3 files changed, 63 insertions, 4 deletions
diff --git a/test/Transforms/TailCallElim/EraseBB.ll b/test/Transforms/TailCallElim/EraseBB.ll
new file mode 100644
index 0000000..c8290d7
--- /dev/null
+++ b/test/Transforms/TailCallElim/EraseBB.ll
@@ -0,0 +1,26 @@
+; RUN: opt -tailcallelim -S < %s 2>&1 | FileCheck %s
+
+; CHECK: add nsw i32
+; CHECK-NEXT: br label
+; CHECK: add nsw i32
+; CHECK-NEXT: br label
+; CHECK-NOT: Uses remain when a value is destroyed
+define i32 @test(i32 %n) {
+entry:
+  %cmp = icmp slt i32 %n, 2
+  br i1 %cmp, label %if.then, label %if.else
+
+if.then:                                          ; preds = %entry
+  %v1 = add nsw i32 %n, -2
+  %call1 = tail call i32 @test(i32 %v1)
+  br label %return
+
+if.else:                                          ; preds = %entry
+  %v2 = add nsw i32 %n, 4
+  %call2 = tail call i32 @test(i32 %v2)
+  br label %return
+
+return:                                           ; preds = %if.end, %if.else
+  %retval = phi i32 [ %call1, %if.then ], [ %call2, %if.else ]
+  ret i32 %retval
+}
diff --git a/test/Transforms/TailCallElim/basic.ll b/test/Transforms/TailCallElim/basic.ll
index 341736d..8e9814b 100644
--- a/test/Transforms/TailCallElim/basic.ll
+++ b/test/Transforms/TailCallElim/basic.ll
@@ -174,3 +174,17 @@ if.end:
 return:
   ret void
 }
+
+declare void @test11_helper1(i8** nocapture, i8*)
+declare void @test11_helper2(i8*)
+define void @test11() {
+; CHECK-LABEL: @test11
+; CHECK-NOT: tail
+  %a = alloca i8*
+  %b = alloca i8
+  call void @test11_helper1(i8** %a, i8* %b)  ; a = &b
+  %c = load i8** %a
+  call void @test11_helper2(i8* %c)
+; CHECK: call void @test11_helper2
+  ret void
+}
diff --git a/test/Transforms/TailCallElim/reorder_load.ll b/test/Transforms/TailCallElim/reorder_load.ll
index 53c65da..2e350d6 100644
--- a/test/Transforms/TailCallElim/reorder_load.ll
+++ b/test/Transforms/TailCallElim/reorder_load.ll
@@ -1,6 +1,8 @@
 ; RUN: opt < %s -tailcallelim -S | FileCheck %s
 ; PR4323
 
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+
 ; Several cases where tail call elimination should move the load above the call,
 ; then eliminate the tail recursion.
 
@@ -12,6 +14,11 @@
 ; This load can be moved above the call because the function won't write to it
 ; and the call has no side effects.
 define fastcc i32 @raise_load_1(i32* %a_arg, i32 %a_len_arg, i32 %start_arg) nounwind readonly {
+; CHECK-LABEL: @raise_load_1(
+; CHECK-NOT: call
+; CHECK: load i32*
+; CHECK-NOT: call
+; CHECK: }
 entry:
 	%tmp2 = icmp sge i32 %start_arg, %a_len_arg		; <i1> [#uses=1]
 	br i1 %tmp2, label %if, label %else
@@ -21,7 +28,6 @@ if:		; preds = %entry
 
 else:		; preds = %entry
 	%tmp7 = add i32 %start_arg, 1		; <i32> [#uses=1]
-; CHECK-NOT: call
 	%tmp8 = call fastcc i32 @raise_load_1(i32* %a_arg, i32 %a_len_arg, i32 %tmp7)		; <i32> [#uses=1]
 	%tmp9 = load i32* %a_arg		; <i32> [#uses=1]
 	%tmp10 = add i32 %tmp9, %tmp8		; <i32> [#uses=1]
@@ -32,6 +38,11 @@ else:		; preds = %entry
 ; This load can be moved above the call because the function won't write to it
 ; and the load provably can't trap.
 define fastcc i32 @raise_load_2(i32* %a_arg, i32 %a_len_arg, i32 %start_arg) readonly {
+; CHECK-LABEL: @raise_load_2(
+; CHECK-NOT: call
+; CHECK: load i32*
+; CHECK-NOT: call
+; CHECK: }
 entry:
 	%tmp2 = icmp sge i32 %start_arg, %a_len_arg		; <i1> [#uses=1]
 	br i1 %tmp2, label %if, label %else
@@ -48,7 +59,6 @@ unwind:		; preds = %else
 
 recurse:		; preds = %else
 	%tmp7 = add i32 %start_arg, 1		; <i32> [#uses=1]
-; CHECK-NOT: call
 	%tmp8 = call fastcc i32 @raise_load_2(i32* %a_arg, i32 %a_len_arg, i32 %tmp7)		; <i32> [#uses=1]
 	%tmp9 = load i32* @global		; <i32> [#uses=1]
 	%tmp10 = add i32 %tmp9, %tmp8		; <i32> [#uses=1]
@@ -59,6 +69,11 @@ recurse:		; preds = %else
 ; This load can be safely moved above the call (even though it's from an
 ; extern_weak global) because the call has no side effects.
 define fastcc i32 @raise_load_3(i32* %a_arg, i32 %a_len_arg, i32 %start_arg) nounwind readonly {
+; CHECK-LABEL: @raise_load_3(
+; CHECK-NOT: call
+; CHECK: load i32*
+; CHECK-NOT: call
+; CHECK: }
 entry:
 	%tmp2 = icmp sge i32 %start_arg, %a_len_arg		; <i1> [#uses=1]
 	br i1 %tmp2, label %if, label %else
@@ -68,7 +83,6 @@ if:		; preds = %entry
 
 else:		; preds = %entry
 	%tmp7 = add i32 %start_arg, 1		; <i32> [#uses=1]
-; CHECK-NOT: call
 	%tmp8 = call fastcc i32 @raise_load_3(i32* %a_arg, i32 %a_len_arg, i32 %tmp7)		; <i32> [#uses=1]
 	%tmp9 = load i32* @extern_weak_global		; <i32> [#uses=1]
 	%tmp10 = add i32 %tmp9, %tmp8		; <i32> [#uses=1]
@@ -80,6 +94,12 @@ else:		; preds = %entry
 ; unknown pointer (which normally means it might trap) because the first load
 ; proves it doesn't trap.
 define fastcc i32 @raise_load_4(i32* %a_arg, i32 %a_len_arg, i32 %start_arg) readonly {
+; CHECK-LABEL: @raise_load_4(
+; CHECK-NOT: call
+; CHECK: load i32*
+; CHECK-NEXT: load i32*
+; CHECK-NOT: call
+; CHECK: }
 entry:
 	%tmp2 = icmp sge i32 %start_arg, %a_len_arg		; <i1> [#uses=1]
 	br i1 %tmp2, label %if, label %else
@@ -97,7 +117,6 @@ unwind:		; preds = %else
 recurse:		; preds = %else
 	%tmp7 = add i32 %start_arg, 1		; <i32> [#uses=1]
 	%first = load i32* %a_arg		; <i32> [#uses=1]
-; CHECK-NOT: call
 	%tmp8 = call fastcc i32 @raise_load_4(i32* %a_arg, i32 %first, i32 %tmp7)		; <i32> [#uses=1]
 	%second = load i32* %a_arg		; <i32> [#uses=1]
 	%tmp10 = add i32 %second, %tmp8		; <i32> [#uses=1]