Convert sqrt functions into sqrt instructions when -ffast-math is in effect.

When -ffast-math is in effect (on Linux, at least), clang defines
__FINITE_MATH_ONLY__ > 0 when including <math.h>. This causes the
preprocessor to include <bits/math-finite.h>, which renames the sqrt functions.
For instance, "sqrt" is renamed as "__sqrt_finite". 

This patch adds the 3 new names in such a way that they will be treated
as equivalent to their respective original names.





git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@182739 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Preston Gurd 2013-05-27 15:44:35 +00:00
parent f594e41ae9
commit b704d23062
4 changed files with 74 additions and 0 deletions

View File

@ -57,6 +57,12 @@ namespace llvm {
dunder_isoc99_sscanf,
/// void *__memcpy_chk(void *s1, const void *s2, size_t n, size_t s1size);
memcpy_chk,
/// double __sqrt_finite(double x);
sqrt_finite,
/// float __sqrt_finite(float x);
sqrtf_finite,
/// long double __sqrt_finite(long double x);
sqrtl_finite,
/// char * __strdup(const char *s);
dunder_strdup,
/// char *__strndup(const char *s, size_t n);
@ -677,6 +683,8 @@ public:
case LibFunc::sin: case LibFunc::sinf: case LibFunc::sinl:
case LibFunc::cos: case LibFunc::cosf: case LibFunc::cosl:
case LibFunc::sqrt: case LibFunc::sqrtf: case LibFunc::sqrtl:
case LibFunc::sqrt_finite: case LibFunc::sqrtf_finite:
case LibFunc::sqrtl_finite:
case LibFunc::floor: case LibFunc::floorf: case LibFunc::floorl:
case LibFunc::nearbyint: case LibFunc::nearbyintf: case LibFunc::nearbyintl:
case LibFunc::ceil: case LibFunc::ceilf: case LibFunc::ceill:

View File

@ -5562,6 +5562,9 @@ void SelectionDAGBuilder::visitCall(const CallInst &I) {
case LibFunc::sqrt:
case LibFunc::sqrtf:
case LibFunc::sqrtl:
case LibFunc::sqrt_finite:
case LibFunc::sqrtf_finite:
case LibFunc::sqrtl_finite:
if (visitUnaryFloatCall(I, ISD::FSQRT))
return;
break;

View File

@ -43,6 +43,9 @@ const char* TargetLibraryInfo::StandardNames[LibFunc::NumLibFuncs] =
"__isoc99_scanf",
"__isoc99_sscanf",
"__memcpy_chk",
"__sqrt_finite",
"__sqrtf_finite",
"__sqrtl_finite",
"__strdup",
"__strndup",
"__strtok_r",

View File

@ -0,0 +1,60 @@
; RUN: llc < %s -mcpu=core2 | FileCheck %s
; generated using "clang -S -O2 -ffast-math -emit-llvm sqrt.c" from
; #include <math.h>
;
; double fd(double d){
; return sqrt(d);
; }
;
; float ff(float f){
; return sqrtf(f);
; }
;
; long double fld(long double ld){
; return sqrtl(ld);
; }
;
; Tests conversion of sqrt function calls into sqrt instructions when
; -ffast-math is in effect.
; ModuleID = 'sqrt.c'
target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"
; Function Attrs: nounwind readnone uwtable
define double @fd(double %d) #0 {
entry:
; CHECK: sqrtsd
%call = tail call double @__sqrt_finite(double %d) #2
ret double %call
}
; Function Attrs: nounwind readnone
declare double @__sqrt_finite(double) #1
; Function Attrs: nounwind readnone uwtable
define float @ff(float %f) #0 {
entry:
; CHECK: sqrtss
%call = tail call float @__sqrtf_finite(float %f) #2
ret float %call
}
; Function Attrs: nounwind readnone
declare float @__sqrtf_finite(float) #1
; Function Attrs: nounwind readnone uwtable
define x86_fp80 @fld(x86_fp80 %ld) #0 {
entry:
; CHECK: fsqrt
%call = tail call x86_fp80 @__sqrtl_finite(x86_fp80 %ld) #2
ret x86_fp80 %call
}
; Function Attrs: nounwind readnone
declare x86_fp80 @__sqrtl_finite(x86_fp80) #1
attributes #0 = { nounwind readnone uwtable "less-precise-fpmad"="false" "no-frame-pointer-elim"="false" "no-frame-pointer-elim-non-leaf"="false" "no-infs-fp-math"="true" "no-nans-fp-math"="true" "unsafe-fp-math"="true" "use-soft-float"="false" }
attributes #1 = { nounwind readnone "less-precise-fpmad"="false" "no-frame-pointer-elim"="false" "no-frame-pointer-elim-non-leaf"="false" "no-infs-fp-math"="true" "no-nans-fp-math"="true" "unsafe-fp-math"="true" "use-soft-float"="false" }
attributes #2 = { nounwind readnone }