[BOLT][AArch64]support inline-small-functions for AArch64 (#120187)

Add some functions in `AArch64MCPlusBuilder.cpp` to support inline for
AArch64.

Author: liusy58

bolt/lib/Passes/Inliner.cpp

@@ -310,13 +310,13 @@ Inliner::inlineCall(BinaryBasicBlock &CallerBB,
       if (MIB.isPseudo(Inst))
         continue;
 
-      MIB.stripAnnotations(Inst, /*KeepTC=*/BC.isX86());
+      MIB.stripAnnotations(Inst, /*KeepTC=*/BC.isX86() || BC.isAArch64());
 
       // Fix branch target. Strictly speaking, we don't have to do this as
       // targets of direct branches will be fixed later and don't matter
       // in the CFG state. However, disassembly may look misleading, and
       // hence we do the fixing.
-      if (MIB.isBranch(Inst)) {
+      if (MIB.isBranch(Inst) && !MIB.isTailCall(Inst)) {
         assert(!MIB.isIndirectBranch(Inst) &&
                "unexpected indirect branch in callee");
         const BinaryBasicBlock *TargetBB =

bolt/lib/Target/AArch64/AArch64MCPlusBuilder.cpp

@@ -133,6 +133,36 @@ class AArch64MCPlusBuilder : public MCPlusBuilder {
 public:
   using MCPlusBuilder::MCPlusBuilder;
 
+  MCPhysReg getStackPointer() const override { return AArch64::SP; }
+
+  bool isPush(const MCInst &Inst) const override { return false; }
+
+  bool isPop(const MCInst &Inst) const override { return false; }
+
+  void createCall(MCInst &Inst, const MCSymbol *Target,
+                  MCContext *Ctx) override {
+    createDirectCall(Inst, Target, Ctx, false);
+  }
+
+  bool convertTailCallToCall(MCInst &Inst) override {
+    int NewOpcode;
+    switch (Inst.getOpcode()) {
+    default:
+      return false;
+    case AArch64::B:
+      NewOpcode = AArch64::BL;
+      break;
+    case AArch64::BR:
+      NewOpcode = AArch64::BLR;
+      break;
+    }
+
+    Inst.setOpcode(NewOpcode);
+    removeAnnotation(Inst, MCPlus::MCAnnotation::kTailCall);
+    clearOffset(Inst);
+    return true;
+  }
+
   bool equals(const MCTargetExpr &A, const MCTargetExpr &B,
               CompFuncTy Comp) const override {
     const auto &AArch64ExprA = cast<AArch64MCExpr>(A);

bolt/test/AArch64/inline-small-function-1.s

@@ -0,0 +1,42 @@
+## This test checks that inline is properly handled by BOLT on aarch64.
+
+# REQUIRES: system-linux
+
+# RUN: llvm-mc -filetype=obj -triple aarch64-unknown-unknown %s -o %t.o
+# RUN: %clang %cflags -O0 %t.o -o %t.exe -Wl,-q
+# RUN: llvm-bolt --inline-small-functions --print-inline  --print-only=_Z3barP1A  \
+# RUN: %t.exe -o %t.bolt  | FileCheck %s
+
+# CHECK: BOLT-INFO: inlined 0 calls at 1 call sites in 2 iteration(s). Change in binary size: 4 bytes.
+# CHECK: Binary Function "_Z3barP1A" after inlining {
+# CHECK-NOT: bl	_Z3fooP1A
+# CHECK: ldr	x8, [x0]
+# CHECK-NEXT: ldr	w0, [x8]
+ 
+	.text
+	.globl	_Z3fooP1A                      
+	.type	_Z3fooP1A,@function
+_Z3fooP1A:                              
+	ldr	x8, [x0]
+	ldr	w0, [x8]
+	ret
+	.size	_Z3fooP1A, .-_Z3fooP1A
+
+	.globl	_Z3barP1A                       
+	.type	_Z3barP1A,@function
+_Z3barP1A:                              
+	stp	x29, x30, [sp, #-16]!           
+	mov	x29, sp
+	bl	_Z3fooP1A
+	mul	w0, w0, w0
+	ldp	x29, x30, [sp], #16             
+	ret
+	.size	_Z3barP1A, .-_Z3barP1A
+
+	.globl	main                            
+	.p2align	2
+	.type	main,@function
+main:                                   
+	mov	w0, wzr
+	ret
+	.size	main, .-main

bolt/test/AArch64/inline-small-function-2.s

@@ -0,0 +1,48 @@
+## This test checks that inline is properly handled by BOLT on aarch64.
+
+# REQUIRES: system-linux
+
+# RUN: llvm-mc -filetype=obj -triple aarch64-unknown-unknown %s -o %t.o
+# RUN: %clang %cflags -O0 %t.o -o %t.exe -Wl,-q
+# RUN: llvm-bolt --inline-small-functions --print-inline  --print-only=test  \
+# RUN: %t.exe -o %t.bolt | FileCheck %s
+
+#CHECK: BOLT-INFO: inlined 0 calls at 1 call sites in 2 iteration(s). Change in binary size: 4 bytes.
+#CHECK: Binary Function "test" after inlining {
+#CHECK-NOT: bl	indirect
+#CHECK: add	w0, w1, w0
+#CHECK-NEXT: blr	x2
+ 
+	.text
+	.globl	indirect                       
+	.type	indirect,@function
+indirect:                               
+	add	w0, w1, w0
+	br	x2
+	.size	indirect, .-indirect
+
+	.globl	test                           
+	.type	test,@function
+test:                                   
+	stp	x29, x30, [sp, #-32]!          
+	stp	x20, x19, [sp, #16]            
+	mov	x29, sp
+	mov	w19, w1
+	mov	w20, w0
+	bl	indirect
+	add	w8, w19, w20
+	cmp	w0, #0
+	csinc	w0, w8, wzr, eq
+	ldp	x20, x19, [sp, #16]             
+	ldp	x29, x30, [sp], #32            
+	ret
+	.size	test, .-test
+
+	.globl	main                            
+	.type	main,@function
+main:                                   
+	mov	w0, wzr
+	ret
+	.size	main, .-main
+
+