1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc -mtriple=riscv64 -mattr=+v -verify-machineinstrs < %s | FileCheck %s --check-prefix NOSUBREG
3 ; RUN: llc -mtriple=riscv64 -mattr=+v -verify-machineinstrs < %s -riscv-enable-subreg-liveness=true | FileCheck %s --check-prefix SUBREG
5 ; This test checks that vrgatherei16 instructions are correctly
6 ; register-allocated. The LMUL=1 destination register groups may not overlap
7 ; with the EMUL=2 source vector register groups.
9 ; FIXME: enabling subregister liveness results in incorrect register
12 define internal void @foo(<vscale x 1 x i16> %v15, <vscale x 1 x i16> %0, <vscale x 1 x i16> %vs12.i.i.i, <vscale x 1 x i16> %1, <vscale x 8 x i8> %v37) {
13 ; NOSUBREG-LABEL: foo:
14 ; NOSUBREG: # %bb.0: # %loopIR.preheader.i.i
15 ; NOSUBREG-NEXT: vsetvli a0, zero, e16, m2, ta, ma
16 ; NOSUBREG-NEXT: vmv.v.i v14, 0
17 ; NOSUBREG-NEXT: vsetvli zero, zero, e8, m1, ta, ma
18 ; NOSUBREG-NEXT: vmv.v.i v9, 0
19 ; NOSUBREG-NEXT: vsetivli zero, 4, e8, m1, tu, ma
20 ; NOSUBREG-NEXT: vmv1r.v v8, v9
21 ; NOSUBREG-NEXT: vrgatherei16.vv v8, v9, v14
22 ; NOSUBREG-NEXT: .LBB0_1: # %loopIR3.i.i
23 ; NOSUBREG-NEXT: # =>This Inner Loop Header: Depth=1
24 ; NOSUBREG-NEXT: vl1r.v v9, (zero)
25 ; NOSUBREG-NEXT: vsetivli zero, 4, e8, m1, tu, ma
26 ; NOSUBREG-NEXT: vmv1r.v v13, v12
27 ; NOSUBREG-NEXT: vrgatherei16.vv v13, v9, v10
28 ; NOSUBREG-NEXT: vsetvli a0, zero, e8, m1, ta, ma
29 ; NOSUBREG-NEXT: vand.vv v9, v8, v13
30 ; NOSUBREG-NEXT: vs1r.v v9, (zero)
31 ; NOSUBREG-NEXT: j .LBB0_1
34 ; SUBREG: # %bb.0: # %loopIR.preheader.i.i
35 ; SUBREG-NEXT: vsetvli a0, zero, e16, m2, ta, ma
36 ; SUBREG-NEXT: vmv.v.i v14, 0
37 ; SUBREG-NEXT: vsetvli zero, zero, e8, m1, ta, ma
38 ; SUBREG-NEXT: vmv.v.i v9, 0
39 ; SUBREG-NEXT: vsetivli zero, 4, e8, m1, tu, ma
40 ; SUBREG-NEXT: vmv1r.v v8, v9
41 ; SUBREG-NEXT: vrgatherei16.vv v8, v9, v14
42 ; SUBREG-NEXT: .LBB0_1: # %loopIR3.i.i
43 ; SUBREG-NEXT: # =>This Inner Loop Header: Depth=1
44 ; SUBREG-NEXT: vl1r.v v9, (zero)
45 ; SUBREG-NEXT: vsetivli zero, 4, e8, m1, tu, ma
46 ; SUBREG-NEXT: vmv1r.v v13, v12
47 ; SUBREG-NEXT: vrgatherei16.vv v13, v9, v10
48 ; SUBREG-NEXT: vsetvli a0, zero, e8, m1, ta, ma
49 ; SUBREG-NEXT: vand.vv v9, v8, v13
50 ; SUBREG-NEXT: vs1r.v v9, (zero)
51 ; SUBREG-NEXT: j .LBB0_1
53 %v18 = tail call <vscale x 8 x i16> @llvm.vector.insert.nxv8i16.nxv1i16(<vscale x 8 x i16> poison, <vscale x 1 x i16> %vs12.i.i.i, i64 0)
56 loopIR3.i.i: ; preds = %loopIR3.i.i, %loopIR.preheader.i.i
57 %v376 = load <vscale x 8 x i8>, ptr addrspace(1) null, align 8
58 %v38 = tail call <vscale x 8 x i8> @llvm.riscv.vrgatherei16.vv.nxv8i8.i64(<vscale x 8 x i8> zeroinitializer, <vscale x 8 x i8> zeroinitializer, <vscale x 8 x i16> zeroinitializer, i64 4)
59 %v40 = tail call <vscale x 8 x i8> @llvm.riscv.vrgatherei16.vv.nxv8i8.i64(<vscale x 8 x i8> %v37, <vscale x 8 x i8> %v376, <vscale x 8 x i16> %v18, i64 4)
60 %v42 = and <vscale x 8 x i8> %v38, %v40
61 store <vscale x 8 x i8> %v42, ptr addrspace(1) null, align 4
65 ; Function Attrs: nocallback nofree nosync nounwind readnone willreturn
66 declare <vscale x 8 x i16> @llvm.vector.insert.nxv8i16.nxv1i16(<vscale x 8 x i16>, <vscale x 1 x i16>, i64 immarg) #0
68 ; Function Attrs: nounwind readnone
69 declare <vscale x 8 x i8> @llvm.riscv.vrgatherei16.vv.nxv8i8.i64(<vscale x 8 x i8>, <vscale x 8 x i8>, <vscale x 8 x i16>, i64) #1
71 attributes #0 = { nocallback nofree nosync nounwind readnone willreturn }
72 attributes #1 = { nounwind readnone }