CopyTextureForBrowser() handles flipY by flipping source texture then
applying copy op. But this is not correct. Instead
it should find the copy rect and apply flipY op in copy rect only,
which is the same as WebGPU GPUImageCopyExternalImage.flipY definition
(https://www.w3.org/TR/webgpu/#dom-gpuimagecopyexternalimage-flipy).
This CL fixed the issue and updated related end2end tests.
Bug: dawn:1635
Change-Id: I3ea7c9de44fb45224bc438486e0e92385446bfb1
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116924
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Shaobo Yan <shaobo.yan@intel.com>
This isn't just failing on Windows Intel but also Windows Nvidia.
Bug: dawn:1617
Change-Id: I9cd492512df4118161deabb9cf3f97eee29c60d4
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116879
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro: Ben Clayton <bclayton@google.com>
Flip the diagnostics so that the trigger location is on the builtin
that requires uniformity.
We also now show the place at which control flow diverges regardless
of where it is in the function call stack.
Change-Id: Id739a137b9011c900649b74165a6600a95d87ca4
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116691
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: James Price <jrprice@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Change-Id: I92c6cf4b52132b806822b9edb163838f474d7c3b
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116876
Commit-Queue: Ben Clayton <bclayton@google.com>
Kokoro: Ben Clayton <bclayton@google.com>
Auto-Submit: Ben Clayton <bclayton@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Emits branch prediction hints.
Give unlikely hints about where we call TINT_ICE.
Change-Id: Ied5bc3d7c8b3a838e96e5a0a64156048f90411c6
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116875
Commit-Queue: Ben Clayton <bclayton@google.com>
Auto-Submit: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
Uniformity analysis can have very deep call stacks, and we're on the cusp of stack overflowing. Reducing these sizes helps avoid cache misses at the cost of more heap allocations.
Change-Id: I4685d1d888d062456c296d9dc25231a48d72e941
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116878
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
Commit-Queue: Ben Clayton <bclayton@google.com>
Deferred exepectations didn't handle device loss gracefully and would
crash, which meant all future tests were skipped. Instead of crashing,
skip the expectation, the test will fail anyway because the device was
lost.
Bug: None
Change-Id: If143f00a5ed9d2ddd5a923da7c771b1f40d80962
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116861
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Loko Kung <lokokung@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Clang was doing a poor job inlining this very hot function and its lambda argument.
Change-Id: Id2005b2ad131a1a1802bc7fb66085395b659ade2
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116867
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
WebGPU specification described that Buffer mapState must be
"unmapped" after destroyed because Destroy() must internally
call Unmap().
But in the current Wire implementation the mapState won't be
"unmapped". The root issue is that Destroy() doesn't update
mMapState in Wire.
This commit fixes the problem by updating mMapState in
Destroy() in Wire.
bug: dawn:1634
Change-Id: I7398510f83ebacfbc6bb33fcc070c2eab10b9c24
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116820
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Takahiro <hogehoge@gachapin.jp>
Shows up in profiling, and doesn't always need to be done.
Change-Id: If8bf061563979d17dea4c48334dab4834770d921
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116865
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
Auto-Submit: Ben Clayton <bclayton@google.com>
If node is shutdown without first explicitly destroying the GPUDevice, the device's destructor will be called, which would automatically destroy the dawn device.
Before doing this, it would first reject the lost-device promise. With newer versions of node, attempting to construct new GC'd objects on NodeJS tear-down triggers a fatal error.
Work around this by adding a new state to promises, that allows promises to be destroyed without first being rejected or resolved, and without raising a 'Promise not resolved or rejected' fatal message.
Change-Id: I810d6894d384511cdb7989b9c6c5b707dd68d8c2
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116864
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Auto-Submit: Ben Clayton <bclayton@google.com>
Kokoro: Ben Clayton <bclayton@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
If a device lists no features, don't zero-index an empty std::vector.
Change-Id: I10d632f0c5d5a162abec180797ad11adee67fc6b
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116863
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Auto-Submit: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
std::array unnecessarily zeros the memory, which shows up in profiles
Change-Id: I2a18ef4fb8ca46bc6fc87504858046d9b7d2fd45
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116866
Commit-Queue: Ben Clayton <bclayton@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
The [selected] was on the second line of an adpater description in the
output of DawnTest which made it a bit difficult to find. Move it at the
end of the first line which is the adapter name.
Bug: None
Change-Id: I686b40588b03aac827c347510525f6c969b7180b
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116860
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
The HLSL `sign` method returns an `int` result (scalar or vector). The
WGSL `sign` expects the result to be the same type as the argument. This
CL injects a cast to the correct type after the `sign` call in the HLSL
generated source.
Bug: tint:1795
Change-Id: I51fed24b5b8b752b6b27fdfb5dd47eb803902793
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116692
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Enable coverage collection when using the test server, which is substantially faster than running in separate, isolated processes.
Use clang's `__llvm_profile_*` APIs to reset the counters between each test case run.
Change-Id: I01f8d0c1b3f215f66cfa50ef0fd51f2522c2ea57
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/113880
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Ben Clayton <bclayton@google.com>
Add the "--adapter" flag to run-cts as a helper for setting this.
Make "--verbose" print the picked adapter.
Rename "dawn-backend" flag to just "backend" - this is already specific to the "gpu provider" (dawn).
Change-Id: Idc8d0eb3ccf5fa23325c06f0f9520aa9b528d9dd
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116295
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
Auto-Submit: Ben Clayton <bclayton@google.com>
Kokoro: Ben Clayton <bclayton@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
printf() can return -1 on error. In this situation, don't adjust msg with a negative offset.
Fixes a spurious crash when emitting lots of text.
Change-Id: Id1e9402bbbe3dd49cf08e660dea0cf67c5369516
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116289
Commit-Queue: Ben Clayton <bclayton@google.com>
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
This patch removes the test
ShaderModuleValidationTest.FragmentOutputLocationExceedsMaxColorAttachments
from dawn_unittests because now the emission of the validation errors about
the maximum value of fragment output location is delayed from
CreateShaderModule() to CreateRenderPipeline(), however this test expects the
error will still be reported in CreateShaderModule().
Previously this test can still pass because kMaxColorAttachments is declared as
an uint8_t, so it will be added as "\b" (ascii code is 8) in string stream when
constructing the shader, then we can still get a shader-creation error because
it is not an integer that is inside the bracket of the location() expression.
Bug: dawn:986
Change-Id: Iaf03f83b2c27e3d9e986f813b2086726b51a0aeb
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116554
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
This patch counts the line position, offset and size of the compilation
message in UTF-16 and saves them to WGPUCompilationMessage to align the
latest WebGPU SPEC.
Bug: dawn:1357
Change-Id: If8f4026bd5b4a64a078e100762b6d1f61da50053
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/115640
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
When validating the image coordinate type of an identifier that may have
been hoisted into a 'var' declaration, spirv-reader correctly checked
the type of the unwrapped reference for scalars but not when the
coordinate type is a vector.
This change applies the vector type checks to the unwrapped type.
Introduced a vector coordinate variant of the
SpvParserHandleTest.ImageCoordinateCanBeHoistedConstant test which
demonstrates the issue and passes with the fix.
Fixed: tint:1712
Change-Id: I9d99a1996e5df71921d6f66d1af02fb5088f1f20
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116371
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: David Neto <dneto@google.com>
Reviewed-by: David Neto <dneto@google.com>
This CL adds diagnostics for the ast nodes which would come before
override substitution. The struct member attributes are marked as ICE as
they should never be encountered. The Struct declaration is ignored as
we will get the `type::Struct` from the semantic usage.
Bug: tint:1718
Change-Id: I8c51787f1455d02e822f222f43a9606a844eed3c
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116549
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Dan Sinclair <dsinclair@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
This Cl updates the flow node tests to compare against a disassembled
output instead of explicit expect statements. This makes it easier to
see the structure of the IR and to determine any changes.
Bug: tint:1718
Change-Id: I5b8ab42ada4ba902e8937099c7058a39533f2010
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116548
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Dan Sinclair <dsinclair@chromium.org>
This CL expands the disassembler output and makes a bit more useful. The
case selector value is changed to an `ir::Constant` instead of the
`constant::Value` to make disassembly easier. The `BuilderImpl` is
updated to not fail in the face of missing implementation but continue
as best as possible.
Bug: tint:1718
Change-Id: I8b275a19bccbb02bb785d311778198bb0c9e0456
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116547
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Dan Sinclair <dsinclair@chromium.org>
When branching to a different flow node, there is a need to pass
arguments to the branch. These arguments could be the value of the
return.
This extracts a `Branch` out to an object with a target and arguments
and then updates the IR to use the new Branch structure.
Bug: tint:1718
Change-Id: Ic8de8046f58056327a04c8afe3b597810c80ccdb
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116546
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Dan Sinclair <dsinclair@chromium.org>
OpenGLES doesn't support 1D textures, so use 2D textures of width x 1
as a workaround (requires dependent Tint change).
Bug: dawn:1301
Change-Id: I99dbccfae497ee86d6f9b9e1ca1608049971016d
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/114820
Reviewed-by: Austin Eng <enga@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Stephen White <senorblanco@chromium.org>
Fix unit and WGSL test results.
Bug: dawn:1301
Change-Id: Idfe046bdb211c8db9724e02c2f9dfb12d04d5c2a
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/114800
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Stephen White <senorblanco@chromium.org>
This is required for GLSL ES, which doesn't support Texture1D.
Bug: dawn:1301
Change-Id: Iba08d04a0bc23c278e65618550ea314ca0cbee1c
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/114363
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Stephen White <senorblanco@chromium.org>
Unsure whether this error is correct, there's an issue open on VVL to
figure that out. Suppress until that's resolved.
Bug: dawn:1627
Change-Id: I9663c1434f6eda48385ff66751432d22b2d8a3d3
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116544
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Auto-Submit: Kai Ninomiya <kainino@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
This Cl updates the IR tests which depend on a built IR to clear out the
program after the IR is generated. This will, hopefully, find accesses
into the old program AST in the sanitizer bots.
Bug: tint:1718
Change-Id: I8c83d8d6aa93d702dac6a1e8068ca8e0a00a7753
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116366
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Dan Sinclair <dsinclair@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
This CL removes the AST pointers from the IR nodes.
Bug: tint:1718
Change-Id: I818e1debc644b366c6e8378f448683b1b7d8fb00
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116365
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Dan Sinclair <dsinclair@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
This CL converts the case selectors over from ast CaseSelectors to IR
CaseSelectors. They work the same way in that a `nullptr` value signals
a `default` selector but they only store the resulting `constant::Value`
instead of the `ast::Expression`.
Bug: tint:1718
Change-Id: Ied62d661e03a7f8da4c1e1bdaccc04f21ab38111
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116364
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Dan Sinclair <dsinclair@chromium.org>
Reviewed-by: Ben Clayton <bclayton@google.com>
This CL cleans up a few comments for spelling and content. The EmitType
method is removed, types should not be emitted into the IR, they should
be cloned, so we should not need EmitType.
Bug: tint:1718
Change-Id: I7879e864e075837f756d47da422b2b129623d900
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116369
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Dan Sinclair <dsinclair@chromium.org>
The call to Queue::WriteTexture was passing a size that was bigger than
the array of data being used. This caused an ASAN failure when running
with the wire, because the serialization of the command would memcpy
past the end of the data.
Bug: dawn:1625
Change-Id: I2a00f2e32e3d8f8ff66d85575d1908480861f153
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116288
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
This patch enables all the dawn_end2end_tests that fail with old
Windows Intel driver but pass on the newer ones after the driver
of the bots being upgraded to 31.0.101.2111.
Bug: dawn:416, dawn:815, dawn:1070
Test: dawn_end2end_tests
Change-Id: I3022ff09a6c2be21f1c99fab0077743f84f83b34
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116133
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Austin Eng <enga@chromium.org>
This Cl adds the ability to clone a constant into a context provided.
This allows the IR to clone the constants out of the Program and into
the IR.
Bug: tint:1718
Change-Id: I78170cdc66b5824a1ab81000976a747b5bffee79
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/116363
Commit-Queue: Dan Sinclair <dsinclair@chromium.org>
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Accept any type in the intrinsics definition, and then manually
validate that there are no atomics in the type. Add manual E2E tests
for composite types.
Use the BuiltinPolyfill transform to implement it for all backends.
Update the uniformity analysis with special-case tags for the builtin.
Fixed: tint:1780
Change-Id: I95786dff4df70a0b16ed1c53b853b5d0ec6bc501
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/114862
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: James Price <jrprice@google.com>
Kokoro: James Price <jrprice@google.com>