[PATCH] D85533: [InstCombine] Optimize select(freeze(icmp eq/ne x, y), x, y)

[Eli Friedman via Phabricator via llvm-commits]

efriedma added inline comments.


================
Comment at: llvm/test/Transforms/InstCombine/select.ll:2588
+  call void @use_i1_i32(i1 %c.fr, i32 %v)
+  ret void
+}
----------------
If I'm understanding correctly, the following transform would be legal:

```
define void @select_freeze_icmp_multuses(i32 %x, i32 %y) {
  %x.fr = freeze i32 %x
  %y.fr = freeze i32 %y
  %c = icmp ne i32 %x.fr, %y.fr
  call void @use_i1_i32(i1 %c, i32 %x.fr)
  ret void
}
```

Is this something we should be looking into?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D85533/new/

https://reviews.llvm.org/D85533