[PATCH v2] configure.ac: use AC_HEADER_MAJOR to detect major()/minor()

[Emil Velikov]

On 13 December 2017 at 07:58, Pekka Paalanen <ppaalanen at gmail.com> wrote:
> On Tue, 12 Dec 2017 15:41:52 +0000
> Emil Velikov <emil.l.velikov at gmail.com> wrote:
>
>> On 12 December 2017 at 08:28, Pekka Paalanen <ppaalanen at gmail.com> wrote:
>> > On Mon, 11 Dec 2017 16:16:47 +0200
>> > Pekka Paalanen <ppaalanen at gmail.com> wrote:
>> >
>> >> On Wed, 31 May 2017 22:17:50 +0100
>> >> Sergei Trofimovich <slyfox at gentoo.org> wrote:
>> >>
>> >> > This change slightly updates c4d7f66c12853b9575366dd9f4a7960ec5694934
>> >> > which added <sys/sysmacros.h> inclusion.
>> >> >
>> >> > Autoconf has AC_HEADER_MAJOR to find out which header defines
>> >> > reqiured macros:
>> >> >     https://www.gnu.org/software/autoconf/manual/autoconf-2.69/html_node/Particular-Headers.html
>> >> >
>> >> > This change should increase portability across other libcs.
>> >> >
>> >> > Bug: https://bugs.gentoo.org/610652
>> >> > Signed-off-by: Sergei Trofimovich <slyfox at gentoo.org>
>> >> > ---
>> >> > Change since v1:
>> >> > - rebased on top of current master
>> >> > - dedented nested '#    include'
>> >> >  configure.ac                       | 1 +
>> >> >  libweston/launcher-direct.c        | 9 ++++++++-
>> >> >  libweston/launcher-logind.c        | 9 ++++++++-
>> >> >  libweston/launcher-weston-launch.c | 8 +++++++-
>> >> >  libweston/weston-launch.c          | 9 ++++++++-
>> >> >  5 files changed, 32 insertions(+), 4 deletions(-)
>> >>
>> >> Hi,
>> >>
>> >> I'm on Gentoo with sys-libs/glibc-2.25-r9:2.2 and I am now getting a
>> >> bunch of warnings like this:
>> >>
>> >> /home/pq/git/weston/libweston/weston-launch.c: In function ‘handle_open’:
>> >> /home/pq/git/weston/libweston/weston-launch.c:333:13: warning: In the GNU C Library, "major" is defined
>> >>  by <sys/sysmacros.h>. For historical compatibility, it is
>> >>  currently defined by <sys/types.h> as well, but we plan to
>> >>  remove this soon. To use "major", include <sys/sysmacros.h>
>> >>  directly. If you did not intend to use a system-defined macro
>> >>  "major", you should undefine it after including <sys/types.h>.
>> >>   if (major(s.st_rdev) != INPUT_MAJOR &&
>> >>
>> >>
>> >> Reverting this patch makes the warnings go away.
>> >>
>> >> What's the right fix here?
>> >>
>> >> In config.h I have:
>> >>
>> >> /* Define to 1 if `major', `minor', and `makedev' are declared in <mkdev.h>.
>> >>    */
>> >> /* #undef MAJOR_IN_MKDEV */
>> >>
>> >> /* Define to 1 if `major', `minor', and `makedev' are declared in
>> >>    <sysmacros.h>. */
>> >> /* #undef MAJOR_IN_SYSMACROS */
>> >>
>> >> I'm not sure what I needed to do get even that far, maybe wipe my
>> >> out-of-tree build dir. Even then, I cannot find any evidence of
>> >> configure actually running the tests. I always start a new build dir by
>> >> running autogen.sh.
>> >
>> > Hi,
>> >
>> > there is an email thread of the problem from a slightly different point
>> > of view:
>> > https://lists.gnu.org/archive/html/autoconf/2016-09/msg00000.html
>> >
>> > If we keep the code as is, the warning will not go away until glibc
>> > actually drops the things from sys/types.h and distros ship it.
>> > Apparently there is a workaround one could make with config.site:
>> > https://lists.gnu.org/archive/html/autoconf/2016-09/msg00013.html
>> >
>> > Or we could wait for a new autoconf release and require it, but the
>> > discussion happened in 2016 and at least Gentoo does not have a
>> > autoconf 2.70 yet. Or we could ship a fixed AC_HEADER_MAJOR.
>> >
>> > Or we could just revert this patch until something in the above changes
>> > or we have migrated to Meson.
>> >
>> I wish the glibc people were a bit more diligent about the deprecation :-(
>> Haven't looked at how Meson does it, but I would suspect it has a
>> workaround similar to this [1].
>>
>> -Emil
>> [1] https://cgit.freedesktop.org/mesa/drm/commit/?id=7040fea0280bad527ed4b3d5eee7d7bfbf303efc
>
> Hi Emil,
>
> the workaround taken in libdrm was discussed in that 2016 email thread,
> concluding it is not a good idea, because autoconf compiler tests have
> horrible code quality: they often provoke warnings anyway, which means
> -Werror would cause false negatives in tests.
>
> Maybe it's fine for AC_HEADER_MAJOR alone?
>
Precisely what I meant to say.
Having a global -Werror is a bad idea, although using it for that
macro alone should be fine.

-Emil