1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 4
2 ; RUN: llc < %s -mtriple=powerpc64le-unknown-linux-gnu -verify-machineinstrs\
3 ; RUN: -mcpu=pwr9 --ppc-enable-pipeliner 2>&1 | FileCheck %s
5 ; Test that the pipeliner schedules the store instructions correctly. Since
6 ; there is a dependence between the store, they cannot be scheduled further than
7 ; MII cycles/instructions apart. That is, the first store cannot occur multiple
8 ; times before the second ctore in the schedule.
9 define dso_local void @comp_method(ptr noalias nocapture noundef readonly %0, ptr nocapture noundef writeonly %1, ptr nocapture noundef writeonly %2, i32 noundef %3, i32 noundef %4, i32 noundef %5, i32 noundef %6, i64 %v1) local_unnamed_addr {
10 ; CHECK-LABEL: comp_method:
12 ; CHECK-NEXT: extsw 7, 8
13 ; CHECK-NEXT: extsw 8, 9
14 ; CHECK-NEXT: clrldi 9, 6, 32
15 ; CHECK-NEXT: addi 6, 3, -1
17 ; CHECK-NEXT: li 11, 0
18 ; CHECK-NEXT: sradi 12, 11, 2
19 ; CHECK-NEXT: add 5, 5, 8
22 ; CHECK-NEXT: addi 11, 7, 0
23 ; CHECK-NEXT: std 30, -16(1) # 8-byte Folded Spill
24 ; CHECK-NEXT: lbzu 9, 1(6)
25 ; CHECK-NEXT: add 12, 12, 10
26 ; CHECK-NEXT: extsb 9, 9
27 ; CHECK-NEXT: stbx 8, 4, 9
28 ; CHECK-NEXT: add 9, 9, 12
29 ; CHECK-NEXT: bdz .LBB0_2
30 ; CHECK-NEXT: .p2align 4
31 ; CHECK-NEXT: .LBB0_1:
32 ; CHECK-NEXT: lbzu 0, 1(6)
33 ; CHECK-NEXT: sradi 12, 11, 2
34 ; CHECK-NEXT: add 11, 11, 7
35 ; CHECK-NEXT: add 12, 12, 10
36 ; CHECK-NEXT: sldi 30, 9, 2
37 ; CHECK-NEXT: add 9, 9, 30
38 ; CHECK-NEXT: extsb 0, 0
39 ; CHECK-NEXT: stbx 3, 5, 9
40 ; CHECK-NEXT: add 9, 0, 12
41 ; CHECK-NEXT: stbx 8, 4, 0
42 ; CHECK-NEXT: bdnz .LBB0_1
43 ; CHECK-NEXT: .LBB0_2:
44 ; CHECK-NEXT: sldi 4, 9, 2
45 ; CHECK-NEXT: ld 30, -16(1) # 8-byte Folded Reload
46 ; CHECK-NEXT: add 4, 9, 4
47 ; CHECK-NEXT: stbx 3, 5, 4
49 %8 = icmp sgt i32 %3, 64
50 tail call void @llvm.assume(i1 %8)
52 %10 = icmp eq i32 %9, 0
53 tail call void @llvm.assume(i1 %10)
54 %11 = sext i32 %5 to i64
55 %12 = sext i32 %6 to i64
56 %13 = zext nneg i32 %3 to i64
57 %14 = getelementptr i8, ptr %2, i64 %12
64 %17 = phi i64 [ 0, %7 ], [ %24, %16 ]
65 %18 = getelementptr inbounds i8, ptr %0, i64 %17
66 %19 = load i8, ptr %18, align 1
67 %20 = sext i8 %19 to i64
68 %21 = getelementptr inbounds i8, ptr %1, i64 %20
69 store i8 2, ptr %21, align 1
70 %22 = mul nsw i64 %17, %11
72 %a2 = add i64 %a1, %v1
73 %a3 = add i64 %20, %a2
74 %a4 = mul nsw i64 %a3, 5
75 %23 = getelementptr i8, ptr %14, i64 %a4
76 store i8 8, ptr %23, align 1
77 %24 = add nuw nsw i64 %17, 1
78 %25 = icmp eq i64 %24, %13
79 br i1 %25, label %15, label %16
82 declare void @llvm.assume(i1 noundef) #1
84 attributes #1 = { mustprogress nocallback nofree nosync nounwind willreturn memory(inaccessiblemem: write) }