This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [COMMITTED] tile: define __NO_LONG_DOUBLE_MATH
- From: Joseph Myers <joseph at codesourcery dot com>
- To: Chris Metcalf <cmetcalf at ezchip dot com>
- Cc: <libc-alpha at sourceware dot org>
- Date: Wed, 25 Nov 2015 18:48:06 +0000
- Subject: Re: [COMMITTED] tile: define __NO_LONG_DOUBLE_MATH
- Authentication-results: sourceware.org; auth=none
- References: <1448399819-2804-1-git-send-email-cmetcalf at ezchip dot com> <alpine dot DEB dot 2 dot 10 dot 1511242230080 dot 10652 at digraph dot polyomino dot org dot uk> <56550015 dot 7020109 at ezchip dot com> <alpine dot DEB dot 2 dot 10 dot 1511250104490 dot 10652 at digraph dot polyomino dot org dot uk> <5655FFFD dot 4010306 at ezchip dot com>
On Wed, 25 Nov 2015, Chris Metcalf wrote:
> But apparently there are subtleties, given the bug I cited in my previous
> email that
> led to __isnanl causing a link failure on hppa. I would have thought that
> would be
> impossible. So I think I will not try tackling this at the moment...
If someone were to write an isnan macro call with an argument whose size
isn't that of float or double (e.g. an integer argument), that would have
resulted in a call to __isnanl without __NO_LONG_DOUBLE_MATH defined.
Such code has undefined behavior - all those macros require arguments of
real floating type - but it could explain the __isnanl call.
--
Joseph S. Myers
joseph@codesourcery.com