[PATCH] D154294: [Windows SEH]: Do not fold branches for MSVC TableSEH function

[Eli Friedman via Phabricator via llvm-commits]

efriedma added a comment.

> the problem might be related to the jumping edge issue @efriedma mentioned on https://reviews.llvm.org/D102817#3487499

Right, that's makes sense.

I'm concerned that if we ignore the bad CFG now, it's going to come back to bite us again.  In particular, for register allocation... although, maybe if we setExposesReturnsTwice(), it would mitigate that particular issue.  Or maybe we'll detect "loops" that aren't really loops, and hoist/sink incorrectly.  Or a number of other similar things.  Basically, I'm concerned we're going to end up spreading classifyEHPersonality() calls all over the backend for various issues.

Can you look into adding EH edges to the relevant blocks?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D154294/new/

https://reviews.llvm.org/D154294