On Sat, 31.05.14 19:43, Samuli Suominen ([email protected]) wrote: > > > On 31/05/14 17:10, Samuli Suominen wrote: > > On 31/05/14 14:14, Samuli Suominen wrote: > >> 1. "libsystemd_network_la_SOURCES =" in Makefile.am includes > >> "src/libsystemd-network/network-internal.h" > >> and there is no anykind of #ifdef -logic behind it > >> > >> 2. "src/libsystemd-network/network-internal.h" has #include <libkmod.h> > >> and there is no anykind of #ifdef > >> -logic behind it > >> > >> so kmod is hardcoded dependency in systemd-213: > >> > >> > > This patch is a half-complete hack that only works with > > --disable-networkd, as it only covers the files > > that are outside of the #if NETWORKD in Makefile.am > > As in, I've used this patch only for the package within Gentoo that > > builds only udev, not rest of systemd, > > that uses --disable-networkd > > So consider this as proof of consept, not for inclusion > > > > - Samuli > > > > > > _______________________________________________ > > systemd-devel mailing list > > [email protected] > > http://lists.freedesktop.org/mailman/listinfo/systemd-devel > > I just realized configure.ac already has, > > AS_IF([test "x$have_networkd" = "xyes" -a "x$have_kmod" != "xyes"], > [AC_MSG_ERROR([networkd requires kmod])]) > > So this patch should be considered for inclusion after all as-is. You > should credit "Mike Gilbert <[email protected]>" for > the patch, not me.
I'd prefer I we could fix this properly and allow networkd build without kmod. I am very cool on linking against kmod from networkd anyway (in particular since I want networkd to run without CAP_SYS_MODULE), so I think this could be the least we should be doing... Lennart -- Lennart Poettering, Red Hat _______________________________________________ systemd-devel mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/systemd-devel
