// Flat Instructions
//===----------------------------------------------------------------------===//
-def FLAT_LOAD_UBYTE : FLAT_Load_Helper <0x8, "flat_load_ubyte", VGPR_32>;
-def FLAT_LOAD_SBYTE : FLAT_Load_Helper <0x9, "flat_load_sbyte", VGPR_32>;
-def FLAT_LOAD_USHORT : FLAT_Load_Helper <0xa, "flat_load_ushort", VGPR_32>;
-def FLAT_LOAD_SSHORT : FLAT_Load_Helper <0xb, "flat_load_sshort", VGPR_32>;
-def FLAT_LOAD_DWORD : FLAT_Load_Helper <0xc, "flat_load_dword", VGPR_32>;
-def FLAT_LOAD_DWORDX2 : FLAT_Load_Helper <0xd, "flat_load_dwordx2", VReg_64>;
-def FLAT_LOAD_DWORDX4 : FLAT_Load_Helper <0xe, "flat_load_dwordx4", VReg_128>;
-def FLAT_LOAD_DWORDX3 : FLAT_Load_Helper <0xf, "flat_load_dwordx3", VReg_96>;
-def FLAT_STORE_BYTE : FLAT_Store_Helper <0x18, "flat_store_byte", VGPR_32>;
-def FLAT_STORE_SHORT : FLAT_Store_Helper <0x1a, "flat_store_short", VGPR_32>;
-def FLAT_STORE_DWORD : FLAT_Store_Helper <0x1c, "flat_store_dword", VGPR_32>;
-def FLAT_STORE_DWORDX2 : FLAT_Store_Helper <
- 0x1d, "flat_store_dwordx2", VReg_64
->;
-def FLAT_STORE_DWORDX4 : FLAT_Store_Helper <
- 0x1e, "flat_store_dwordx4", VReg_128
->;
-def FLAT_STORE_DWORDX3 : FLAT_Store_Helper <
- 0x1f, "flat_store_dwordx3", VReg_96
->;
-defm FLAT_ATOMIC_SWAP : FLAT_ATOMIC <0x30, "flat_atomic_swap", VGPR_32>;
+defm FLAT_LOAD_UBYTE : FLAT_Load_Helper <
+ flat<0x8, 0x10>, "flat_load_ubyte", VGPR_32
+>;
+defm FLAT_LOAD_SBYTE : FLAT_Load_Helper <
+ flat<0x9, 0x11>, "flat_load_sbyte", VGPR_32
+>;
+defm FLAT_LOAD_USHORT : FLAT_Load_Helper <
+ flat<0xa, 0x12>, "flat_load_ushort", VGPR_32
+>;
+defm FLAT_LOAD_SSHORT : FLAT_Load_Helper <
+ flat<0xb, 0x13>, "flat_load_sshort", VGPR_32>
+;
+defm FLAT_LOAD_DWORD : FLAT_Load_Helper <
+ flat<0xc, 0x14>, "flat_load_dword", VGPR_32
+>;
+defm FLAT_LOAD_DWORDX2 : FLAT_Load_Helper <
+ flat<0xd, 0x15>, "flat_load_dwordx2", VReg_64
+>;
+defm FLAT_LOAD_DWORDX4 : FLAT_Load_Helper <
+ flat<0xe, 0x17>, "flat_load_dwordx4", VReg_128
+>;
+defm FLAT_LOAD_DWORDX3 : FLAT_Load_Helper <
+ flat<0xf, 0x16>, "flat_load_dwordx3", VReg_96
+>;
+defm FLAT_STORE_BYTE : FLAT_Store_Helper <
+ flat<0x18>, "flat_store_byte", VGPR_32
+>;
+defm FLAT_STORE_SHORT : FLAT_Store_Helper <
+ flat <0x1a>, "flat_store_short", VGPR_32
+>;
+defm FLAT_STORE_DWORD : FLAT_Store_Helper <
+ flat<0x1c>, "flat_store_dword", VGPR_32
+>;
+defm FLAT_STORE_DWORDX2 : FLAT_Store_Helper <
+ flat<0x1d>, "flat_store_dwordx2", VReg_64
+>;
+defm FLAT_STORE_DWORDX4 : FLAT_Store_Helper <
+ flat<0x1e, 0x1f>, "flat_store_dwordx4", VReg_128
+>;
+defm FLAT_STORE_DWORDX3 : FLAT_Store_Helper <
+ flat<0x1f, 0x1e>, "flat_store_dwordx3", VReg_96
+>;
+defm FLAT_ATOMIC_SWAP : FLAT_ATOMIC <
+ flat<0x30, 0x40>, "flat_atomic_swap", VGPR_32
+>;
defm FLAT_ATOMIC_CMPSWAP : FLAT_ATOMIC <
- 0x31, "flat_atomic_cmpswap", VGPR_32, VReg_64
->;
-defm FLAT_ATOMIC_ADD : FLAT_ATOMIC <0x32, "flat_atomic_add", VGPR_32>;
-defm FLAT_ATOMIC_SUB : FLAT_ATOMIC <0x33, "flat_atomic_sub", VGPR_32>;
-defm FLAT_ATOMIC_RSUB : FLAT_ATOMIC <0x34, "flat_atomic_rsub", VGPR_32>;
-defm FLAT_ATOMIC_SMIN : FLAT_ATOMIC <0x35, "flat_atomic_smin", VGPR_32>;
-defm FLAT_ATOMIC_UMIN : FLAT_ATOMIC <0x36, "flat_atomic_umin", VGPR_32>;
-defm FLAT_ATOMIC_SMAX : FLAT_ATOMIC <0x37, "flat_atomic_smax", VGPR_32>;
-defm FLAT_ATOMIC_UMAX : FLAT_ATOMIC <0x38, "flat_atomic_umax", VGPR_32>;
-defm FLAT_ATOMIC_AND : FLAT_ATOMIC <0x39, "flat_atomic_and", VGPR_32>;
-defm FLAT_ATOMIC_OR : FLAT_ATOMIC <0x3a, "flat_atomic_or", VGPR_32>;
-defm FLAT_ATOMIC_XOR : FLAT_ATOMIC <0x3b, "flat_atomic_xor", VGPR_32>;
-defm FLAT_ATOMIC_INC : FLAT_ATOMIC <0x3c, "flat_atomic_inc", VGPR_32>;
-defm FLAT_ATOMIC_DEC : FLAT_ATOMIC <0x3d, "flat_atomic_dec", VGPR_32>;
-defm FLAT_ATOMIC_FCMPSWAP : FLAT_ATOMIC <
- 0x3e, "flat_atomic_fcmpswap", VGPR_32, VReg_64
+ flat<0x31, 0x41>, "flat_atomic_cmpswap", VGPR_32, VReg_64
+>;
+defm FLAT_ATOMIC_ADD : FLAT_ATOMIC <
+ flat<0x32, 0x42>, "flat_atomic_add", VGPR_32
+>;
+defm FLAT_ATOMIC_SUB : FLAT_ATOMIC <
+ flat<0x33, 0x43>, "flat_atomic_sub", VGPR_32
+>;
+defm FLAT_ATOMIC_SMIN : FLAT_ATOMIC <
+ flat<0x35, 0x44>, "flat_atomic_smin", VGPR_32
+>;
+defm FLAT_ATOMIC_UMIN : FLAT_ATOMIC <
+ flat<0x36, 0x45>, "flat_atomic_umin", VGPR_32
+>;
+defm FLAT_ATOMIC_SMAX : FLAT_ATOMIC <
+ flat<0x37, 0x46>, "flat_atomic_smax", VGPR_32
+>;
+defm FLAT_ATOMIC_UMAX : FLAT_ATOMIC <
+ flat<0x38, 0x47>, "flat_atomic_umax", VGPR_32
+>;
+defm FLAT_ATOMIC_AND : FLAT_ATOMIC <
+ flat<0x39, 0x48>, "flat_atomic_and", VGPR_32
+>;
+defm FLAT_ATOMIC_OR : FLAT_ATOMIC <
+ flat<0x3a, 0x49>, "flat_atomic_or", VGPR_32
+>;
+defm FLAT_ATOMIC_XOR : FLAT_ATOMIC <
+ flat<0x3b, 0x4a>, "flat_atomic_xor", VGPR_32
+>;
+defm FLAT_ATOMIC_INC : FLAT_ATOMIC <
+ flat<0x3c, 0x4b>, "flat_atomic_inc", VGPR_32
+>;
+defm FLAT_ATOMIC_DEC : FLAT_ATOMIC <
+ flat<0x3d, 0x4c>, "flat_atomic_dec", VGPR_32
+>;
+defm FLAT_ATOMIC_SWAP_X2 : FLAT_ATOMIC <
+ flat<0x50, 0x60>, "flat_atomic_swap_x2", VReg_64
>;
-defm FLAT_ATOMIC_FMIN : FLAT_ATOMIC <0x3f, "flat_atomic_fmin", VGPR_32>;
-defm FLAT_ATOMIC_FMAX : FLAT_ATOMIC <0x40, "flat_atomic_fmax", VGPR_32>;
-defm FLAT_ATOMIC_SWAP_X2 : FLAT_ATOMIC <0x50, "flat_atomic_swap_x2", VReg_64>;
defm FLAT_ATOMIC_CMPSWAP_X2 : FLAT_ATOMIC <
- 0x51, "flat_atomic_cmpswap_x2", VReg_64, VReg_128
->;
-defm FLAT_ATOMIC_ADD_X2 : FLAT_ATOMIC <0x52, "flat_atomic_add_x2", VReg_64>;
-defm FLAT_ATOMIC_SUB_X2 : FLAT_ATOMIC <0x53, "flat_atomic_sub_x2", VReg_64>;
-defm FLAT_ATOMIC_RSUB_X2 : FLAT_ATOMIC <0x54, "flat_atomic_rsub_x2", VReg_64>;
-defm FLAT_ATOMIC_SMIN_X2 : FLAT_ATOMIC <0x55, "flat_atomic_smin_x2", VReg_64>;
-defm FLAT_ATOMIC_UMIN_X2 : FLAT_ATOMIC <0x56, "flat_atomic_umin_x2", VReg_64>;
-defm FLAT_ATOMIC_SMAX_X2 : FLAT_ATOMIC <0x57, "flat_atomic_smax_x2", VReg_64>;
-defm FLAT_ATOMIC_UMAX_X2 : FLAT_ATOMIC <0x58, "flat_atomic_umax_x2", VReg_64>;
-defm FLAT_ATOMIC_AND_X2 : FLAT_ATOMIC <0x59, "flat_atomic_and_x2", VReg_64>;
-defm FLAT_ATOMIC_OR_X2 : FLAT_ATOMIC <0x5a, "flat_atomic_or_x2", VReg_64>;
-defm FLAT_ATOMIC_XOR_X2 : FLAT_ATOMIC <0x5b, "flat_atomic_xor_x2", VReg_64>;
-defm FLAT_ATOMIC_INC_X2 : FLAT_ATOMIC <0x5c, "flat_atomic_inc_x2", VReg_64>;
-defm FLAT_ATOMIC_DEC_X2 : FLAT_ATOMIC <0x5d, "flat_atomic_dec_x2", VReg_64>;
-defm FLAT_ATOMIC_FCMPSWAP_X2 : FLAT_ATOMIC <
- 0x5e, "flat_atomic_fcmpswap_x2", VReg_64, VReg_128
+ flat<0x51, 0x61>, "flat_atomic_cmpswap_x2", VReg_64, VReg_128
+>;
+defm FLAT_ATOMIC_ADD_X2 : FLAT_ATOMIC <
+ flat<0x52, 0x62>, "flat_atomic_add_x2", VReg_64
+>;
+defm FLAT_ATOMIC_SUB_X2 : FLAT_ATOMIC <
+ flat<0x53, 0x63>, "flat_atomic_sub_x2", VReg_64
+>;
+defm FLAT_ATOMIC_SMIN_X2 : FLAT_ATOMIC <
+ flat<0x55, 0x64>, "flat_atomic_smin_x2", VReg_64
+>;
+defm FLAT_ATOMIC_UMIN_X2 : FLAT_ATOMIC <
+ flat<0x56, 0x65>, "flat_atomic_umin_x2", VReg_64
+>;
+defm FLAT_ATOMIC_SMAX_X2 : FLAT_ATOMIC <
+ flat<0x57, 0x66>, "flat_atomic_smax_x2", VReg_64
+>;
+defm FLAT_ATOMIC_UMAX_X2 : FLAT_ATOMIC <
+ flat<0x58, 0x67>, "flat_atomic_umax_x2", VReg_64
+>;
+defm FLAT_ATOMIC_AND_X2 : FLAT_ATOMIC <
+ flat<0x59, 0x68>, "flat_atomic_and_x2", VReg_64
+>;
+defm FLAT_ATOMIC_OR_X2 : FLAT_ATOMIC <
+ flat<0x5a, 0x69>, "flat_atomic_or_x2", VReg_64
+>;
+defm FLAT_ATOMIC_XOR_X2 : FLAT_ATOMIC <
+ flat<0x5b, 0x6a>, "flat_atomic_xor_x2", VReg_64
+>;
+defm FLAT_ATOMIC_INC_X2 : FLAT_ATOMIC <
+ flat<0x5c, 0x6b>, "flat_atomic_inc_x2", VReg_64
+>;
+defm FLAT_ATOMIC_DEC_X2 : FLAT_ATOMIC <
+ flat<0x5d, 0x6c>, "flat_atomic_dec_x2", VReg_64
>;
-defm FLAT_ATOMIC_FMIN_X2 : FLAT_ATOMIC <0x5f, "flat_atomic_fmin_x2", VReg_64>;
-defm FLAT_ATOMIC_FMAX_X2 : FLAT_ATOMIC <0x60, "flat_atomic_fmax_x2", VReg_64>;
} // End SubtargetPredicate = isCIVI
-//===----------------------------------------------------------------------===//
-// Flat Patterns
-//===----------------------------------------------------------------------===//
+// CI Only flat instructions
-let Predicates = [HasFlatAddressSpace] in {
+let SubtargetPredicate = isCI, VIAssemblerPredicate = DisableInst in {
-class FLATLoad_Pattern <FLAT Instr_ADDR64, ValueType vt,
- PatFrag flat_ld> :
- Pat <(vt (flat_ld i64:$ptr)),
- (Instr_ADDR64 $ptr, 0, 0, 0)
+defm FLAT_ATOMIC_FCMPSWAP : FLAT_ATOMIC <
+ flat<0x3e>, "flat_atomic_fcmpswap", VGPR_32, VReg_64
+>;
+defm FLAT_ATOMIC_FMIN : FLAT_ATOMIC <
+ flat<0x3f>, "flat_atomic_fmin", VGPR_32
+>;
+defm FLAT_ATOMIC_FMAX : FLAT_ATOMIC <
+ flat<0x40>, "flat_atomic_fmax", VGPR_32
+>;
+defm FLAT_ATOMIC_FCMPSWAP_X2 : FLAT_ATOMIC <
+ flat<0x5e>, "flat_atomic_fcmpswap_x2", VReg_64, VReg_128
+>;
+defm FLAT_ATOMIC_FMIN_X2 : FLAT_ATOMIC <
+ flat<0x5f>, "flat_atomic_fmin_x2", VReg_64
+>;
+defm FLAT_ATOMIC_FMAX_X2 : FLAT_ATOMIC <
+ flat<0x60>, "flat_atomic_fmax_x2", VReg_64
>;
-def : FLATLoad_Pattern <FLAT_LOAD_SBYTE, i32, sextloadi8_flat>;
-def : FLATLoad_Pattern <FLAT_LOAD_UBYTE, i32, az_extloadi8_flat>;
-def : FLATLoad_Pattern <FLAT_LOAD_SSHORT, i32, sextloadi16_flat>;
-def : FLATLoad_Pattern <FLAT_LOAD_USHORT, i32, az_extloadi16_flat>;
-def : FLATLoad_Pattern <FLAT_LOAD_DWORD, i32, flat_load>;
-def : FLATLoad_Pattern <FLAT_LOAD_DWORDX2, i64, flat_load>;
-def : FLATLoad_Pattern <FLAT_LOAD_DWORDX2, i64, az_extloadi32_flat>;
-def : FLATLoad_Pattern <FLAT_LOAD_DWORDX2, v2i32, flat_load>;
-def : FLATLoad_Pattern <FLAT_LOAD_DWORDX4, v4i32, flat_load>;
-
-class FLATStore_Pattern <FLAT Instr, ValueType vt, PatFrag st> :
- Pat <(st vt:$value, i64:$ptr),
- (Instr $value, $ptr, 0, 0, 0)
- >;
-
-def : FLATStore_Pattern <FLAT_STORE_BYTE, i32, truncstorei8_flat>;
-def : FLATStore_Pattern <FLAT_STORE_SHORT, i32, truncstorei16_flat>;
-def : FLATStore_Pattern <FLAT_STORE_DWORD, i32, flat_store>;
-def : FLATStore_Pattern <FLAT_STORE_DWORDX2, i64, flat_store>;
-def : FLATStore_Pattern <FLAT_STORE_DWORDX2, v2i32, flat_store>;
-def : FLATStore_Pattern <FLAT_STORE_DWORDX4, v4i32, flat_store>;
-
-} // End HasFlatAddressSpace predicate
+} // End let SubtargetPredicate = isCI, VIAssemblerPredicate = DisableInst
let Predicates = [isCI] in {
>;
} // End Predicates = [isCI]
+
+
+//===----------------------------------------------------------------------===//
+// Flat Patterns
+//===----------------------------------------------------------------------===//
+
+let Predicates = [isCIVI] in {
+
+// Patterns for global loads with no offset
+class FlatLoadPat <FLAT inst, SDPatternOperator node, ValueType vt> : Pat <
+ (vt (node i64:$addr)),
+ (inst $addr, 0, 0, 0)
+>;
+
+def : FlatLoadPat <FLAT_LOAD_UBYTE, flat_az_extloadi8, i32>;
+def : FlatLoadPat <FLAT_LOAD_SBYTE, flat_sextloadi8, i32>;
+def : FlatLoadPat <FLAT_LOAD_USHORT, flat_az_extloadi16, i32>;
+def : FlatLoadPat <FLAT_LOAD_SSHORT, flat_sextloadi16, i32>;
+def : FlatLoadPat <FLAT_LOAD_DWORD, flat_load, i32>;
+def : FlatLoadPat <FLAT_LOAD_DWORDX2, flat_load, v2i32>;
+def : FlatLoadPat <FLAT_LOAD_DWORDX4, flat_load, v4i32>;
+
+class FlatStorePat <FLAT inst, SDPatternOperator node, ValueType vt> : Pat <
+ (node vt:$data, i64:$addr),
+ (inst $data, $addr, 0, 0, 0)
+>;
+
+def : FlatStorePat <FLAT_STORE_BYTE, flat_truncstorei8, i32>;
+def : FlatStorePat <FLAT_STORE_SHORT, flat_truncstorei16, i32>;
+def : FlatStorePat <FLAT_STORE_DWORD, flat_store, i32>;
+def : FlatStorePat <FLAT_STORE_DWORDX2, flat_store, v2i32>;
+def : FlatStorePat <FLAT_STORE_DWORDX4, flat_store, v4i32>;
+
+class FlatAtomicPat <FLAT inst, SDPatternOperator node, ValueType vt> : Pat <
+ (vt (node i64:$addr, vt:$data)),
+ (inst $addr, $data, 0, 0)
+>;
+
+def : FlatAtomicPat <FLAT_ATOMIC_ADD_RTN, atomic_add_global, i32>;
+def : FlatAtomicPat <FLAT_ATOMIC_AND_RTN, atomic_and_global, i32>;
+def : FlatAtomicPat <FLAT_ATOMIC_SUB_RTN, atomic_sub_global, i32>;
+def : FlatAtomicPat <FLAT_ATOMIC_SMAX_RTN, atomic_max_global, i32>;
+def : FlatAtomicPat <FLAT_ATOMIC_UMAX_RTN, atomic_umax_global, i32>;
+def : FlatAtomicPat <FLAT_ATOMIC_SMIN_RTN, atomic_min_global, i32>;
+def : FlatAtomicPat <FLAT_ATOMIC_UMIN_RTN, atomic_umin_global, i32>;
+def : FlatAtomicPat <FLAT_ATOMIC_OR_RTN, atomic_or_global, i32>;
+def : FlatAtomicPat <FLAT_ATOMIC_SWAP_RTN, atomic_swap_global, i32>;
+def : FlatAtomicPat <FLAT_ATOMIC_XOR_RTN, atomic_xor_global, i32>;
+
+} // End Predicates = [isCIVI]
projects / oota-llvm.git / blobdiff