This replaces the entry point IO component of the HLSL sanitizing
transform, and completes support for the new entry point IO syntax.
Struct emission in the HLSL writer is updated to use the correct
attributes depending on the pipeline stage usage.
Fixed: tint:511
Change-Id: I6a30ed2182ee19b2f25262a30a83685ffcb5ef25
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/46521
Auto-Submit: James Price <jrprice@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: James Price <jrprice@google.com>
This replaces the entry point IO component of the MSL sanitizing
transform, and completes support for the new entry point IO syntax.
Struct emission in the MSL writer is updated to use the correct
attributes depending on the pipeline stage usage.
Fixed: tint:510
Change-Id: I7ace33568c5a6e1bbf5c90f67b920579ba14be78
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/46520
Auto-Submit: James Price <jrprice@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: James Price <jrprice@google.com>
After the transform, an entry point's parameters will be aggregated
into a single struct, and its return type will either be a struct or
void. All structs in the module that have entry point IO decorations
will have exactly one pipeline stage usage.
This will be used to sanitize entry points for the MSL and HLSL
generators.
Change-Id: I7d2fcfba961d2e20326086964207747d573b6e05
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/46500
Auto-Submit: James Price <jrprice@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
* With this change, ProgramBuilder::WrapInStatement(expr), which
produced an "expr = expr" assignment expression, would fail validation
in some cases like for call expressions. Replaced this with a
declaration of a variable with type inferred from expr.
* Moved existing validation tests to resolver\assignment_validation.cc,
and added missing tests: AssignFromPointer_Fail.
* Fixed broken tests as a result of this change.
Bug: tint:642
Change-Id: I6a738b67edb0e116a46e936d652c2dcb4389281e
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/46161
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
There is still no way to spell this out in WGSL, but this adds support
for VariableDecls with an ast::Variable that has nullptr type. In this
case, the Resolver uses the type of the rhs (constructor expression),
which is stored in semantic::Variable.
Added tests for resolving inferred types from constructor, arithmetic,
and call expressions.
Bug: tint:672
Change-Id: I3dcfd18adecebc8b969373d2ac72c21891c21a87
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/46160
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
Use the semantic StorageClassUsage() to determine whether the structure is used only for storage buffer usage. If it is, don't emit a struct definition for it.
This fixes issues with attempting to generate runtime arrays - they're only legal for storage buffer usage. Storage buffers use ByteAddressBuffer instead of structured loads / stores.
Bug: tint:185
Fixed: tint:682
Change-Id: I5b58a133eee2fe036a84e028fa85b611e4895b1a
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/46382
Commit-Queue: Ben Clayton <bclayton@google.com>
Reviewed-by: James Price <jrprice@google.com>
[[align(n)]], [[size(n)]] are valid decorations that should not cause the writer to fail.
Instead of allow-listing these, just handle the cases it actually cares about.
Fixed: tint:686
Change-Id: I7c60d251fcaee424fe1c1a1f1f58123eea8c99aa
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/46380
Commit-Queue: Ben Clayton <bclayton@google.com>
Reviewed-by: James Price <jrprice@google.com>
In anticipation of adding support for type inference, no longer use
ast::Variable::type() everywhere, as it will eventually return nullptr
for type-inferred variables. Instead, the Resolver now stores the final
resolved type into the semantic::Variable, and nearly all code now makes
use of that.
ast::Variable::type() has been renamed to ast::Variable::declared_type()
to help make its usage clear, and to distinguish it from
semantic::Variable::Type().
Fixed tests that failed after this change because variables were missing
VariableDeclStatements, so there was no path to the variables during
resolving, and thus no semantic info generated for them.
Bug: tint:672
Change-Id: I0125e2f555839a4892248dc6739a72e9c7f51b1e
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/46100
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
* Formerly, we reported the same error message if we detected no default
clause or more than one. I made it so that we output a different error
message for each. This makes it more clear, and in the case of more than
one, the error source location points at the second default clause,
rather than at the switch statement.
* Add functions to ProgramBuilder to more easily define switch and case
statements.
* Fix broken tests as a result of this change.
Bug: tint:642
Change-Id: Iab4e610a563165862d9bc190772d32a4dd24ac45
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/45880
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
Recursively hoist struct members out to module-scope variables, and
redeclare the structs without entry point IO decorations. Generate a
function for storing entry point outputs to the corresponding
module-scope variables and replace return statements with calls to
this function.
Fixed: tint:509
Change-Id: I8977f384b3c7425f844e9346dbbde33b750ea920
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/45821
Auto-Submit: James Price <jrprice@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: James Price <jrprice@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Call validation was already implemented in Resolver. This change
completes it by deleting the relevant code in Validator, and moving and
updating the builtins validation test to use the Resolver.
Also added the "v-0004" error code for when detecting recursion, as was
done for the similar error in the Validator.
Bug: tint:642
Bug: tint:487
Change-Id: If7973bfd2d19681a0cbf48c6d427e17a3b927cde
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/45463
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
This was mostly already implemented in the Resolver, except for adding a
variable scope for blocks.
Moved tests and improved them to only add Source on the error node.
Bug: tint:642
Change-Id: I175dd22c873df5933133bc92276101aeab3021ed
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/45460
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
ast::IdentifierExpression nodes can appear outside of functions
(e.g. as initializers for module-scope variables), so we cannot assume
that current_block_ is not nullptr.
We already have several tests that do this, but for some reason the
nullptr dereference does not cause problems on our presubmits.
Change-Id: I612f3eb0d5711a0b1d0bb71663be7cca388b2b3c
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/45580
Auto-Submit: James Price <jrprice@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
This isn't in the WGSL spec, nor is it generated by readers.
This was only used inside the SPIR-V writer, but this remaining usage was removed in the parent change.
Change-Id: I1bbfde67dc760b761af010a7a144dccb52369148
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/45343
Commit-Queue: Ben Clayton <bclayton@google.com>
Reviewed-by: David Neto <dneto@google.com>
Requiring a temporary stack-allocated ast::Literal is an unpleasant requirement to generate a SPIR-V constant value.
GenerateU32Literal() was also creating an invalid AST - the type was U32, yet an an ast::SintLiteral was used.
Instead add Constant for holding a constant value, and use this as the map key.
This also removes the last remaining use of ast::NullLiteral, which will be removed in the next change.
Change-Id: Ia85732784075f153503dbef101ba95018eaa4bf5
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/45342
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: David Neto <dneto@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
