WGSL and SPIR-V modules can contain multiple entrypoints, for different
shader stages, that the pipelines can choose from. This is the first CL
in a stack that will change Dawn internals to not rely on ShaderModules
having a single entrypoint.
EntryPointMetadata is introduced that will contain all reflection data
for an entrypoint of a shader module. To ease review this CL doesn't
introduce any functional changes and doesn't expose the
EntryPointMetadata at the ShaderModuleBase interface. Instead
ShaderModuleBase contains a single metadata object for its single entry
point, and layout-related queries and proxied to the EntryPointMetadata
object.
Finally some small renames and formatting changes are done.
Bug: dawn:216
Change-Id: I0f4d12a5075ba14c5e8fd666be4073d34288f6f9
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/27240
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
This CL moves the global variables checks to a function
Adds tests and checks for validation rule v-0022:
Global variables must have a storage class.
Bug: tint: 6
Change-Id: I2f2cd7df6e849bfd1ddfbca35568c6fc3345efa6
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/27283
Commit-Queue: Sarah Mashayekhi <sarahmashay@google.com>
Reviewed-by: dan sinclair <dsinclair@chromium.org>
The `const_expr` can recurse into itself if there are type declarations
inside the const_expr (e.g. vec2<f32>(f32(1.0), f32(2.0))). Currently
there is no limit on the amount of recursion which can be triggered.
This CL sets a limit of 128 nested type declarations at which point an
error will be emitted.
Bug: chromium:1112144
Change-Id: Ifae45034dc9de35aed78ba8eddf584a46c7a55ce
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/27340
Reviewed-by: dan sinclair <dsinclair@chromium.org>
Reviewed-by: Ryan Harrison <rharrison@chromium.org>
Commit-Queue: dan sinclair <dsinclair@chromium.org>
This patch cleans up some issues in the end2end tests that will cause
test failures when we enable buffer lazy initialization by default.
This patch also skips a test that always fails with Vulkan validation
layer.
BUG=dawn:414
TEST=dawn_end2end_tests
Change-Id: I40f643615b3fec4e52c90d576285534a99950915
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/26960
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
This CL validates the following rule. ie. As functions must be defined before use (v-0005), self-recursion is only case that has to be invalidated.
v-0004: Recursions are not allowed.
Bug: tint: 6
Change-Id: Icfb040907c5ea0abb6359dade74dcfc30a0db7d9
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/26980
Commit-Queue: Sarah Mashayekhi <sarahmashay@google.com>
Reviewed-by: dan sinclair <dsinclair@chromium.org>
In the test DepthStencilCopyTests/FromStencilAspect we are writing into
gl_FragDepth but we disable the depth writing when creating the render
pipeline, which causes Intel Mesa driver hangs when the test is running
on it. In this patch we try to add workaround to this issue by enabling
the depth writing.
BUG=dawn:439
TEST=dawn_end2end_tests
Change-Id: I618246097a26ee62c830f0ac8dc1aa0ab4d224a9
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/27000
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
When using a dynamic uploader we didn't align the offset
that the allocated memory might have already had.
That fixes WriteTexture, WriteBuffer, ClearTexture and
on D3D12 ClearBuffer.
Bug: dawn:512
Change-Id: I64c7511ad6b0d3d6a28a494e1324a10ad4d38091
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/27020
Commit-Queue: Tomek Ponitka <tommek@google.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Jiawei Shao <jiawei.shao@intel.com>
- Implement QuerySet on Vulkan backend.
- Enable end2end tests on Vulkan. The timestamp tests will be enabled
in following CL which implement timestamp query.
Bug: dawn:434
Change-Id: I7ee04380c5f6b5af561cc23e28637dcae70bc7b9
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/26360
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Hao Li <hao.x.li@intel.com>
This CL adds the beginning of intrinsic emission for the HLSL backend.
The `outer_product`, `is_normal` and `select` intrinsics are currently
missing.
Bug: tint:7
Change-Id: Ice7a2b285eeb52041e3accd9751e127d6c5a0177
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/26927
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Reviewed-by: David Neto <dneto@google.com>
You can only typedef builtin types in HLSL. This Cl updates the struct
emission to emit named structs instead of typedef'd structs.
Bug: tint:7
Change-Id: I835b7f4d23bc225c730ef3f39c4572c043a58156
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/26921
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Reviewed-by: David Neto <dneto@google.com>
This CL adds emission of uniform storage class variables to the HLSL
backend. If the variable is a base type (float, int, etc) it is emitted
as a `cbuffer`. If the variable is a struct it will emit as a
`ConstantBuffer`.
Bug: tint:7
Change-Id: I9932d30df24c023c58d3a2ba4167bcb7ccb85dae
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/26920
Commit-Queue: dan sinclair <dsinclair@chromium.org>
Reviewed-by: David Neto <dneto@google.com>
Prior versions of GN had a bug (gn:22) where visibility rules
for configs weren't enforced properly.
This CL tweaks the visibility settings of some configs to conform
to the latest version.
Change-Id: If0e9f06667d3d89bcd0bbfc938e159e590e11e27
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/26929
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Previously glFrontFace was called only if some cull mode was set. This
was incorrect because the front face also influences whether a triangle
uses stencilFront or stencilBack.
Because OpenGL default to GL_CCW (which with the Y-flip is the inverse
of wgpu::FrontFace::CCW that's default in the descriptor), if
stencilFront != stencilBack and cull mode is none, then the incorrect
stencil face descriptor was used.
Also adds a regression test for this issue.
Bug: dawn:508
Change-Id: I00d93bda6d4f030cf9db472a9f2b0deefc72707f
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/26880
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
When emitting a global constant we'd early exit the emission code before
emitting the debug statement for the constants name. This Cl adds the
needed code to emit names for global constants.
Change-Id: I1a7a3660b4ff31879393a6a776a9f00e895de216
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/26923
Commit-Queue: Ryan Harrison <rharrison@chromium.org>
Reviewed-by: Ryan Harrison <rharrison@chromium.org>
Previously, lazy clearing always added DEPTH_STENCIL or RENDER_TARGET to
textures because we cleared using ClearDepthStencilView or
ClearRenderTargetView. Now, we're able to clear using copies.
This also allows textures to actually use the small resource heap
placement optimization. Doing so generates debug layer warnings when the
small alignment is first tried but rejected. This CL silences those
warnings.
Bug: dawn:145
Change-Id: Id385846536b337cddcfdadc5739561c7adc30c8c
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/26840
Commit-Queue: Austin Eng <enga@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
