xref: /llvm-project/llvm/test/Transforms/CodeGenPrepare/AMDGPU/bypass-slow-div-debug-info.ll (revision f5b5a30858f32e237636acd296b6d0f87c1dfe97)
1; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -codegenprepare %s | FileCheck %s
3; Make sure BypassSlowDivision doesn't drop debug info
4
5define i64 @sdiv64(i64 %a, i64 %b) {
6; CHECK-LABEL: @sdiv64(
7; CHECK-NEXT:    [[TMP1:%.*]] = or i64 [[A:%.*]], [[B:%.*]], !dbg !6
8; CHECK-NEXT:    [[TMP2:%.*]] = and i64 [[TMP1]], -4294967296, !dbg !6
9; CHECK-NEXT:    [[TMP3:%.*]] = icmp eq i64 [[TMP2]], 0, !dbg !6
10; CHECK-NEXT:    br i1 [[TMP3]], label [[TMP4:%.*]], label [[TMP9:%.*]], !dbg !6
11; CHECK:       4:
12; CHECK-NEXT:    [[TMP5:%.*]] = trunc i64 [[B]] to i32, !dbg !6
13; CHECK-NEXT:    [[TMP6:%.*]] = trunc i64 [[A]] to i32, !dbg !6
14; CHECK-NEXT:    [[TMP7:%.*]] = udiv i32 [[TMP6]], [[TMP5]], !dbg !6
15; CHECK-NEXT:    [[TMP8:%.*]] = zext i32 [[TMP7]] to i64, !dbg !6
16; CHECK-NEXT:    br label [[TMP11:%.*]], !dbg !6
17; CHECK:       9:
18; CHECK-NEXT:    [[TMP10:%.*]] = sdiv i64 [[A]], [[B]], !dbg !6
19; CHECK-NEXT:    br label [[TMP11]], !dbg !6
20; CHECK:       11:
21; CHECK-NEXT:    [[TMP12:%.*]] = phi i64 [ [[TMP8]], [[TMP4]] ], [ [[TMP10]], [[TMP9]] ], !dbg !6
22; CHECK-NEXT:    ret i64 [[TMP12]]
23;
24  %d = sdiv i64 %a, %b, !dbg !6
25  ret i64 %d
26}
27
28; FIXME: The debugloc for the rem parts end up with the dbg of the
29; division.
30define <2 x i64> @sdivrem64(i64 %a, i64 %b) {
31; CHECK-LABEL: @sdivrem64(
32; CHECK-NEXT:    [[TMP1:%.*]] = or i64 [[A:%.*]], [[B:%.*]], !dbg !6
33; CHECK-NEXT:    [[TMP2:%.*]] = and i64 [[TMP1]], -4294967296, !dbg !6
34; CHECK-NEXT:    [[TMP3:%.*]] = icmp eq i64 [[TMP2]], 0, !dbg !6
35; CHECK-NEXT:    br i1 [[TMP3]], label [[TMP4:%.*]], label [[TMP11:%.*]], !dbg !6
36; CHECK:       4:
37; CHECK-NEXT:    [[TMP5:%.*]] = trunc i64 [[B]] to i32, !dbg !6
38; CHECK-NEXT:    [[TMP6:%.*]] = trunc i64 [[A]] to i32, !dbg !6
39; CHECK-NEXT:    [[TMP7:%.*]] = udiv i32 [[TMP6]], [[TMP5]], !dbg !6
40; CHECK-NEXT:    [[TMP8:%.*]] = urem i32 [[TMP6]], [[TMP5]], !dbg !6
41; CHECK-NEXT:    [[TMP9:%.*]] = zext i32 [[TMP7]] to i64, !dbg !6
42; CHECK-NEXT:    [[TMP10:%.*]] = zext i32 [[TMP8]] to i64, !dbg !6
43; CHECK-NEXT:    br label [[TMP14:%.*]], !dbg !6
44; CHECK:       11:
45; CHECK-NEXT:    [[TMP12:%.*]] = sdiv i64 [[A]], [[B]], !dbg !6
46; CHECK-NEXT:    [[TMP13:%.*]] = srem i64 [[A]], [[B]], !dbg !6
47; CHECK-NEXT:    br label [[TMP14]], !dbg !6
48; CHECK:       14:
49; CHECK-NEXT:    [[TMP15:%.*]] = phi i64 [ [[TMP9]], [[TMP4]] ], [ [[TMP12]], [[TMP11]] ], !dbg !6
50; CHECK-NEXT:    [[TMP16:%.*]] = phi i64 [ [[TMP10]], [[TMP4]] ], [ [[TMP13]], [[TMP11]] ], !dbg !6
51; CHECK-NEXT:    [[INS0:%.*]] = insertelement <2 x i64> undef, i64 [[TMP15]], i32 0
52; CHECK-NEXT:    [[INS1:%.*]] = insertelement <2 x i64> [[INS0]], i64 [[TMP16]], i32 1
53; CHECK-NEXT:    ret <2 x i64> [[INS1]]
54;
55  %d = sdiv i64 %a, %b, !dbg !6
56  %r = srem i64 %a, %b, !dbg !10
57  %ins0 = insertelement <2 x i64> undef, i64 %d, i32 0
58  %ins1 = insertelement <2 x i64> %ins0, i64 %r, i32 1
59  ret <2 x i64> %ins1
60}
61
62!llvm.dbg.cu = !{!0}
63!llvm.module.flags = !{!3, !4}
64!llvm.ident = !{!5}
65
66!0 = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 3.5 ", isOptimized: false, runtimeVersion: 0, emissionKind: NoDebug, enums: !2, retainedTypes: !2, globals: !2, imports: !2)
67!1 = !DIFile(filename: "basic.c", directory: ".")
68!2 = !{}
69!3 = !{i32 2, !"Dwarf Version", i32 4}
70!4 = !{i32 1, !"Debug Info Version", i32 3}
71!5 = !{!"clang version 3.5 "}
72!6 = !DILocation(line: 3, scope: !7)
73!7 = distinct !DILexicalBlock(scope: !8, file: !1, line: 3)
74!8 = distinct !DISubprogram(name: "foo", scope: !1, file: !1, line: 1, type: !9, scopeLine: 1, virtualIndex: 6, flags: DIFlagPrototyped, spFlags: DISPFlagDefinition, unit: !0, retainedNodes: !2)
75!9 = !DISubroutineType(types: !2)
76!10 = !DILocation(line: 4, scope: !7)
77