Bug 13854

Summary: tan() is incorrect for large inputs on x86_64 and x86
Product: glibc Reporter: Andreas Jaeger <aj>
Component: mathAssignee: Not yet assigned to anyone <unassigned>
Status: RESOLVED FIXED    
Severity: normal CC: aj, bugdal, carlos, ppluzhnikov, vincent-srcware
Priority: P2 Keywords: glibc_2.15
Version: 2.13Flags: fweimer: security-
Target Milestone: ---   
Host: Target:
Build: Last reconfirmed:
Bug Depends on: 13658, 15563    
Bug Blocks:    

Description Andreas Jaeger 2012-03-15 15:09:05 UTC 
+++ This bug was initially created as a clone of Bug #13658 +++

tan is inaccurate, see bug #13658 for a reference.

Test program:
#define _GNU_SOURCE
#include <stdio.h>
#include <math.h>


int main (void)
{
  volatile float xf = 1e22;
  float s1f, s2f;
  volatile double xd = 1e22;
  double s1d, s2d;
  volatile long double xl = 1e22;
  long double s1l, s2l;

  s1d = tan (xd);
  printf ("double: \n");
  printf ("x = %.17g\n", xd);
  printf ("s1 = %.17g\n", s1d);


  s1f = tanf (xf);
  printf ("\nfloat: \n");
  printf ("x = %.17g\n", xf);
  printf ("s1 = %.17g\n", s1f);

  s1l = tanl (xl);
  printf ("\nlong double: \n");
  printf ("x = %.17Lg\n", xl);
  printf ("s1 = %.17Lg\n", s1l);


  return 0;
}


On x86-64:
double: 
x = 1e+22
s1 = -1.6287782256068988

float: 
x = 9.9999997781963084e+21
s1 = -1.0810239315032959

long double: 
x = 1e+22
s1 = -0.52180662513577755

And on x86:
double: 
x = 1e+22
s1 = -0.52180662513577758

float: 
x = 9.9999997781963084e+21
s1 = -0.29305437207221985

long double: 
x = 1e+22
s1 = -0.52180662513577755

The correct result of tan (1e22) is:
s1 = -1.6287782256068988
Comment 1 Joseph Myers 2012-03-16 20:11:02 UTC 
Fixed by:

commit 11b90b9f504df5b2da91ce3a06c1657d99e4a95f
Author: Joseph Myers <joseph@codesourcery.com>
Date:   Fri Mar 16 20:05:04 2012 +0000

    Fix tan, tanl for large inputs.