[InstCombine] Fix wrong folding of constant comparison involving ahsr and negative quantities (PR20945).

Example:
define i1 @foo(i32 %a) {
  %shr = ashr i32 -9, %a
  %cmp = icmp ne i32 %shr, -5
  ret i1 %cmp
}

Before this fix, the instruction combiner wrongly thought that %shr
could have never been equal to -5. Therefore, %cmp was always folded to 'true'.
However, when %a is equal to 1, then %cmp evaluates to 'false'. Therefore,
in this example, it is not valid to fold %cmp to 'true'.
The problem was only affecting the case where the comparison was between
negative quantities where one of the quantities was obtained from arithmetic
shift of a negative constant.

This patch fixes the problem with the wrong folding (fixes PR20945).
With this patch, the 'icmp' from the example is now simplified to a
comparison between %a and 1. This still allows us to get rid of the arithmetic
shift (%shr).

llvm-svn: 217950

1 files changed, 15 insertions, 0 deletions

diff --git a/llvm/test/Transforms/InstCombine/icmp-shr.ll b/llvm/test/Transforms/InstCombine/icmp-shr.ll
index 36490e5d10a..41009d24c31 100644
--- a/llvm/test/Transforms/InstCombine/icmp-shr.ll
+++ b/llvm/test/Transforms/InstCombine/icmp-shr.ll
@@ -675,3 +675,18 @@ define i1 @nonexact_ashr_ne_noexactlog(i8 %a) {
  %cmp = icmp ne i8 %shr, -30
  ret i1 %cmp
 }
+
+; Don't try to fold the entire body of function @PR20945 into a
+; single `ret i1 true` statement.
+; If %B is equal to 1, then this function would return false.
+; As a consequence, the instruction combiner is not allowed to fold %cmp
+; to 'true'. Instead, it should replace %cmp with a simpler comparison
+; between %B and 1.
+
+; CHECK-LABEL: @PR20945(
+; CHECK: icmp ne i32 %B, 1
+define i1 @PR20945(i32 %B) {
+  %shr = ashr i32 -9, %B
+  %cmp = icmp ne i32 %shr, -5
+  ret i1 %cmp
+}