[PATCH] D129629: [llvm-reduce] Fix crash when reducing integer vectors to 1

[Fraser Cormack via Phabricator via llvm-commits]

frasercrmck marked 2 inline comments as done.
frasercrmck added inline comments.


================
Comment at: llvm/test/tools/llvm-reduce/remove-operands-int.ll:1
+; Test that llvm-reduce can reduce floating point operands
+;
----------------
arsenm wrote:
> frasercrmck wrote:
> > note: the delta seems to be called "reduce operands" but the floating-point test was called "remove" so I followed suit. Maybe they should both be renamed `reduce-operands-*.ll`?
> Probably should be reduce, the operands are replaced, not removed
Renamed this and the existing tests.


================
Comment at: llvm/test/tools/llvm-reduce/remove-operands-int.ll:61
+  %add8 = add <2 x i32> %arg2, <i32 1, i32 1>
+  %add9 = add <2 x i32> %arg2, undef
+  ret void
----------------
arsenm wrote:
> Nonsplat constant is another case worth testing
Good idea. I've added that now


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D129629/new/

https://reviews.llvm.org/D129629