[CXX_FAST_TLS] Disable tail call when calling conventions are mismatched.

Since CXX_FAST_TLS has a bigger set of CSRs, we don't tail call when caller
and callee have mismatched calling conventions.

llvm-svn: 263856

lib/Target/AArch64/AArch64ISelLowering.cpp

@@ -2817,6 +2817,13 @@ bool AArch64TargetLowering::isEligibleForTailCallOptimization(
   CallingConv::ID CallerCC = CallerF->getCallingConv();
   bool CCMatch = CallerCC == CalleeCC;
 
+  // Disable tailcall for CXX_FAST_TLS when callee and caller have different
+  // calling conventions, given that CXX_FAST_TLS has a bigger CSR set.
+  if (!CCMatch &&
+      (CallerCC == CallingConv::CXX_FAST_TLS ||
+       CalleeCC == CallingConv::CXX_FAST_TLS))
+    return false;
+
   // Byval parameters hand the function a pointer directly into the stack area
   // we want to reuse during a tail call. Working around this *is* possible (see
   // X86) but less efficient and uglier in LowerCall.

lib/Target/ARM/ARMISelLowering.cpp

@@ -2100,6 +2100,13 @@ ARMTargetLowering::IsEligibleForTailCallOptimization(SDValue Callee,
   CallingConv::ID CallerCC = CallerF->getCallingConv();
   bool CCMatch = CallerCC == CalleeCC;
 
+  // Disable tailcall for CXX_FAST_TLS when callee and caller have different
+  // calling conventions, given that CXX_FAST_TLS has a bigger CSR set.
+  if (!CCMatch &&
+      (CallerCC == CallingConv::CXX_FAST_TLS ||
+       CalleeCC == CallingConv::CXX_FAST_TLS))
+    return false;
+
   assert(Subtarget->supportsTailCall());
 
   // Look for obvious safe cases to perform tail call optimization that do not

lib/Target/X86/X86ISelLowering.cpp

@@ -3767,6 +3767,13 @@ bool X86TargetLowering::IsEligibleForTailCallOptimization(
   if (IsCalleeWin64 != IsCallerWin64)
     return false;
 
+  // Disable tailcall for CXX_FAST_TLS when callee and caller have different
+  // calling conventions, given that CXX_FAST_TLS has a bigger CSR set.
+  if (!CCMatch &&
+      (CallerCC == CallingConv::CXX_FAST_TLS ||
+       CalleeCC == CallingConv::CXX_FAST_TLS))
+    return false;
+
   if (DAG.getTarget().Options.GuaranteedTailCallOpt) {
     if (canGuaranteeTCO(CalleeCC) && CCMatch)
       return true;

test/CodeGen/AArch64/cxx-tlscc.ll

@@ -205,5 +205,18 @@ define cxx_fast_tlscc %class.C* @tls_test2() #1 {
   call cxx_fast_tlscc void @tls_helper()
   ret %class.C* @tC
 }
+
+; Make sure we do not allow tail call when caller and callee have different
+; calling conventions.
+declare %class.C* @_ZN1CD1Ev(%class.C* readnone returned %this)
+; CHECK-LABEL: tls_test
+; CHECK: bl __tlv_atexit
+define cxx_fast_tlscc void @__tls_test() {
+entry:
+  store i32 0, i32* getelementptr inbounds (%class.C, %class.C* @tC, i64 0, i32 0), align 4
+  %0 = tail call i32 @_tlv_atexit(void (i8*)* bitcast (%class.C* (%class.C*)* @_ZN1CD1Ev to void (i8*)*), i8* bitcast (%class.C* @tC to i8*), i8* nonnull @__dso_handle) #1
+  ret void
+}
+
 attributes #0 = { nounwind "no-frame-pointer-elim"="true" }
 attributes #1 = { nounwind }

test/CodeGen/ARM/cxx-tlscc.ll

@@ -101,5 +101,18 @@ define cxx_fast_tlscc %class.C* @tls_test2() #1 {
   call cxx_fast_tlscc void @tls_helper()
   ret %class.C* @tC
 }
+
+; Make sure we do not allow tail call when caller and callee have different
+; calling conventions.
+declare %class.C* @_ZN1CD1Ev(%class.C* readnone returned %this)
+; CHECK-LABEL: tls_test
+; CHECK: bl __tlv_atexit
+define cxx_fast_tlscc void @__tls_test() {
+entry:
+  store i32 0, i32* getelementptr inbounds (%class.C, %class.C* @tC, i64 0, i32 0), align 4
+  %0 = tail call i32 @_tlv_atexit(void (i8*)* bitcast (%class.C* (%class.C*)* @_ZN1CD1Ev to void (i8*)*), i8* bitcast (%class.C* @tC to i8*), i8* nonnull @__dso_handle) #1
+  ret void
+}
+
 attributes #0 = { nounwind "no-frame-pointer-elim"="true" }
 attributes #1 = { nounwind }

test/CodeGen/X86/cxx_tlscc64.ll

@@ -138,5 +138,18 @@ define cxx_fast_tlscc %class.C* @tls_test2() #1 {
   call cxx_fast_tlscc void @tls_helper()
   ret %class.C* @tC
 }
+
+; Make sure we do not allow tail call when caller and callee have different
+; calling conventions.
+declare %class.C* @_ZN1CD1Ev(%class.C* readnone returned %this)
+; CHECK-LABEL: tls_test
+; CHECK: callq {{.*}}tlv_atexit
+define cxx_fast_tlscc void @tls_test() {
+entry:
+  store i32 0, i32* getelementptr inbounds (%class.C, %class.C* @tC, i64 0, i32 0), align 4
+  %0 = tail call i32 @_tlv_atexit(void (i8*)* bitcast (%class.C* (%class.C*)* @_ZN1CD1Ev to void (i8*)*), i8* bitcast (%class.C* @tC to i8*), i8* nonnull @__dso_handle) #1
+  ret void
+}
+
 attributes #0 = { nounwind "no-frame-pointer-elim"="true" }
 attributes #1 = { nounwind }