On Mon, Jan 07, 2019 at 03:49:47PM +0000, Stuart Henderson wrote:
> On 2019/01/07 16:42, Antoine Jacoutot wrote:
> > On Mon, Jan 07, 2019 at 04:35:51PM +0100, Matthieu Herrb wrote:
> > > Hi,
> > >
> > > Trying to build some ports on macppc, glib2 is now broken:
> > >
> > > FAILED: glib/4430778@@glib-2.0@sha/gatomic.c.o
> > > cc -Iglib/4430778@@glib-2.0@sha -Iglib -I../glib-2.58.2/glib -I. -I../glib-2.58.2/ -I/usr/local/include -pipe -D_FILE_OFFSET_BITS=64 -std=gnu89 -D_GNU_SOURCE -fno-strict-aliasing -Wall -Wstrict-prototypes -Wunused -Wno-bad-function-cast -Werror=declaration-after-statement -Werror=format=2 -Werror=implicit-function-declaration -Werror=init-self -Werror=missing-include-dirs -Werror=missing-prototypes -Werror=pointer-arith -O2 -g -fPIC -pthread '-DG_LOG_DOMAIN="GLib"' -DGLIB_COMPILATION -fvisibility=hidden -MD -MQ 'glib/4430778@@glib-2.0@sha/gatomic.c.o' -MF 'glib/4430778@@glib-2.0@sha/gatomic.c.o.d' -o 'glib/4430778@@glib-2.0@sha/gatomic.c.o' -c ../glib-2.58.2/glib/gatomic.c
> > > ../glib-2.58.2/glib/gatomic.c:675:2: error: #error G_ATOMIC_LOCK_FREE defined, but incapable of lock-free atomics.
> > > ninja: build stopped: subcommand failed.
> >
> > Probably a broken test at configure time?
> >
> > --
> > Antoine
> >
>
> This is the one landry ran into the other day I think?
Yes that's the one.
https://gitlab.gnome.org/GNOME/glib/blob/master/glib/gatomic.c#L675 is
where it barfs.
https://gitlab.gnome.org/GNOME/glib/blob/master/meson.build#L1578
might have to be tested for OpenBSD. from what i understood reading
https://gitlab.gnome.org/GNOME/glib/blob/master/meson.build#L1562 meson links
int main() {
volatile int atomic = 2;
__sync_bool_compare_and_swap (&atomic, 2, 3);
return 0;
}
to check that atomic ops are supported but later on
__GCC_HAVE_SYNC_COMPARE_AND_SWAP_4 isnt defined for some reason and the
#ifdef explodes in gatomic.c.
note that it also affects sparc64, and probably all non-clang archs.
on the bright side the current sparc64 & macppc bulks will have a fast
turnaround, as not having glib2 cuts a good chunk of the tree.
Landry
No comments:
Post a Comment