This change ensures that when errors are raised from WebGPU, including
lost context errors, they can include any associated D3D12 validation
layer messages in the message text if backend validation is enabled.
This will allow these messages to be surfaced in places like when
running CTS tests on browser build infrastructure.
Also makes a minor adjustment to how HRESULT codes are reported so that
they're easier to read.
Bug: dawn:1396
Change-Id: Ib5c039157c57e6926bc82941a68be03e33e9084c
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/88044
Commit-Queue: Brandon Jones <bajones@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Austin Eng <enga@chromium.org>
ES requires GL_EXT_texture_format_BGRA8888 or GL_APPLE_texture_format_BGRA8888 to create BGRA8 internalFormat textures, and GL_EXT_read_format_bgra to read from them. Desktop GL can swizzle back and forth from RGBA8, so keep using RGBA8 if the extension is unavailable.
Intel's implementation of GL_EXT_texture_format_BGRA8888 on ES is broken, and won't create GL_BGRA8_EXT or GL_BGRA internalFormat textures, so disable the test there and modify another test to not use BGRA textures.
Change-Id: Ia81d9ff20e2849b00379f8e01fb5d2ecfa34bd53
Bug: dawn:596, dawn:1393
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/86744
Commit-Queue: Stephen White <senorblanco@google.com>
Reviewed-by: Austin Eng <enga@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
In follow-up commits these workarounds will be used in more than just
the CommandBuffer but also for lazy-texture clearing. In this CL the
helpers are just moved and changed to take `encodeInside` as a
std::function.
Bug: dawn:1071
Change-Id: Ib3b90826c89e12edca41223b3c8b5a4cfc6fed04
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/87863
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
In this patch, `i` suffix for SInt literal is added, making `-234i`
valid. Meanwhile, validation of UInt literal must not be negative is
add, making `-42u` invalid with proper error message.
Bugs: tint:1514, tint:1515
Change-Id: Idd0e9ae5649a62e7d348c29f95c2570049653bd8
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/88140
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Zhaoming Jiang <zhaoming.jiang@intel.com>
In this CL the enable directive is implemented.
1. Add AST node for enable directive, assign a ExtensionKind (enum) for
each supported extension.
2. Use an unorder_set in ast::Module to record all required extensions'
kind.
3. Provide inspector methods for getting names of used extension, and
getting all used enable directives' extension names and location.
4. For different writer, the extension nodes are handled in different
ways. MSL and HLSL writers will just ignore the extension nodes, while
SPIRV and GLSL writers will emit corresponding code.
5. Implement unittests and end2end test for enable directive and
inspector, using a reserved extension name `InternalExtensionForTesting`.
Bug: tint:1472
Change-Id: I40cb4061554deb477bc2005d7e38c9718385f825
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/86623
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Zhaoming Jiang <zhaoming.jiang@intel.com>
Currently the entire block of output gets coloured in the CTS output.
This can lead to giant blocks of red and green on the screen which is
hard to read. This PR changes the output to only colour the title and
leave the message the default colour.
The skip colour of blue was changed to cyan as the blue was showing as
red on my terminal which makes it appear as an error instead of a skip.
Change-Id: Idb4fc053bb83c710871f75b997e1fc3e3fdea695
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/87679
Auto-Submit: Dan Sinclair <dsinclair@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
Previously the error messages about buffer usages would also include
the internal ones that are not set by the base WebGPU API (e.g.
kInternalStorageBuffer is showed when the buffer usage includes Indirect
or QueryResolve), which may confuse the developers.
This patch fixes this issue by implementing Buffer::GetUsageExternalOnly()
for front-end validations which only return the usages with which the
buffer was created using the base WebGPU API.
BUG=dawn:1382
Change-Id: Ie13135efaf2c3dab729d75a80eff486efa525c7e
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/87382
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
The OpenGL ES 3.1 spec says:
"The point size is taken from the shader built-in gl_PointSize written by the vertex shader, and clamped to the implementation-dependent point size range. If the value written to gl_PointSize is less than or equal to zero, or if no value is written, the point size is undefined."
ANGLE/D3D11 will emit a warning on this and abort processing, causing dawn_end2end tests to fail.
Bug: dawn:1390
Change-Id: I6a56467842ba80d8a070af7240ee07bb111089dd
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/87720
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Stephen White <senorblanco@chromium.org>
Before this change, we duplicated line break parsing in both
Source::File and Lexer. This change makes it so that the Lexer no longer
looks for line breaks, instead relying on Source::File for this info.
This de-duplication will also help in implementing the latest spec
changes with respect to line breaks (CRLF vs CR, etc).
Bug: tint:1505
Bug: tint:1513
Change-Id: Ifa820f75ede7e82822525282127e05d2fea047e1
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/87604
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
remove_stale_autogen_files will remove files in
gen/third_party/daw/src/tint/fuzzers, so build
failed flakey with no such file under this directory
if remove_stale_autogen_files run after tint_generator_wsg_corpus
action.
add deps to remove_stale_autogen_files to make sure
tint_generator_wgsl_corpus runs after remove_stale_autogen_files run.
Bug: crbug.com/1314527
Change-Id: Ic072d1c80719cd1c1cf29b1ad4463792018ebb94
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/87680
Auto-Submit: Fumitoshi Ukai <ukai@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
In [Mach engine](https://machengine.org), we compile Dawn's C++ sources using Zig as the C/C++
compiler and Zig's build system (we don't use Dawn's regular build system.) Zig uses clang
with UBSan enabled by default: this caught this instance of undefined behavior, where passing
no dynamic offsets would result in memcpy being called with size 0.
We've encountered at least [one other instance of undefined behavior](https://github.com/hexops/mach/issues/239)
via UBSan in addition to this one, which I will send another CL to fix soon. I'll also verify
we don't see any other instances of UB.
Bug: dawn:1371
Signed-off-by: Stephen Gutekanst <stephen@hexops.com>
Change-Id: Ifc3b938e990b6a0a3b203796ea99ee5abb9a4a09
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/87380
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
This CL adds names to the TODOs which did not have them and enables
the readability/todo lint. The names are based on the git blame for
the TODO lines.
Bug: dawn:1339
Change-Id: I25a2920bc8fa9606f5dda67a629fdef1c10f8948
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/87600
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Dan Sinclair <dsinclair@chromium.org>
- To be used for pipeline cache testing.
- Plumbs overriding the platform to the adapters for testing.
- Restructures build a little bit so that the test infrastructure can have full access to dawn native internals. Also differentiates end2end and white_box a bit more to make it clear that end2end should not have access to dawn native internals.
Bug: dawn:549, dawn:1374
Change-Id: Ibcc6c44a116c7967ee2317c74409f613e896eb0a
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/86841
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Loko Kung <lokokung@google.com>
- Note: goal is to merge the two into the same class, but to avoid breaking/fixing the current D3D12 shader caching, introducing a new one, and will merge the old one later on.
- Adds instance of BlobCache to the Instance, and passthrough for the Device to get to it. (We want the cache to be at the Instance so that eventually different devices can still use the cache (assuming same isolation key)).
Bug: dawn:549
Change-Id: Ib7c82a439f32b306f0cccd3db424fb0e427a90a1
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/86400
Reviewed-by: Austin Eng <enga@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Loko Kung <lokokung@google.com>
D3D12 requires more buffer storage than it should. If the buffer used
for B2T/T2B copy is big enough according to WebGPU spec but it doesn't
meet D3D12's requirement, we need to workaround it via split the copy
into two copies, in order to make B2T/T2B copy being done correctly on
D3D12.
Bug: dawn:1289, dawn:1278
Change-Id: Id88c580dce00553845817677a2d356690ef4acfe
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/86860
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Yunchao He <yunchao.he@intel.com>
