[PATCH] D48330: [GVN] Avoid casting a vector of size less than 8 bits to i8

[Eli Friedman via Phabricator via llvm-commits]

efriedma added a comment.

For the test, I mean something like the following:

  define <7 x i1> @f(<7 x i1>* %a) {
  entry:
    %vec = load <7 x i1>, <7 x i1>* %a
    %0 = getelementptr inbounds <7 x i1>, <7 x i1>* %a, i64 0, i64 0
    %val = load i1, i1* %0, align 2
    br i1 %val, label %cond.true, label %cond.false
  
  cond.true:
    ret <7 x i1> %vec
  
  cond.false:
    ret <7 x i1> <i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false>
  }


Repository:
  rL LLVM

https://reviews.llvm.org/D48330