dawn-cmake/src/dawn/tests/end2end/BufferTests.cpp

// Copyright 2017 The Dawn Authors
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//     http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

#include <array>
#include <cstring>
#include <limits>
#include <vector>

#include "dawn/tests/DawnTest.h"

class BufferMappingTests : public DawnTest {
  protected:
    void MapAsyncAndWait(const wgpu::Buffer& buffer,
                         wgpu::MapMode mode,
                         size_t offset,
                         size_t size) {
        bool done = false;
        buffer.MapAsync(
            mode, offset, size,
            [](WGPUBufferMapAsyncStatus status, void* userdata) {
                ASSERT_EQ(WGPUBufferMapAsyncStatus_Success, status);
                *static_cast<bool*>(userdata) = true;
            },
            &done);

        while (!done) {
            WaitABit();
        }
    }

    wgpu::Buffer CreateMapReadBuffer(uint64_t size) {
        wgpu::BufferDescriptor descriptor;
        descriptor.size = size;
        descriptor.usage = wgpu::BufferUsage::MapRead | wgpu::BufferUsage::CopyDst;
        return device.CreateBuffer(&descriptor);
    }

    wgpu::Buffer CreateMapWriteBuffer(uint64_t size) {
        wgpu::BufferDescriptor descriptor;
        descriptor.size = size;
        descriptor.usage = wgpu::BufferUsage::MapWrite | wgpu::BufferUsage::CopySrc;
        return device.CreateBuffer(&descriptor);
    }
};

void CheckMapping(const void* actual, const void* expected, size_t size) {
    EXPECT_NE(actual, nullptr);
    if (actual != nullptr) {
        EXPECT_EQ(0, memcmp(actual, expected, size));
    }
}

// Test that the simplest map read works
TEST_P(BufferMappingTests, MapRead_Basic) {
    wgpu::Buffer buffer = CreateMapReadBuffer(4);

    uint32_t myData = 0x01020304;
    constexpr size_t kSize = sizeof(myData);
    queue.WriteBuffer(buffer, 0, &myData, kSize);

    MapAsyncAndWait(buffer, wgpu::MapMode::Read, 0, 4);
    CheckMapping(buffer.GetConstMappedRange(), &myData, kSize);
    CheckMapping(buffer.GetConstMappedRange(0, kSize), &myData, kSize);
    buffer.Unmap();
}

// Test map-reading a zero-sized buffer.
TEST_P(BufferMappingTests, MapRead_ZeroSized) {
    wgpu::Buffer buffer = CreateMapReadBuffer(0);

    MapAsyncAndWait(buffer, wgpu::MapMode::Read, 0, wgpu::kWholeMapSize);
    ASSERT_NE(buffer.GetConstMappedRange(), nullptr);
    buffer.Unmap();
}

// Test map-reading with a non-zero offset
TEST_P(BufferMappingTests, MapRead_NonZeroOffset) {
    wgpu::Buffer buffer = CreateMapReadBuffer(12);

    uint32_t myData[3] = {0x01020304, 0x05060708, 0x090A0B0C};
    queue.WriteBuffer(buffer, 0, &myData, sizeof(myData));

    MapAsyncAndWait(buffer, wgpu::MapMode::Read, 8, 4);
    ASSERT_EQ(myData[2], *static_cast<const uint32_t*>(buffer.GetConstMappedRange(8)));
    buffer.Unmap();
}

// Map read and unmap twice. Test that both of these two iterations work.
TEST_P(BufferMappingTests, MapRead_Twice) {
    wgpu::Buffer buffer = CreateMapReadBuffer(4);

    uint32_t myData = 0x01020304;
    queue.WriteBuffer(buffer, 0, &myData, sizeof(myData));

    MapAsyncAndWait(buffer, wgpu::MapMode::Read, 0, 4);
    ASSERT_EQ(myData, *static_cast<const uint32_t*>(buffer.GetConstMappedRange()));
    buffer.Unmap();

    myData = 0x05060708;
    queue.WriteBuffer(buffer, 0, &myData, sizeof(myData));

    MapAsyncAndWait(buffer, wgpu::MapMode::Read, 0, 4);
    ASSERT_EQ(myData, *static_cast<const uint32_t*>(buffer.GetConstMappedRange()));
    buffer.Unmap();
}

// Map read and test multiple get mapped range data
TEST_P(BufferMappingTests, MapRead_MultipleMappedRange) {
    wgpu::Buffer buffer = CreateMapReadBuffer(12);

    uint32_t myData[] = {0x00010203, 0x04050607, 0x08090a0b};
    queue.WriteBuffer(buffer, 0, &myData, 12);

    MapAsyncAndWait(buffer, wgpu::MapMode::Read, 0, 12);
    ASSERT_EQ(myData[0], *static_cast<const uint32_t*>(buffer.GetConstMappedRange(0)));
    ASSERT_EQ(myData[1], *(static_cast<const uint32_t*>(buffer.GetConstMappedRange(0)) + 1));
    ASSERT_EQ(myData[2], *(static_cast<const uint32_t*>(buffer.GetConstMappedRange(0)) + 2));
    ASSERT_EQ(myData[2], *static_cast<const uint32_t*>(buffer.GetConstMappedRange(8)));
    buffer.Unmap();
}

// Test map-reading a large buffer.
TEST_P(BufferMappingTests, MapRead_Large) {
    constexpr uint32_t kDataSize = 1000 * 1000;
    constexpr size_t kByteSize = kDataSize * sizeof(uint32_t);
    wgpu::Buffer buffer = CreateMapReadBuffer(kByteSize);

    std::vector<uint32_t> myData;
    for (uint32_t i = 0; i < kDataSize; ++i) {
        myData.push_back(i);
    }
    queue.WriteBuffer(buffer, 0, myData.data(), kByteSize);

    MapAsyncAndWait(buffer, wgpu::MapMode::Read, 0, kByteSize);
    EXPECT_EQ(nullptr, buffer.GetConstMappedRange(0, kByteSize + 4));
    EXPECT_EQ(0, memcmp(buffer.GetConstMappedRange(), myData.data(), kByteSize));
    EXPECT_EQ(0, memcmp(buffer.GetConstMappedRange(8), myData.data() + 2, kByteSize - 8));
    EXPECT_EQ(
        0, memcmp(buffer.GetConstMappedRange(8, kByteSize - 8), myData.data() + 2, kByteSize - 8));
    buffer.Unmap();

    MapAsyncAndWait(buffer, wgpu::MapMode::Read, 16, kByteSize - 16);
    // Size is too big.
    EXPECT_EQ(nullptr, buffer.GetConstMappedRange(16, kByteSize - 12));
    // Offset defaults to 0 which is less than 16
    EXPECT_EQ(nullptr, buffer.GetConstMappedRange());
    // Offset less than 8 is less than 16
    EXPECT_EQ(nullptr, buffer.GetConstMappedRange(8));

    // Test a couple values.
    EXPECT_EQ(0, memcmp(buffer.GetConstMappedRange(16), myData.data() + 4, kByteSize - 16));
    EXPECT_EQ(0, memcmp(buffer.GetConstMappedRange(24), myData.data() + 6, kByteSize - 24));

    buffer.Unmap();
}

// Test that GetConstMappedRange works inside map-read callback
TEST_P(BufferMappingTests, MapRead_InCallback) {
    constexpr size_t kBufferSize = 12;
    wgpu::Buffer buffer = CreateMapReadBuffer(kBufferSize);

    uint32_t myData[3] = {0x01020304, 0x05060708, 0x090A0B0C};
    static constexpr size_t kSize = sizeof(myData);
    queue.WriteBuffer(buffer, 0, &myData, kSize);

    struct UserData {
        bool done;
        wgpu::Buffer buffer;
        void* expected;
    };
    UserData user{false, buffer, &myData};

    buffer.MapAsync(
        wgpu::MapMode::Read, 0, kBufferSize,
        [](WGPUBufferMapAsyncStatus status, void* userdata) {
            UserData* user = static_cast<UserData*>(userdata);

            EXPECT_EQ(WGPUBufferMapAsyncStatus_Success, status);
            if (status == WGPUBufferMapAsyncStatus_Success) {
                CheckMapping(user->buffer.GetConstMappedRange(), user->expected, kSize);
                CheckMapping(user->buffer.GetConstMappedRange(0, kSize), user->expected, kSize);

                CheckMapping(user->buffer.GetConstMappedRange(8, 4),
                             static_cast<const uint32_t*>(user->expected) + 2, sizeof(uint32_t));

                user->buffer.Unmap();
            }
            user->done = true;
        },
        &user);

    while (!user.done) {
        WaitABit();
    }
}

// Test that the simplest map write works.
TEST_P(BufferMappingTests, MapWrite_Basic) {
    wgpu::Buffer buffer = CreateMapWriteBuffer(4);

    uint32_t myData = 2934875;
    MapAsyncAndWait(buffer, wgpu::MapMode::Write, 0, 4);
    ASSERT_NE(nullptr, buffer.GetMappedRange());
    ASSERT_NE(nullptr, buffer.GetConstMappedRange());
    memcpy(buffer.GetMappedRange(), &myData, sizeof(myData));
    buffer.Unmap();

    EXPECT_BUFFER_U32_EQ(myData, buffer, 0);
}

// Test that the simplest map write works with a range.
TEST_P(BufferMappingTests, MapWrite_BasicRange) {
    wgpu::Buffer buffer = CreateMapWriteBuffer(4);

    uint32_t myData = 2934875;
    MapAsyncAndWait(buffer, wgpu::MapMode::Write, 0, 4);
    ASSERT_NE(nullptr, buffer.GetMappedRange(0, 4));
    ASSERT_NE(nullptr, buffer.GetConstMappedRange(0, 4));
    memcpy(buffer.GetMappedRange(), &myData, sizeof(myData));
    buffer.Unmap();

    EXPECT_BUFFER_U32_EQ(myData, buffer, 0);
}

// Test map-writing a zero-sized buffer.
TEST_P(BufferMappingTests, MapWrite_ZeroSized) {
    wgpu::Buffer buffer = CreateMapWriteBuffer(0);

    MapAsyncAndWait(buffer, wgpu::MapMode::Write, 0, wgpu::kWholeMapSize);
    ASSERT_NE(buffer.GetConstMappedRange(), nullptr);
    ASSERT_NE(buffer.GetMappedRange(), nullptr);
    buffer.Unmap();
}

// Test map-writing with a non-zero offset.
TEST_P(BufferMappingTests, MapWrite_NonZeroOffset) {
    wgpu::Buffer buffer = CreateMapWriteBuffer(12);

    uint32_t myData = 2934875;
    MapAsyncAndWait(buffer, wgpu::MapMode::Write, 8, 4);
    memcpy(buffer.GetMappedRange(8), &myData, sizeof(myData));
    buffer.Unmap();

    EXPECT_BUFFER_U32_EQ(myData, buffer, 8);
}

// Map, write and unmap twice. Test that both of these two iterations work.
TEST_P(BufferMappingTests, MapWrite_Twice) {
    wgpu::Buffer buffer = CreateMapWriteBuffer(4);

    uint32_t myData = 2934875;
    MapAsyncAndWait(buffer, wgpu::MapMode::Write, 0, 4);
    memcpy(buffer.GetMappedRange(), &myData, sizeof(myData));
    buffer.Unmap();

    EXPECT_BUFFER_U32_EQ(myData, buffer, 0);

    myData = 9999999;
    MapAsyncAndWait(buffer, wgpu::MapMode::Write, 0, 4);
    memcpy(buffer.GetMappedRange(), &myData, sizeof(myData));
    buffer.Unmap();

    EXPECT_BUFFER_U32_EQ(myData, buffer, 0);
}

// Map write and unmap twice with different ranges and make sure the first write is preserved
TEST_P(BufferMappingTests, MapWrite_TwicePreserve) {
    wgpu::Buffer buffer = CreateMapWriteBuffer(12);

    uint32_t data1 = 0x08090a0b;
    size_t offset1 = 8;
    MapAsyncAndWait(buffer, wgpu::MapMode::Write, offset1, wgpu::kWholeMapSize);
    memcpy(buffer.GetMappedRange(offset1), &data1, sizeof(data1));
    buffer.Unmap();

    uint32_t data2 = 0x00010203;
    size_t offset2 = 0;
    MapAsyncAndWait(buffer, wgpu::MapMode::Write, offset2, wgpu::kWholeMapSize);
    memcpy(buffer.GetMappedRange(offset2), &data2, sizeof(data2));
    buffer.Unmap();

    EXPECT_BUFFER_U32_EQ(data1, buffer, offset1);
    EXPECT_BUFFER_U32_EQ(data2, buffer, offset2);
}

// Map write and unmap twice with overlapping ranges and make sure data is updated correctly
TEST_P(BufferMappingTests, MapWrite_TwiceRangeOverlap) {
    wgpu::Buffer buffer = CreateMapWriteBuffer(16);

    uint32_t data1[] = {0x01234567, 0x89abcdef};
    size_t offset1 = 8;
    MapAsyncAndWait(buffer, wgpu::MapMode::Write, offset1, 8);
    memcpy(buffer.GetMappedRange(offset1, 8), data1, 8);
    buffer.Unmap();

    EXPECT_BUFFER_U32_EQ(0x00000000, buffer, 0);
    EXPECT_BUFFER_U32_EQ(0x00000000, buffer, 4);
    EXPECT_BUFFER_U32_EQ(0x01234567, buffer, 8);
    EXPECT_BUFFER_U32_EQ(0x89abcdef, buffer, 12);

    uint32_t data2[] = {0x01234567, 0x89abcdef, 0x55555555};
    size_t offset2 = 0;
    MapAsyncAndWait(buffer, wgpu::MapMode::Write, offset2, 12);
    memcpy(buffer.GetMappedRange(offset2, 12), data2, 12);
    buffer.Unmap();

    EXPECT_BUFFER_U32_EQ(0x01234567, buffer, 0);
    EXPECT_BUFFER_U32_EQ(0x89abcdef, buffer, 4);
    EXPECT_BUFFER_U32_EQ(0x55555555, buffer, 8);
    EXPECT_BUFFER_U32_EQ(0x89abcdef, buffer, 12);
}

// Map write and test multiple mapped range data get updated correctly
TEST_P(BufferMappingTests, MapWrite_MultipleMappedRange) {
    wgpu::Buffer buffer = CreateMapWriteBuffer(12);

    uint32_t data1 = 0x08090a0b;
    size_t offset1 = 8;
    uint32_t data2 = 0x00010203;
    size_t offset2 = 0;
    MapAsyncAndWait(buffer, wgpu::MapMode::Write, 0, 12);
    memcpy(buffer.GetMappedRange(offset1), &data1, sizeof(data1));
    memcpy(buffer.GetMappedRange(offset2), &data2, sizeof(data2));
    buffer.Unmap();

    EXPECT_BUFFER_U32_EQ(data1, buffer, offset1);
    EXPECT_BUFFER_U32_EQ(data2, buffer, offset2);
}

// Test mapping a large buffer.
TEST_P(BufferMappingTests, MapWrite_Large) {
    constexpr uint32_t kDataSize = 1000 * 1000;
    constexpr size_t kByteSize = kDataSize * sizeof(uint32_t);
    wgpu::Buffer buffer = CreateMapWriteBuffer(kDataSize * sizeof(uint32_t));

    std::vector<uint32_t> myData;
    for (uint32_t i = 0; i < kDataSize; ++i) {
        myData.push_back(i);
    }

    MapAsyncAndWait(buffer, wgpu::MapMode::Write, 16, kByteSize - 20);
    EXPECT_EQ(nullptr, buffer.GetMappedRange());
    EXPECT_EQ(nullptr, buffer.GetMappedRange(0));
    EXPECT_EQ(nullptr, buffer.GetMappedRange(8));
    EXPECT_EQ(nullptr, buffer.GetMappedRange(16, kByteSize - 8));
    memcpy(buffer.GetMappedRange(16, kByteSize - 20), myData.data(), kByteSize - 20);
    buffer.Unmap();
    EXPECT_BUFFER_U32_RANGE_EQ(myData.data(), buffer, 16, kDataSize - 5);
}

// Stress test mapping many buffers.
TEST_P(BufferMappingTests, MapWrite_ManySimultaneous) {
    constexpr uint32_t kDataSize = 1000;
    std::vector<uint32_t> myData;
    for (uint32_t i = 0; i < kDataSize; ++i) {
        myData.push_back(i);
    }

    constexpr uint32_t kBuffers = 100;
    std::array<wgpu::Buffer, kBuffers> buffers;
    uint32_t mapCompletedCount = 0;

    // Create buffers and request mapping them.
    wgpu::BufferDescriptor descriptor;
    descriptor.size = static_cast<uint32_t>(kDataSize * sizeof(uint32_t));
    descriptor.usage = wgpu::BufferUsage::MapWrite | wgpu::BufferUsage::CopySrc;
    for (uint32_t i = 0; i < kBuffers; ++i) {
        buffers[i] = device.CreateBuffer(&descriptor);

        buffers[i].MapAsync(
            wgpu::MapMode::Write, 0, descriptor.size,
            [](WGPUBufferMapAsyncStatus status, void* userdata) {
                ASSERT_EQ(WGPUBufferMapAsyncStatus_Success, status);
                (*static_cast<uint32_t*>(userdata))++;
            },
            &mapCompletedCount);
    }

    // Wait for all mappings to complete
    while (mapCompletedCount != kBuffers) {
        WaitABit();
    }

    // All buffers are mapped, write into them and unmap them all.
    for (uint32_t i = 0; i < kBuffers; ++i) {
        memcpy(buffers[i].GetMappedRange(0, descriptor.size), myData.data(), descriptor.size);
        buffers[i].Unmap();
    }

    // Check the content of the buffers.
    for (uint32_t i = 0; i < kBuffers; ++i) {
        EXPECT_BUFFER_U32_RANGE_EQ(myData.data(), buffers[i], 0, kDataSize);
    }
}

// Test that the map offset isn't updated when the call is an error.
TEST_P(BufferMappingTests, OffsetNotUpdatedOnError) {
    uint32_t data[3] = {0xCA7, 0xB0A7, 0xBA7};
    wgpu::Buffer buffer = CreateMapReadBuffer(sizeof(data));
    queue.WriteBuffer(buffer, 0, data, sizeof(data));

    // Map the buffer but do not wait on the result yet.
    bool done1 = false;
    bool done2 = false;
    buffer.MapAsync(
        wgpu::MapMode::Read, 8, 4,
        [](WGPUBufferMapAsyncStatus status, void* userdata) {
            ASSERT_EQ(WGPUBufferMapAsyncStatus_Success, status);
            *static_cast<bool*>(userdata) = true;
        },
        &done1);

    // Call MapAsync another time, the callback will be rejected with error status
    // (but doesn't produce a validation error) and mMapOffset is not updated
    // because the buffer is already being mapped and it doesn't allow multiple
    // MapAsync requests.
    buffer.MapAsync(
        wgpu::MapMode::Read, 0, 4,
        [](WGPUBufferMapAsyncStatus status, void* userdata) {
            ASSERT_EQ(WGPUBufferMapAsyncStatus_Error, status);
            *static_cast<bool*>(userdata) = true;
        },
        &done2);

    while (!done1 || !done2) {
        WaitABit();
    }

    // mMapOffset has not been updated so it should still be 4, which is data[1]
    ASSERT_EQ(0, memcmp(buffer.GetConstMappedRange(8), &data[2], sizeof(uint32_t)));
}

// Test that Get(Const)MappedRange work inside map-write callback.
TEST_P(BufferMappingTests, MapWrite_InCallbackDefault) {
    wgpu::Buffer buffer = CreateMapWriteBuffer(4);

    static constexpr uint32_t myData = 2934875;
    static constexpr size_t kSize = sizeof(myData);

    struct UserData {
        bool done;
        wgpu::Buffer buffer;
    };
    UserData user{false, buffer};

    buffer.MapAsync(
        wgpu::MapMode::Write, 0, kSize,
        [](WGPUBufferMapAsyncStatus status, void* userdata) {
            UserData* user = static_cast<UserData*>(userdata);

            EXPECT_EQ(WGPUBufferMapAsyncStatus_Success, status);
            if (status == WGPUBufferMapAsyncStatus_Success) {
                EXPECT_NE(nullptr, user->buffer.GetConstMappedRange());
                void* ptr = user->buffer.GetMappedRange();
                EXPECT_NE(nullptr, ptr);
                if (ptr != nullptr) {
                    uint32_t data = myData;
                    memcpy(ptr, &data, kSize);
                }

                user->buffer.Unmap();
            }
            user->done = true;
        },
        &user);

    while (!user.done) {
        WaitABit();
    }

    EXPECT_BUFFER_U32_EQ(myData, buffer, 0);
}

// Test that Get(Const)MappedRange with range work inside map-write callback.
TEST_P(BufferMappingTests, MapWrite_InCallbackRange) {
    wgpu::Buffer buffer = CreateMapWriteBuffer(4);

    static constexpr uint32_t myData = 2934875;
    static constexpr size_t kSize = sizeof(myData);

    struct UserData {
        bool done;
        wgpu::Buffer buffer;
    };
    UserData user{false, buffer};

    buffer.MapAsync(
        wgpu::MapMode::Write, 0, kSize,
        [](WGPUBufferMapAsyncStatus status, void* userdata) {
            UserData* user = static_cast<UserData*>(userdata);

            EXPECT_EQ(WGPUBufferMapAsyncStatus_Success, status);
            if (status == WGPUBufferMapAsyncStatus_Success) {
                EXPECT_NE(nullptr, user->buffer.GetConstMappedRange(0, kSize));
                void* ptr = user->buffer.GetMappedRange(0, kSize);
                EXPECT_NE(nullptr, ptr);
                if (ptr != nullptr) {
                    uint32_t data = myData;
                    memcpy(ptr, &data, kSize);
                }

                user->buffer.Unmap();
            }
            user->done = true;
        },
        &user);

    while (!user.done) {
        WaitABit();
    }

    EXPECT_BUFFER_U32_EQ(myData, buffer, 0);
}

// Regression test for crbug.com/dawn/969 where this test
// produced invalid barriers.
TEST_P(BufferMappingTests, MapWrite_ZeroSizedTwice) {
    wgpu::Buffer buffer = CreateMapWriteBuffer(0);

    MapAsyncAndWait(buffer, wgpu::MapMode::Write, 0, wgpu::kWholeMapSize);
    buffer.Unmap();

    MapAsyncAndWait(buffer, wgpu::MapMode::Write, 0, wgpu::kWholeMapSize);
}

// Regression test for crbug.com/1421170 where dropping a buffer which needs
// padding bytes initialization resulted in a use-after-free.
TEST_P(BufferMappingTests, RegressChromium1421170) {
    // Create a mappable buffer of size 7. It will be internally
    // aligned such that the padding bytes need to be zero initialized.
    wgpu::BufferDescriptor descriptor;
    descriptor.size = 7;
    descriptor.usage = wgpu::BufferUsage::MapWrite;
    descriptor.mappedAtCreation = false;
    wgpu::Buffer buffer = device.CreateBuffer(&descriptor);

    // Drop the buffer. The pending commands to zero initialize the
    // padding bytes should stay valid.
    buffer = nullptr;
    // Flush pending commands.
    device.Tick();
}

DAWN_INSTANTIATE_TEST(BufferMappingTests,
                      D3D11Backend(),
                      D3D12Backend(),
                      MetalBackend(),
                      OpenGLBackend(),
                      OpenGLESBackend(),
                      VulkanBackend());

class BufferMappingCallbackTests : public BufferMappingTests {
  protected:
    void SubmitCommandBuffer(wgpu::Buffer buffer) {
        wgpu::CommandEncoder encoder = device.CreateCommandEncoder();

        {
            // Record enough commands to make sure the submission cannot be completed by GPU too
            // quick.
            constexpr int kRepeatCount = 50;
            constexpr int kBufferSize = 1024 * 1024 * 10;
            wgpu::Buffer tempWriteBuffer = CreateMapWriteBuffer(kBufferSize);
            wgpu::Buffer tempReadBuffer = CreateMapReadBuffer(kBufferSize);
            for (int i = 0; i < kRepeatCount; ++i) {
                encoder.CopyBufferToBuffer(tempWriteBuffer, 0, tempReadBuffer, 0, kBufferSize);
            }
        }

        if (buffer) {
            if (buffer.GetUsage() & wgpu::BufferUsage::CopyDst) {
                encoder.ClearBuffer(buffer);
            } else {
                wgpu::Buffer tempBuffer = CreateMapReadBuffer(buffer.GetSize());
                encoder.CopyBufferToBuffer(buffer, 0, tempBuffer, 0, buffer.GetSize());
            }
        }
        wgpu::CommandBuffer commandBuffer = encoder.Finish();
        queue.Submit(1, &commandBuffer);
    }

    void Wait(std::vector<bool>& done) {
        do {
            WaitABit();
        } while (std::any_of(done.begin(), done.end(), [](bool done) { return !done; }));
    }
};

TEST_P(BufferMappingCallbackTests, EmptySubmissionAndThenMap) {
    wgpu::Buffer buffer = CreateMapWriteBuffer(4);
    MapAsyncAndWait(buffer, wgpu::MapMode::Write, 0, wgpu::kWholeMapSize);
    buffer.Unmap();

    std::vector<bool> done = {false, false};

    // 1. submission without using buffer.
    SubmitCommandBuffer({});
    queue.OnSubmittedWorkDone(
        0,
        [](WGPUQueueWorkDoneStatus status, void* userdata) {
            EXPECT_EQ(status, WGPUQueueWorkDoneStatus_Success);
            auto& done = *static_cast<std::vector<bool>*>(userdata);
            done[0] = true;
            // Step 2 callback should be called first, this is the second.
            const std::vector<bool> kExpected = {true, true};
            EXPECT_EQ(done, kExpected);
        },
        &done);

    // 2.
    buffer.MapAsync(
        wgpu::MapMode::Write, 0, wgpu::kWholeMapSize,
        [](WGPUBufferMapAsyncStatus status, void* userdata) {
            EXPECT_EQ(status, WGPUBufferMapAsyncStatus_Success);
            auto& done = *static_cast<std::vector<bool>*>(userdata);
            done[1] = true;
            // The buffer is not used by step 1, so this callback is called first.
            const std::vector<bool> kExpected = {false, true};
            EXPECT_EQ(done, kExpected);
        },
        &done);

    Wait(done);
}

TEST_P(BufferMappingCallbackTests, UseTheBufferAndThenMap) {
    wgpu::Buffer buffer = CreateMapWriteBuffer(4);
    MapAsyncAndWait(buffer, wgpu::MapMode::Write, 0, wgpu::kWholeMapSize);
    buffer.Unmap();

    std::vector<bool> done = {false, false};

    // 1. Submit a command buffer which uses the buffer
    SubmitCommandBuffer(buffer);
    queue.OnSubmittedWorkDone(
        0,
        [](WGPUQueueWorkDoneStatus status, void* userdata) {
            EXPECT_EQ(status, WGPUQueueWorkDoneStatus_Success);
            auto& done = *static_cast<std::vector<bool>*>(userdata);
            done[0] = true;
            // This callback should be called first
            const std::vector<bool> kExpected = {true, false};
            EXPECT_EQ(done, kExpected);
        },
        &done);

    // 2.
    buffer.MapAsync(
        wgpu::MapMode::Write, 0, wgpu::kWholeMapSize,
        [](WGPUBufferMapAsyncStatus status, void* userdata) {
            EXPECT_EQ(status, WGPUBufferMapAsyncStatus_Success);
            auto& done = *static_cast<std::vector<bool>*>(userdata);
            done[1] = true;
            // The buffer is used by step 1, so this callback is called second.
            const std::vector<bool> kExpected = {true, true};
            EXPECT_EQ(done, kExpected);
        },
        &done);

    Wait(done);

    buffer.Unmap();
}

TEST_P(BufferMappingCallbackTests, EmptySubmissionWriteAndThenMap) {
    wgpu::Buffer buffer = CreateMapReadBuffer(4);
    MapAsyncAndWait(buffer, wgpu::MapMode::Read, 0, wgpu::kWholeMapSize);
    buffer.Unmap();

    std::vector<bool> done = {false, false};

    // 1. submission without using buffer.
    SubmitCommandBuffer({});
    queue.OnSubmittedWorkDone(
        0,
        [](WGPUQueueWorkDoneStatus status, void* userdata) {
            EXPECT_EQ(status, WGPUQueueWorkDoneStatus_Success);
            auto& done = *static_cast<std::vector<bool>*>(userdata);
            done[0] = true;
            // Step 2 callback should be called first, this is the second.
            const std::vector<bool> kExpected = {true, false};
            EXPECT_EQ(done, kExpected);
        },
        &done);

    int32_t data = 0x12345678;
    queue.WriteBuffer(buffer, 0, &data, sizeof(data));

    // 2.
    buffer.MapAsync(
        wgpu::MapMode::Read, 0, wgpu::kWholeMapSize,
        [](WGPUBufferMapAsyncStatus status, void* userdata) {
            EXPECT_EQ(status, WGPUBufferMapAsyncStatus_Success);
            auto& done = *static_cast<std::vector<bool>*>(userdata);
            done[1] = true;
            // The buffer is not used by step 1, so this callback is called first.
            const std::vector<bool> kExpected = {true, true};
            EXPECT_EQ(done, kExpected);
        },
        &done);

    Wait(done);

    buffer.Unmap();
}

DAWN_INSTANTIATE_TEST(BufferMappingCallbackTests,
                      D3D11Backend(),
                      D3D12Backend(),
                      MetalBackend(),
                      VulkanBackend());

class BufferMappedAtCreationTests : public DawnTest {
  protected:
    static void MapCallback(WGPUBufferMapAsyncStatus status, void* userdata) {
        EXPECT_EQ(WGPUBufferMapAsyncStatus_Success, status);
        *static_cast<bool*>(userdata) = true;
    }

    const void* MapAsyncAndWait(const wgpu::Buffer& buffer, wgpu::MapMode mode, size_t size) {
        bool done = false;
        buffer.MapAsync(mode, 0, size, MapCallback, &done);

        while (!done) {
            WaitABit();
        }

        return buffer.GetConstMappedRange(0, size);
    }

    void UnmapBuffer(const wgpu::Buffer& buffer) { buffer.Unmap(); }

    wgpu::Buffer BufferMappedAtCreation(wgpu::BufferUsage usage, uint64_t size) {
        wgpu::BufferDescriptor descriptor;
        descriptor.size = size;
        descriptor.usage = usage;
        descriptor.mappedAtCreation = true;
        return device.CreateBuffer(&descriptor);
    }

    wgpu::Buffer BufferMappedAtCreationWithData(wgpu::BufferUsage usage,
                                                const std::vector<uint32_t>& data) {
        size_t byteLength = data.size() * sizeof(uint32_t);
        wgpu::Buffer buffer = BufferMappedAtCreation(usage, byteLength);
        memcpy(buffer.GetMappedRange(), data.data(), byteLength);
        return buffer;
    }
};

// Test that the simplest mappedAtCreation works for MapWrite buffers.
TEST_P(BufferMappedAtCreationTests, MapWriteUsageSmall) {
    uint32_t myData = 230502;
    wgpu::Buffer buffer = BufferMappedAtCreationWithData(
        wgpu::BufferUsage::MapWrite | wgpu::BufferUsage::CopySrc, {myData});
    UnmapBuffer(buffer);
    EXPECT_BUFFER_U32_EQ(myData, buffer, 0);
}

// Test that the simplest mappedAtCreation works for MapRead buffers.
TEST_P(BufferMappedAtCreationTests, MapReadUsageSmall) {
    uint32_t myData = 230502;
    wgpu::Buffer buffer = BufferMappedAtCreationWithData(wgpu::BufferUsage::MapRead, {myData});
    UnmapBuffer(buffer);

    const void* mappedData = MapAsyncAndWait(buffer, wgpu::MapMode::Read, 4);
    ASSERT_EQ(myData, *reinterpret_cast<const uint32_t*>(mappedData));
    UnmapBuffer(buffer);
}

// Test that the simplest mappedAtCreation works for non-mappable buffers.
TEST_P(BufferMappedAtCreationTests, NonMappableUsageSmall) {
    uint32_t myData = 4239;
    wgpu::Buffer buffer = BufferMappedAtCreationWithData(wgpu::BufferUsage::CopySrc, {myData});
    UnmapBuffer(buffer);

    EXPECT_BUFFER_U32_EQ(myData, buffer, 0);
}

// Test mappedAtCreation for a large MapWrite buffer
TEST_P(BufferMappedAtCreationTests, MapWriteUsageLarge) {
    constexpr uint64_t kDataSize = 1000 * 1000;
    std::vector<uint32_t> myData;
    for (uint32_t i = 0; i < kDataSize; ++i) {
        myData.push_back(i);
    }

    wgpu::Buffer buffer = BufferMappedAtCreationWithData(
        wgpu::BufferUsage::MapWrite | wgpu::BufferUsage::CopySrc, {myData});
    UnmapBuffer(buffer);

    EXPECT_BUFFER_U32_RANGE_EQ(myData.data(), buffer, 0, kDataSize);
}

// Test mappedAtCreation for a large MapRead buffer
TEST_P(BufferMappedAtCreationTests, MapReadUsageLarge) {
    constexpr uint64_t kDataSize = 1000 * 1000;
    std::vector<uint32_t> myData;
    for (uint32_t i = 0; i < kDataSize; ++i) {
        myData.push_back(i);
    }

    wgpu::Buffer buffer = BufferMappedAtCreationWithData(wgpu::BufferUsage::MapRead, myData);
    UnmapBuffer(buffer);

    const void* mappedData =
        MapAsyncAndWait(buffer, wgpu::MapMode::Read, kDataSize * sizeof(uint32_t));
    ASSERT_EQ(0, memcmp(mappedData, myData.data(), kDataSize * sizeof(uint32_t)));
    UnmapBuffer(buffer);
}

// Test mappedAtCreation for a large non-mappable buffer
TEST_P(BufferMappedAtCreationTests, NonMappableUsageLarge) {
    constexpr uint64_t kDataSize = 1000 * 1000;
    std::vector<uint32_t> myData;
    for (uint32_t i = 0; i < kDataSize; ++i) {
        myData.push_back(i);
    }

    wgpu::Buffer buffer = BufferMappedAtCreationWithData(wgpu::BufferUsage::CopySrc, {myData});
    UnmapBuffer(buffer);

    EXPECT_BUFFER_U32_RANGE_EQ(myData.data(), buffer, 0, kDataSize);
}

// Test destroying a non-mappable buffer mapped at creation.
// This is a regression test for an issue where the D3D12 backend thought the buffer was actually
// mapped and tried to unlock the heap residency (when actually the buffer was using a staging
// buffer)
TEST_P(BufferMappedAtCreationTests, DestroyNonMappableWhileMappedForCreation) {
    wgpu::Buffer buffer = BufferMappedAtCreation(wgpu::BufferUsage::CopySrc, 4);
    buffer.Destroy();
}

// Test destroying a mappable buffer mapped at creation.
TEST_P(BufferMappedAtCreationTests, DestroyMappableWhileMappedForCreation) {
    wgpu::Buffer buffer = BufferMappedAtCreation(wgpu::BufferUsage::MapRead, 4);
    buffer.Destroy();
}

// Test that mapping a buffer is valid after mappedAtCreation and Unmap
TEST_P(BufferMappedAtCreationTests, CreateThenMapSuccess) {
    static uint32_t myData = 230502;
    wgpu::Buffer buffer = BufferMappedAtCreationWithData(
        wgpu::BufferUsage::MapWrite | wgpu::BufferUsage::CopySrc, {myData});
    UnmapBuffer(buffer);

    EXPECT_BUFFER_U32_EQ(myData, buffer, 0);

    bool done = false;
    buffer.MapAsync(
        wgpu::MapMode::Write, 0, 4,
        [](WGPUBufferMapAsyncStatus status, void* userdata) {
            ASSERT_EQ(WGPUBufferMapAsyncStatus_Success, status);
            *static_cast<bool*>(userdata) = true;
        },
        &done);

    while (!done) {
        WaitABit();
    }

    UnmapBuffer(buffer);
}

// Test that is is invalid to map a buffer twice when using mappedAtCreation
TEST_P(BufferMappedAtCreationTests, CreateThenMapBeforeUnmapFailure) {
    uint32_t myData = 230502;
    wgpu::Buffer buffer = BufferMappedAtCreationWithData(
        wgpu::BufferUsage::MapWrite | wgpu::BufferUsage::CopySrc, {myData});

    ASSERT_DEVICE_ERROR([&]() {
        bool done = false;
        buffer.MapAsync(
            wgpu::MapMode::Write, 0, 4,
            [](WGPUBufferMapAsyncStatus status, void* userdata) {
                ASSERT_EQ(WGPUBufferMapAsyncStatus_Error, status);
                *static_cast<bool*>(userdata) = true;
            },
            &done);

        while (!done) {
            WaitABit();
        }
    }());

    // mappedAtCreation is unaffected by the MapWrite error.
    UnmapBuffer(buffer);
}

// Test that creating a zero-sized buffer mapped is allowed.
TEST_P(BufferMappedAtCreationTests, ZeroSized) {
    wgpu::BufferDescriptor descriptor;
    descriptor.size = 0;
    descriptor.usage = wgpu::BufferUsage::Vertex;
    descriptor.mappedAtCreation = true;
    wgpu::Buffer buffer = device.CreateBuffer(&descriptor);

    ASSERT_NE(nullptr, buffer.GetMappedRange());

    // Check that unmapping the buffer works too.
    UnmapBuffer(buffer);
}

// Test that creating a zero-sized mapppable buffer mapped. (it is a different code path)
TEST_P(BufferMappedAtCreationTests, ZeroSizedMappableBuffer) {
    wgpu::BufferDescriptor descriptor;
    descriptor.size = 0;
    descriptor.usage = wgpu::BufferUsage::MapWrite;
    descriptor.mappedAtCreation = true;
    wgpu::Buffer buffer = device.CreateBuffer(&descriptor);

    ASSERT_NE(nullptr, buffer.GetMappedRange());

    // Check that unmapping the buffer works too.
    UnmapBuffer(buffer);
}

// Test that creating a zero-sized error buffer mapped. (it is a different code path)
TEST_P(BufferMappedAtCreationTests, ZeroSizedErrorBuffer) {
    DAWN_TEST_UNSUPPORTED_IF(HasToggleEnabled("skip_validation"));

    wgpu::BufferDescriptor descriptor;
    descriptor.size = 0;
    descriptor.usage = wgpu::BufferUsage::MapWrite | wgpu::BufferUsage::Storage;
    descriptor.mappedAtCreation = true;
    wgpu::Buffer buffer;
    ASSERT_DEVICE_ERROR(buffer = device.CreateBuffer(&descriptor));

    ASSERT_NE(nullptr, buffer.GetMappedRange());
}

// Test the result of GetMappedRange when mapped at creation.
TEST_P(BufferMappedAtCreationTests, GetMappedRange) {
    wgpu::BufferDescriptor descriptor;
    descriptor.size = 4;
    descriptor.usage = wgpu::BufferUsage::CopyDst;
    descriptor.mappedAtCreation = true;
    wgpu::Buffer buffer = device.CreateBuffer(&descriptor);

    ASSERT_EQ(buffer.GetMappedRange(), buffer.GetConstMappedRange());
    ASSERT_NE(buffer.GetMappedRange(), nullptr);
    buffer.Unmap();
}

// Test the result of GetMappedRange when mapped at creation for a zero-sized buffer.
TEST_P(BufferMappedAtCreationTests, GetMappedRangeZeroSized) {
    wgpu::BufferDescriptor descriptor;
    descriptor.size = 0;
    descriptor.usage = wgpu::BufferUsage::CopyDst;
    descriptor.mappedAtCreation = true;
    wgpu::Buffer buffer = device.CreateBuffer(&descriptor);

    ASSERT_EQ(buffer.GetMappedRange(), buffer.GetConstMappedRange());
    ASSERT_NE(buffer.GetMappedRange(), nullptr);
    buffer.Unmap();
}

DAWN_INSTANTIATE_TEST(BufferMappedAtCreationTests,
                      D3D11Backend(),
                      D3D12Backend(),
                      D3D12Backend({}, {"use_d3d12_resource_heap_tier2"}),
                      MetalBackend(),
                      OpenGLBackend(),
                      OpenGLESBackend(),
                      VulkanBackend());

class BufferTests : public DawnTest {};

// Test that creating a zero-buffer is allowed.
TEST_P(BufferTests, ZeroSizedBuffer) {
    wgpu::BufferDescriptor desc;
    desc.size = 0;
    desc.usage = wgpu::BufferUsage::CopyDst;
    device.CreateBuffer(&desc);
}

// Test that creating a very large buffers fails gracefully.
TEST_P(BufferTests, CreateBufferOOM) {
    // TODO(http://crbug.com/dawn/749): Missing support.
    DAWN_TEST_UNSUPPORTED_IF(IsOpenGL());
    DAWN_TEST_UNSUPPORTED_IF(IsOpenGLES());
    DAWN_TEST_UNSUPPORTED_IF(IsAsan());

    wgpu::BufferDescriptor descriptor;
    descriptor.usage = wgpu::BufferUsage::CopyDst;

    descriptor.size = std::numeric_limits<uint64_t>::max();
    ASSERT_DEVICE_ERROR(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));

    // Validation errors should always be prior to OOM.
    descriptor.usage = wgpu::BufferUsage::MapRead | wgpu::BufferUsage::Uniform;
    ASSERT_DEVICE_ERROR(device.CreateBuffer(&descriptor));
}

// Test that a very large buffer mappedAtCreation fails gracefully.
TEST_P(BufferTests, BufferMappedAtCreationOOM) {
    // TODO(crbug.com/dawn/1506): new (std::nothrow) crashes on OOM on Mac ARM64 because libunwind
    // doesn't see the previous catchall try-catch.
    DAWN_SUPPRESS_TEST_IF(DAWN_PLATFORM_IS(MACOS) && DAWN_PLATFORM_IS(ARM64));

    // TODO(http://crbug.com/dawn/749): Missing support.
    DAWN_TEST_UNSUPPORTED_IF(IsOpenGL());
    DAWN_TEST_UNSUPPORTED_IF(IsOpenGLES());
    DAWN_TEST_UNSUPPORTED_IF(IsAsan());

    // Test non-mappable buffer
    {
        wgpu::BufferDescriptor descriptor;
        descriptor.size = 4;
        descriptor.usage = wgpu::BufferUsage::CopyDst;
        descriptor.mappedAtCreation = true;

        // Control: test a small buffer works.
        device.CreateBuffer(&descriptor);

        // Test an enormous buffer fails
        descriptor.size = std::numeric_limits<uint64_t>::max();
        if (UsesWire()) {
            wgpu::Buffer buffer = device.CreateBuffer(&descriptor);
            ASSERT_EQ(nullptr, buffer.Get());
        } else {
            ASSERT_DEVICE_ERROR(device.CreateBuffer(&descriptor));
        }

        // UINT64_MAX may be special cased. Test a smaller, but really large buffer also fails
        descriptor.size = 1ull << 50;
        if (UsesWire()) {
            wgpu::Buffer buffer = device.CreateBuffer(&descriptor);
            ASSERT_EQ(nullptr, buffer.Get());
        } else {
            ASSERT_DEVICE_ERROR(device.CreateBuffer(&descriptor));
        }
    }

    // Test mappable buffer
    {
        wgpu::BufferDescriptor descriptor;
        descriptor.size = 4;
        descriptor.usage = wgpu::BufferUsage::CopySrc | wgpu::BufferUsage::MapWrite;
        descriptor.mappedAtCreation = true;

        // Control: test a small buffer works.
        device.CreateBuffer(&descriptor);

        // Test an enormous buffer fails
        descriptor.size = std::numeric_limits<uint64_t>::max();
        if (UsesWire()) {
            wgpu::Buffer buffer = device.CreateBuffer(&descriptor);
            ASSERT_EQ(nullptr, buffer.Get());
        } else {
            ASSERT_DEVICE_ERROR(device.CreateBuffer(&descriptor));
        }

        // UINT64_MAX may be special cased. Test a smaller, but really large buffer also fails
        descriptor.size = 1ull << 50;
        if (UsesWire()) {
            wgpu::Buffer buffer = device.CreateBuffer(&descriptor);
            ASSERT_EQ(nullptr, buffer.Get());
        } else {
            ASSERT_DEVICE_ERROR(device.CreateBuffer(&descriptor));
        }
    }
}

// Test that mapping an OOM buffer fails gracefully
TEST_P(BufferTests, CreateBufferOOMMapAsync) {
    // TODO(http://crbug.com/dawn/749): Missing support.
    DAWN_TEST_UNSUPPORTED_IF(IsOpenGL());
    DAWN_TEST_UNSUPPORTED_IF(IsOpenGLES());
    DAWN_TEST_UNSUPPORTED_IF(IsAsan());

    auto RunTest = [this](const wgpu::BufferDescriptor& descriptor) {
        wgpu::Buffer buffer;
        ASSERT_DEVICE_ERROR(buffer = device.CreateBuffer(&descriptor));

        bool done = false;
        ASSERT_DEVICE_ERROR(buffer.MapAsync(
            wgpu::MapMode::Write, 0, 4,
            [](WGPUBufferMapAsyncStatus status, void* userdata) {
                EXPECT_EQ(status, WGPUBufferMapAsyncStatus_Error);
                *static_cast<bool*>(userdata) = true;
            },
            &done));

        while (!done) {
            WaitABit();
        }
    };

    wgpu::BufferDescriptor descriptor;
    descriptor.usage = wgpu::BufferUsage::CopySrc | wgpu::BufferUsage::MapWrite;

    // Test an enormous buffer
    descriptor.size = std::numeric_limits<uint64_t>::max();
    RunTest(descriptor);

    // UINT64_MAX may be special cased. Test a smaller, but really large buffer also fails
    descriptor.size = 1ull << 50;
    RunTest(descriptor);
}

DAWN_INSTANTIATE_TEST(BufferTests,
                      D3D11Backend(),
                      D3D12Backend(),
                      MetalBackend(),
                      OpenGLBackend(),
                      OpenGLESBackend(),
                      VulkanBackend());

class BufferNoSuballocationTests : public DawnTest {};

// Regression test for crbug.com/1313172
// This tests a buffer. It then performs writeBuffer and immediately destroys
// it. Though writeBuffer references a destroyed buffer, it should not crash.
TEST_P(BufferNoSuballocationTests, WriteBufferThenDestroy) {
    uint32_t myData = 0x01020304;

    wgpu::BufferDescriptor desc;
    desc.size = 1024;
    desc.usage = wgpu::BufferUsage::CopyDst;
    wgpu::Buffer buffer = device.CreateBuffer(&desc);

    // Enqueue a pending write into the buffer.
    constexpr size_t kSize = sizeof(myData);
    queue.WriteBuffer(buffer, 0, &myData, kSize);

    // Destroy the buffer.
    buffer.Destroy();

    // Flush and wait for all commands.
    queue.Submit(0, nullptr);
    WaitForAllOperations();
}

DAWN_INSTANTIATE_TEST(BufferNoSuballocationTests,
                      D3D11Backend({"disable_resource_suballocation"}),
                      D3D12Backend({"disable_resource_suballocation"}),
                      MetalBackend({"disable_resource_suballocation"}),
                      OpenGLBackend({"disable_resource_suballocation"}),
                      OpenGLESBackend({"disable_resource_suballocation"}),
                      VulkanBackend({"disable_resource_suballocation"}));