It's not necessary to do rounding for alloca operations when the requested
alignment is equal to the stack alignment.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40004 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/test/CodeGen/ARM/2007-03-07-CombinerCrash.ll b/test/CodeGen/ARM/2007-03-07-CombinerCrash.ll
new file mode 100644
index 0000000..7317e62
--- /dev/null
+++ b/test/CodeGen/ARM/2007-03-07-CombinerCrash.ll
@@ -0,0 +1,21 @@
+; RUN: llvm-as < %s | llc -mtriple=arm-apple-darwin -mattr=+v6,+vfp2
+
+define fastcc i8* @read_sleb128(i8* %p, i32* %val) {
+ br label %bb
+
+bb: ; preds = %bb, %0
+ %p_addr.0 = getelementptr i8* %p, i32 0 ; <i8*> [#uses=1]
+ %tmp2 = load i8* %p_addr.0 ; <i8> [#uses=2]
+ %tmp4.rec = add i32 0, 1 ; <i32> [#uses=1]
+ %tmp4 = getelementptr i8* %p, i32 %tmp4.rec ; <i8*> [#uses=1]
+ %tmp56 = zext i8 %tmp2 to i32 ; <i32> [#uses=1]
+ %tmp7 = and i32 %tmp56, 127 ; <i32> [#uses=1]
+ %tmp9 = shl i32 %tmp7, 0 ; <i32> [#uses=1]
+ %tmp11 = or i32 %tmp9, 0 ; <i32> [#uses=1]
+ icmp slt i8 %tmp2, 0 ; <i1>:1 [#uses=1]
+ br i1 %1, label %bb, label %cond_next28
+
+cond_next28: ; preds = %bb
+ store i32 %tmp11, i32* %val
+ ret i8* %tmp4
+}