maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

AMDGPU: split ret/noret patterns for global atomics 

Differential Revision: https://reviews.llvm.org/D27989

llvm-svn: 290435
This commit is contained in:
Jan Vesely 2016-12-23 15:34:51 +00:00
parent bbd8536321
commit 206a510e54
3 changed files with 50 additions and 20 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

62
llvm/lib/Target/AMDGPU/AMDGPUInstructions.td
View File

@ -363,24 +363,54 @@ multiclass AtomicCmpSwapLocal <SDNode cmp_swap_node> {
defm atomic_cmp_swap : AtomicCmpSwapLocal <atomic_cmp_swap>;
class global_binary_atomic_op<SDNode atomic_op> : PatFrag<
(ops node:$ptr, node:$value),
(atomic_op node:$ptr, node:$value),
[{return cast<MemSDNode>(N)->getAddressSpace() == AMDGPUAS::GLOBAL_ADDRESS;}]
>;
multiclass global_binary_atomic_op<SDNode atomic_op> {
def "" : PatFrag<
(ops node:$ptr, node:$value),
(atomic_op node:$ptr, node:$value),
[{return cast<MemSDNode>(N)->getAddressSpace() == AMDGPUAS::GLOBAL_ADDRESS;}]>;
def atomic_swap_global : global_binary_atomic_op<atomic_swap>;
def atomic_add_global : global_binary_atomic_op<atomic_load_add>;
def atomic_and_global : global_binary_atomic_op<atomic_load_and>;
def atomic_max_global : global_binary_atomic_op<atomic_load_max>;
def atomic_min_global : global_binary_atomic_op<atomic_load_min>;
def atomic_or_global : global_binary_atomic_op<atomic_load_or>;
def atomic_sub_global : global_binary_atomic_op<atomic_load_sub>;
def atomic_umax_global : global_binary_atomic_op<atomic_load_umax>;
def atomic_umin_global : global_binary_atomic_op<atomic_load_umin>;
def atomic_xor_global : global_binary_atomic_op<atomic_load_xor>;
def _noret : PatFrag<
(ops node:$ptr, node:$value),
(atomic_op node:$ptr, node:$value),
[{return cast<MemSDNode>(N)->getAddressSpace() == AMDGPUAS::GLOBAL_ADDRESS && (SDValue(N, 0).use_empty());}]>;
def atomic_cmp_swap_global : global_binary_atomic_op<AMDGPUatomic_cmp_swap>;
def _ret : PatFrag<
(ops node:$ptr, node:$value),
(atomic_op node:$ptr, node:$value),
[{return cast<MemSDNode>(N)->getAddressSpace() == AMDGPUAS::GLOBAL_ADDRESS && (!SDValue(N, 0).use_empty());}]>;
}
defm atomic_swap_global : global_binary_atomic_op<atomic_swap>;
defm atomic_add_global : global_binary_atomic_op<atomic_load_add>;
defm atomic_and_global : global_binary_atomic_op<atomic_load_and>;
defm atomic_max_global : global_binary_atomic_op<atomic_load_max>;
defm atomic_min_global : global_binary_atomic_op<atomic_load_min>;
defm atomic_or_global : global_binary_atomic_op<atomic_load_or>;
defm atomic_sub_global : global_binary_atomic_op<atomic_load_sub>;
defm atomic_umax_global : global_binary_atomic_op<atomic_load_umax>;
defm atomic_umin_global : global_binary_atomic_op<atomic_load_umin>;
defm atomic_xor_global : global_binary_atomic_op<atomic_load_xor>;
//legacy
def AMDGPUatomic_cmp_swap_global : PatFrag<
(ops node:$ptr, node:$value),
(AMDGPUatomic_cmp_swap node:$ptr, node:$value),
[{return cast<MemSDNode>(N)->getAddressSpace() == AMDGPUAS::GLOBAL_ADDRESS;}]>;
def atomic_cmp_swap_global : PatFrag<
(ops node:$ptr, node:$cmp, node:$value),
(atomic_cmp_swap node:$ptr, node:$cmp, node:$value),
[{return cast<MemSDNode>(N)->getAddressSpace() == AMDGPUAS::GLOBAL_ADDRESS;}]>;
def atomic_cmp_swap_global_noret : PatFrag<
(ops node:$ptr, node:$cmp, node:$value),
(atomic_cmp_swap node:$ptr, node:$cmp, node:$value),
[{return cast<MemSDNode>(N)->getAddressSpace() == AMDGPUAS::GLOBAL_ADDRESS && (SDValue(N, 0).use_empty());}]>;
def atomic_cmp_swap_global_ret : PatFrag<
(ops node:$ptr, node:$cmp, node:$value),
(atomic_cmp_swap node:$ptr, node:$cmp, node:$value),
[{return cast<MemSDNode>(N)->getAddressSpace() == AMDGPUAS::GLOBAL_ADDRESS && (!SDValue(N, 0).use_empty());}]>;
//===----------------------------------------------------------------------===//
// Misc Pattern Fragments

4
llvm/lib/Target/AMDGPU/FLATInstructions.td
View File

@ -372,7 +372,7 @@ 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_CMPSWAP_RTN, atomic_cmp_swap_global, i32, v2i32>;
def : FlatAtomicPat <FLAT_ATOMIC_CMPSWAP_RTN, AMDGPUatomic_cmp_swap_global, i32, v2i32>;
def : FlatAtomicPat <FLAT_ATOMIC_XOR_RTN, atomic_xor_global, i32>;
def : FlatAtomicPat <FLAT_ATOMIC_ADD_X2_RTN, atomic_add_global, i64>;
@ -386,7 +386,7 @@ def : FlatAtomicPat <FLAT_ATOMIC_SMIN_X2_RTN, atomic_min_global, i64>;
def : FlatAtomicPat <FLAT_ATOMIC_UMIN_X2_RTN, atomic_umin_global, i64>;
def : FlatAtomicPat <FLAT_ATOMIC_OR_X2_RTN, atomic_or_global, i64>;
def : FlatAtomicPat <FLAT_ATOMIC_SWAP_X2_RTN, atomic_swap_global, i64>;
def : FlatAtomicPat <FLAT_ATOMIC_CMPSWAP_X2_RTN, atomic_cmp_swap_global, i64, v2i64>;
def : FlatAtomicPat <FLAT_ATOMIC_CMPSWAP_X2_RTN, AMDGPUatomic_cmp_swap_global, i64, v2i64>;
def : FlatAtomicPat <FLAT_ATOMIC_XOR_X2_RTN, atomic_xor_global, i64>;
} // End Predicates = [isCIVI]

4
llvm/lib/Target/AMDGPU/SIInstrInfo.td
View File

@ -94,8 +94,8 @@ def SIpc_add_rel_offset : SDNode<"AMDGPUISD::PC_ADD_REL_OFFSET",
// PatFrags for global memory operations
//===----------------------------------------------------------------------===//
def atomic_inc_global : global_binary_atomic_op<SIatomic_inc>;
def atomic_dec_global : global_binary_atomic_op<SIatomic_dec>;
defm atomic_inc_global : global_binary_atomic_op<SIatomic_inc>;
defm atomic_dec_global : global_binary_atomic_op<SIatomic_dec>;
//===----------------------------------------------------------------------===//
// SDNodes and PatFrag for local loads and stores to enable s_mov_b32 m0, -1