[PowerPC][GISEL]support for float point and integer convertion

[shchenz]

Summary

Add support for fptosi,fptoui,sitofp,uitofp

For now only handle 64 bit integer to make it does not depend on any other patches. 32 bit integer needs handling for G_SEXT/G_ZEXT.

[arsenm]

Seems a shame to have to have all that manual selection for such a simple instruction

llvm/lib/Target/PowerPC/GISel/PPCInstructionSelector.cpp

136

Typo inteter

[shchenz]

In D139174#3966811, @arsenm wrote:

Seems a shame to have to have all that manual selection for such a simple instruction

Thanks for review. Yeah, we can improve this at least for simple patterns when direct move and fpcvt are available on the targets in future.

llvm/lib/Target/PowerPC/GISel/PPCInstructionSelector.cpp

 //===- PPCInstructionSelector.cpp --------------------------------*- C++ -*-==//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 /// \file
 /// This file implements the targeting of the InstructionSelector class for
 /// PowerPC.
 //===----------------------------------------------------------------------===//
 
+#include "PPC.h"
 #include "PPCInstrInfo.h"
 #include "PPCRegisterBankInfo.h"
 #include "PPCSubtarget.h"
 #include "PPCTargetMachine.h"
 #include "llvm/CodeGen/GlobalISel/InstructionSelector.h"
 #include "llvm/CodeGen/GlobalISel/InstructionSelectorImpl.h"
 #include "llvm/CodeGen/MachineFunction.h"
 #include "llvm/IR/IntrinsicsPowerPC.h"
public:
   bool select(MachineInstr &I) override;
   static const char *getName() { return DEBUG_TYPE; }
 
 private:
   /// tblgen generated 'select' implementation that is used as the initial
   /// selector for the patterns that do not require complex C++.
   bool selectImpl(MachineInstr &I, CodeGenCoverage &CoverageInfo) const;
 
+  bool selectFPToInt(MachineInstr &I, MachineBasicBlock &MBB,
+                  MachineRegisterInfo &MRI) const;
+  bool selectIntToFP(MachineInstr &I, MachineBasicBlock &MBB,
+                  MachineRegisterInfo &MRI) const;
+
+  const PPCSubtarget &STI;
   const PPCInstrInfo &TII;
   const PPCRegisterInfo &TRI;
   const PPCRegisterBankInfo &RBI;
 
 #define GET_GLOBALISEL_PREDICATES_DECL
 #include "PPCGenGlobalISel.inc"
 #undef GET_GLOBALISEL_PREDICATES_DECL
 
 #define GET_GLOBALISEL_TEMPORARIES_DECL
 #include "PPCGenGlobalISel.inc"
 #undef GET_GLOBALISEL_TEMPORARIES_DECL
 };
 
 } // end anonymous namespace
 
 #define GET_GLOBALISEL_IMPL
 #include "PPCGenGlobalISel.inc"
 #undef GET_GLOBALISEL_IMPL
 
 PPCInstructionSelector::PPCInstructionSelector(const PPCTargetMachine &TM,
                                                const PPCSubtarget &STI,
                                                const PPCRegisterBankInfo &RBI)
-    : TII(*STI.getInstrInfo()), TRI(*STI.getRegisterInfo()), RBI(RBI),
+    : STI(STI), TII(*STI.getInstrInfo()), TRI(*STI.getRegisterInfo()), RBI(RBI),
 #define GET_GLOBALISEL_PREDICATES_INIT
 #include "PPCGenGlobalISel.inc"
 #undef GET_GLOBALISEL_PREDICATES_INIT
 #define GET_GLOBALISEL_TEMPORARIES_INIT
 #include "PPCGenGlobalISel.inc"
 #undef GET_GLOBALISEL_TEMPORARIES_INIT
 {
 }
if (!RBI.constrainGenericRegister(DstReg, *DstRC, MRI)) {
     LLVM_DEBUG(dbgs() << "Failed to constrain " << TII.getName(I.getOpcode())
                       << " operand\n");
     return false;
   }
 
   return true;
 }
 
+bool PPCInstructionSelector::selectIntToFP(MachineInstr &I,
+                                           MachineBasicBlock &MBB,
+                                           MachineRegisterInfo &MRI) const {
+  if (!STI.hasDirectMove() || !STI.isPPC64() || !STI.hasFPCVT())
+    return false;
+
+  const DebugLoc &DbgLoc = I.getDebugLoc();
+  const Register DstReg = I.getOperand(0).getReg();
+  const Register SrcReg = I.getOperand(1).getReg();
+
+  Register MoveReg = MRI.createVirtualRegister(&PPC::VSFRCRegClass);
+
+  // For now, only handle the case for 64 bit inteter.
+  BuildMI(MBB, I, DbgLoc, TII.get(PPC::MTVSRD), MoveReg).addReg(SrcReg);
+
+  bool IsSingle = MRI.getType(DstReg).getSizeInBits() == 32;
+  bool IsSigned = I.getOpcode() == TargetOpcode::G_SITOFP;
+  unsigned ConvOp = IsSingle ? (IsSigned ? PPC::XSCVSXDSP : PPC::XSCVUXDSP)
+                             : (IsSigned ? PPC::XSCVSXDDP : PPC::XSCVUXDDP);
+
+  MachineInstr *MI =
+      BuildMI(MBB, I, DbgLoc, TII.get(ConvOp), DstReg).addReg(MoveReg);
+
+  I.eraseFromParent();
+  return constrainSelectedInstRegOperands(*MI, TII, TRI, RBI);
+}
+
+bool PPCInstructionSelector::selectFPToInt(MachineInstr &I,
+                                           MachineBasicBlock &MBB,
+                                           MachineRegisterInfo &MRI) const {
+  if (!STI.hasDirectMove() || !STI.isPPC64() || !STI.hasFPCVT())
+    return false;
+
+  const DebugLoc &DbgLoc = I.getDebugLoc();
+  const Register DstReg = I.getOperand(0).getReg();
+  const Register SrcReg = I.getOperand(1).getReg();
+
+  Register CopyReg = MRI.createVirtualRegister(&PPC::VSFRCRegClass);
+  BuildMI(MBB, I, DbgLoc, TII.get(TargetOpcode::COPY), CopyReg).addReg(SrcReg);
+
+  Register ConvReg = MRI.createVirtualRegister(&PPC::VSFRCRegClass);
+
+  bool IsSigned = I.getOpcode() == TargetOpcode::G_FPTOSI;
+
+  // single-precision on PPC is stored as double-precision on PPC in registers,
+  // so always use double-precision convertions.
+  unsigned ConvOp = IsSigned ? PPC::XSCVDPSXDS : PPC::XSCVDPUXDS;
+
+  BuildMI(MBB, I, DbgLoc, TII.get(ConvOp), ConvReg).addReg(CopyReg);
+
+  MachineInstr *MI =
+      BuildMI(MBB, I, DbgLoc, TII.get(PPC::MFVSRD), DstReg).addReg(ConvReg);
+
+  I.eraseFromParent();
+  return constrainSelectedInstRegOperands(*MI, TII, TRI, RBI);
+}
+
 bool PPCInstructionSelector::select(MachineInstr &I) {
   auto &MBB = *I.getParent();
   auto &MF = *MBB.getParent();
   auto &MRI = MF.getRegInfo();
 
   if (!isPreISelGenericOpcode(I.getOpcode())) {
     if (I.isCopy())
       return selectCopy(I, TII, MRI, TRI, RBI);
 
     return true;
   }
 
   if (selectImpl(I, *CoverageInfo))
     return true;
+
+  unsigned Opcode = I.getOpcode();
+
+  switch (Opcode) {
+  default:
+    return false;
+  case TargetOpcode::G_SITOFP:
+  case TargetOpcode::G_UITOFP:
+    return selectIntToFP(I, MBB, MRI);
+  case TargetOpcode::G_FPTOSI:
+  case TargetOpcode::G_FPTOUI:
+    return selectFPToInt(I, MBB, MRI);
+  }
   return false;
 }
 
 namespace llvm {
 InstructionSelector *
 createPPCInstructionSelector(const PPCTargetMachine &TM,
                              const PPCSubtarget &Subtarget,
                              const PPCRegisterBankInfo &RBI) {
   return new PPCInstructionSelector(TM, Subtarget, RBI);
 }
 } // end namespace llvm

llvm/lib/Target/PowerPC/GISel/PPCLegalizerInfo.cpp

getActionDefinitionsBuilder({G_AND, G_OR, G_XOR})
       .clampScalar(0, S64, S64);
   getActionDefinitionsBuilder({G_ADD, G_SUB})
       .legalFor({S64})
       .clampScalar(0, S64, S64);
 
   getActionDefinitionsBuilder({G_FADD, G_FSUB, G_FMUL, G_FDIV})
       .legalFor({S32, S64});
 
+  getActionDefinitionsBuilder({G_FPTOSI, G_FPTOUI})
+      .legalForCartesianProduct({S64}, {S32, S64});
+
+  getActionDefinitionsBuilder({G_SITOFP, G_UITOFP})
+      .legalForCartesianProduct({S32, S64}, {S64});
+
   getLegacyLegalizerInfo().computeTables();
 }

llvm/lib/Target/PowerPC/GISel/PPCRegisterBankInfo.cpp

case TargetOpcode::G_FDIV: {
 
     assert((Size == 32 || Size == 64) && "Unsupported floating point types!\n");
     OperandsMapping = getValueMapping(Size == 32 ? PMI_FPR32 : PMI_FPR64);
     break;
   }
   case TargetOpcode::G_CONSTANT:
     OperandsMapping = getOperandsMapping({getValueMapping(PMI_GPR64), nullptr});
     break;
+  case TargetOpcode::G_FPTOUI:
+  case TargetOpcode::G_FPTOSI: {
+    Register SrcReg = MI.getOperand(1).getReg();
+    unsigned Size = getSizeInBits(SrcReg, MRI, TRI);
+
+    OperandsMapping = getOperandsMapping(
+        {getValueMapping(PMI_GPR64),
+         getValueMapping(Size == 32 ? PMI_FPR32 : PMI_FPR64)});
+    break;
+  }
+  case TargetOpcode::G_UITOFP:
+  case TargetOpcode::G_SITOFP: {
+    Register SrcReg = MI.getOperand(0).getReg();
+    unsigned Size = getSizeInBits(SrcReg, MRI, TRI);
+
+    OperandsMapping =
+        getOperandsMapping({getValueMapping(Size == 32 ? PMI_FPR32 : PMI_FPR64),
+                            getValueMapping(PMI_GPR64)});
+    break;
+  }
   default:
     return getInvalidInstructionMapping();
   }
 
   return getInstructionMapping(MappingID, Cost, OperandsMapping, NumOperands);
 }
 
 RegisterBankInfo::InstructionMappings
 PPCRegisterBankInfo::getInstrAlternativeMappings(const MachineInstr &MI) const {
   // TODO Implement.
   return RegisterBankInfo::getInstrAlternativeMappings(MI);
 }

llvm/test/CodeGen/PowerPC/GlobalISel/float-integer-conv.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+
+; RUN: llc -mtriple=powerpc64le-unknown-linux-gnu -global-isel  -ppc-vsr-nums-as-vr \
+; RUN:   -ppc-asm-full-reg-names -verify-machineinstrs -o - < %s | FileCheck %s
+
+define i64 @fptosi_float_i64(float %i) {
+; CHECK-LABEL: fptosi_float_i64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    xscvdpsxds f0, f1
+; CHECK-NEXT:    mffprd r3, f0
+; CHECK-NEXT:    blr
+entry:
+  %conv = fptosi float %i to i64
+  ret i64 %conv
+}
+
+define i64 @fptosi_double_i64(double %i) {
+; CHECK-LABEL: fptosi_double_i64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    xscvdpsxds f0, f1
+; CHECK-NEXT:    mffprd r3, f0
+; CHECK-NEXT:    blr
+entry:
+  %conv = fptosi double %i to i64
+  ret i64 %conv
+}
+
+define i64 @fptoui_float_i64(float %i) {
+; CHECK-LABEL: fptoui_float_i64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    xscvdpuxds f0, f1
+; CHECK-NEXT:    mffprd r3, f0
+; CHECK-NEXT:    blr
+entry:
+  %conv = fptoui float %i to i64
+  ret i64 %conv
+}
+
+define i64 @fptoui_double_i64(double %i) {
+; CHECK-LABEL: fptoui_double_i64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    xscvdpuxds f0, f1
+; CHECK-NEXT:    mffprd r3, f0
+; CHECK-NEXT:    blr
+entry:
+  %conv = fptoui double %i to i64
+  ret i64 %conv
+}
+
+define float @sitofp_i64_float(i64 %i) {
+; CHECK-LABEL: sitofp_i64_float:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    mtfprd f0, r3
+; CHECK-NEXT:    xscvsxdsp f1, f0
+; CHECK-NEXT:    blr
+entry:
+  %conv = sitofp i64 %i to float
+  ret float %conv
+}
+
+define double @sitofp_i64_double(i64 %i) {
+; CHECK-LABEL: sitofp_i64_double:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    mtfprd f0, r3
+; CHECK-NEXT:    xscvsxddp f1, f0
+; CHECK-NEXT:    blr
+entry:
+  %conv = sitofp i64 %i to double
+  ret double %conv
+}
+
+define float @uitofp_i64_float(i64 %i) {
+; CHECK-LABEL: uitofp_i64_float:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    mtfprd f0, r3
+; CHECK-NEXT:    xscvuxdsp f1, f0
+; CHECK-NEXT:    blr
+entry:
+  %conv = uitofp i64 %i to float
+  ret float %conv
+}
+
+define double @uitofp_i64_double(i64 %i) {
+; CHECK-LABEL: uitofp_i64_double:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    mtfprd f0, r3
+; CHECK-NEXT:    xscvuxddp f1, f0
+; CHECK-NEXT:    blr
+entry:
+  %conv = uitofp i64 %i to double
+  ret double %conv
+}