Emit a single _udivmodsi4 libcall instead of two separate _udivsi3 and
_umodsi3 libcalls if they have the same arguments. This optimization
was apparently broken if one of the node was replaced in place.
rdar://11714607
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@158900 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/test/CodeGen/ARM/divmod.ll b/test/CodeGen/ARM/divmod.ll
index 49c4103..7fbf8f4 100644
--- a/test/CodeGen/ARM/divmod.ll
+++ b/test/CodeGen/ARM/divmod.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s -mtriple=arm-apple-ios5.0 | FileCheck %s
+; RUN: llc < %s -mtriple=arm-apple-ios5.0 -mcpu=cortex-a8 | FileCheck %s
define void @foo(i32 %x, i32 %y, i32* nocapture %P) nounwind ssp {
entry:
@@ -56,3 +56,17 @@
declare i32 @llvm.objectsize.i32(i8*, i1) nounwind readnone
declare i8* @__memset_chk(i8*, i32, i32, i32) nounwind
+
+; rdar://11714607
+define i32 @howmany(i32 %x, i32 %y) nounwind {
+entry:
+; CHECK: howmany:
+; CHECK: bl ___udivmodsi4
+; CHECK-NOT: ___udivsi3
+ %rem = urem i32 %x, %y
+ %div = udiv i32 %x, %y
+ %not.cmp = icmp ne i32 %rem, 0
+ %add = zext i1 %not.cmp to i32
+ %cond = add i32 %add, %div
+ ret i32 %cond
+}