From 446ff28df1242fcd11ce1df18af9c944afdcbdd3 Mon Sep 17 00:00:00 2001
From: Evan Cheng <evan.cheng@apple.com>
Date: Tue, 25 Sep 2012 05:32:34 +0000
Subject: [PATCH] Fix an illegal tailcall opt where the callee returns a double
 via xmm while caller returns x86_fp80 via st0. rdar://12229511

llvm-svn: 164588
---
 llvm/lib/Target/X86/X86ISelLowering.cpp | 10 ++++++++-
 llvm/lib/Target/X86/X86ISelLowering.h   |  1 +
 llvm/test/CodeGen/X86/tailcall-64.ll    | 29 ++++++++++++++++++++++---
 3 files changed, 36 insertions(+), 4 deletions(-)

diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index b2b94794f205..cb803f891576 100644
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -2198,7 +2198,7 @@ X86TargetLowering::LowerCall(TargetLowering::CallLoweringInfo &CLI,
     // Check if it's really possible to do a tail call.
     isTailCall = IsEligibleForTailCallOptimization(Callee, CallConv,
                     isVarArg, SR != NotStructReturn,
-                    MF.getFunction()->hasStructRetAttr(),
+                    MF.getFunction()->hasStructRetAttr(), CLI.RetTy,
                     Outs, OutVals, Ins, DAG);
 
     // Sibcalls are automatically detected tailcalls which do not require
@@ -2718,6 +2718,7 @@ X86TargetLowering::IsEligibleForTailCallOptimization(SDValue Callee,
                                                      bool isVarArg,
                                                      bool isCalleeStructRet,
                                                      bool isCallerStructRet,
+                                                     Type *RetTy,
                                     const SmallVectorImpl<ISD::OutputArg> &Outs,
                                     const SmallVectorImpl<SDValue> &OutVals,
                                     const SmallVectorImpl<ISD::InputArg> &Ins,
@@ -2729,6 +2730,13 @@ X86TargetLowering::IsEligibleForTailCallOptimization(SDValue Callee,
   // If -tailcallopt is specified, make fastcc functions tail-callable.
   const MachineFunction &MF = DAG.getMachineFunction();
   const Function *CallerF = DAG.getMachineFunction().getFunction();
+
+  // If the function return type is x86_fp80 and the callee return type is not,
+  // then the FP_EXTEND of the call result is not a nop. It's not safe to
+  // perform a tailcall optimization here.
+  if (CallerF->getReturnType()->isX86_FP80Ty() && !RetTy->isX86_FP80Ty())
+    return false;
+
   CallingConv::ID CallerCC = CallerF->getCallingConv();
   bool CCMatch = CallerCC == CalleeCC;
 
diff --git a/llvm/lib/Target/X86/X86ISelLowering.h b/llvm/lib/Target/X86/X86ISelLowering.h
index 653654f28ac4..a5773294f8c2 100644
--- a/llvm/lib/Target/X86/X86ISelLowering.h
+++ b/llvm/lib/Target/X86/X86ISelLowering.h
@@ -744,6 +744,7 @@ namespace llvm {
                                            bool isVarArg,
                                            bool isCalleeStructRet,
                                            bool isCallerStructRet,
+                                           Type *RetTy,
                                     const SmallVectorImpl<ISD::OutputArg> &Outs,
                                     const SmallVectorImpl<SDValue> &OutVals,
                                     const SmallVectorImpl<ISD::InputArg> &Ins,
diff --git a/llvm/test/CodeGen/X86/tailcall-64.ll b/llvm/test/CodeGen/X86/tailcall-64.ll
index adc862006082..ecc253ba587e 100644
--- a/llvm/test/CodeGen/X86/tailcall-64.ll
+++ b/llvm/test/CodeGen/X86/tailcall-64.ll
@@ -1,6 +1,4 @@
-; RUN: llc < %s | FileCheck %s
-target datalayout = "e-p:64:64:64-S128-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f16:16:16-f32:32:32-f64:64:64-f128:128:128-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64"
-target triple = "x86_64-apple-darwin11.4.0"
+; RUN: llc -mtriple=x86_64-apple-macosx -mcpu=core2 < %s | FileCheck %s
 
 declare i64 @testi()
 
@@ -132,3 +130,28 @@ entry:
   %call = tail call i32 (i8*, ...)* %0(i8* null, i32 0, i32 0, i32 0, i32 0, i32 0) nounwind
   ret i32 %call
 }
+
+define x86_fp80 @fp80_call(x86_fp80 %x) nounwind  {
+entry:
+; CHECK: fp80_call:
+; CHECK: jmp _fp80_callee
+  %call = tail call x86_fp80 @fp80_callee(x86_fp80 %x) nounwind
+  ret x86_fp80 %call
+}
+
+declare x86_fp80 @fp80_callee(x86_fp80)
+
+; rdar://12229511
+define x86_fp80 @trunc_fp80(x86_fp80 %x) nounwind  {
+entry:
+; CHECK: trunc_fp80
+; CHECK: callq _trunc
+; CHECK-NOT: jmp _trunc
+; CHECK: ret
+  %conv = fptrunc x86_fp80 %x to double
+  %call = tail call double @trunc(double %conv) nounwind readnone
+  %conv1 = fpext double %call to x86_fp80
+  ret x86_fp80 %conv1
+}
+
+declare double @trunc(double) nounwind readnone