[X86] Make the subus special case in LowerVSETCC self contained

[craig.topper]

Summary

Previously this code overrode the flags and opcode used by the later code in LowerVSETCC. This makes the code difficult to read and follow.

This patch moves all the SUBUS code into its own function and makes it responsible for creating its own SDNodes on success.

[spatel]

LGTM. Thanks for the clean-up!

lib/Target/X86/X86ISelLowering.cpp

18136–18139

No need to duplicate the function comment here. Also, if you use the doxygen special '///' on top of the function, then some IDEs show that comment as part of auto-completion options. It's nice when it works. :)

lib/Target/X86/X86ISelLowering.cpp

if (Val == 0)
       return SDValue();
 
     ULTOp1.push_back(DAG.getConstant(Val - 1, dl, EVT));
   }
 
   return DAG.getBuildVector(VT, dl, ULTOp1);
 }
 
+// As another special case, use PSUBUS[BW] when it's profitable. E.g. for
+// Op0 u<= Op1:
+//   t = psubus Op0, Op1
+//   pcmpeq t, <0..0>
+static SDValue LowerVSETCCWithSUBUS(SDValue Op0, SDValue Op1, MVT VT,
+                                    ISD::CondCode Cond, const SDLoc &dl,
+                                    const X86Subtarget &Subtarget,
+                                    SelectionDAG &DAG) {
+  if (!Subtarget.hasSSE2())
+    return SDValue();
+
+  MVT VET = VT.getVectorElementType();
+  if (VET != MVT::i8 && VET != MVT::i16)
+    return SDValue();
+
+  switch (Cond) {
+  default:
+    return SDValue();
+  case ISD::SETULT: {
+    // If the comparison is against a constant we can turn this into a
+    // setule.  With psubus, setule does not require a swap.  This is
+    // beneficial because the constant in the register is no longer
+    // destructed as the destination so it can be hoisted out of a loop.
+    // Only do this pre-AVX since vpcmp* is no longer destructive.
+    if (Subtarget.hasAVX())
+      return SDValue();
+    SDValue ULEOp1 = ChangeVSETULTtoVSETULE(dl, Op1, DAG);
+    if (!ULEOp1)
+      return SDValue();
+    Op1 = ULEOp1;
+    break;
+  }
+  // Psubus is better than flip-sign because it requires no inversion.
+  case ISD::SETUGE:
+    std::swap(Op0, Op1);
+    break;
+  case ISD::SETULE:
+    break;
+  }
+
+  SDValue Result = DAG.getNode(X86ISD::SUBUS, dl, VT, Op0, Op1);
+  return DAG.getNode(X86ISD::PCMPEQ, dl, VT, Result,
+                     getZeroVector(VT, Subtarget, DAG, dl));
+}
+
 static SDValue LowerVSETCC(SDValue Op, const X86Subtarget &Subtarget,
                            SelectionDAG &DAG) {
   SDValue Op0 = Op.getOperand(0);
   SDValue Op1 = Op.getOperand(1);
   SDValue CC = Op.getOperand(2);
   MVT VT = Op.getSimpleValueType();
   ISD::CondCode Cond = cast<CondCodeSDNode>(CC)->get();
   bool isFP = Op.getOperand(1).getSimpleValueType().isFloatingPoint();
if (ISD::isUnsignedIntSetCC(Cond) &&
 
     // If the logical-not of the result is required, perform that now.
     if (Invert)
       Result = DAG.getNOT(dl, Result, VT);
 
     return Result;
   }
 
+  // As another special case, use PSUBUS[BW] when it's profitable. E.g. for
+  // Op0 u<= Op1:
+  //   t = psubus Op0, Op1
+  //   pcmpeq t, <0..0>
+  if (SDValue V = LowerVSETCCWithSUBUS(Op0, Op1, VT, Cond, dl, Subtarget, DAG))
+    return V;
+
   // We are handling one of the integer comparisons here. Since SSE only has
   // GT and EQ comparisons for integer, swapping operands and multiple
   // operations may be required for some comparisons.
   unsigned Opc = (Cond == ISD::SETEQ || Cond == ISD::SETNE) ? X86ISD::PCMPEQ
                                                             : X86ISD::PCMPGT;
   bool Swap = Cond == ISD::SETLT || Cond == ISD::SETULT ||
               Cond == ISD::SETGE || Cond == ISD::SETUGE;
   bool Invert = Cond == ISD::SETNE ||
                 (Cond != ISD::SETEQ && ISD::isTrueWhenEqual(Cond));
 
-  MVT VET = VT.getVectorElementType();
-  bool HasSubus = Subtarget.hasSSE2() && (VET == MVT::i8 || VET == MVT::i16);
-  bool Subus = false;
-  if (HasSubus) {
-    // As another special case, use PSUBUS[BW] when it's profitable. E.g. for
-    // Op0 u<= Op1:
-    //   t = psubus Op0, Op1
-    //   pcmpeq t, <0..0>
-    switch (Cond) {
-    default: break;
-    case ISD::SETULT: {
-      // If the comparison is against a constant we can turn this into a
-      // setule.  With psubus, setule does not require a swap.  This is
-      // beneficial because the constant in the register is no longer
-      // destructed as the destination so it can be hoisted out of a loop.
-      // Only do this pre-AVX since vpcmp* is no longer destructive.
-      if (Subtarget.hasAVX())
-        break;
-      if (SDValue ULEOp1 = ChangeVSETULTtoVSETULE(dl, Op1, DAG)) {
-        Op1 = ULEOp1;
-        Subus = true; Invert = false; Swap = false;
-      }
-      break;
-    }
-    // Psubus is better than flip-sign because it requires no inversion.
-    case ISD::SETUGE: Subus = true; Invert = false; Swap = true;  break;
-    case ISD::SETULE: Subus = true; Invert = false; Swap = false; break;
-    }
-
-    if (Subus) {
-      Opc = X86ISD::SUBUS;
-      FlipSigns = false;
-    }
-  }
-
   if (Swap)
     std::swap(Op0, Op1);
 
   // Check that the operation in question is available (most are plain SSE2,
   // but PCMPGTQ and PCMPEQQ have different requirements).
   if (VT == MVT::v2i64) {
     if (Opc == X86ISD::PCMPGT && !Subtarget.hasSSE42()) {
       assert(Subtarget.hasSSE2() && "Don't know how to lower!");
#endif
   }
 
   SDValue Result = DAG.getNode(Opc, dl, VT, Op0, Op1);
 
   // If the logical-not of the result is required, perform that now.
   if (Invert)
     Result = DAG.getNOT(dl, Result, VT);
 
-  if (Subus)
-    Result = DAG.getNode(X86ISD::PCMPEQ, dl, VT, Result,
-                         getZeroVector(VT, Subtarget, DAG, dl));
-
   return Result;
 }
 
 // Try to select this as a KTEST+SETCC if possible.
 static SDValue EmitKTEST(SDValue Op0, SDValue Op1, ISD::CondCode CC,
                          const SDLoc &dl, SelectionDAG &DAG,
                          const X86Subtarget &Subtarget) {
   // Only support equality comparisons.