Index: test/GPGPU/test.ll
===================================================================
--- /dev/null
+++ test/GPGPU/test.ll
@@ -0,0 +1,52 @@
+; RUN: opt %loadPolly -polly-codegen-ppcg -polly-acc-fail-on-verify-module-failure \
+; RUN: -disable-output < %s
+
+; ModuleID = 'reduced.ll'
+source_filename = "bugpoint-output-cb5513b.bc"
+target datalayout = "e-p:64:64:64-S128-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f16:16:16-f32:32:32-f64:64:64-f128:128:128-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64"
+target triple = "x86_64-unknown-linux-gnu"
+
+declare void @fn_to_fence(i8 *%val)
+
+; Function Attrs: nounwind uwtable
+define void @f(i32* noalias %nprocx, i32* noalias %igroup_cart) #1 {
+entry:
+  %tmp = tail call noalias i8* @malloc(i64 undef) #2
+  br label %for.init
+
+for.init:                                             ; preds = %for.end, %entry.split
+  %tmp1 = phi i32 [ %tmp13, %for.end ], [ 0, %entry ]
+  %tmp2 = load i32, i32* %nprocx, align 4
+  br label %for2.body
+
+for2.body:                                             ; preds = %"65", %"64"
+  %tmp4 = phi i32 [ %tmp11, %for2.body ], [ 0, %for.init ]
+  %tmp5 = sext i32 %tmp4 to i64
+  %tmp8 = bitcast i8* %tmp to i32*
+  %tmp9 = getelementptr i32, i32* %tmp8, i64 %tmp5
+  store i32 %tmp1, i32* %tmp9, align 4
+  %tmp10 = icmp eq i32 %tmp4, %tmp2
+  %tmp11 = add i32 %tmp4, 1
+  br i1 %tmp10, label %for2.body.fence, label %for2.body
+
+for2.body.fence:                                             ; preds = %"65"
+  call void @fn_to_fence(i8* %tmp) #2
+  br i1 undef, label %for.end, label %exit
+
+exit:                                             ; preds = %"69"
+  ret void
+
+for.end:                                             ; preds = %"69"
+  %tmp13 = add i32 %tmp1, 1
+  br label %for.init
+}
+
+; Function Attrs: nounwind
+declare noalias i8* @malloc(i64) #2
+
+declare void @used_fn(...)
+
+
+attributes #0 = { argmemonly nounwind }
+attributes #1 = { nounwind uwtable }
+attributes #2 = { nounwind }