This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [COMMITTED] PowerPC: fix hypot/hypotf check for -INF
- From: Adhemerval Zanella <azanella at linux dot vnet dot ibm dot com>
- To: "Joseph S. Myers" <joseph at codesourcery dot com>
- Cc: "GNU C. Library" <libc-alpha at sourceware dot org>
- Date: Fri, 17 May 2013 12:37:18 -0300
- Subject: Re: [COMMITTED] PowerPC: fix hypot/hypotf check for -INF
- References: <51962EFA dot 6000007 at linux dot vnet dot ibm dot com> <Pine dot LNX dot 4 dot 64 dot 1305171450320 dot 21342 at digraph dot polyomino dot org dot uk>
On 05/17/2013 11:54 AM, Joseph S. Myers wrote:
> On Fri, 17 May 2013, Adhemerval Zanella wrote:
>
>> Testing on POWER4/5/6 I found an issue with recent hypot (-INF, ...) and
>> PPC code. The fix is trivial: the return FP->INT transformation should
>> return the absolute value of the number so the following comparison does
>> not fail for -INF.
> As usual, if a patch fixes a bug that was user-visible in previous
> releases but doesn't already have a bug filed in Bugzilla, please file a
> bug in Bugzilla for it, add the bug number to the ChangeLog entry and NEWS
> and resolve the bug as FIXED.
>
> In some cases of architecture-specific libm fixes for bugs shown up by
> testsuite changes, a bug may already have been filed and marked fixed for
> the same issue in other versions of the code - in which case you can
> simply reuse that bug number in your ChangeLog entry. In this case,
> however, the bug was shown up (I presume) by my testsuite fix to remove
> bogus uses of IGNORE_ZERO_INF_SIGN - which didn't get a bug filed because
> it was purely a testsuite change rather than changing anything
> user-visible about the library itself.
>
I'll fill a bugzilla and update NEWS and Changelog.