| Eric Christopher | cee313d | 2019-04-17 04:52:47 +0000 | [diff] [blame] | 1 | ; RUN: opt -S -instcombine -o - %s | FileCheck %s |
| 2 | |
| 3 | ; Test that fast math lib call simplification of double math function to float |
| 4 | ; equivalent doesn't occur when the calling function matches the float |
| 5 | ; equivalent math function. Otherwise this can cause the generation of infinite |
| 6 | ; loops when compiled with -O2/3 and fast math. |
| 7 | |
| 8 | ; Test case C source: |
| 9 | ; |
| 10 | ; extern double exp(double x); |
| 11 | ; inline float expf(float x) { return (float) exp((double) x); } |
| 12 | ; float fn(float f) { return expf(f); } |
| 13 | ; |
| 14 | ; IR generated with command: |
| 15 | ; |
| 16 | ; clang -cc1 -O2 -ffast-math -emit-llvm -disable-llvm-passes -triple x86_64-unknown-unknown -o - <srcfile> |
| 17 | |
| 18 | target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" |
| 19 | target triple = "x86_64-unknown-unknown" |
| 20 | |
| 21 | ; Function Attrs: nounwind |
| 22 | define float @fn(float %f) #0 { |
| 23 | ; CHECK: define float @fn( |
| 24 | ; CHECK: call fast float @expf( |
| 25 | %f.addr = alloca float, align 4 |
| 26 | store float %f, float* %f.addr, align 4, !tbaa !1 |
| 27 | %1 = load float, float* %f.addr, align 4, !tbaa !1 |
| 28 | %call = call fast float @expf(float %1) #3 |
| 29 | ret float %call |
| 30 | } |
| 31 | |
| 32 | ; Function Attrs: inlinehint nounwind readnone |
| 33 | define available_externally float @expf(float %x) #1 { |
| 34 | ; CHECK: define available_externally float @expf( |
| 35 | ; CHECK: fpext float |
| 36 | ; CHECK: call fast double @exp( |
| 37 | ; CHECK: fptrunc double |
| 38 | %x.addr = alloca float, align 4 |
| 39 | store float %x, float* %x.addr, align 4, !tbaa !1 |
| 40 | %1 = load float, float* %x.addr, align 4, !tbaa !1 |
| 41 | %conv = fpext float %1 to double |
| 42 | %call = call fast double @exp(double %conv) #3 |
| 43 | %conv1 = fptrunc double %call to float |
| 44 | ret float %conv1 |
| 45 | } |
| 46 | |
| 47 | ; Function Attrs: nounwind readnone |
| 48 | declare double @exp(double) #2 |
| 49 | |
| 50 | attributes #0 = { nounwind "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "no-frame-pointer-elim"="false" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-features"="+mmx,+sse,+sse2,+x87" "unsafe-fp-math"="false" "use-soft-float"="false" } |
| 51 | attributes #1 = { inlinehint nounwind readnone "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "no-frame-pointer-elim"="false" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-features"="+mmx,+sse,+sse2,+x87" "unsafe-fp-math"="false" "use-soft-float"="false" } |
| 52 | attributes #2 = { nounwind readnone } |
| 53 | |
| 54 | !llvm.ident = !{!0} |
| 55 | |
| 56 | !0 = !{!"clang version 5.0.0"} |
| 57 | !1 = !{!2, !2, i64 0} |
| 58 | !2 = !{!"float", !3, i64 0} |
| 59 | !3 = !{!"omnipotent char", !4, i64 0} |
| 60 | !4 = !{!"Simple C/C++ TBAA"} |