[PATCH] libstdc++: Use GLIBCXX_CHECK_LINKER_FEATURES for cross-builds (PR111238)

[Hans-Peter Nilsson]

> From: Hans-Peter Nilsson <hp@axis.com>
> Date: Thu, 31 Aug 2023 19:05:19 +0200

> > Date: Thu, 31 Aug 2023 17:25:45 +0200
> > From: Christophe Lyon via Gcc-patches <gcc-patches@gcc.gnu.org>

> > However, this would hide the fact that libstdc++ somehow forces the
> > user to use -Wl,-gc-sections to avoid undefined references to chdir,
> > mkdir, chmod, pathconf, ... so maybe it's better to keep the status
> > quo and not apply this patch?

I agree with the sentiment, but maybe --gc-sections should
instead be passed by default for arm-eabi when linking, with
way to opt-out; as for cris-elf per below.

> Datapoint: no failures for cris-elf in the listed tests -
> but it instead passes --gc-sections if -O2 or -O3 is seen
> for linking; see cris/cris.h.  It's been like that forever,
> modulo a patch in 2002 not passing it if "-r" is seen.
> 
> Incidentally, I've been sort-of investigating a recent-ish
> commit to newlib (8/8) that added a stub for getpriority,
> which was apparently added due to testsuite breakage for
> libstdc++ and arm-eabi, but that instead broke testsuite
> results for *other* targets, as warning at link-time.  Film
> at 11.
> 
> > 2023-08-31  Christophe Lyon  <christophe.lyon@linaro.org>
> > 
> > libstdc++-v3/ChangeLog:
> > 
> >         PR libstdc++/111238
> >         * configure: Regenerate.
> >         * configure.ac: Call GLIBCXX_CHECK_LINKER_FEATURES in cross,
> >         non-Canadian builds.
> 
> On this actual patch, I can't say yay or nay though (but
> leaning towards yay), but I'll test for cris-elf.  Would you
> mind holding off committing for a day or two?

No regressions for cris-elf with this patch.  Still, on one
thought I'm also not wild about libstdc++ this way
overriding the target, and on the other hand, I'll likely to
suggest something similar (adding options) to "improve"
GCC_TRY_COMPILE_OR_LINK (more targets actually linking).

brgds, H-P