AMDGPU/GlobalISel: Don't reject shader types

[arsenm]

Summary

I'm not sure what complications these present, but the current
argument lowering is pretty much directly copied from the DAG
lowering, so I assume these work as they should.

No tests because I'm lazy and things are getting pretty close to the
point where the existing calling-conventions.ll can be shared with
SelectionDAG.

[nhaehnle]

Sure, why not.

[arsenm]

r367870

lib/Target/AMDGPU/AMDGPUCallLowering.cpp

bool AMDGPUCallLowering::lowerFormalArguments(
   CallingConv::ID CC = F.getCallingConv();
 
   // The infrastructure for normal calling convention lowering is essentially
   // useless for kernels. We want to avoid any kind of legalization or argument
   // splitting.
   if (CC == CallingConv::AMDGPU_KERNEL)
     return lowerFormalArgumentsKernel(MIRBuilder, F, VRegs);
 
-  // AMDGPU_GS and AMDGP_HS are not supported yet.
-  if (CC == CallingConv::AMDGPU_GS || CC == CallingConv::AMDGPU_HS)
-    return false;
-
   const bool IsShader = AMDGPU::isShader(CC);
   const bool IsEntryFunc = AMDGPU::isEntryFunctionCC(CC);
 
   MachineFunction &MF = MIRBuilder.getMF();
   MachineBasicBlock &MBB = MIRBuilder.getMBB();
   MachineRegisterInfo &MRI = MF.getRegInfo();
   SIMachineFunctionInfo *Info = MF.getInfo<SIMachineFunctionInfo>();
   const GCNSubtarget &Subtarget = MF.getSubtarget<GCNSubtarget>();