[Outliner] Add tail call support

This commit adds tail call support to the MachineOutliner pass. This allows
the outliner to insert jumps rather than calls in areas where tail calling is
possible. Outlined tail calls include the return or terminator of the basic
block being outlined from.

Tail call support allows the outliner to take returns and terminators into
consideration while finding candidates to outline. It also allows the outliner
to save more instructions. For example, in the X86-64 outliner, a tail called
outlined function saves one instruction since no return has to be inserted.

llvm-svn: 297653

2 files changed, 41 insertions, 6 deletions

diff --git a/llvm/test/CodeGen/X86/machine-outliner-tailcalls.ll b/llvm/test/CodeGen/X86/machine-outliner-tailcalls.ll
new file mode 100644
index 00000000000..020f7eeaaff
--- /dev/null
+++ b/llvm/test/CodeGen/X86/machine-outliner-tailcalls.ll
@@ -0,0 +1,35 @@
+; RUN: llc -enable-machine-outliner -mtriple=x86_64-apple-darwin < %s | FileCheck %s
+
+@x = common local_unnamed_addr global i32 0, align 4
+
+define i32 @foo0(i32) local_unnamed_addr #0 {
+; CHECK-LABEL: _foo0:
+; CHECK: jmp l_OUTLINED_FUNCTION_0
+; CHECK-NEXT: .cfi_endproc
+  store i32 0, i32* @x, align 4, !tbaa !2
+  %2 = tail call i32 @ext(i32 1) #2
+  ret i32 undef
+}
+
+declare i32 @ext(i32) local_unnamed_addr #1
+
+define i32 @foo1(i32) local_unnamed_addr #0 {
+; CHECK-LABEL: _foo1:
+; CHECK: jmp l_OUTLINED_FUNCTION_0
+; CHECK-NEXT: .cfi_endproc
+  store i32 0, i32* @x, align 4, !tbaa !2
+  %2 = tail call i32 @ext(i32 1) #2
+  ret i32 undef
+}
+
+attributes #0 = { noredzone nounwind ssp uwtable "no-frame-pointer-elim"="false" }
+
+!2 = !{!3, !3, i64 0}
+!3 = !{!"int", !4, i64 0}
+!4 = !{!"omnipotent char", !5, i64 0}
+!5 = !{!"Simple C/C++ TBAA"}
+
+; CHECK-LABEL: l_OUTLINED_FUNCTION_0:
+; CHECK: movl  $0, (%rax)
+; CHECK-NEXT: movl  $1, %edi
+; CHECK-NEXT: jmp _ext 
\ No newline at end of file
diff --git a/llvm/test/CodeGen/X86/machine-outliner.ll b/llvm/test/CodeGen/X86/machine-outliner.ll
index 9246348c563..9f8e6ec298f 100644
--- a/llvm/test/CodeGen/X86/machine-outliner.ll
+++ b/llvm/test/CodeGen/X86/machine-outliner.ll
@@ -15,7 +15,7 @@ define i32 @check_boundaries() #0 {
   %7 = icmp ne i32 %6, 0
   br i1 %7, label %9, label %8
 
-  ; CHECK: callq l_OUTLINED_FUNCTION_1
+  ; CHECK: callq [[OFUNC1:l_OUTLINED_FUNCTION_[0-9]+]]
   ; CHECK: cmpl  $0, -{{[0-9]+}}(%rbp)
   store i32 1, i32* %2, align 4
   store i32 2, i32* %3, align 4
@@ -30,7 +30,7 @@ define i32 @check_boundaries() #0 {
   %12 = icmp ne i32 %11, 0
   br i1 %12, label %14, label %13
 
-  ; CHECK: callq l_OUTLINED_FUNCTION_1
+  ; CHECK: callq [[OFUNC1]]
   store i32 1, i32* %2, align 4
   store i32 2, i32* %3, align 4
   store i32 3, i32* %4, align 4
@@ -79,13 +79,13 @@ define i32 @main() #0 {
 
   store i32 0, i32* %1, align 4
   store i32 0, i32* @x, align 4
-  ; CHECK: callq l_OUTLINED_FUNCTION_0
+  ; CHECK: callq [[OFUNC2:l_OUTLINED_FUNCTION_[0-9]+]]
   store i32 1, i32* %2, align 4
   store i32 2, i32* %3, align 4
   store i32 3, i32* %4, align 4
   store i32 4, i32* %5, align 4
   store i32 1, i32* @x, align 4
-  ; CHECK: callq l_OUTLINED_FUNCTION_0
+  ; CHECK: callq [[OFUNC2]]
   store i32 1, i32* %2, align 4
   store i32 2, i32* %3, align 4
   store i32 3, i32* %4, align 4
@@ -95,14 +95,14 @@ define i32 @main() #0 {
 
 attributes #0 = { noredzone nounwind ssp uwtable "no-frame-pointer-elim"="true" }
 
-; CHECK-LABEL: l_OUTLINED_FUNCTION_0:
+; CHECK-LABEL: l_OUTLINED_FUNCTION_1:
 ; CHECK:      movl  $1, -{{[0-9]+}}(%rbp)
 ; CHECK-NEXT: movl  $2, -{{[0-9]+}}(%rbp)
 ; CHECK-NEXT: movl  $3, -{{[0-9]+}}(%rbp)
 ; CHECK-NEXT: movl  $4, -{{[0-9]+}}(%rbp)
 ; CHECK-NEXT: retq
 
-; CHECK-LABEL: l_OUTLINED_FUNCTION_1:
+; CHECK-LABEL: l_OUTLINED_FUNCTION_0:
 ; CHECK:      movl  $1, -{{[0-9]+}}(%rbp)
 ; CHECK-NEXT: movl  $2, -{{[0-9]+}}(%rbp)
 ; CHECK-NEXT: movl  $3, -{{[0-9]+}}(%rbp)