Bug 15429

Summary:	ldbl-128 cos() inaccurate new pi/2
Product:	glibc	Reporter:	David S. Miller <davem>
Component:	math	Assignee:	Not yet assigned to anyone <unassigned>
Status:	RESOLVED FIXED
Severity:	normal	CC:	carlos
Priority:	P2	Flags:	fweimer: security-
Version:	unspecified
Target Milestone:	---
Host:		Target:
Build:		Last reconfirmed:

Description David S. Miller 2013-05-02 20:34:51 UTC

On sparc, which uses ldbl-128, the tests using cos() on pi/2 give
inaccurate results.  For example:

Test "cos (pi/2) == 0x1.cd129024e088a67cc74020bbea64p-115":
ildouble: 267676
ldouble: 267676
...
Test "sincos (pi/2, &sin_res, &cos_res) puts 0x1.cd129024e088a67cc74020bbea64p-115 in cos_res":
ildouble: 267676
ldouble: 267676
...
Function "cos":
ildouble: 267676
ldouble: 267676
...
Function "sincos":
ildouble: 267676
ldouble: 267676

Comment 1 Carlos O'Donell 2013-05-09 13:14:50 UTC

I'll be disabling this test for ldbl-128 until we fix this.

Comment 2 Joseph Myers 2013-05-09 21:31:18 UTC

Fixed for 2.18 by:

commit d0213cd0b629cae95ed9607fcf26793a3a0d686c
Author: Joseph Myers <joseph@codesourcery.com>
Date:   Thu May 9 21:28:54 2013 +0000

    Fix ldbl-128 cos range reduction near pi/2 (bug 15429).