This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: Fix csqrt overflow/underflow (bug 13841)
On Tuesday, March 13, 2012 22:30:12 Joseph S. Myers wrote:
> This patch fixes overflow and underflow in csqrt functions. These use
> hypot internally, which inevitably means overflows for large inputs
> and loss of precision for subnormal inputs, even though the final
> correct results of csqrt will not overflow. The fix is to scale the
> inputs appropriately before calling hypot, then scale the result
> afterwards, in the potentially affected cases.
>
> This patch depends on my hypotf fix
> <http://sourceware.org/ml/libc-alpha/2012-03/msg00357.html> (that is,
> some new tests will fail without that fix and it's through those
> failures that I found the hypotf bug).
>
> Tested x86_64 and x86 and ulps updated based on that testing.
Thanks, this is ok,
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