[VE] Support copy of vector mask registers

[kaz7]

Summary

Support VM and VMP registers in copyPhysReg() function. Also add
regression tests.

llvm/lib/Target/VE/VEInstrInfo.cpp

for (unsigned Idx = 0; Idx != NumSubRegs; ++Idx) {
     Register SubSrc = TRI->getSubReg(SrcReg, SubRegIdx[Idx]);
     assert(SubDest && SubSrc && "Bad sub-register");
 
     if (MCID.getOpcode() == VE::ORri) {
       // generate "ORri, dest, src, 0" instruction.
       MachineInstrBuilder MIB =
           BuildMI(MBB, I, DL, MCID, SubDest).addReg(SubSrc).addImm(0);
       MovMI = MIB.getInstr();
+    } else if (MCID.getOpcode() == VE::ANDMmm) {
+      // generate "ANDM, dest, vm0, src" instruction.
+      MachineInstrBuilder MIB =
+          BuildMI(MBB, I, DL, MCID, SubDest).addReg(VE::VM0).addReg(SubSrc);
+      MovMI = MIB.getInstr();
     } else {
       llvm_unreachable("Unexpected reg-to-reg copy instruction");
     }
   }
   // Add implicit super-register defs and kills to the last MovMI.
   MovMI->addRegisterDefined(DestReg, TRI);
   if (KillSrc)
     MovMI->addRegisterKilled(SrcReg, TRI, true);
BuildMI(MBB, I, DL, get(VE::LEAzii), TmpReg)
         .addImm(0)
         .addImm(0)
         .addImm(256);
     MachineInstrBuilder MIB = BuildMI(MBB, I, DL, get(VE::VORmvl), DestReg)
                                   .addImm(M1(0)) // Represent (0)1.
                                   .addReg(SrcReg, getKillRegState(KillSrc))
                                   .addReg(SubTmp, getKillRegState(true));
     MIB.getInstr()->addRegisterKilled(TmpReg, TRI, true);
+  } else if (VE::VMRegClass.contains(DestReg, SrcReg)) {
+    BuildMI(MBB, I, DL, get(VE::ANDMmm), DestReg)
+        .addReg(VE::VM0)
+        .addReg(SrcReg, getKillRegState(KillSrc));
+  } else if (VE::VM512RegClass.contains(DestReg, SrcReg)) {
+    // Use two instructions.
+    const unsigned SubRegIdx[] = {VE::sub_vm_even, VE::sub_vm_odd};
+    unsigned int NumSubRegs = 2;
+    copyPhysSubRegs(MBB, I, DL, DestReg, SrcReg, KillSrc, get(VE::ANDMmm),
+                    NumSubRegs, SubRegIdx, &getRegisterInfo());
   } else if (VE::F128RegClass.contains(DestReg, SrcReg)) {
     // Use two instructions.
     const unsigned SubRegIdx[] = {VE::sub_even, VE::sub_odd};
     unsigned int NumSubRegs = 2;
     copyPhysSubRegs(MBB, I, DL, DestReg, SrcReg, KillSrc, get(VE::ORri),
                     NumSubRegs, SubRegIdx, &getRegisterInfo());
   } else {
     const TargetRegisterInfo *TRI = &getRegisterInfo();

llvm/test/CodeGen/VE/Vector/fastcc_callee.ll

 ; TODO: Uncomment test when vector loads are upstream (vreg stack passing).
 ; define <256 x i32> @vreg_arg_v256i32_r7(<256 x i32> %p0, <256 x i32> %p1, <256 x i32> %p2, <256 x i32> %p3, <256 x i32> %p4, <256 x i32> %p5, <256 x i32> %p6, <256 x i32> %p7) {
 ;   ret <256 x i32> %p7
 ; }
 
 ; define <256 x i32> @vreg_arg_v256i32_r8(<256 x i32> %p0, <256 x i32> %p1, <256 x i32> %p2, <256 x i32> %p3, <256 x i32> %p4, <256 x i32> %p5, <256 x i32> %p6, <256 x i32> %p7, <256 x i32> %p8) {
 ;   ret <256 x i32> %p8
 ; }
+
+define fastcc <256 x i1> @vreg_arg_v256i1_vm7(<256 x i1> %vm1, <256 x i1> %vm2, <256 x i1> %vm3, <256 x i1> %vm4, <256 x i1> %vm5, <256 x i1> %vm6, <256 x i1> %vm7, <256 x i1> %vm8) {
+; CHECK-LABEL: vreg_arg_v256i1_vm7:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    andm %vm1, %vm0, %vm6
+; CHECK-NEXT:    b.l.t (, %s10)
+  ret <256 x i1> %vm6
+}
+
+define fastcc <512 x i1> @vreg_arg_v512i1_vmp3(<512 x i1> %vmp1, <512 x i1> %vmp2, <512 x i1> %vmp3, <512 x i1> %vmp4) {
+; CHECK-LABEL: vreg_arg_v512i1_vmp3:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    andm %vm2, %vm0, %vm6
+; CHECK-NEXT:    andm %vm3, %vm0, %vm7
+; CHECK-NEXT:    b.l.t (, %s10)
+  ret <512 x i1> %vmp3
+}