On Sun, Jun 09, 2013 at 05:11:19PM +0200, Zbigniew Jędrzejewski-Szmek wrote: > On Sun, Jun 09, 2013 at 08:52:01AM +0100, Ross Lagerwall wrote: > > On Sat, Jun 08, 2013 at 05:06:50PM +0200, Zbigniew Jędrzejewski-Szmek wrote: > > > Maybe mention that systemd-shutdown is statically linked (I know it > > > can be inferred from the text, but being explicit might be better). > > > > At least on Arch, it is still statically linked to libc and udev: > > $ ldd /usr/lib/systemd/systemd-shutdown > > linux-vdso.so.1 (0x00007fff80bff000) > > libudev.so.1 => /usr/lib/libudev.so.1 (0x00007f9a43582000) > > librt.so.1 => /usr/lib/librt.so.1 (0x00007f9a4337a000) > > libc.so.6 => /usr/lib/libc.so.6 (0x00007f9a42fcd000) > > /lib64/ld-linux-x86-64.so.2 (0x00007f9a43794000) > > libdl.so.2 => /usr/lib/libdl.so.2 (0x00007f9a42dc9000) > > libpthread.so.0 => /usr/lib/libpthread.so.0 (0x00007f9a42bad000) > Stupid me, I typed 'ldd build/systemd-shutdown', saw > 'not a dynamic executable' and didn't think that ldd > is looking at a shell script. > > > And if I run "pacman -S glibc" and then shutdown: > > -.mount mount process exited, code=exited status=32 > > -.mount changed unmounting -> mounted > > Job -.mount/stop finished, result=failed > > Failed unmounting /. > > > > Should it be completely statically linked? > systemd-shutdown should be using the updated libraries, so this > shouldn't be an issue. Don't you see this message otherwise? >
I have just tried it several times in an Arch vm with systemd 204 and it seems fairly reproducible. If I reinstall glibc, it fails to unmount, if I do not, it unmounts successfully. It is just a plain single partition root filesystem with ext4. Any tips for debugging this? -- Ross Lagerwall _______________________________________________ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel