[WinEH] Emit int3 after noreturn calls on Win64

[oota-llvm.git] / lib / CodeGen / SelectionDAG / SelectionDAGBuilder.cpp

diff --git a/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp b/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
index 2885776a30649a19bd1e885ae540962217d24b49..a48de572113750efe9a1c2804b4f92236064b2df 100644 (file)
--- a/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
+++ b/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
@@ -80,7 +80,7 @@ LimitFPPrecision("limit-float-precision",
                  cl::init(0));
 
 static cl::opt<bool>
-EnableFMFInDAG("enable-fmf-dag", cl::init(false), cl::Hidden,
+EnableFMFInDAG("enable-fmf-dag", cl::init(true), cl::Hidden,
                 cl::desc("Enable fast-math-flags for DAG nodes"));
 
 // Limit the width of DAG chains. This is important in general to prevent
@@ -1182,6 +1182,7 @@ void SelectionDAGBuilder::visitCleanupPad(const CleanupPadInst &CPI) {
   // Don't emit any special code for the cleanuppad instruction. It just marks
   // the start of a funclet.
   FuncInfo.MBB->setIsEHFuncletEntry();
+  FuncInfo.MBB->setIsCleanupFuncletEntry();
 }
 
 /// When an invoke or a cleanupret unwinds to the next EH pad, there are
@@ -2204,10 +2205,7 @@ void SelectionDAGBuilder::visitIndirectBr(const IndirectBrInst &I) {
                           getValue(I.getAddress())));
 }
 
-void SelectionDAGBuilder::visitUnreachable(const UnreachableInst &I) {
-  if (DAG.getTarget().Options.TrapUnreachable)
-    DAG.setRoot(DAG.getNode(ISD::TRAP, getCurSDLoc(), MVT::Other, DAG.getRoot()));
-}
+void SelectionDAGBuilder::visitUnreachable(const UnreachableInst &I) {}
 
 void SelectionDAGBuilder::visitFSub(const User &I) {
   // -0.0 - X --> fneg
@@ -2347,6 +2345,10 @@ void SelectionDAGBuilder::visitFCmp(const User &I) {
   SDValue Op1 = getValue(I.getOperand(0));
   SDValue Op2 = getValue(I.getOperand(1));
   ISD::CondCode Condition = getFCmpCondCode(predicate);
+  
+  // FIXME: Fcmp instructions have fast-math-flags in IR, so we should use them.
+  // FIXME: We should propagate the fast-math-flags to the DAG node itself for
+  // further optimization, but currently FMF is only applicable to binary nodes.
   if (TM.Options.NoNaNsFPMath)
     Condition = getFCmpCodeWithoutNaN(Condition);
   EVT DestVT = DAG.getTargetLoweringInfo().getValueType(DAG.getDataLayout(),
@@ -3629,6 +3631,8 @@ getF32Constant(SelectionDAG &DAG, unsigned Flt, SDLoc dl) {
 
 static SDValue getLimitedPrecisionExp2(SDValue t0, SDLoc dl,
                                        SelectionDAG &DAG) {
+  // TODO: What fast-math-flags should be set on the floating-point nodes?
+
   //   IntegerPartOfX = ((int32_t)(t0);
   SDValue IntegerPartOfX = DAG.getNode(ISD::FP_TO_SINT, dl, MVT::i32, t0);
 
@@ -3727,6 +3731,8 @@ static SDValue expandExp(SDLoc dl, SDValue Op, SelectionDAG &DAG,
     //
     //   #define LOG2OFe 1.4426950f
     //   t0 = Op * LOG2OFe
+
+    // TODO: What fast-math-flags should be set here?
     SDValue t0 = DAG.getNode(ISD::FMUL, dl, MVT::f32, Op,
                              getF32Constant(DAG, 0x3fb8aa3b, dl));
     return getLimitedPrecisionExp2(t0, dl, DAG);
@@ -3740,6 +3746,9 @@ static SDValue expandExp(SDLoc dl, SDValue Op, SelectionDAG &DAG,
 /// limited-precision mode.
 static SDValue expandLog(SDLoc dl, SDValue Op, SelectionDAG &DAG,
                          const TargetLowering &TLI) {
+ 
+  // TODO: What fast-math-flags should be set on the floating-point nodes?
+
   if (Op.getValueType() == MVT::f32 &&
       LimitFloatPrecision > 0 && LimitFloatPrecision <= 18) {
     SDValue Op1 = DAG.getNode(ISD::BITCAST, dl, MVT::i32, Op);
@@ -3836,6 +3845,9 @@ static SDValue expandLog(SDLoc dl, SDValue Op, SelectionDAG &DAG,
 /// limited-precision mode.
 static SDValue expandLog2(SDLoc dl, SDValue Op, SelectionDAG &DAG,
                           const TargetLowering &TLI) {
+  
+  // TODO: What fast-math-flags should be set on the floating-point nodes?
+
   if (Op.getValueType() == MVT::f32 &&
       LimitFloatPrecision > 0 && LimitFloatPrecision <= 18) {
     SDValue Op1 = DAG.getNode(ISD::BITCAST, dl, MVT::i32, Op);
@@ -3931,6 +3943,9 @@ static SDValue expandLog2(SDLoc dl, SDValue Op, SelectionDAG &DAG,
 /// limited-precision mode.
 static SDValue expandLog10(SDLoc dl, SDValue Op, SelectionDAG &DAG,
                            const TargetLowering &TLI) {
+
+  // TODO: What fast-math-flags should be set on the floating-point nodes?
+
   if (Op.getValueType() == MVT::f32 &&
       LimitFloatPrecision > 0 && LimitFloatPrecision <= 18) {
     SDValue Op1 = DAG.getNode(ISD::BITCAST, dl, MVT::i32, Op);
@@ -4040,6 +4055,7 @@ static SDValue expandPow(SDLoc dl, SDValue LHS, SDValue RHS,
     }
   }
 
+  // TODO: What fast-math-flags should be set on the FMUL node?
   if (IsExp10) {
     // Put the exponent in the right bit position for later addition to the
     // final result:
@@ -4083,6 +4099,8 @@ static SDValue ExpandPowI(SDLoc DL, SDValue LHS, SDValue RHS,
       // the benefit of being both really simple and much better than a libcall.
       SDValue Res;  // Logically starts equal to 1.0
       SDValue CurSquare = LHS;
+      // TODO: Intrinsics should have fast-math-flags that propagate to these
+      // nodes.
       while (Val) {
         if (Val & 1) {
           if (Res.getNode())
@@ -4373,15 +4391,9 @@ SelectionDAGBuilder::visitIntrinsicCall(const CallInst &I, unsigned Intrinsic) {
                                    N);
           return nullptr;
         }
-      } else if (AI)
+      } else {
         SDV = DAG.getDbgValue(Variable, Expression, N.getNode(), N.getResNo(),
                               true, 0, dl, SDNodeOrder);
-      else {
-        // Can't do anything with other non-AI cases yet.
-        DEBUG(dbgs() << "Dropping debug info for " << DI << "\n");
-        DEBUG(dbgs() << "non-AllocaInst issue for Address: \n\t");
-        DEBUG(Address->dump());
-        return nullptr;
       }
       DAG.AddDbgValue(SDV, N.getNode(), isParameter);
     } else {
@@ -4736,6 +4748,7 @@ SelectionDAGBuilder::visitIntrinsicCall(const CallInst &I, unsigned Intrinsic) {
                                getValue(I.getArgOperand(1)),
                                getValue(I.getArgOperand(2))));
     } else {
+      // TODO: Intrinsic calls should have fast-math-flags.
       SDValue Mul = DAG.getNode(ISD::FMUL, sdl,
                                 getValue(I.getArgOperand(0)).getValueType(),
                                 getValue(I.getArgOperand(0)),
@@ -5241,7 +5254,13 @@ SelectionDAGBuilder::lowerInvokable(TargetLowering::CallLoweringInfo &CLI,
     DAG.setRoot(DAG.getEHLabel(getCurSDLoc(), getRoot(), EndLabel));
 
     // Inform MachineModuleInfo of range.
-    MMI.addInvoke(FuncInfo.MBBMap[EHPadBB], BeginLabel, EndLabel);
+    if (MMI.hasEHFunclets()) {
+      WinEHFuncInfo &EHInfo =
+          MMI.getWinEHFuncInfo(DAG.getMachineFunction().getFunction());
+      EHInfo.addIPToStateRange(EHPadBB, BeginLabel, EndLabel);
+    } else {
+      MMI.addInvoke(FuncInfo.MBBMap[EHPadBB], BeginLabel, EndLabel);
+    }
   }
 
   return Result;
@@ -6112,7 +6131,7 @@ void SelectionDAGBuilder::visitInlineAsm(ImmutableCallSite CS) {
       SDISelAsmOperandInfo &Input = ConstraintOperands[OpInfo.MatchingInput];
 
       if (OpInfo.ConstraintVT != Input.ConstraintVT) {
-       const TargetRegisterInfo *TRI = DAG.getSubtarget().getRegisterInfo();
+        const TargetRegisterInfo *TRI = DAG.getSubtarget().getRegisterInfo();
         std::pair<unsigned, const TargetRegisterClass *> MatchRC =
             TLI.getRegForInlineAsmConstraint(TRI, OpInfo.ConstraintCode,
                                              OpInfo.ConstraintVT);
@@ -8125,13 +8144,13 @@ void SelectionDAGBuilder::lowerWorkItem(SwitchWorkListItem W, Value *Cond,
         uint32_t JumpWeight = I->Weight;
         uint32_t FallthroughWeight = UnhandledWeights;
 
-        // If Fallthrough is a target of the jump table, we evenly distribute
-        // the weight on the edge to Fallthrough to successors of CurMBB.
-        // Also update the weight on the edge from JumpMBB to Fallthrough.
+        // If the default statement is a target of the jump table, we evenly
+        // distribute the default weight to successors of CurMBB. Also update
+        // the weight on the edge from JumpMBB to Fallthrough.
         for (MachineBasicBlock::succ_iterator SI = JumpMBB->succ_begin(),
                                               SE = JumpMBB->succ_end();
              SI != SE; ++SI) {
-          if (*SI == Fallthrough) {
+          if (*SI == DefaultMBB) {
             JumpWeight += DefaultWeight / 2;
             FallthroughWeight -= DefaultWeight / 2;
             JumpMBB->setSuccWeight(SI, DefaultWeight / 2);