1; RUN: llc -verify-machineinstrs -mcpu=pwr8 -mtriple=powerpc64le-unknown-linux-gnu < %s | FileCheck %s
2
3%class.FullMatrix = type { i8 }
4%class.Vector = type { float* }
5
6$test = comdat any
7
8define weak_odr void @test(%class.FullMatrix* %this, %class.Vector* dereferenceable(8) %p1, %class.Vector* dereferenceable(8), i1 zeroext) {
9entry:
10  %call = tail call signext i32 @fn1(%class.FullMatrix* %this)
11  %cmp10 = icmp sgt i32 %call, 0
12  br i1 %cmp10, label %for.body.lr.ph, label %for.cond.cleanup
13
14for.body.lr.ph:                                   ; preds = %entry
15  %val.i = getelementptr inbounds %class.Vector, %class.Vector* %p1, i64 0, i32 0
16  %2 = load float*, float** %val.i, align 8
17  %wide.trip.count = zext i32 %call to i64
18  %min.iters.check = icmp ult i32 %call, 4
19  br i1 %min.iters.check, label %for.body.preheader, label %min.iters.checked
20
21for.body.preheader:                               ; preds = %middle.block, %min.iters.checked, %for.body.lr.ph
22  %indvars.iv.ph = phi i64 [ 0, %min.iters.checked ], [ 0, %for.body.lr.ph ], [ %n.vec, %middle.block ]
23  br label %for.body
24
25min.iters.checked:                                ; preds = %for.body.lr.ph
26  %3 = and i32 %call, 3
27  %n.mod.vf = zext i32 %3 to i64
28  %n.vec = sub nsw i64 %wide.trip.count, %n.mod.vf
29  %cmp.zero = icmp eq i64 %n.vec, 0
30  br i1 %cmp.zero, label %for.body.preheader, label %vector.body.preheader
31
32vector.body.preheader:                            ; preds = %min.iters.checked
33  br label %vector.body
34
35vector.body:                                      ; preds = %vector.body.preheader, %vector.body
36  %index = phi i64 [ %index.next, %vector.body ], [ 0, %vector.body.preheader ]
37  %4 = getelementptr inbounds float, float* %2, i64 %index
38  %5 = bitcast float* %4 to <4 x float>*
39  %wide.load = load <4 x float>, <4 x float>* %5, align 4
40  %6 = fpext <4 x float> %wide.load to <4 x ppc_fp128>
41  %7 = fadd <4 x ppc_fp128> %6, %6
42  %8 = fptrunc <4 x ppc_fp128> %7 to <4 x float>
43  %9 = bitcast float* %4 to <4 x float>*
44  store <4 x float> %8, <4 x float>* %9, align 4
45  %index.next = add i64 %index, 4
46  %10 = icmp eq i64 %index.next, %n.vec
47  br i1 %10, label %middle.block, label %vector.body
48
49middle.block:                                     ; preds = %vector.body
50  %cmp.n = icmp eq i32 %3, 0
51  br i1 %cmp.n, label %for.cond.cleanup, label %for.body.preheader
52
53for.cond.cleanup.loopexit:                        ; preds = %for.body
54  br label %for.cond.cleanup
55
56for.cond.cleanup:                                 ; preds = %for.cond.cleanup.loopexit, %middle.block, %entry
57  ret void
58
59for.body:                                         ; preds = %for.body.preheader, %for.body
60  %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ %indvars.iv.ph, %for.body.preheader ]
61  %arrayidx.i = getelementptr inbounds float, float* %2, i64 %indvars.iv
62  %11 = load float, float* %arrayidx.i, align 4
63  %conv = fpext float %11 to ppc_fp128
64  %add = fadd ppc_fp128 %conv, %conv
65  %conv4 = fptrunc ppc_fp128 %add to float
66  store float %conv4, float* %arrayidx.i, align 4
67  %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
68  %exitcond = icmp eq i64 %indvars.iv.next, %wide.trip.count
69  br i1 %exitcond, label %for.cond.cleanup.loopexit, label %for.body
70; CHECK: stfdx
71; CHECK: lxvd2x
72}
73
74declare signext i32 @fn1(%class.FullMatrix*) local_unnamed_addr #1
75