Add the new classes:
* `ast::Let`
* `ast::Override`
* `ast::Parameter`
* `ast::Var`
Limit the fields to those that are only applicable for their type.
Note: The resolver and validator is a tangled mess for each of the
variable types. This CL tries to keep the functionality exactly the
same. I'll clean this up in another change.
Bug: tint:1582
Change-Id: Iee83324167ffd4d92ae3032b2134677629c90079
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/93780
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
Commit-Queue: Ben Clayton <bclayton@chromium.org>
Kokoro: Kokoro <noreply+kokoro@google.com>
Parameters don't have storage classes or access qualifiers. This was
just (ab)using the fact that a parameter uses the same AST type as a
'var'.
Also simplify the parameter disable validation logic.
Bug: tint:1582
Change-Id: Ic218078a410f991e7956e6cb23621a94a69b75a3
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/93603
Commit-Queue: Ben Clayton <bclayton@google.com>
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
Drop the vector typename from the initializer lists. These don't really
provide any significant help in understanding the arguments types, as
the list of element types can be easily inferred.
This is done to simplify the refactor ast::VariableList ->
ast::ParameterList.
Bug: tint:1580
Change-Id: Ibf8564ca9b2bafd2eaa2e4aa54c29be6bbe7a682
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/93600
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Commit-Queue: Ben Clayton <bclayton@chromium.org>
In this patch NumWorkgroupsFromUniform::Config changed to storage
std::optional<sem::BindingPoint>, and if it has no value,
NumWorkgroupsFromUniform will choose a free binding group, i.e.
binding 0 of the largest used group plus 1 is used if at least one
resource is bound, otherwise group 0 binding 0 is used. Tint CLI
is also changed to provide a --hlsl-root-constant-binding-point
option allowing user to specify the binding point for num_workgroups
uniform buffer.
Bug: tint:1566
Change-Id: I3b8c22a4276bab722d901f5b07d23a268786c417
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/91980
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Zhaoming Jiang <zhaoming.jiang@intel.com>
Also fixed implementation of this atomic in GLSL. It was emitting code
that would not compile because, as for HLSL, we must pass in the
variable directly to atomic funcs, not via an in/out arg to a function.
Bug: tint:1185
Change-Id: Id0e9f99d6368717511ef3a94473634c512e10cb8
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/91881
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
Instead of vector-of-variant.
This:
• Makes it impossible to produce a mix of scalar variant types, which
would make no sense.
• Reduces the size of a Constant, by removing the union-tag from each
element.
Also clean up terminology. Rename 'Constant::Scalar' to
'Constant::Element'. Scalars are well-defined in WGSL, and with the
introduction of abstract-numerics, this no longer makes sense.
Bug: tint:1504
Change-Id: I599aa97ad1ea798b7db8e512a5990ba75827faad
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/91304
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Ben Clayton <bclayton@chromium.org>
Commit-Queue: Ben Clayton <bclayton@google.com>
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>
* 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>
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>
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 is done via a new extension, which in the future could also be
used by shader authors as an escape hatch while we are still refining
the analysis.
The transform is run by the sanitizers for all of the non-WGSL
backends.
Bug: tint:880
Change-Id: Ibe90d7437d34c741a91eda65dff6d21d8469b9c7
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/88464
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Unsuffixed float literals are currently treated as f32,
but will shortly become AbstractFloat. To keep tests behaving
identically to how they are currently, change all float literals
to explicitly use the f32 '_f' suffix.
Bug: tint:1504
Change-Id: I2a00725ee1b34a6efbe15ac4ba438c00c4416dd8
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/89402
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
Commit-Queue: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Unsuffixed integer literals are currently treated as i32,
but will shortly become AbstractInteger. To keep tests behaving
identically to how they are currently, change all test literals
to using either 'i' or 'u' suffixes.
Bug: tint:1504
Change-Id: Ic373d18ce1c718a16b6905568aec89da3641d36b
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/88845
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
Commit-Queue: Ben Clayton <bclayton@google.com>
Merge SintLiteralExpression and UintLiteralExpression with
IntLiteralExpression. IntLiteralExpression has a new Suffix field which
indicates whether the literal is either a:
• 'i' suffixed integer literal
• 'u' suffixed integer literal
• no-suffix integer literal
Have the SPIR-V reader produce no-suffixed literals for i32 types, to
keep this change small. In future changes the SPIR-V reader will
produce 'i' suffixed types for these.
Have all consumers of IntLiteralExpression treat unsuffixed integers the
same as 'i'-suffixed literals. Unsuffixed will be treated as abstract in
future changes.
Removed SemHelper::TypeOf(const ast::LiteralExpression* lit).
Fixed: tint:1510
Bug: tint:1504
Change-Id: I443f41984e637ddd948182ee756af1010c5f8226
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/88842
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
Commit-Queue: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Instead of using an `if` node that has a list of `else` statements,
make each `if` statement have a single optional `else` statement,
which may itself be an `if` statement (or just a block statement).
This better matches the WGSL grammar (now that we have removed
`elseif`), and simplifies various pieces of code that handle these
statements.
Change-Id: Ie4272f1422224490ac598a03aa8b4dd00ba03010
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/87940
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Commit-Queue: James Price <jrprice@google.com>
These methods produce `let` declarations.
With creation-time expressions, we'll need to add `const` declarations.
Note that module-scope `let` declarations have been removed in the spec (for `const`). ProgramBuilder::GlobalConst() has not been renamed, although it still currently produces 'let' declarations.
Bug: tint:1504
Change-Id: I34f6d62236f0572163fc9c2d8fddfe4503817422
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/88305
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
Reviewed-by: Dan Sinclair <dsinclair@chromium.org>
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>
Remove transform::Glsl and replace with a Sanitize function. Cleans up
the code, reduces allocations and copies, and makes it consistent with
the other backends.
Also add a copy of src/tint/.clang-format to include/tint/ to keep files
in there formatted as per the tint standard.
Bug: tint:1495
Change-Id: I8a44ffecc6b3d244517bceb374ed93063e96f218
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/86205
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro-Run: Antonio Maiorano <amaiorano@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
In our backends, we std::move this unordered_set into the Results
struct, but because we'd get a const reference to a const object,
std::move would cast this to a const r-value, which binds to const ref,
and thus would end up copying instead of moving. This change just
ensures that we get a non-const ref to a non-const unordered_set where
we do the move.
Bug: tint:1495
Change-Id: I8d1cfe66b8343bae4146fc85fc86d72bac41a1a1
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/86202
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
Refactor the ExpandCompoundAssignment transform to handle these
statements, which delivers support for all of the non-WGSL backends.
Fixed: tint:1488
Change-Id: I96cdc31851c61f6d92d296447d0b0637907d5fe5
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/86004
Reviewed-by: Ben Clayton <bclayton@google.com>
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 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>
Added a new transform::RemoveContinueInSwitch that replaces continue
statements in switch cases with setting a bool variable, and checking if
the variable is set after the switch to continue.
Bug: tint:1080
Change-Id: I3c0a6c790e1bb612fac3f927a4bd5beb2d0d4ed1
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/84960
Reviewed-by: James Price <jrprice@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
