumentDescs = indirect_argument_descs;
+    command_signature_desc.ByteStride = sizeof(D3D12_DRAW_INDEXED_ARGUMENTS) + sizeof(D3D12_VERTEX_BUFFER_VIEW);
+    indirect_argument_descs[0].Type = D3D12_INDIRECT_ARGUMENT_TYPE_DRAW_INDEXED;
+    indirect_argument_descs[1].Type = D3D12_INDIRECT_ARGUMENT_TYPE_VERTEX_BUFFER_VIEW;
+    hr = ID3D12Device_CreateCommandSignature(context.device, &command_signature_desc, NULL,
+            &IID_ID3D12CommandSignature, (void**)&command_signature);
+    ok(hr == E_INVALIDARG, "Unexpected hr #%x.\n", hr);
+
+    command_signature_desc.ByteStride = sizeof(D3D12_DRAW_INDEXED_ARGUMENTS) + sizeof(D3D12_DRAW_INDEXED_ARGUMENTS);
+    indirect_argument_descs[0].Type = D3D12_INDIRECT_ARGUMENT_TYPE_DRAW_INDEXED;
+    indirect_argument_descs[1].Type = D3D12_INDIRECT_ARGUMENT_TYPE_DRAW_INDEXED;
+    hr = ID3D12Device_CreateCommandSignature(context.device, &command_signature_desc, NULL,
+            &IID_ID3D12CommandSignature, (void**)&command_signature);
+    ok(hr == E_INVALIDARG, "Unexpected hr #%x.\n", hr);
+
+    for (i = 0; i < ARRAY_SIZE(tests); i++)
+    {
+        struct vec4 expect_reset_state[2];
+        const struct vec4 *expect, *v;
+        uint32_t expected_output_size;
+        uint32_t clear_vbo_mask;
+        bool clear_ibo;
+        uint32_t size;
+
+        vkd3d_test_set_context("Test %u", i);
+
+        command_signature_desc.ByteStride = tests[i].stride;
+        command_signature_desc.pArgumentDescs = tests[i].indirect_arguments;
+        command_signature_desc.NumArgumentDescs = tests[i].indirect_argument_count;
+        command_signature_desc.NodeMask = 0;
+        hr = ID3D12Device_CreateCommandSignature(context.device, &command_signature_desc,
+                tests[i].needs_root_sig ? root_signatures[tests[i].pso_index] : NULL,
+                &IID_ID3D12CommandSignature, (void**)&command_signature);
+        ok(SUCCEEDED(hr), "Failed to create command signature, hr #%x.\n", hr);
+
+        argument_buffer = create_upload_buffer(context.device, 256 * 1024, NULL);
+        argument_buffer_late = create_default_buffer(context.device, 256 * 1024,
+                D3D12_RESOURCE_FLAG_NONE, D3D12_RESOURCE_STATE_COPY_DEST);
+        {
+            void *ptr;
+            ID3D12Resource_Map(argument_buffer, 0, NULL, &ptr);
+            memcpy(ptr, tests[i].argument_buffer_data, tests[i].argument_buffer_size);
+            ID3D12Resource_Unmap(argument_buffer, 0, NULL);
+        }
+
+        count_buffer = create_upload_buffer(context.device, sizeof(tests[i].api_max_count), &tests[i].api_max_count);
+        streamout_buffer = create_default_buffer(context.device, 64 * 1024,
+                D3D12_RESOURCE_FLAG_NONE, D3D12_RESOURCE_STATE_STREAM_OUT);
+
+        ID3D12GraphicsCommandList_SetGraphicsRootSignature(command_list, root_signatures[tests[i].pso_index]);
+        ID3D12GraphicsCommandList_SetPipelineState(command_list, psos[tests[i].pso_index]);
+        sov.SizeInBytes = 64 * 1024 - sizeof(struct vec4);
+        sov.BufferLocation = ID3D12Resource_GetGPUVirtualAddress(streamout_buffer) + sizeof(struct vec4);
+        sov.BufferFilledSizeLocation = ID3D12Resource_GetGPUVirtualAddress(streamout_buffer);
+        ID3D12GraphicsCommandList_SOSetTargets(command_list, 0, 1, &sov);
+
+        /* Set up default rendering state. */
+        ibv.BufferLocation = ID3D12Resource_GetGPUVirtualAddress(ibo[0]);
+        ibv.SizeInBytes = sizeof(ibo_data[0]);
+        ibv.Format = DXGI_FORMAT_R32_UINT;
+        vbvs[0].BufferLocation = ID3D12Resource_GetGPUVirtualAddress(vbo[0]);
+        vbvs[0].SizeInBytes = sizeof(vbo_data[0]);
+        vbvs[0].StrideInBytes = 4;
+        vbvs[1].BufferLocation = ID3D12Resource_GetGPUVirtualAddress(vbo[1]);
+        vbvs[1].SizeInBytes = sizeof(vbo_data[1]);
+        vbvs[1].StrideInBytes = 4;
+
+        ID3D12GraphicsCommandList_IASetIndexBuffer(command_list, &ibv);
+        ID3D12GraphicsCommandList_IASetPrimitiveTopology(command_list, D3D_PRIMITIVE_TOPOLOGY_POINTLIST);
+        ID3D12GraphicsCommandList_IASetVertexBuffers(command_list, 0, 2, vbvs);
+
+        for (j = 0; j < (tests[i].pso_index ? 12 : 1); j++)
+            ID3D12GraphicsCommandList_SetGraphicsRoot32BitConstants(command_list, 0, 4, &values, 4 * j);
+
+        ID3D12GraphicsCommandList_SetGraphicsRootConstantBufferView(command_list, 1,
+                ID3D12Resource_GetGPUVirtualAddress(cbv));
+        ID3D12GraphicsCommandList_SetGraphicsRootShaderResourceView(command_list, 2,
+                ID3D12Resource_GetGPUVirtualAddress(srv));
+        ID3D12GraphicsCommandList_SetGraphicsRootUnorderedAccessView(command_list, 3,
+                ID3D12Resource_GetGPUVirtualAddress(uav));
+        ID3D12GraphicsCommandList_ExecuteIndirect(command_list, command_signature, tests[i].api_max_count,
+                argument_buffer, 0, NULL, 0);
+        /* Test equivalent call with indirect count. */
+        ID3D12GraphicsCommandList_ExecuteIndirect(command_list, command_signature, 1024,
+                argument_buffer, 0, count_buffer, 0);
+        /* Test equivalent, but now with late transition to INDIRECT. */
+        ID3D12GraphicsCommandList_CopyResource(command_list, argument_buffer_late, argument_buffer);
+        transition_resource_state(command_list, argument_buffer_late, D3D12_RESOURCE_STATE_COPY_DEST,
+                D3D12_RESOURCE_STATE_INDIRECT_ARGUMENT);
+        ID3D12GraphicsCommandList_ExecuteIndirect(command_list, command_signature, 1024,
+                argument_buffer_late, 0, count_buffer, 0);
+
+        /* Root descriptors which are part of the state block are cleared to NULL. Recover them here
+         * since attempting to draw next test will crash GPU. */
+        ID3D12GraphicsCommandList_SetGraphicsRootConstantBufferView(command_list, 1,
+                ID3D12Resource_GetGPUVirtualAddress(cbv));
+        ID3D12GraphicsCommandList_SetGraphicsRootShaderResourceView(command_list, 2,
+                ID3D12Resource_GetGPUVirtualAddress(srv));
+        ID3D12GraphicsCommandList_SetGraphicsRootUnorderedAccessView(command_list, 3,
+                ID3D12Resource_GetGPUVirtualAddress(uav));
+
+        /* Other state is cleared to 0. */
+
+        ID3D12GraphicsCommandList_DrawInstanced(command_list, 2, 1, 0, 0);
+        transition_resource_state(command_list, streamout_buffer, D3D12_RESOURCE_STATE_STREAM_OUT, D3D12_RESOURCE_STATE_COPY_SOURCE);
+
+        get_buffer_readback_with_command_list(streamout_buffer, DXGI_FORMAT_R32G32B32A32_FLOAT, &rb, queue, command_list);
+        reset_command_list(command_list, context.allocator);
+
+        expected_output_size = (tests[i].expected_output_count * 3 + 2) * sizeof(struct vec4);
+        size = get_readback_uint(&rb, 0, 0, 0);
+        ok(size == expected_output_size, "Expected size %u, got %u.\n", expected_output_size, size);
+
+        for (j = 0; j < tests[i].expected_output_count; j++)
+        {
+            expect = &tests[i].expected_output[j];
+            v = get_readback_vec4(&rb, j + 1, 0);
+            ok(compare_vec4(v, expect, 0), "Element (direct count) %u failed: (%f, %f, %f, %f) != (%f, %f, %f, %f)\n",
+                    j, v->x, v->y, v->z, v->w, expect->x, expect->y, expect->z, expect->w);
+
+            v = get_readback_vec4(&rb, j + tests[i].expected_output_count + 1, 0);
+            ok(compare_vec4(v, expect, 0), "Element (indirect count) %u failed: (%f, %f, %f, %f) != (%f, %f, %f, %f)\n",
+                    j, v->x, v->y, v->z, v->w, expect->x, expect->y, expect->z, expect->w);
+
+            v = get_readback_vec4(&rb, j + 2 * tests[i].expected_output_count + 1, 0);
+            ok(compare_vec4(v, expect, 0), "Element (late latch) %u failed: (%f, %f, %f, %f) != (%f, %f, %f, %f)\n",
+                    j, v->x, v->y, v->z, v->w, expect->x, expect->y, expect->z, expect->w);
+        }
+
+        clear_vbo_mask = 0;
+        expect_reset_state[0] = values;
+
+        /* Root constant state is cleared to zero if it's part of the signature. */
+        for (j = 0; j < tests[i].indirect_argument_count; j++)
+        {
+            if (tests[i].indirect_arguments[j].Type == D3D12_INDIRECT_ARGUMENT_TYPE_CONSTANT)
+            {
+                for (k = 0; k < tests[i].indirect_arguments[j].Constant.Num32BitValuesToSet; k++)
+                    (&expect_reset_state[0].x)[(tests[i].indirect_arguments[j].Constant.DestOffsetIn32BitValues + k) % 4] = 0.0f;
+            }
+            else if (tests[i].indirect_arguments[j].Type == D3D12_INDIRECT_ARGUMENT_TYPE_VERTEX_BUFFER_VIEW)
+                clear_vbo_mask |= 1u << tests[i].indirect_arguments[j].VertexBuffer.Slot;
+        }
+
+        expect_reset_state[1] = expect_reset_state[0];
+
+        /* VBO/IBO state is cleared to zero if it's part of the signature.
+         * A NULL IBO should be seen as a IBO which only reads 0 index. */
+        if (!(clear_vbo_mask & (1u << 0)))
+            expect_reset_state[1].x += 1.0f;
+
+        if (!(clear_vbo_mask & (1u << 1)))
+        {
+            expect_reset_state[0].y += 64.0f;
+            expect_reset_state[1].y += 65.0f;
+        }
+
+        for (j = 0; j < 2; j++)
+        {
+            v = get_readback_vec4(&rb, j + 1 + 3 * tests[i].expected_output_count, 0);
+            expect = &expect_reset_state[j];
+            ok(compare_vec4(v, expect, 0), "Post-reset element %u failed: (%f, %f, %f, %f) != (%f, %f, %f, %f)\n",
+                    j, v->x, v->y, v->z, v->w, expect->x, expect->y, expect->z, expect->w);
+        }
+
+        ID3D12CommandSignature_Release(command_signature);
+        ID3D12Resource_Release(argument_buffer);
+        ID3D12Resource_Release(argument_buffer_late);
+        ID3D12Resource_Release(count_buffer);
+        ID3D12Resource_Release(streamout_buffer);
+        release_resource_readback(&rb);
+    }
+    vkd3d_test_set_context(NULL);
+
+    for (i = 0; i < ARRAY_SIZE(psos); i++)
+        ID3D12PipelineState_Release(psos[i]);
+    for (i = 0; i < ARRAY_SIZE(root_signatures); i++)
+        ID3D12RootSignature_Release(root_signatures[i]);
+    for (i = 0; i < ARRAY_SIZE(vbo); i++)
+        ID3D12Resource_Release(vbo[i]);
+    for (i = 0; i < ARRAY_SIZE(ibo); i++)
+        ID3D12Resource_Release(ibo[i]);
+    ID3D12Resource_Release(cbv);
+    ID3D12Resource_Release(srv);
+    ID3D12Resource_Release(uav);
+
+    destroy_test_context(&context);
+}
+
 void test_execute_indirect(void)
 {
     ID3D12Resource *argument_buffer, *count_buffer, *uav;

tests/d3d12_tests.h

@@ -135,6 +135,7 @@ decl_test(test_resolve_non_issued_query_data);
 decl_test(test_resolve_query_data_in_different_command_list);
 decl_test(test_resolve_query_data_in_reordered_command_list);
 decl_test(test_execute_indirect);
+decl_test(test_execute_indirect_state);
 decl_test(test_dispatch_zero_thread_groups);
 decl_test(test_unaligned_vertex_stride);
 decl_test(test_zero_vertex_stride);