llvm.org GIT mirror llvm / 63abb57
Merging r250324: ------------------------------------------------------------------------ r250324 | wschmidt | 2015-10-14 16:45:00 -0400 (Wed, 14 Oct 2015) | 10 lines [PowerPC] Fix invalid lxvdsx optimization (PR25157) PR25157 identifies a bug where a load plus a vector shuffle is incorrectly converted into an LXVDSX instruction. That optimization is only valid if the load is of a doubleword, and in the noted case, it was not. This corrects that problem. Joint patch with Eric Schweitz, who provided the bugpoint-reduced test case. ------------------------------------------------------------------------ git-svn-id: https://llvm.org/svn/llvm-project/llvm/branches/release_37@252483 91177308-0d34-0410-b5e6-96231b3b80d8 Tom Stellard 3 years ago
2 changed file(s) with 60 addition(s) and 0 deletion(s). Raw diff Collapse all Expand all
27982798 SDValue Base, Offset;
27992799
28002800 if (LD->isUnindexed() &&
2801 (LD->getMemoryVT() == MVT::f64 ||
2802 LD->getMemoryVT() == MVT::i64) &&
28012803 SelectAddrIdxOnly(LD->getBasePtr(), Base, Offset)) {
28022804 SDValue Chain = LD->getChain();
28032805 SDValue Ops[] = { Base, Offset, Chain };
0 ; RUN: llc -mcpu=pwr8 -mtriple=powerpc64le-unknown-linux-gnu < %s | FileCheck %s
1
2 ; Verify correct generation of an lxsspx rather than an invalid optimization
3 ; to lxvdsx. Bugpoint-reduced test from Eric Schweitz.
4
5 %struct.BSS38.51.4488.9911.14348.16813.20264.24701.28152.31603.35054.39491.44914.45407.46393.46886.47872.49351.49844.50830.51323.52309.53295.53788.54281.55267.55760.59211.61625 = type <{ [28 x i8] }>
6 %struct_main1_2_.491.4928.10351.14788.17253.20704.25141.28592.32043.35494.39931.45354.45847.46833.47326.48312.49791.50284.51270.51763.52749.53735.54228.54721.55707.56200.59651.61626 = type <{ [64 x i8] }>
7
8 @.BSS38 = external global %struct.BSS38.51.4488.9911.14348.16813.20264.24701.28152.31603.35054.39491.44914.45407.46393.46886.47872.49351.49844.50830.51323.52309.53295.53788.54281.55267.55760.59211.61625, align 32
9 @_main1_2_ = external global %struct_main1_2_.491.4928.10351.14788.17253.20704.25141.28592.32043.35494.39931.45354.45847.46833.47326.48312.49791.50284.51270.51763.52749.53735.54228.54721.55707.56200.59651.61626, section ".comm", align 16
10
11 define void @aercalc_() {
12 L.entry:
13 br i1 undef, label %L.LB38_2426, label %L.LB38_2911
14
15 L.LB38_2911:
16 br i1 undef, label %L.LB38_2140, label %L.LB38_2640
17
18 L.LB38_2640:
19 unreachable
20
21 L.LB38_2426:
22 br i1 undef, label %L.LB38_2438, label %L.LB38_2920
23
24 L.LB38_2920:
25 br i1 undef, label %L.LB38_2438, label %L.LB38_2921
26
27 L.LB38_2921:
28 br label %L.LB38_2140
29
30 L.LB38_2140:
31 ret void
32
33 L.LB38_2438:
34 br i1 undef, label %L.LB38_2451, label %L.LB38_2935
35
36 L.LB38_2935:
37 br i1 undef, label %L.LB38_2451, label %L.LB38_2936
38
39 L.LB38_2936:
40 unreachable
41
42 L.LB38_2451:
43 br i1 undef, label %L.LB38_2452, label %L.LB38_2937
44
45 L.LB38_2937:
46 unreachable
47
48 L.LB38_2452:
49 %0 = load float, float* bitcast (i8* getelementptr inbounds (%struct.BSS38.51.4488.9911.14348.16813.20264.24701.28152.31603.35054.39491.44914.45407.46393.46886.47872.49351.49844.50830.51323.52309.53295.53788.54281.55267.55760.59211.61625, %struct.BSS38.51.4488.9911.14348.16813.20264.24701.28152.31603.35054.39491.44914.45407.46393.46886.47872.49351.49844.50830.51323.52309.53295.53788.54281.55267.55760.59211.61625* @.BSS38, i64 0, i32 0, i64 16) to float*), align 16
50 %1 = fpext float %0 to double
51 %2 = insertelement <2 x double> undef, double %1, i32 1
52 store <2 x double> %2, <2 x double>* bitcast (i8* getelementptr inbounds (%struct_main1_2_.491.4928.10351.14788.17253.20704.25141.28592.32043.35494.39931.45354.45847.46833.47326.48312.49791.50284.51270.51763.52749.53735.54228.54721.55707.56200.59651.61626, %struct_main1_2_.491.4928.10351.14788.17253.20704.25141.28592.32043.35494.39931.45354.45847.46833.47326.48312.49791.50284.51270.51763.52749.53735.54228.54721.55707.56200.59651.61626* @_main1_2_, i64 0, i32 0, i64 32) to <2 x double>*), align 16
53 unreachable
54 }
55
56 ; CHECK-LABEL: @aercalc_
57 ; CHECK: lxsspx