Enable tail calls on ARM by default, with some
basic tests.
This has been well tested on Darwin but not elsewhere.
It should work provided the linker correctly resolves
B.W <label in other function>
which it has not seen before, at least from llvm-based
compilers. I'm leaving the arm-tail-calls switch in
until I see if there's any problems because of that;
it might need to be disabled for some environments.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@106299 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/test/CodeGen/ARM/ifcvt6-tc.ll b/test/CodeGen/ARM/ifcvt6-tc.ll
new file mode 100644
index 0000000..5b28804
--- /dev/null
+++ b/test/CodeGen/ARM/ifcvt6-tc.ll
@@ -0,0 +1,23 @@
+; RUN: llc < %s -march=arm -mtriple=arm-apple-darwin | \
+; RUN: grep cmpne | count 1
+; RUN: llc < %s -march=arm -mtriple=arm-apple-darwin | \
+; RUN: grep bhi | count 1
+; Here, tail call wins over eliminating branches. It is 1 fewer instruction
+; and removes all stack accesses, so seems like a win.
+
+define void @foo(i32 %X, i32 %Y) {
+entry:
+ %tmp1 = icmp ult i32 %X, 4 ; <i1> [#uses=1]
+ %tmp4 = icmp eq i32 %Y, 0 ; <i1> [#uses=1]
+ %tmp7 = or i1 %tmp4, %tmp1 ; <i1> [#uses=1]
+ br i1 %tmp7, label %cond_true, label %UnifiedReturnBlock
+
+cond_true: ; preds = %entry
+ %tmp10 = tail call i32 (...)* @bar( ) ; <i32> [#uses=0]
+ ret void
+
+UnifiedReturnBlock: ; preds = %entry
+ ret void
+}
+
+declare i32 @bar(...)