TableGen: Support physical register inputs > 255

This was truncating register value that didn't fit in unsigned char.
Switch AMDGPU sendmsg intrinsics to using a tablegen pattern.

llvm-svn: 366695

llvm/include/llvm/CodeGen/SelectionDAGISel.h

@@ -162,6 +162,7 @@ public:
     OPC_EmitMergeInputChains1_1,
     OPC_EmitMergeInputChains1_2,
     OPC_EmitCopyToReg,
+    OPC_EmitCopyToReg2,
     OPC_EmitNodeXForm,
     OPC_EmitNode,
     // Space-optimized forms that implicitly encode number of result VTs.

llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp

@@ -3323,10 +3323,13 @@ void SelectionDAGISel::SelectCodeCommon(SDNode *NodeToMatch,
       continue;
     }
 
-    case OPC_EmitCopyToReg: {
+    case OPC_EmitCopyToReg:
+    case OPC_EmitCopyToReg2: {
       unsigned RecNo = MatcherTable[MatcherIndex++];
       assert(RecNo < RecordedNodes.size() && "Invalid EmitCopyToReg");
       unsigned DestPhysReg = MatcherTable[MatcherIndex++];
+      if (Opcode == OPC_EmitCopyToReg2)
+        DestPhysReg |= MatcherTable[MatcherIndex++] << 8;
 
       if (!InputChain.getNode())
         InputChain = CurDAG->getEntryNode();

llvm/utils/TableGen/DAGISelMatcher.h

@@ -932,13 +932,15 @@ private:
 ///
 class EmitCopyToRegMatcher : public Matcher {
   unsigned SrcSlot; // Value to copy into the physreg.
-  Record *DestPhysReg;
+  const CodeGenRegister *DestPhysReg;
+
 public:
-  EmitCopyToRegMatcher(unsigned srcSlot, Record *destPhysReg)
+  EmitCopyToRegMatcher(unsigned srcSlot,
+                       const CodeGenRegister *destPhysReg)
     : Matcher(EmitCopyToReg), SrcSlot(srcSlot), DestPhysReg(destPhysReg) {}
 
   unsigned getSrcSlot() const { return SrcSlot; }
-  Record *getDestPhysReg() const { return DestPhysReg; }
+  const CodeGenRegister *getDestPhysReg() const { return DestPhysReg; }
 
   static bool classof(const Matcher *N) {
     return N->getKind() == EmitCopyToReg;

llvm/utils/TableGen/DAGISelMatcherEmitter.cpp

@@ -670,12 +670,22 @@ EmitMatcher(const Matcher *N, unsigned Indent, unsigned CurrentIdx,
     OS << '\n';
     return 2+MN->getNumNodes();
   }
-  case Matcher::EmitCopyToReg:
-    OS << "OPC_EmitCopyToReg, "
-       << cast<EmitCopyToRegMatcher>(N)->getSrcSlot() << ", "
-       << getQualifiedName(cast<EmitCopyToRegMatcher>(N)->getDestPhysReg())
-       << ",\n";
-    return 3;
+  case Matcher::EmitCopyToReg: {
+    const auto *C2RMatcher = cast<EmitCopyToRegMatcher>(N);
+    int Bytes = 3;
+    const CodeGenRegister *Reg = C2RMatcher->getDestPhysReg();
+    if (Reg->EnumValue > 255) {
+      assert(isUInt<16>(Reg->EnumValue) && "not handled");
+      OS << "OPC_EmitCopyToReg2, " << C2RMatcher->getSrcSlot() << ", "
+         << "TARGET_VAL(" << getQualifiedName(Reg->TheDef) << "),\n";
+      ++Bytes;
+    } else {
+      OS << "OPC_EmitCopyToReg, " << C2RMatcher->getSrcSlot() << ", "
+         << getQualifiedName(Reg->TheDef) << ",\n";
+    }
+
+    return Bytes;
+  }
   case Matcher::EmitNodeXForm: {
     const EmitNodeXFormMatcher *XF = cast<EmitNodeXFormMatcher>(N);
     OS << "OPC_EmitNodeXForm, " << getNodeXFormID(XF->getNodeXForm()) << ", "

llvm/utils/TableGen/DAGISelMatcherGen.cpp

@@ -867,9 +867,13 @@ EmitResultInstructionAsOperand(const TreePatternNode *N,
   if (isRoot && !PhysRegInputs.empty()) {
     // Emit all of the CopyToReg nodes for the input physical registers.  These
     // occur in patterns like (mul:i8 AL:i8, GR8:i8:$src).
-    for (unsigned i = 0, e = PhysRegInputs.size(); i != e; ++i)
+    for (unsigned i = 0, e = PhysRegInputs.size(); i != e; ++i) {
+      const CodeGenRegister *Reg =
+        CGP.getTargetInfo().getRegBank().getReg(PhysRegInputs[i].first);
       AddMatcher(new EmitCopyToRegMatcher(PhysRegInputs[i].second,
-                                          PhysRegInputs[i].first));
+                                          Reg));
+    }
+
     // Even if the node has no other glue inputs, the resultant node must be
     // glued to the CopyFromReg nodes we just generated.
     TreeHasInGlue = true;