Add end2end tests to expose buffer size error on D3D12

When we do B2T or T2B copy from/to a buffer with paddings,
D3D12 may wrongly calculate the required buffer size.

For example, if copySize = {1, 1, 2}, offset = 0, bytesPerRow =
256, and rowsPerImage = 2 (there is 1-row padding for every image),
and we are copying a non-compressed format like rgba8unorm,
the required minimum buffer size should be:
offset + bytesPerRow * rowsPerImage * (copySize.depthOrArrayLayers - 1)
+ bytesPerRow * (copySize.height - 1) + bytesPerBlock * copySize.width.
It is 0 + 256 * 2 * (2 - 1) + 256 * (1 - 1) + 4 * 1 = 516.

The required minimum buffer on D3D12 (including WARP) size is:
offset + bytesPerRow * rowsPerImage * (copySize.depthOrArrayLayers - 1)
+ bytesPerRow * (rowsPerImage - 1) + bytesPerBlock * copySize.width.
Or offset + bytesPerRow * rowsPerImage * copySize.depthOrArrayLayers
+ bytesPerBlock * copySize.width - bytesPerRow.
It is 0 + 256 * 2 * (2 - 1) + 256 * (2 - 1) + 4 * 1 = 772.

It looks like D3D12 requires unnecessary buffer storage for
rowsPerImagePadding in the last image. It does respect
bytesPerRowPadding in the last row and doesn't require storage for
that part, though.

You can verify the buffer size requirement on D3D12 backend with the
new tests via --enable-backend-validation. The validation layer
says that D3D12 requires 772 bytes but we only provide a 516-bytes
buffer, and leads to E_INVALIDARG (Error code 0x80070057) when run
mD3d12CommandList->Close() in CommandRecordingContext::ExecuteCommandList
and causes device lost.

Bug: dawn:1278, dawn:1288, dawn:1289
Change-Id: Icfb792dec60ff7444cb20b3c283709cdb165f80a
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/85341
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Yunchao He <yunchao.he@intel.com>
2 files changed
tree: 84a03b2be5d5f84b20fda0c1621e608646c168f0
  1. .vscode/
  2. build_overrides/
  3. docs/
  4. generator/
  5. include/
  6. infra/
  7. samples/
  8. scripts/
  9. src/
  10. test/
  11. third_party/
  12. tools/
  13. webgpu-cts/
  14. .clang-format
  15. .gitattributes
  16. .gitignore
  17. .gn
  18. AUTHORS
  19. BUILD.gn
  20. CMakeLists.txt
  21. codereview.settings
  22. dawn.json
  23. dawn_wire.json
  24. DEPS
  25. DIR_METADATA
  26. LICENSE
  27. OWNERS
  28. PRESUBMIT.py
  29. README.chromium
  30. README.md
README.md

Dawn's logo: a sun rising behind a stylized mountain inspired by the WebGPU logo. The text Dawn is written below it.

Dawn, a WebGPU implementation

Dawn is an open-source and cross-platform implementation of the work-in-progress WebGPU standard. More precisely it implements webgpu.h that is a one-to-one mapping with the WebGPU IDL. Dawn is meant to be integrated as part of a larger system and is the underlying implementation of WebGPU in Chromium.

Dawn provides several WebGPU building blocks:

  • WebGPU C/C++ headers that applications and other building blocks use.
    • The webgpu.h version that Dawn implements.
    • A C++ wrapper for the webgpu.h.
  • A “native” implementation of WebGPU using platforms' GPU APIs:
    • D3D12 on Windows 10
    • Metal on macOS and iOS
    • Vulkan on Windows, Linux, ChromeOS, Android and Fuchsia
    • OpenGL as best effort where available
  • A client-server implementation of WebGPU for applications that are in a sandbox without access to native drivers

Helpful links:

Documentation table of content

Developer documentation:

User documentation: (TODO, figure out what overlaps with the webgpu.h docs)

Status

(TODO)

License

Apache 2.0 Public License, please see LICENSE.

Disclaimer

This is not an officially supported Google product.