AMDGPU/GlobalISel: Select llvm.amdgcn.sffbh

llvm-svn: 371538

3 files changed, 68 insertions, 2 deletions

diff --git a/llvm/lib/Target/AMDGPU/AMDGPUInstrInfo.td b/llvm/lib/Target/AMDGPU/AMDGPUInstrInfo.td
index f205f16ef35..9b5976318e4 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUInstrInfo.td
+++ b/llvm/lib/Target/AMDGPU/AMDGPUInstrInfo.td
@@ -284,7 +284,7 @@ def AMDGPUbfi : SDNode<"AMDGPUISD::BFI", AMDGPUDTIntTernaryOp>;
 def AMDGPUbfm : SDNode<"AMDGPUISD::BFM", SDTIntBinOp>;
 
 def AMDGPUffbh_u32 : SDNode<"AMDGPUISD::FFBH_U32", SDTIntUnaryOp>;
-def AMDGPUffbh_i32 : SDNode<"AMDGPUISD::FFBH_I32", SDTIntUnaryOp>;
+def AMDGPUffbh_i32_impl : SDNode<"AMDGPUISD::FFBH_I32", SDTIntUnaryOp>;
 
 def AMDGPUffbl_b32 : SDNode<"AMDGPUISD::FFBL_B32", SDTIntUnaryOp>;
 
@@ -445,3 +445,7 @@ def AMDGPUfp_class : PatFrags<(ops node:$src0, node:$src1),
 def AMDGPUfmed3 : PatFrags<(ops node:$src0, node:$src1, node:$src2),
   [(int_amdgcn_fmed3 node:$src0, node:$src1, node:$src2),
    (AMDGPUfmed3_impl node:$src0, node:$src1, node:$src2)]>;
+
+def AMDGPUffbh_i32 : PatFrags<(ops node:$src),
+  [(int_amdgcn_sffbh node:$src),
+   (AMDGPUffbh_i32_impl node:$src)]>;
diff --git a/llvm/lib/Target/AMDGPU/VOP1Instructions.td b/llvm/lib/Target/AMDGPU/VOP1Instructions.td
index 553e55ee41b..76abda9218f 100644
--- a/llvm/lib/Target/AMDGPU/VOP1Instructions.td
+++ b/llvm/lib/Target/AMDGPU/VOP1Instructions.td
@@ -237,7 +237,7 @@ defm V_NOT_B32 : VOP1Inst <"v_not_b32", VOP_I32_I32>;
 defm V_BFREV_B32 : VOP1Inst <"v_bfrev_b32", VOP_I32_I32, bitreverse>;
 defm V_FFBH_U32 : VOP1Inst <"v_ffbh_u32", VOP_I32_I32>;
 defm V_FFBL_B32 : VOP1Inst <"v_ffbl_b32", VOP_I32_I32>;
-defm V_FFBH_I32 : VOP1Inst <"v_ffbh_i32", VOP_I32_I32>;
+defm V_FFBH_I32 : VOP1Inst <"v_ffbh_i32", VOP_I32_I32, AMDGPUffbh_i32>;
 
 let SchedRW = [WriteDoubleAdd] in {
 defm V_FREXP_EXP_I32_F64 : VOP1Inst <"v_frexp_exp_i32_f64", VOP_I32_F64, int_amdgcn_frexp_exp>;
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-amdgcn.sffbh.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-amdgcn.sffbh.mir
new file mode 100644
index 00000000000..84e52db1508
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-amdgcn.sffbh.mir
@@ -0,0 +1,62 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -march=amdgcn -mcpu=tahiti -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck %s
+
+---
+name: sffbh_s32_ss
+legalized: true
+regBankSelected: true
+tracksRegLiveness: true
+
+body: |
+  bb.0:
+    liveins: $sgpr0
+
+    ; CHECK-LABEL: name: sffbh_s32_ss
+    ; CHECK: liveins: $sgpr0
+    ; CHECK: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr0
+    ; CHECK: [[S_FLBIT_I32_:%[0-9]+]]:sreg_32 = S_FLBIT_I32 [[COPY]]
+    ; CHECK: S_ENDPGM 0, implicit [[S_FLBIT_I32_]]
+    %0:sgpr(s32) = COPY $sgpr0
+    %1:sgpr(s32) = G_INTRINSIC intrinsic(@llvm.amdgcn.sffbh), %0
+    S_ENDPGM 0, implicit %1
+...
+
+---
+name: sffbh_s32_vs
+legalized: true
+regBankSelected: true
+tracksRegLiveness: true
+
+body: |
+  bb.0:
+    liveins: $sgpr0
+
+    ; CHECK-LABEL: name: sffbh_s32_vs
+    ; CHECK: liveins: $sgpr0
+    ; CHECK: [[COPY:%[0-9]+]]:sreg_32_xm0 = COPY $sgpr0
+    ; CHECK: [[V_FFBH_I32_e64_:%[0-9]+]]:vgpr_32 = V_FFBH_I32_e64 [[COPY]], implicit $exec
+    ; CHECK: S_ENDPGM 0, implicit [[V_FFBH_I32_e64_]]
+    %0:sgpr(s32) = COPY $sgpr0
+    %1:vgpr(s32) = G_INTRINSIC intrinsic(@llvm.amdgcn.sffbh), %0
+    S_ENDPGM 0, implicit %1
+...
+
+---
+name: sffbh_s32_vv
+legalized: true
+regBankSelected: true
+tracksRegLiveness: true
+
+body: |
+  bb.0:
+    liveins: $vgpr0
+
+    ; CHECK-LABEL: name: sffbh_s32_vv
+    ; CHECK: liveins: $vgpr0
+    ; CHECK: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
+    ; CHECK: [[V_FFBH_I32_e64_:%[0-9]+]]:vgpr_32 = V_FFBH_I32_e64 [[COPY]], implicit $exec
+    ; CHECK: S_ENDPGM 0, implicit [[V_FFBH_I32_e64_]]
+    %0:vgpr(s32) = COPY $vgpr0
+    %1:vgpr(s32) = G_INTRINSIC intrinsic(@llvm.amdgcn.sffbh), %0
+    S_ENDPGM 0, implicit %1
+...