FXC has trouble dealing with these.
This was originally added to handle returning arrays as structures.
HLSL supports typedefs, which is a much simpiler solution, and doesn't upset FXC.
Bug: tint:848
Bug: tint:904
Change-Id: Ie841c9c454461a885a35c41476fd4d05d3f34cbf
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/56774
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
Use the new transforms to try and simplify loops into for-loops.
Emit loops when the initialiser, condition and continuing are simple enough to do so.
Bug: tint:952
Change-Id: I5b3c225b245ffa72996abf6a70f52a9cd25b748e
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/56772
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: David Neto <dneto@google.com>
And migrate the WGSL writer over to the new APIs.
TextBuffer allows text to be written to different buffers.
Helps with the complexities around for loops.
Bug: tint:952
Change-Id: I094a726254f2a97b5830a8d6e07af7649c91e083
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/56771
Reviewed-by: David Neto <dneto@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
A Transform that attempts to convert WGSL `loop {}` statements into a for-loop statement.
For-loops cause less broken behavior with FXC than our current loop constructs.
Bug: tint:952
Change-Id: I0b7b1dbec9df4c004b0419d3feae53a195ec79bb
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/56767
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: David Neto <dneto@google.com>
UniqueIdentifier() will generate a program-global unique symbol.
MslGeneratorImplTest.AttemptTintPadSymbolCollision tests for collisions with the field names.
TextGeneratorTest.UniqueIdentifier_ConflictWithExisting tests for collisions between general symbols.
Fixed: tint:654
Change-Id: If2ba75d04ff0e2a9975e878596ac114d51adcd46
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/56580
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Sarah Mashayekhi <sarahmashay@google.com>
Reviewed-by: James Price <jrprice@google.com>
Commit-Queue: James Price <jrprice@google.com>
Auto-Submit: Ben Clayton <bclayton@google.com>
When indexing into vectors in a loop, FXC sometimes fails to determine
the max number of iterations when attempting to unroll the loop,
resulting in "error X3511: forced to unroll loop, but unrolling
failed.". We work around this by calling a function that sets the input
value at the input index into an inout vector. This seems to nudge FXC
enough for it to determine the number of loop iterations to unroll.
Bug: tint:534
Change-Id: I52cb209be29fcad8fbb91283c7be8c6e22e00656
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/56140
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Real problem was that we relied on SymbolTable::New() to be called with
the default arg of "tint_symbol", which isn't ergonomic when forwarding
to this function from others, like TextGenerator::UniqueIdentifier().
Instead, make New() take emptry string by default, and make it use
"tint_symbol" if input arg is empty string.
Also made it so that SymbolTable::Register() must not take an empty
string, so we now assert, rather than return an invalid symbol.
Change-Id: I386ece318c86d1d399f1dd1557a95fecac01f7ec
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/56461
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Antonio Maiorano <amaiorano@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
A function parameter of pointer type must be in one of the following storage classes:
- function
- private
- workgroup
A function parameter must one the following types:
- atomic-free plain type
- a pointer type
- a texture type
- a sampler type
Bug: tint:896 tint:894
Change-Id: Id8cec1bdc8e5be2c8c18a8420cec8f13f6aeddd0
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/55940
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
They are only valid on entry point parameters and return types, and
struct members. They must only be used on floating point scalar and
vector types. If the interpolation type is flat, the sampling type
must not be specified.
Bug: tint:746
Change-Id: Iab17816bc9947a74593a5937bdf513ac9ec664f1
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/56241
Auto-Submit: James Price <jrprice@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
As all loops in the parser require `synchronized_` to be true, don't set this to true if we've hit the maximum number of error diagnostics.
This lets the parser bubble up as soon as the limit was reached, instead of only aborting once we finally reach the top level loop.
ClusterFuzz has uncovered a number of cases where, in a loop, it can produce an error, then resynchronize, then error again on the next loop iteration. This produces more than the max limit of errors, and can stall the tests long enough to time out.
No unit tests for this, as it requires a really contrived input to trigger, and to exhaustively catch all the places we now call maybe_set_synchronized() would result in a large colleciton of horrible tests. Instead, let ClusterFuzz do the testing.
Fixed: chromium:1224031
Fixed: chromium:1224032
Fixed: chromium:1224042
Fixed: chromium:1224049
Fixed: chromium:1224050
Fixed: chromium:1224130
Fixed: chromium:1224131
Fixed: chromium:1224132
Fixed: chromium:1224144
Fixed: chromium:1224191
Change-Id: I63f160e28d13b85ee92e1b921239d7fb8ac22a08
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/56070
Auto-Submit: Ben Clayton <bclayton@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
Reviewed-by: James Price <jrprice@google.com>
Add `out` parameters to expression and type generators.
Use the new helper classes in TextGenerator.
Cleans up bad formatting.
Prepares the writer generating 'pre' statements, required for atomics.
If-else statements are generated slightly differently. This is done so that 'pre' statements for the else conditions are scoped correctly. This is identical to the HLSL writer.
Bug tint:892
Change-Id: I4c6e96c90673ba30898b3682bf3198497d63a2d4
Reviewed-on: https://dawn-review.googlesource.com/c/tint/+/56067
Kokoro: Kokoro <noreply+kokoro@google.com>
Reviewed-by: James Price <jrprice@google.com>
Commit-Queue: Ben Clayton <bclayton@google.com>
