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/Integer/testmemory_bt.ll b/test/Integer/testmemory_bt.ll
new file mode 100644
index 0000000..e503c56
--- /dev/null
+++ b/test/Integer/testmemory_bt.ll
@@ -0,0 +1,45 @@
+; RUN: llvm-as %s -o - | llvm-dis > %t1.ll
+; RUN: llvm-as %t1.ll -o - | llvm-dis > %t2.ll
+; RUN: diff %t1.ll %t2.ll
+
+
+%struct = type { i31 , {float, {i9 } } , i64 }
+%complexty = type {i31, {[4 x i9 *], float}, double}
+
+
+define i31 @"main"()
+begin
+ call i31 @testfunction(i64 0, i64 1)
+ ret i31 0
+end
+
+define i31 @"testfunction"(i64 %i0, i64 %j0)
+begin
+ %array0 = malloc [4 x i9] ; yields {[4 x i9]*}:array0
+ %size = add i32 2, 2 ; yields {i31}:size = i31 %4
+ %array1 = malloc i9, i32 4 ; yields {i9*}:array1
+ %array2 = malloc i9, i32 %size ; yields {i9*}:array2
+
+ %idx = getelementptr [4 x i9]* %array0, i64 0, i64 2
+ store i9 123, i9* %idx
+ free [4x i9]* %array0
+ free i9* %array1
+ free i9* %array2
+
+
+ %aa = alloca %complexty, i32 5
+ %idx2 = getelementptr %complexty* %aa, i64 %i0, i32 1, i32 0, i64 %j0
+ store i9 *null, i9** %idx2
+
+ %ptr = alloca i31 ; yields {i31*}:ptr
+ store i31 3, i31* %ptr ; yields {void}
+ %val = load i31* %ptr ; yields {i31}:val = i31 %3
+
+ %sptr = alloca %struct ; yields {%struct*}:sptr
+ %ubsptr = getelementptr %struct * %sptr, i64 0, i32 1, i32 1 ; yields {{i9}*}:ubsptr
+ %idx3 = getelementptr {i9} * %ubsptr, i64 0, i32 0
+ store i9 4, i9* %idx3
+
+ ret i31 3
+end
+