This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: Fix powerpc round, roundf spurious "inexact" (bug 19238) [committed]
- From: Adhemerval Zanella <adhemerval dot zanella at linaro dot org>
- To: libc-alpha at sourceware dot org
- Date: Thu, 17 Dec 2015 15:22:34 -0200
- Subject: Re: Fix powerpc round, roundf spurious "inexact" (bug 19238) [committed]
- Authentication-results: sourceware.org; auth=none
- References: <alpine dot DEB dot 2 dot 10 dot 1511121900250 dot 6391 at digraph dot polyomino dot org dot uk> <1450369489 dot 7571 dot 10 dot camel at oc7878010663> <alpine dot DEB dot 2 dot 10 dot 1512171650550 dot 3080 at digraph dot polyomino dot org dot uk>
On 17-12-2015 14:55, Joseph Myers wrote:
> On Thu, 17 Dec 2015, Steven Munroe wrote:
>
>>> This issue and fix do not apply to builds using power5+ versions of
>>> round and roundf, which use the frin instruction and avoid "inexact"
>>> exceptions that way.
>>>
>> Joseph this gets confusing when we introduce the multiarch IFUNC
>> support. In a modern system (POWER6/7/8 the code below should never be
>> used. The code from:
>
> Yes, I already said this code isn't used when the power5+ versions are
> used. My 64-bit powerpc testing is on a POWER4 system and my 32-bit
> testing is on a 476 system; although I now have access to a little-endian
> POWER8 system I haven't tried glibc testing there.
>
You can also trigger this kind of failures with --disable-multi-arch --with-cpu=power4
(for instance) on newer chips.