[PATCH] math: Exclude internal math symbols for tests [BZ #32414]

[H.J. Lu]

On Thu, Dec 5, 2024 at 6:29 PM Joseph Myers <josmyers@redhat.com> wrote:
>
> On Thu, 5 Dec 2024, H.J. Lu wrote:
>
> > On Thu, Dec 5, 2024 at 9:04 AM H.J. Lu <hjl.tools@gmail.com> wrote:
> > >
> > > Since internal tests don't have access to internal symbols in libm,
> > > exclude them for internal tests.  Also make tst-strtod5 and tst-strtod5i
> > > depend on $(libm) to support older versions of GCC which can't inline
> > > copysign family functions.  This fixes BZ #32414.
> > >
> > >
> > > --
> > > H.J.
> >
> > It is used in glibc build for some targets:
> >
> > FAIL: glibcs-powerpc-linux-gnu build
> > FAIL: glibcs-powerpc-linux-gnu-power4 build
> > FAIL: glibcs-powerpc64le-linux-gnu build
> > FAIL: glibcs-powerpc-linux-gnu-soft build
> > FAIL: glibcs-powerpc64-linux-gnu build
> > FAIL: glibcs-powerpc64le-linux-gnu-disable-multi-arch build
> > FAIL: glibcs-sparcv9-linux-gnu build
> > FAIL: glibcs-sparc64-linux-gnu build
> >
> > But they shouldn't be declared in the installed header.
>
> Please give more details of the failures you see (with this patch, as
> opposed to the "[PATCH] math: Don't declare __XXX math functions [BZ
> #32418]" one).  It seems rather odd for a change adding !IS_IN
> (testsuite_internal) conditionals to break the build of glibc itself.
> (And this patch is only changing non-installed files, it's nothing to do
> with what's in installed headers.)
>

I followed it up with the wrong patch.   This patch seems OK.


-- 
H.J.