+1 on zero's comments. Once they are implemented: LGTM!

There is no documentation on this operation. However, following the dialect [[documentation|https://mlir.llvm.org/docs/Dialects/LLVM/]]:

Unless explicitly stated otherwise, the semantics of the LLVM dialect operations must correspond to the semantics of LLVM IR instructions and any divergence is considered a bug.

It seems a bit wasteful to walk all the landingpads for each of the landingpads, but there might not be another way. It would be nice if one could compare the types with the result type of the personality function, but it does not exist for non-llvm functions.

LGTM! Thanks for pushing on this.

LGTM, modulo minor comments.

I only have one small comment apart from that this LGTM.

LGTM!

Dropped some comments.

address reviewer comments

After looking a bit closer over the tblgen part, I conclude that this looks good to me.

LGTM module minor comment.

I'm not familiar with this tblgen part, so I suggest to wait for answers of the downstream users. The rest is looking nice.

fix nullptr comparison

Thanks for splitting this into a separate revision. LGTM!

It might make sense to split the change to have the LLVMInterfaces changes in a separate commit.

address review comments

Make now potential null attr optional

LGTM module minor comment.

address review comments

In D144253#4134704, @gysit wrote:

Also should this be an NFC commit?

Address review comments

Thanks for all the comment addressing. LGTM!

rebased

address review comments

LGTM!

address review comments and move a builder to the cpp file.

LGTM!

fix test label name

LGTM!

LGTM

Added a helper vector to ensure that the graph elements have a deterministic order. Otherwise the cycle error message was non-deterministic, which could break tests.

address review comments

Thanks for pushing on this, LGTM!