Integrates Tint repo into Dawn
KIs:
- Building docs for Tint is turned off, because it fails due to lack
of annotations in Dawn source files.
- Dawn CQ needs to be updated to run Tint specific tests
- Significant post-merge cleanup needed
R=bclayton,cwallez
BUG=dawn:1339
Change-Id: I6c9714a0030934edd6c51f3cac4684dcd59d1ea3
With this change, the backend sanitizers always run the
MultiplanarExternalTexture transform. If the new option is enabled, it
auto-generates bindings for this transform.
This change also enables this auto-generation for the Tint commandline
application, as well as for the fuzzers.
Bug: chromium:1310623
Change-Id: I3c661c4753dc67c0212051d09024cbeda3939f8c
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/85542
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
This adds a test, and a toggle disable_resource_suballocation.
This enables testing the behavior discovered in the bug without
creating enormous resources.
Bug: chromium:1313172
Change-Id: I779aad50c051e5022a9c85ebfbf33c18173a748f
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/85861
Reviewed-by: Loko Kung <lokokung@google.com>
Reviewed-by: Brandon Jones <bajones@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
Heaps were destroyed immediately instead of deferring destruction
until after all work using the buffer was complete. This is only
a problem on D3D12. Vulkan allocations already have deferred
deletion, and Metal allocations are managed by the driver.
Bug: chromium:1313172
Change-Id: I0ef43709949c9e86c40e766f7f2029b14c8a2e97
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/85840
Reviewed-by: Brandon Jones <bajones@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
This reverts commit 0126761de8.
Reason for revert:
Causing Dawn->Chromium roll to fail
BUG=dawn:1346
Original change's description:
> Enable Queue, Device labels to be set.
>
> Queue labels can be set by the defaultQueue.label member of the device
> descriptor or the setQueue method.
>
> Device labels can be set label member of the device
> descriptor or the setQueue method.
>
> D3D12 and VK backend label support included.
>
> Change-Id: Id12dd6e1fc8f1519c55e4efb35e1ead67c085e46
> Bug: dawn:1323
> Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/85540
> Reviewed-by: Kai Ninomiya <kainino@chromium.org>
> Commit-Queue: Brandon Jones <bajones@chromium.org>
# Not skipping CQ checks because original CL landed > 1 day ago.
Bug: dawn:1323
Change-Id: I62e4b508d2c55fd89f2f4c5cbe5d04d22681aeef
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/85700
Reviewed-by: Brandon Jones <bajones@chromium.org>
Auto-Submit: Ryan Harrison <rharrison@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
Textures as function parameters should not have the "uniform"
qualifier. Fixed by handling StorageClass::kUniformConstant the
same as StorageClass::kUniform, and removing the unconditional
"uniform" qualifier output. (Global texture variables have
StorageClass::kUniformConstant set, while function parameters don't.)
Change-Id: I9d380550ac4554917527ff330171a76a90a290e8
Bug: tint:1492
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/85820
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Commit-Queue: Stephen White <senorblanco@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
This removes a little bit of noise from the reflection of shader
metadata in Dawn. Tests are added to make sure that Tint does the
correct validation (it does).
Bug: None
Change-Id: I334e7c23b723cf5b5985c9914cc9f8d79a7c0568
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/85502
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Loko Kung <lokokung@google.com>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
A list of errors, `infringingLimits`, is added to EntryPointMetadata.
During shader reflection, instead of directly bubbling limit errors up,
they are stored in this list and check only later during pipeline
creation.
Several ShaderModule tests are reworked to create a pipeline to check
for the validation of these limits. For the IO variable limits the tests
needed to be reworked to check for strings in the error messages because
since IO structs have to match between VS and FS, if one failed the
other failed too. (so it's no possible to target the validation of one
of these in particular)
Bug: dawn:986
Change-Id: I689e16454488d4a3c746ece53828555ed72ed561
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/85501
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
These are implemented by wrapping the integer types in transparent
ClampedInteger<> and EnforceRangeInteger<> structures.
Some parts of the core needed to be updated after this, either to
disambiguate conversions, or because of bugs (u32 vs u64).
To make the CTS tests checking for this pass, the errors returned when
conversion FromJS failed needed to be updated to TypeError and not just
the generic Napi::Error.
Bug: dawn:1123
Change-Id: Ife1d0baa7687e43d735a1814ec41883c49ae74a6
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/85640
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Queue labels can be set by the defaultQueue.label member of the device
descriptor or the setQueue method.
Device labels can be set label member of the device
descriptor or the setQueue method.
D3D12 and VK backend label support included.
Change-Id: Id12dd6e1fc8f1519c55e4efb35e1ead67c085e46
Bug: dawn:1323
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/85540
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Brandon Jones <bajones@chromium.org>
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>
This remove some interfaces like OffscreenCanvas from the global scope
because dawn.node is not capable to usefully create them. The CTS
uses the absence of these interfaces to skip tests when needed.
Bug: dawn:1123
Change-Id: I6d57600ba6b41be58c541d1f8091e7e88781f04f
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/85364
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Previously a new promise was created and new promises were never
resolved on creation, only on loss/destruction of the device. This made
the following code wait forever:
device.destroy();
await device.lost();
Bug: dawn:1123
Change-Id: I1e31cf9ccd466672eed4cad464c38cb9f8b3d724
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/85362
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This is done by moving the set up of these attributes to the place where
the wrapper objects are created, by doing:
jsObject.DefineProperty("foo", impl->getOnceFoo());
Three alternatives that weren't chosen are:
- Caching a weak reference to the member's Napi::Value on the wrapper
struct, and recreate it only as needed. This is good because it
doesn't risk using the value after it is GCed, but it can result in
multiple calls to the getters, which could be unexpected (for example
for GPUDevice.lost in a follow-up CL).
- Caching a persistent reference to the member's Napi::Value on the
wrapper struct. This calls the getter once and doesn't risk using the
value after it is GCed. However if Javascript does something like
`myGPUDevice.limits.device = myGPUDevice`, a cycle is created that
the GC doesn't have visibility into, and that can't be collected.
(the origin of the edge of the reference graph that persistent
references make is unknown to the GC).
- Caching the member on a hidden variable of the JS object. I didn't
find a way to do this. The closest would have been to do
jsObject[Symbol(...)] = cachedValue but even symbols can be retrieved
with Object.getOwnPropertySymbols.
Bug: dawn:1123
Fixed: dawn:1144
Change-Id: I1bc82dd9d10be95bf2bdca73bdfb843bc556d2df
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/85361
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
This patch sets depthWriteEnabled to its default value (false) in the
helper function DawnTest::ExpectAttachmentDepthStencilTestData() as all
the tests can pass with depthWriteEnabled == false on the Linux Intel
bots now.
Note that previously using depthWriteEnabled == false and writing into
FragDepth will cause Linux Intel Mesa driver crash on Mesa 19.0.2.
BUG=dawn:821
TEST=dawn_end2end_tests
Change-Id: I22cc0dcdb8521fd8eae436d99a7c06167af89b09
Change-Id: Id0dd1c31099c6aafad175bde038ba9662b02a160
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/85322
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
This transform converts compound assignment statements into regular
assignments, hoisting LHS expressions and converting for-loops and
else-if statements if necessary.
The vector-component case needs particular care, as we cannot take the
address of a vector component. We need to capture a pointer to the
whole vector and also the component index expression:
// Before
vector_array[foo()][bar()] *= 2.0;
// After:
let _vec = &vector_array[foo()];
let _idx = bar();
(*_vec)[_idx] = (*_vec)[_idx] * 2.0;
Bug: tint:1325
Change-Id: I8b9b31fc9ac4b3697f954100ceb4be24d063bca6
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/85282
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Reuse the logic for resolving binary operator result types that was
implemented for binary expressions. This validates that the LHS and
RHS are compatible for the target operator. We then try to match the
resolved result type against the LHS store type.
Bug: tint:1325
Change-Id: If80a883079bb71fa6c4eb5545654279fefffacb4
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/74362
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
