AMDGPU/GlobalISel: Avoid repeating 32-bit type lists

[kerbowa]

LGTM

[arsenm]

r371156

lib/Target/AMDGPU/BUFInstructions.td

 defm : MUBUFScratchLoadPat <BUFFER_LOAD_SBYTE_OFFEN, BUFFER_LOAD_SBYTE_OFFSET, i16, sextloadi8_private>;
 defm : MUBUFScratchLoadPat <BUFFER_LOAD_UBYTE_OFFEN, BUFFER_LOAD_UBYTE_OFFSET, i16, extloadi8_private>;
 defm : MUBUFScratchLoadPat <BUFFER_LOAD_UBYTE_OFFEN, BUFFER_LOAD_UBYTE_OFFSET, i16, zextloadi8_private>;
 defm : MUBUFScratchLoadPat <BUFFER_LOAD_SSHORT_OFFEN, BUFFER_LOAD_SSHORT_OFFSET, i32, sextloadi16_private>;
 defm : MUBUFScratchLoadPat <BUFFER_LOAD_USHORT_OFFEN, BUFFER_LOAD_USHORT_OFFSET, i32, extloadi16_private>;
 defm : MUBUFScratchLoadPat <BUFFER_LOAD_USHORT_OFFEN, BUFFER_LOAD_USHORT_OFFSET, i32, zextloadi16_private>;
 defm : MUBUFScratchLoadPat <BUFFER_LOAD_USHORT_OFFEN, BUFFER_LOAD_USHORT_OFFSET, i16, load_private>;
 
-foreach vt = [i32, f32, v2i16, v2f16, p2, p3, p5, p6] in {
+foreach vt = Reg32Types.types in {
 defm : MUBUFScratchLoadPat <BUFFER_LOAD_DWORD_OFFEN, BUFFER_LOAD_DWORD_OFFSET, i32, load_private>;
 }
 defm : MUBUFScratchLoadPat <BUFFER_LOAD_DWORDX2_OFFEN, BUFFER_LOAD_DWORDX2_OFFSET, v2i32, load_private>;
 defm : MUBUFScratchLoadPat <BUFFER_LOAD_DWORDX3_OFFEN, BUFFER_LOAD_DWORDX3_OFFSET, v3i32, load_private>;
 defm : MUBUFScratchLoadPat <BUFFER_LOAD_DWORDX4_OFFEN, BUFFER_LOAD_DWORDX4_OFFSET, v4i32, load_private>;
 
 let OtherPredicates = [D16PreservesUnusedBits] in {
 defm : MUBUFScratchLoadPat_D16<BUFFER_LOAD_SHORT_D16_HI_OFFEN, BUFFER_LOAD_SHORT_D16_HI_OFFSET, v2i16, load_d16_hi_private>;
multiclass MUBUFScratchStorePat <MUBUF_Pseudo InstrOffen,
   >;
 }
 
 defm : MUBUFScratchStorePat <BUFFER_STORE_BYTE_OFFEN, BUFFER_STORE_BYTE_OFFSET, i32, truncstorei8_private>;
 defm : MUBUFScratchStorePat <BUFFER_STORE_SHORT_OFFEN, BUFFER_STORE_SHORT_OFFSET, i32, truncstorei16_private>;
 defm : MUBUFScratchStorePat <BUFFER_STORE_BYTE_OFFEN, BUFFER_STORE_BYTE_OFFSET, i16, truncstorei8_private>;
 defm : MUBUFScratchStorePat <BUFFER_STORE_SHORT_OFFEN, BUFFER_STORE_SHORT_OFFSET, i16, store_private>;
 
-foreach vt = [i32, f32, v2i16, v2f16, p2, p3, p5, p6] in {
+foreach vt = Reg32Types.types in {
 defm : MUBUFScratchStorePat <BUFFER_STORE_DWORD_OFFEN, BUFFER_STORE_DWORD_OFFSET, vt, store_private>;
 }
 
 defm : MUBUFScratchStorePat <BUFFER_STORE_DWORDX2_OFFEN, BUFFER_STORE_DWORDX2_OFFSET, v2i32, store_private, VReg_64>;
 defm : MUBUFScratchStorePat <BUFFER_STORE_DWORDX3_OFFEN, BUFFER_STORE_DWORDX3_OFFSET, v3i32, store_private, VReg_96>;
 defm : MUBUFScratchStorePat <BUFFER_STORE_DWORDX4_OFFEN, BUFFER_STORE_DWORDX4_OFFSET, v4i32, store_private, VReg_128>;
 
 

lib/Target/AMDGPU/DSInstructions.td

 defm : DSReadPat_mc <DS_READ_U8,  i16, "extloadi8_local">;
 defm : DSReadPat_mc <DS_READ_U8,  i16, "zextloadi8_local">;
 defm : DSReadPat_mc <DS_READ_I16, i32, "sextloadi16_local">;
 defm : DSReadPat_mc <DS_READ_I16, i32, "sextloadi16_local">;
 defm : DSReadPat_mc <DS_READ_U16, i32, "extloadi16_local">;
 defm : DSReadPat_mc <DS_READ_U16, i32, "zextloadi16_local">;
 defm : DSReadPat_mc <DS_READ_U16, i16, "load_local">;
 
-foreach vt = [i32, f32, v2i16, v2f16, p2, p3, p5, p6] in {
+foreach vt = Reg32Types.types in {
 defm : DSReadPat_mc <DS_READ_B32, vt, "load_local">;
 }
 
 defm : DSReadPat_mc <DS_READ_B32, i32, "atomic_load_32_local">;
 defm : DSReadPat_mc <DS_READ_B64, i64, "atomic_load_64_local">;
 
 let AddedComplexity = 100 in {
 

lib/Target/AMDGPU/FLATInstructions.td

 def : FlatLoadPat <FLAT_LOAD_DWORDX4, load_flat, v4i32>;
 
 def : FlatLoadAtomicPat <FLAT_LOAD_DWORD, atomic_load_32_flat, i32>;
 def : FlatLoadAtomicPat <FLAT_LOAD_DWORDX2, atomic_load_64_flat, i64>;
 
 def : FlatStorePat <FLAT_STORE_BYTE, truncstorei8_flat, i32>;
 def : FlatStorePat <FLAT_STORE_SHORT, truncstorei16_flat, i32>;
 
-foreach vt = [i32, f32, v2i16, v2f16, p2, p3, p5, p6] in {
+foreach vt = Reg32Types.types in {
 def : FlatLoadPat <FLAT_LOAD_DWORD, load_flat, vt>;
 def : FlatStorePat <FLAT_STORE_DWORD, store_flat, vt>;
 }
 
 foreach vt = VReg_64.RegTypes in {
 def : FlatStorePat <FLAT_STORE_DWORDX2, store_flat, vt, VReg_64>;
 def : FlatLoadPat <FLAT_LOAD_DWORDX2, load_flat, vt>;
 }

 def : FlatLoadSignedPat <GLOBAL_LOAD_UBYTE, extloadi8_global, i16>;
 def : FlatLoadSignedPat <GLOBAL_LOAD_UBYTE, zextloadi8_global, i16>;
 def : FlatLoadSignedPat <GLOBAL_LOAD_SBYTE, sextloadi8_global, i16>;
 def : FlatLoadSignedPat <GLOBAL_LOAD_USHORT, extloadi16_global, i32>;
 def : FlatLoadSignedPat <GLOBAL_LOAD_USHORT, zextloadi16_global, i32>;
 def : FlatLoadSignedPat <GLOBAL_LOAD_SSHORT, sextloadi16_global, i32>;
 def : FlatLoadSignedPat <GLOBAL_LOAD_USHORT, load_global, i16>;
 
-foreach vt = [i32, f32, v2i16, v2f16, p2, p3, p5, p6] in {
+foreach vt = Reg32Types.types in {
 def : FlatLoadSignedPat <GLOBAL_LOAD_DWORD, load_global, vt>;
 def : FlatStoreSignedPat <GLOBAL_STORE_DWORD, store_global, vt, VGPR_32>;
 }
 
 foreach vt = VReg_64.RegTypes in {
 def : FlatLoadSignedPat <GLOBAL_LOAD_DWORDX2, load_global, vt>;
 def : FlatStoreSignedPat <GLOBAL_STORE_DWORDX2, store_global, vt, VReg_64>;
 }

lib/Target/AMDGPU/SIRegisterInfo.td

 
 def TTMP0_TTMP1_TTMP2_TTMP3_TTMP4_TTMP5_TTMP6_TTMP7_TTMP8_TTMP9_TTMP10_TTMP11_TTMP12_TTMP13_TTMP14_TTMP15_gfx9_gfx10 :
   TmpRegTuplesBase<0, 16,
                    [TTMP0_gfx9_gfx10, TTMP1_gfx9_gfx10, TTMP2_gfx9_gfx10, TTMP3_gfx9_gfx10,
                     TTMP4_gfx9_gfx10, TTMP5_gfx9_gfx10, TTMP6_gfx9_gfx10, TTMP7_gfx9_gfx10,
                     TTMP8_gfx9_gfx10, TTMP9_gfx9_gfx10, TTMP10_gfx9_gfx10, TTMP11_gfx9_gfx10,
                     TTMP12_gfx9_gfx10, TTMP13_gfx9_gfx10, TTMP14_gfx9_gfx10, TTMP15_gfx9_gfx10]>;
 
+class RegisterTypes<list<ValueType> reg_types> {
+  list<ValueType> types = reg_types;
+}
+
+def Reg16Types : RegisterTypes<[i16, f16]>;
+def Reg32Types : RegisterTypes<[i32, f32, v2i16, v2f16, p2, p3, p5, p6]>;
+
+
 // VGPR 32-bit registers
 // i16/f16 only on VI+
-def VGPR_32 : RegisterClass<"AMDGPU", [i32, f32, i16, f16, v2i16, v2f16, p2, p3, p5, p6], 32,
+def VGPR_32 : RegisterClass<"AMDGPU", !listconcat(Reg32Types.types, Reg16Types.types), 32,
                             (add (sequence "VGPR%u", 0, 255))> {
   let AllocationPriority = 1;
   let Size = 32;
 }
 
 // VGPR 64-bit registers
 def VGPR_64 : SIRegisterTuples<getSubRegs<2>.ret, VGPR_32, 255, 1, 2, "v">;