Add maxBufferSize limit

Also emits a warning when create buffer size exceeds
the maxBufferSize limit during the deprecation period.

Bug: dawn:1525
Change-Id: I7b47ae5c85b116035fdcea8b68fb574c0a550729
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/103660
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Auto-Submit: Shrek Shao <shrekshao@google.com>
Commit-Queue: Shrek Shao <shrekshao@google.com>

dawn.json

@@ -1263,6 +1263,7 @@
             {"name": "min uniform buffer offset alignment", "type": "uint32_t", "default": "WGPU_LIMIT_U32_UNDEFINED"},
             {"name": "min storage buffer offset alignment", "type": "uint32_t", "default": "WGPU_LIMIT_U32_UNDEFINED"},
             {"name": "max vertex buffers", "type": "uint32_t", "default": "WGPU_LIMIT_U32_UNDEFINED"},
+            {"name": "max buffer size", "type": "uint64_t", "default": "WGPU_LIMIT_U64_UNDEFINED"},
             {"name": "max vertex attributes", "type": "uint32_t", "default": "WGPU_LIMIT_U32_UNDEFINED"},
             {"name": "max vertex buffer array stride", "type": "uint32_t", "default": "WGPU_LIMIT_U32_UNDEFINED"},
             {"name": "max inter stage shader components", "type": "uint32_t", "default": "WGPU_LIMIT_U32_UNDEFINED"},

docs/support.md

@@ -10,7 +10,7 @@ Vulkan is supported as best effort on other platforms (e.g. Windows and macOS).
 
 **Required features**: `depthBiasClamp`, `fragmentStoresAndAtomics`, `fullDrawIndexUint32`, `imageCubeArray`, `independentBlend`, `sampleRateShading`, and either `textureCompressionBC` or both of `textureCompressionETC` and `textureCompressionASTC_LDR`.
 
-**Required limites**: they are too detailed to describe here, but in general should be wildly supported.
+**Required limits**: they are too detailed to describe here, but in general should be wildly supported.
 See the [WebGPU limits](https://gpuweb.github.io/gpuweb/#limits) that mostly correspond to Vulkan limits.
 
 **Operating system support**:

src/dawn/common/Constants.h

@@ -27,6 +27,8 @@ static constexpr uint8_t kMaxColorAttachments = 8u;
 static constexpr uint32_t kTextureBytesPerRowAlignment = 256u;
 static constexpr uint32_t kMaxInterStageShaderComponents = 60u;
 static constexpr uint32_t kMaxInterStageShaderVariables = 16u;
+static constexpr uint64_t kAssumedMaxBufferSize =
+    0x80000000u;  // Use 2 GB when the limit is unavailable
 
 // Per stage limits
 static constexpr uint32_t kMaxSampledTexturesPerShaderStage = 16;

src/dawn/native/Adapter.cpp

@@ -49,7 +49,8 @@ MaybeError AdapterBase::Initialize() {
         "backend=%s type=%s)",
         mName, mDriverDescription, mVendorId, mDeviceId, mBackend, mAdapterType);
 
-    // Enforce internal Dawn constants.
+    // Enforce internal Dawn constants for some limits to ensure they don't go over fixed-size
+    // arrays in Dawn's internal code.
     mLimits.v1.maxVertexBufferArrayStride =
         std::min(mLimits.v1.maxVertexBufferArrayStride, kMaxVertexBufferArrayStride);
     mLimits.v1.maxColorAttachments =

src/dawn/native/Device.cpp

@@ -1432,6 +1432,14 @@ ResultOrError<Ref<BufferBase>> DeviceBase::CreateBuffer(const BufferDescriptor*
     DAWN_TRY(ValidateIsAlive());
     if (IsValidationEnabled()) {
         DAWN_TRY_CONTEXT(ValidateBufferDescriptor(this, descriptor), "validating %s", descriptor);
+
+        // TODO(dawn:1525): Change to validation error after the deprecation period.
+        if (descriptor->size > mLimits.v1.maxBufferSize) {
+            std::string warning =
+                absl::StrFormat("Buffer size (%u) exceeds the max buffer size limit (%u).",
+                                descriptor->size, mLimits.v1.maxBufferSize);
+            EmitDeprecationWarning(warning.c_str());
+        }
     }
 
     Ref<BufferBase> buffer;

src/dawn/native/Limits.cpp

@@ -28,6 +28,10 @@
 #define LIMITS_STORAGE_BUFFER_BINDING_SIZE(X)                                             \
     X(Maximum, maxStorageBufferBindingSize, 134217728, 1073741824, 2147483647, 4294967295)
 
+// Tiers are 256Mb, 1Gb, 2Gb.
+#define LIMITS_MAX_BUFFER_SIZE(X)                                             \
+    X(Maximum, maxBufferSize, 0x10000000, 0x40000000, 0x80000000)
+
 // TODO(crbug.com/dawn/685):
 // These limits don't have tiers yet. Define two tiers with the same values since the macros
 // in this file expect more than one tier.
@@ -49,6 +53,7 @@
     X(Alignment,           minUniformBufferOffsetAlignment,       256,        256) \
     X(Alignment,           minStorageBufferOffsetAlignment,       256,        256) \
     X(Maximum,                            maxVertexBuffers,         8,          8) \
+    X(Maximum,                               maxBufferSize, 268435456,  268435456) \
     X(Maximum,                         maxVertexAttributes,        16,         16) \
     X(Maximum,                  maxVertexBufferArrayStride,      2048,       2048) \
     X(Maximum,               maxInterStageShaderComponents,        60,         60) \
@@ -64,11 +69,13 @@
 #define LIMITS_EACH_GROUP(X)              \
     X(LIMITS_WORKGROUP_STORAGE_SIZE)      \
     X(LIMITS_STORAGE_BUFFER_BINDING_SIZE) \
+    X(LIMITS_MAX_BUFFER_SIZE)             \
     X(LIMITS_OTHER)
 
 #define LIMITS(X)                         \
     LIMITS_WORKGROUP_STORAGE_SIZE(X)      \
     LIMITS_STORAGE_BUFFER_BINDING_SIZE(X) \
+    LIMITS_MAX_BUFFER_SIZE(X)             \
     LIMITS_OTHER(X)
 
 namespace dawn::native {

src/dawn/native/d3d12/AdapterD3D12.cpp

@@ -304,6 +304,8 @@ MaybeError Adapter::InitializeSupportedLimitsImpl(CombinedLimits* limits) {
     limits->v1.maxUniformBufferBindingSize = D3D12_REQ_CONSTANT_BUFFER_ELEMENT_COUNT * 16;
     // D3D12 has no documented limit on the size of a storage buffer binding.
     limits->v1.maxStorageBufferBindingSize = 4294967295;
+    // D3D12 has no documented limit on the buffer size.
+    limits->v1.maxBufferSize = kAssumedMaxBufferSize;
 
     // Using base limits for:
     // TODO(crbug.com/dawn/685):

src/dawn/native/metal/BackendMTL.mm

@@ -598,6 +598,7 @@ class Adapter : public AdapterBase {
         limits->v1.minStorageBufferOffsetAlignment = mtlLimits.minBufferOffsetAlignment;
 
         uint64_t maxBufferSize = Buffer::QueryMaxBufferLength(*mDevice);
+        limits->v1.maxBufferSize = maxBufferSize;
 
         // Metal has no documented limit on the size of a binding. Use the maximum
         // buffer size.

src/dawn/native/vulkan/AdapterVk.cpp

@@ -309,6 +309,16 @@ MaybeError Adapter::InitializeSupportedLimitsImpl(CombinedLimits* limits) {
         return DAWN_INTERNAL_ERROR("Insufficient Vulkan limits for framebufferDepthSampleCounts");
     }
 
+    if (mDeviceInfo.HasExt(DeviceExt::Maintenance3)) {
+        limits->v1.maxBufferSize = mDeviceInfo.propertiesMaintenance3.maxMemoryAllocationSize;
+        if (mDeviceInfo.propertiesMaintenance3.maxMemoryAllocationSize <
+            baseLimits.v1.maxBufferSize) {
+            return DAWN_INTERNAL_ERROR("Insufficient Vulkan maxBufferSize limit");
+        }
+    } else {
+        limits->v1.maxBufferSize = kAssumedMaxBufferSize;
+    }
+
     // Only check maxFragmentCombinedOutputResources on mobile GPUs. Desktop GPUs drivers seem
     // to put incorrect values for this limit with things like 8 or 16 when they can do bindless
     // storage buffers. Mesa llvmpipe driver also puts 8 here.

src/dawn/native/vulkan/VulkanExtensions.cpp

@@ -134,6 +134,8 @@ static constexpr std::array<DeviceExtInfo, kDeviceExtCount> sDeviceExtInfos{{
     //
     {DeviceExt::BindMemory2, "VK_KHR_bind_memory2", VulkanVersion_1_1},
     {DeviceExt::Maintenance1, "VK_KHR_maintenance1", VulkanVersion_1_1},
+    {DeviceExt::Maintenance2, "VK_KHR_maintenance2", VulkanVersion_1_1},
+    {DeviceExt::Maintenance3, "VK_KHR_maintenance3", VulkanVersion_1_1},
     {DeviceExt::StorageBufferStorageClass, "VK_KHR_storage_buffer_storage_class",
      VulkanVersion_1_1},
     {DeviceExt::GetPhysicalDeviceProperties2, "VK_KHR_get_physical_device_properties2",
@@ -210,6 +212,7 @@ DeviceExtSet EnsureDependencies(const DeviceExtSet& advertisedExts,
             case DeviceExt::BindMemory2:
             case DeviceExt::GetMemoryRequirements2:
             case DeviceExt::Maintenance1:
+            case DeviceExt::Maintenance2:
             case DeviceExt::ImageFormatList:
             case DeviceExt::StorageBufferStorageClass:
                 hasDependencies = true;
@@ -224,6 +227,7 @@ DeviceExtSet EnsureDependencies(const DeviceExtSet& advertisedExts,
             // advertises the extension. So if we didn't have this check, we'd risk a calling
             // a nullptr.
             case DeviceExt::GetPhysicalDeviceProperties2:
+            case DeviceExt::Maintenance3:
                 hasDependencies = instanceExts[InstanceExt::GetPhysicalDeviceProperties2];
                 break;
             case DeviceExt::ExternalMemoryCapabilities:

src/dawn/native/vulkan/VulkanExtensions.h

@@ -76,6 +76,8 @@ enum class DeviceExt {
     // Promoted to 1.1
     BindMemory2,
     Maintenance1,
+    Maintenance2,
+    Maintenance3,
     StorageBufferStorageClass,
     GetPhysicalDeviceProperties2,
     GetMemoryRequirements2,

src/dawn/native/vulkan/VulkanInfo.cpp

@@ -223,9 +223,12 @@ ResultOrError<VulkanDeviceInfo> GatherDeviceInfo(const Adapter& adapter) {
 
     VkPhysicalDeviceProperties2 properties2 = {};
     properties2.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PROPERTIES_2;
-    features2.pNext = nullptr;
+
     PNextChainBuilder propertiesChain(&properties2);
 
+    propertiesChain.Add(&info.propertiesMaintenance3,
+                        VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_MAINTENANCE_3_PROPERTIES);
+
     if (info.extensions[DeviceExt::ShaderFloat16Int8]) {
         featuresChain.Add(&info.shaderFloat16Int8Features,
                           VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SHADER_FLOAT16_INT8_FEATURES_KHR);

src/dawn/native/vulkan/VulkanInfo.h

@@ -61,6 +61,7 @@ struct VulkanDeviceKnobs {
 
 struct VulkanDeviceInfo : VulkanDeviceKnobs {
     VkPhysicalDeviceProperties properties;
+    VkPhysicalDeviceMaintenance3Properties propertiesMaintenance3;
     VkPhysicalDeviceDriverProperties driverProperties;
     VkPhysicalDeviceSubgroupSizeControlPropertiesEXT subgroupSizeControlProperties;
     VkPhysicalDeviceShaderIntegerDotProductPropertiesKHR shaderIntegerDotProductProperties;

src/dawn/tests/end2end/BufferTests.cpp

@@ -812,11 +812,13 @@ TEST_P(BufferTests, CreateBufferOOM) {
     descriptor.usage = wgpu::BufferUsage::CopyDst;
 
     descriptor.size = std::numeric_limits<uint64_t>::max();
-    ASSERT_DEVICE_ERROR(device.CreateBuffer(&descriptor));
+    // TODO(dawn:1525): remove warning expectation after the deprecation period.
+    ASSERT_DEVICE_ERROR(EXPECT_DEPRECATION_WARNING(device.CreateBuffer(&descriptor)));
 
     // UINT64_MAX may be special cased. Test a smaller, but really large buffer also fails
     descriptor.size = 1ull << 50;
-    ASSERT_DEVICE_ERROR(device.CreateBuffer(&descriptor));
+    // TODO(dawn:1525): remove warning expectation after the deprecation period.
+    ASSERT_DEVICE_ERROR(EXPECT_DEPRECATION_WARNING(device.CreateBuffer(&descriptor)));
 }
 
 // Test that a very large buffer mappedAtCreation fails gracefully.
@@ -846,7 +848,8 @@ TEST_P(BufferTests, BufferMappedAtCreationOOM) {
 
         // UINT64_MAX may be special cased. Test a smaller, but really large buffer also fails
         descriptor.size = 1ull << 50;
-        ASSERT_DEVICE_ERROR(device.CreateBuffer(&descriptor));
+        // TODO(dawn:1525): remove warning expectation after the deprecation period.
+        ASSERT_DEVICE_ERROR(EXPECT_DEPRECATION_WARNING(device.CreateBuffer(&descriptor)));
     }
 
     // Test mappable buffer
@@ -865,7 +868,8 @@ TEST_P(BufferTests, BufferMappedAtCreationOOM) {
 
         // UINT64_MAX may be special cased. Test a smaller, but really large buffer also fails
         descriptor.size = 1ull << 50;
-        ASSERT_DEVICE_ERROR(device.CreateBuffer(&descriptor));
+        // TODO(dawn:1525): remove warning expectation after the deprecation period.
+        ASSERT_DEVICE_ERROR(EXPECT_DEPRECATION_WARNING(device.CreateBuffer(&descriptor)));
     }
 }
 
@@ -878,7 +882,8 @@ TEST_P(BufferTests, CreateBufferOOMMapAsync) {
 
     auto RunTest = [this](const wgpu::BufferDescriptor& descriptor) {
         wgpu::Buffer buffer;
-        ASSERT_DEVICE_ERROR(buffer = device.CreateBuffer(&descriptor));
+        // TODO(dawn:1525): remove warning expectation after the deprecation period.
+        ASSERT_DEVICE_ERROR(EXPECT_DEPRECATION_WARNING(buffer = device.CreateBuffer(&descriptor)));
 
         bool done = false;
         ASSERT_DEVICE_ERROR(buffer.MapAsync(

src/dawn/tests/end2end/DeprecatedAPITests.cpp

@@ -168,6 +168,22 @@ TEST_P(DeprecationTests, Dispatch) {
     pass.End();
 }
 
+// Test that creating a buffer with size exceeding the maximum buffer size limit should emits a
+// warning. (dawn:1525)
+TEST_P(DeprecationTests, MaxBufferSizeValidation) {
+    wgpu::BufferDescriptor descriptor;
+    descriptor.usage = wgpu::BufferUsage::CopySrc | wgpu::BufferUsage::CopyDst;
+
+    descriptor.size = 256;
+    device.CreateBuffer(&descriptor);
+
+    descriptor.size = GetSupportedLimits().limits.maxBufferSize;
+    device.CreateBuffer(&descriptor);
+
+    descriptor.size = GetSupportedLimits().limits.maxBufferSize + 1;
+    EXPECT_DEPRECATION_WARNING(device.CreateBuffer(&descriptor));
+}
+
 DAWN_INSTANTIATE_TEST(DeprecationTests,
                       D3D12Backend(),
                       MetalBackend(),

src/dawn/tests/unittests/validation/BufferValidationTests.cpp

@@ -992,8 +992,9 @@ TEST_F(BufferValidationTest, CreationParameterReflectionForOOMBuffer) {
     desc.size = kAmazinglyLargeSize;
 
     // OOM!
+    // TODO(dawn:1525): remove warning expectation after the deprecation period.
     wgpu::Buffer buf;
-    ASSERT_DEVICE_ERROR(buf = device.CreateBuffer(&desc));
+    ASSERT_DEVICE_ERROR(EXPECT_DEPRECATION_WARNING(buf = device.CreateBuffer(&desc)));
 
     // Reflection data is still exactly what was in the descriptor.
     EXPECT_EQ(wgpu::BufferUsage::Storage, buf.GetUsage());