fix(clang-mlir): prevent llvm.mlir.undef on scalar brace initialization - #446
Open
Jimmy-Hu wants to merge 4 commits into
Open
fix(clang-mlir): prevent llvm.mlir.undef on scalar brace initialization#446Jimmy-Hu wants to merge 4 commits into
Jimmy-Hu wants to merge 4 commits into
Conversation
Intercepts empty `InitListExpr` for scalar types to explicitly emit `arith.constant 0` instead of allocating uninitialized memory (`llvm.mlir.undef`). This fixes a bug where modern C++ brace initialization (`{}`) for basic types would crash strict MLIR consumers like CIRCT during High-Level Synthesis.
…ization
When parsing modern C++ zero-initialization for scalars (e.g., `std::size_t index{};`), Clang does not always produce an empty `InitListExpr`. Instead, it often produces an `InitListExpr` with exactly one element: an `ImplicitValueInitExpr`.
Previously, this bypassed the `getNumInits() == 0` check in `MLIRScanner::VisitInitListExpr`, falling back to the default memory allocation path without initialization, which yielded an un-synthesizable `llvm.mlir.undef`.
This commit extends the scalar fast-path to also intercept single-element `InitListExpr`s containing an `ImplicitValueInitExpr`. It safely emits `arith.constant 0` (or `0.0`) as an RValue, preventing `undef` propagation and satisfying strict MLIR consumers like CIRCT/Calyx during High-Level Synthesis.
In `MLIRScanner::VisitCXXScalarValueInitExpr`, scalar zero-initialization only checked for `melem.isa<mlir::IntegerType>()`. When types like `std::size_t` or array indices are lowered to `mlir::IndexType`, they bypass this check and fall through to the floating-point fallback path, causing type mismatches or uninitialized values. Replaced `isa<mlir::IntegerType>()` with `melem.isIntOrIndex()` to ensure all index and integer scalar zero-initializations emit valid zero constants.
Similar to `CXXScalarValueInitExpr`, scalar zero-initialization represented by `ImplicitValueInitExpr` previously used a strict `dyn_cast<mlir::IntegerType>`. This caused `mlir::IndexType` (e.g., `std::size_t`) to bypass the constant zero emission, leading to synthesis failures. Replaced the strict `IntegerType` check with `Mty.isIntOrIndex()` to handle both integers and array indices gracefully, maintaining the original coding style.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Intercepts empty
InitListExprfor scalar types to explicitly emitarith.constant 0instead of allocating uninitialized memory (llvm.mlir.undef). This fixes a bug where modern C++ brace initialization ({}) for basic types would crash strict MLIR consumers like CIRCT during High-Level Synthesis.