Guojin

Thanks, @seven-mile for good direction. Here what I'm thinking after looking at Flang's way a bit. Flang makes those target-related attributes as part of their LLVM lowering pass options, and...

> [@ghehg](https://github.com/ghehg) Your plan sounds nice ; ) Given a ClangIR module corresponds to a TU, it should not lose any expressiveness for a module-level attribute. > > As for...

As to VecShuffleOp, it's just just implementation Value *CreateShuffleVector(Value *V1, Value *V2, Value *Mask, const Twine &Name = "") in LLVM. Of course as you mention, LLVM has the following...

> @ghehg looks like you missed some bits for parser support? Updated docs also seem to be lacking. Can you take a look? sure, looking at it now.

> It looks like the direct reason is that, MLIR doesn't offer mlir::LLVM::VAArgOp in `mlir/include/mlir/Dialect/LLVMIR/LLVMIntrinsicOps.td` while it offers `mlir::LLVM::VaStartOp`, `mlir::LLVM::VaEndOp` and `mlir::LLVM::VaCopyOp`. So if we want to support va_arg in...

acknowledged, will take a look today and tomorrow.

yes, [PR885](https://github.com/llvm/clangir/pull/885) fixed the comdat issue here, and updated the exact test case. Alignment was fixed by https://github.com/llvm/clangir/pull/810 So, yes only remaining part is constant attr

> Since this issue has been stale for over a month could I take it up? Sure, feel free to take it up. Thanks for willing to work on it.

Thanks a lot! Andy, since you are adding array cookie support, I'd like share with you about a very related issue that I just created https://github.com/llvm/clangir/issues/1285 It's about array delete,...

@ljtodo Thank you for reporting the issue! To be clear, I don't think its Clang front end's problem. From crash stackframe, it shows its a problem generating ClangIR for DefaultArgExpr...