Add:
• sem::Type::is_abstract_or_scalar()
• sem::Type::ElementOf()
Use these to clean up some code in src/tint/sem/constant.cc.
Bug: tint:1504
Change-Id: I78e06b580a750c97ac654af4b0b364ddd3de6596
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90534
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
With abstract materialization, any ast::Expression may map to the new
sem::Materialize node. Because of this, we can't assume that an
ast::CallExpression maps to a sem::Call, as it might be wrapped by a
sem::Materialize.
Remove the mapping, and fix up all the code that was relying on this.
Fixes are done by either:
• Calling `UnwrapMaterialize()->As<sem::Call>()` on the semantic
expression. This is done when the logic may assume it's possible for
the expression to be a Materialize node.
• Using the explicit sem::Info::Get<sem::Call>() template argument to
cast the semantic type to sem::Call. This is done when the logic
either knows it is impossible for the expression to be a Materialize.
The backends have been stubbed, as we'll want to emit the constant value
for these nodes. It's likely that we'll just use the FoldConstants
transform to strip all Materialize nodes from the tree. For now, be
defensive.
Bug: tint:1504
Change-Id: If9231b300fc30c7fe886c17a804ead8ee2988285
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90533
Commit-Queue: Ben Clayton <bclayton@chromium.org>
Reviewed-by: David Neto <dneto@google.com>
• Split type_ctors_ into array_ctors_ and struct_ctors_. These are the
only two type constructors that we need to cache in the Resovler. All
other types are cached in the IntrinsicTable. By making these two
separate fields, there's no way we can accidentally put other types in
there. This is slightly more optimized too, as each map is smaller,
and the keys hold less data.
• Drop the 'const' on vectors that are std::move()'d. A const object
cannot be moved, and this results in a silent copy. Fix the logic of
BuiltinCall() and FunctionCall(), as these were happily using the
vectors that had been moved.
• Extract the messy Texture-Sampling collection logic out into two
functions.
The rest of the changes are tweaks required to handle abstract numerics
and materialization, which will be put up for review shortly. Landing
the no-op refactoring now reduces review noise later.
Bug: tint:1504
Change-Id: Iffc8039360d6138c3ac9b456be6ca7b8451ede9f
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90532
Commit-Queue: Ben Clayton <bclayton@chromium.org>
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
A new semantic expression node that wraps another semantic node. Used to
indicate the point at which compile-time, abstract numeric typed
expressions are implicitly converted to a concrete type.
Bug: tint:1504
Change-Id: I52e256bbbdeaa9d9eff4cb93b6f937dd00bdc5cb
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90531
Reviewed-by: David Neto <dneto@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Ben Clayton <bclayton@chromium.org>
These aren't used by anything (yet). Baby steps.
Bug: tint:1504
Change-Id: Icf0261ec9c6802f004d9f1bc4780a6376ebb8dfb
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90530
Reviewed-by: David Neto <dneto@google.com>
Commit-Queue: Ben Clayton <bclayton@chromium.org>
TypeMatcher::String() was not respecting the [[display]] decoration of
the matcher's sub-types. By calling TypeMatcher::String() on the sub-types,
we can display the custom type names in diagnostics.
Bug: tint:1504
Change-Id: I0856fee31231f9c048d2e3028d25c4d261fbb008
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90529
Reviewed-by: David Neto <dneto@google.com>
Commit-Queue: Ben Clayton <bclayton@chromium.org>
When we initialize the dst texture before T2T copy, we clear a
buffer with zeros and do a B2T copy in order to initialize the
texture. The T2T copy may only copy a subrect of the texture's
subresource, so does the initialization via B2T copy. As a result,
the buffer can store compact zeros for that exact subrect and then
do an exact B2T copy to initialize that area.
However, the current B2T copy uses incorrect data layout for the
buffer, making some part of the dst texture uninitialized, and
even leading to D3D12 runtime validation errors sometimes if it
goes beyond the boundary of staging buffer pool (4MB).
This change fixes this bug via using the exact height of copy area
to replace dataLayout.rowsPerImage (unnecessary padding rows included)
for the height of each depth image for 3D textures.
Bug: dawn:1288
Change-Id: I303e0d363039a6a87e352a8445898031e673cf4e
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90780
Commit-Queue: Yunchao He <yunchao.he@intel.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
We already have third_party/webgpu-cts. This came in during the tint ->
Dawn merge, and is not necessary (and confusing).
Change-Id: I1a4ce4463e55624b5f720573c54f45d1d1dc57e7
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90900
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
According to the spec, when call draw or drawIndexed, vertex step mode
vertex buffer never OOB if (vertexCount + firstVertex) = 0, and instance
step mode vertex buffer never OOB if (instanceCount + firstInstance) = 0.
Modify the validation implementation to be aligned with the spec, and
add corresponding unit tests.
This patch also add unit test case for (strideCount - 1) * arrayStride +
lastStride <= bound buffer size < strideCount * arrayStride.
Bug: dawn:1287
Change-Id: If444e400f5ac24f86ca12ff59fb886d8ef70e8c7
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90584
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Commit-Queue: Zhaoming Jiang <zhaoming.jiang@intel.com>
* Extract ast::Enable::ExtensionKind to ast::Extension.
* Move the parsing out of ast::Enable and next to ast/extension.h
* Change the ast::Enable constructor to take the Extension, instead of
a std::string. It's the WGSL parser's responsibility to parse, not the
AST nodes.
* Add ProgramBuilder::Enable() helper.
* Keep ast::Module simple - keep track of the declared AST Enable nodes,
don't do any deduplicating of the enabled extensions.
* Add the de-duplicated ast::Extensions to the sem::Module.
* Remove the kInternalExtensionForTesting enum value - we have kF16
now, which can be used instead for testing.
* Rename kNoExtension to kNone.
Bug: tint:1472
Change-Id: I9af635e95d36991ea468e6e0bf6798bb50937edc
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90523
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
Commit-Queue: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Support overload resolution of abstract-numeric argument types,
allowing them to implicitly convert down to concrete parameter
types (and in the near future, abstract parameter types).
Major kudos to cwallez for the suggested algorithm which is a
minor adaption of what we had already.
Bug: tint:1504
Change-Id: I85fa8e70ab0b6aa643caec4c51433f15784af55f
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90522
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: David Neto <dneto@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
Add support for a [[precedence(N)]] decoration on intrinsic table type
declarations. This will be used to ensure the type with the lowest
conversion rank is matched when a matcher could match multiple types
for a given abstract numeric argument type.
Bug: tint:1504
Change-Id: I96475b000c0917bbfa4e2873b1731ce048b96a7d
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90664
Reviewed-by: David Neto <dneto@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Relying on the config inside //build breaks in projects that use their
own version of //build instead of Chromium's.
Bug: dawn:1405
Change-Id: I18bbf5c6ddce18e6900f5f4b937d91a152bb2b32
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90524
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Austin Eng <enga@chromium.org>
Adds ForTesting APIs to the instance to track the number of devices.
Bug: dawn:1164
Change-Id: Ib743afb1e86ef16740d49613f43f9e2f009232bc
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90583
Reviewed-by: Loko Kung <lokokung@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Austin Eng <enga@chromium.org>
The concept of 'closing' an open type or number made sense when these
were immutable once the first type/number had been matched during
overload resolution.
In order to support abstract numerics, these template parameters need to
be constrained as the arguments are evaluated, so there's no longer a
binary open / closed state.
Give up on this concept, and rename everything to 'template type' and
'template number'. This is likely easier for people to understand
anyway.
Also fix a small typo in the ICE message printed when there's an
ambiguous overload resolution (should never happen with the current
entries in the table).
Bug: tint:1504
Change-Id: I2bf043c71e5afa757259968eae4af830c50f38e0
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90662
Commit-Queue: Ben Clayton <bclayton@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
Reviewed-by: David Neto <dneto@google.com>
Migrate from a hand-rolled tagged-union of [i32, u32, f32, f16, bool]
types. Instead use a std::variant of [AInt, AFloat, bool]. The Constant
holds the actual type, so no information is lost with the reduced types.
Note: Currently integer constants are still limited to 32-bits in size.
This is enforced by the frontend.
Bug: tint:1504
Change-Id: I316957787649c454fffb532334159d726cd1fb2d
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90643
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: James Price <jrprice@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
Remove the ProgramBuilder from ClosedState and use a pointer for the
'overload' field instead of a reference. Let's the Candidate be
copy-assignable, which in turn, allows the Candidates vector to be
sorted directly, instead of jumping through hoops to use moves.
Replace random mix of 'int', 'uint8_t' with 'size_t' (externally to the
constant table data). Reduces fragile weak binding between distant code.
Swap the overload scoring order (high-best -> low-best). Remove the
'matched' field - we can now just check whether the 'score' is 0.
Further simplifies sorting.
Change-Id: I4a4b7934be337306202647d096c546eab5c8498f
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90641
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: David Neto <dneto@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
If the CallSiteRequiredToBeUniform tag is present, add the edge from
RequiredToBeUniform to a new diagnostic node for the function call,
instead of to the control flow coming out of the function call. Doing
the latter causes a false positive when a function both requires
uniform control flow and causes non-uniform control flow.
Bug: tint:880
Change-Id: Icade8f76302e8c21529502f5f945f1981acfc45a
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90582
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Auto-Submit: James Price <jrprice@google.com>
This reverts commit 63cea3f8c0.
...except it skips adding the the constructor/destructor/copy
definitions for dawn::native::Format. Adding them triggers "constexpr
variable cannot have non-literal type" (or "declaration requires a
global constructor" if it's not marked constexpr), unless they're
explicitly marked as =default in the header - which just triggers a
different chromium-style warning, so there's no point.
A better solution to the chromium-style warning on Format may be to just
make the class physically smaller:
https://www.chromium.org/developers/coding-style/chromium-style-checker-errors/#constructordestructor-errors
Bug: dawn:1405
Change-Id: Ied6e9d0abff6bf1330131a40c6583bab18888b67
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90303
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
Eliminates a static initializer in kUnusedFormat (in Texture.cpp) by
making dawn::native::Format constexpr-constructible.
kUnusedFormat doesn't actually have to be constexpr to fix this, but it
adds explicit enforcement that it's constexpr-constructible.
Includes some extra initializers as a workaround for a bug in the old
version of MSVC (14.26) that's on Kokoro. amaiorano figured out how to
reproduce it locally, with a local install of VS2019 and this magic
CMake incantation:
$ cmake -G "Visual Studio 16 2019" -T v142,version=14.26 ..
Bug: dawn:1405
Change-Id: Ic94324fc624fd720671dec35dcc5ea8ad77ee46d
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/89863
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
For all types except for arrays and structures, which are explicitly
typed and have trivial overloads.
This will simplify maintenance of type functions, unifies diagnostic
messages and will greatly simplify the [AbstractInt -> i32|u32]
[AbstractFloat -> f32|f16] logic.
Bug: tint:1504
Change-Id: I2b17ed530d1cece22adcbfc6de0bec4fbda4c7bd
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90248
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: David Neto <dneto@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
This will allow the bots to specify a particular backend.
Change-Id: I9486019c3aef3f8aafb79e30dfc62d23ae9eefac
Bug: dawn:1420
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90320
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Stephen White <senorblanco@chromium.org>
When importing Dawn into Tint:
error: [chromium-style] auto variable type must not deduce to a raw pointer type.
Change-Id: I6ff4451a5519c38b18eb8d96f6bc82b8090077f8
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90500
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: James Price <jrprice@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
The bitwise-and and bitwise-or binary operators on booleans result in an integer.
Explicitly cast this back to a boolean.
Fixed: tint:1540
Fixed: tint:1541
Change-Id: I395176f291e6080c88b8cff18e14ed6cd1234074
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90501
Commit-Queue: Ben Clayton <bclayton@google.com>
Reviewed-by: James Price <jrprice@google.com>
Fixes a stack overflow when traversing particularly deep graphs.
This changes the iteration order which changes some of the diagnostic
notes, but the diagnostics are still correct and deterministic so it
does not matter.
Fixed: chromium:47418
Change-Id: I2def633b6d96d1525027f7929b0fa0a6fba0efeb
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90140
Auto-Submit: James Price <jrprice@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Add additional nodes to capture places where control flow is changed
(if, switch, for, short-circuiting op, function call), and use these
to show the actual point at which control flow became non-uniform as a
result of a non-uniform value.
Do this recursively, to capture cases where control flow becomes
non-uniform after a function call statement.
Bug: tint:880
Change-Id: Ied92d690f98a5c11a1892eef500a50d0f123943d
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/89862
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
This patch adds a workaround on a DXC error in the translation of DP4a
functions that the third parameter "acc" of both dot4add_i8packed() and
dot4add_u8packed() can only be a variable, or a compile error will be
generated.
Bug: tint:1497
Test: tint_unittests
Change-Id: I263d27fb18bd354e0c9110f60cbc98860cf7afe5
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90027
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
If the WGSL provided over 255 arguments, then we'd overflow the uint8_t,
and a bad candidate could be matched.
We have validation in place to ensure that user functions are not
declared with more than 255 parameters, but no validation for number of
arguments provided.
Fixed: chromium:1323605
Change-Id: I048709e8b02bdd95d5082128a1186080c79b2fc7
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90249
Commit-Queue: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
GetOrCreateAttachmentState access mCaches in Device even if device is
already destroyed. Gate the function by checking if encoder is
destroyed.
Bug: chromium:1323310
Change-Id: I8151c5ec696e4da28b8296d9142f3120379782ef
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/89860
Reviewed-by: Austin Eng <enga@chromium.org>
Reviewed-by: Loko Kung <lokokung@google.com>
Commit-Queue: Shrek Shao <shrekshao@google.com>
Remove the space between operator"" and the name of the function, as per
the C++ spec. cppreference says as much, but its example code inserts a
space, confusingly.
Fixed: tint:1535
Change-Id: I46ac8fab74290db1a3c0085b56bc66a4c1d318f4
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90280
Reviewed-by: James Price <jrprice@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
The shader source can be very large and this is causing excessively
large error messages that get truncated in some applications that use
Dawn native. We don't need to see the whole shader in the error
message.
Change-Id: I15ad7fa7814d19875a7c28b39e5fa9a24f265b98
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90161
Reviewed-by: Austin Eng <enga@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
These notes often contain additional useful information that is not
currently being surfaced to the developer console.
Change-Id: I342bfc3147593755d6addf5466f27b083f954826
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90160
Reviewed-by: Austin Eng <enga@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
Fix the namespace - this should have been tint::resolver.
Use a single u32 bitset for overload flags instead of multiple fields.
Bug: tint:1504
Change-Id: I633b21ce14e20fc9aeeed5221886c1d22e327bdc
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90241
Commit-Queue: Ben Clayton <bclayton@chromium.org>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Do the partitioning of unary and binary operators in the intrinsic table
generators, instead of searching all operators at runtime.
Will allow code to be simplified.
Bug: tint:1504
Change-Id: I67246b954e530e0542b1b67c99fb34a756cf532a
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90240
Commit-Queue: Ben Clayton <bclayton@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
This patch:
1. Add the `f16` WGSL extension.
2. Add `f16` as keyword, and remove it from reserved word list.
3. Add ast::f16 and sem::f16, and implement validation that using `f16`
type must be with `f16` extension enabled.
4. Add `Number<NumberKindF16>` for f16 literal and constant, and add
`ast::FloatLiteralExpression::Suffix::kH`.
5. Add placeholder in all writer which report error when try to emit f16
type.
Bugs: tint:1473, tint:1502
Change-Id: Ifb363beeb2699ed7cac57e07227d1b2cfa8050b4
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/89922
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Zhaoming Jiang <zhaoming.jiang@intel.com>
This patch make `sem::Vector::Size()` and `Align()` more general to
prepare for subtypes of 16bits, e.g. f16. Previously these methods
assert that the subtypes are of 4bytes.
Change-Id: Ic939f177e87faabdf17d1d404cee82a91072a179
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90163
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Zhaoming Jiang <zhaoming.jiang@intel.com>
The WebGPU spec says that a validation error is produced if there is a
color target in the pipeline descriptor that has writeMask!=0 while
there are no shader outputs for this target. In Dawn format==undefined
is used as a tag that the color target is not present, so writeMask is
allowed to be 0 if format==undefined. (the validation in the case where
format!=undefined was already present).
Fixed: dawn:1376
Change-Id: I7737b6557223e0fc31740fd4ec2cbfaa54b77a71
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/90040
Auto-Submit: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Loko Kung <lokokung@google.com>
Reviewed-by: Loko Kung <lokokung@google.com>
Ben Clayton
Austin Eng
Brandon Jones
Yunchao He
Antonio Maiorano
Zhaoming Jiang
Corentin Wallez
James Price
Kai Ninomiya
Stephen White
Jiawei Shao
Shrek Shao