Fix tan, tanl for large inputs

[Andreas Jaeger]

On 03/16/2012 04:02 PM, Joseph S. Myers wrote:
> This patch fixes the problems with range reduction for tan (x86) and
> tanl (x86, x86_64), using C implementations instead of fptan
> instructions and adding such an implementation for ldbl-96 based on
> that for ldbl-128 (again, rational function approximation not
> optimized for ldbl-96, but the representation of pi/4 as a sum of two
> long doubles was recomputed for ldbl-96).
>
> Tested x86 and x86_64 and ulps updated accordingly.

Thanks a lot, this is fine!

Andreas
-- 
  Andreas Jaeger aj@{suse.com,opensuse.org} Twitter/Identica: jaegerandi
   SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
    GF: Jeff Hawn,Jennifer Guild,Felix Imendörffer,HRB16746 (AG Nürnberg)
     GPG fingerprint = 93A3 365E CE47 B889 DF7F  FED1 389A 563C C272 A126