[PATCH] D155267: [DAGCombiner] Change foldAndOrOfSETCC() to optimize and/or patterns with floating points.

[Martin Storsjö via Phabricator via llvm-commits]

mstorsjo added a comment.

In D155267#4615903 <https://reviews.llvm.org/D155267#4615903>, @kmitropoulou wrote:

> Thank you for the heads up. I reverted the patch : 48fa79a503a7cf380f98b6335fbd349afae1bd86 <https://reviews.llvm.org/rG48fa79a503a7cf380f98b6335fbd349afae1bd86>
>
> Please let me know once you have the reproducer.

I also ran into this bug. I've reduced the issue down to this reproducer:

  float a, b;
  int c;
  void d() {
    for (;;)
      while (b < 0 && a < 0)
        c--;
  }

Compiled like this: `clang -target aarch64-linux-gnu -c -O2 -ffast-math analyse-reduced.c`


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D155267/new/

https://reviews.llvm.org/D155267