git: 05eac56a0432 - main - lib/msun: Fix x86 GCC6 build after 221622ec0c8e184

[Mark Millard]

John Baldwin jhb at FreeBSD.org wrote on
Fri Mar 12 19:45:17 UTC 2021 :

> On 3/12/21 11:06 AM, Dimitry Andric wrote:
> > On 12 Mar 2021, at 19:45, Alex Richardson <arichardson at FreeBSD.org> wrote:
> >>
> >> The branch main has been updated by arichardson:
> >>
> >> URL: https://cgit.FreeBSD.org/src/commit/?id=05eac56a0432d07acd7f159125855437a4dd6259
> >>
> >> commit 05eac56a0432d07acd7f159125855437a4dd6259
> >> Author:     Alex Richardson <arichardson at FreeBSD.org>
> >> AuthorDate: 2021-03-12 18:44:42 +0000
> >> Commit:     Alex Richardson <arichardson at FreeBSD.org>
> >> CommitDate: 2021-03-12 18:44:44 +0000
> >>
> >>     lib/msun: Fix x86 GCC6 build after 221622ec0c8e184
> >>
> >>     Apparently GCC only supports arithmetic expressions that use static
> >>     const variables in initializers starting with GCC8. To keep older
> >>     versions happy use a macro instead.
> >>
> >>     Fixes:          221622ec0c ("lib/msun: Avoid FE_INEXACT for x86 log2l/log10l")
> >>     Reported by:    Jenkins
> >>     Reviewed By:    imp
> >>     Differential Revision: https://reviews.freebsd.org/D29233
> > 
> > Since gcc 6 is EOLed by upstream, why are we still running CI with it?
> > Isn't it time to upgrade the CI compiler(s), so these kludges don't have
> > to be used anymore?
> 
> I'm trying, but we still don't build with GCC 9 yet everywhere.

Given how much of the time GCC 6 based CI builds spend broken
(build after build), does it really matter if it is switched
to broken GCC 9 builds where the GCC 9 related issues would
then be visible to those that would look on occasion to
monitor the status? Is there a significant value to having
the GCC 6 based CI build results in the current context?

===
Mark Millard
marklmi at yahoo.com
( dsl-only.net went
away in early 2018-Mar)