This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: math test-float.out
- From: "Joseph S. Myers" <joseph at codesourcery dot com>
- To: Klaus Dittrich <kladit at arcor dot de>
- Cc: libc-alpha <libc-alpha at sourceware dot org>
- Date: Sun, 19 May 2013 18:13:12 +0000
- Subject: Re: math test-float.out
- References: <5198F0A4 dot 7090000 at df1tl dot local dot here>
On Sun, 19 May 2013, Klaus Dittrich wrote:
> testing float (without inline functions)
> Failure: nearbyint (0.5) == 0.0: Exception "Inexact" set
> Failure: nearbyint (-0.5) == -0: Exception "Inexact" set
> Did I miss something ?
You didn't mention whether this was a 32-bit or 64-bit build of glibc, or
how you configured glibc (and /proc/cpuinfo information about processor
features is also helpful in such cases where there are multiple
implementations of a function that may be selected at runtime using IFUNC;
information from /proc/cpuinfo is easier to use than a processor model
number). I can reproduce this on x86_64 with --disable-multi-arch to
avoid the SSE4.1 version of nearbyintf being used, and have filed bug
15490 for it.
--
Joseph S. Myers
joseph@codesourcery.com