udev-224 is linking to stuff in /usr/lib64 which it should not and did not before. That breaks things like lvm if you have / not on a lvm partition but /usr is on a lvm partition. wlt wlt # emerge -pv udev [ebuild R ] sys-fs/udev-220-r3::gentoo USE="gudev introspection kmod -acl -doc (-selinux) -static-libs" ABI_X86="(64) -32 (-x32)" 3,945 KiB wlt wlt # ldd /lib64/libusb-1.0.so.0 linux-vdso.so.1 (0x00007ffdb4afc000) libudev.so.1 => /lib64/libudev.so.1 (0x00007f7414b2d000) libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f7414911000) libc.so.6 => /lib64/libc.so.6 (0x00007f741456e000) librt.so.1 => /lib64/librt.so.1 (0x00007f7414366000) libm.so.6 => /lib64/libm.so.6 (0x00007f7414061000) /lib64/ld-linux-x86-64.so.2 (0x00007f7414f97000) dev ~ # emerge -pv udev [ebuild R ] sys-fs/udev-224::gentoo USE="kmod -acl (-selinux) -static-libs" 0 KiB dev ~ # ldd /lib64/libusb-1.0.so.0 linux-vdso.so.1 (0x00007ffdb9bc4000) libudev.so.1 => /lib64/libudev.so.1 (0x00007ffaf89d6000) libpthread.so.0 => /lib64/libpthread.so.0 (0x00007ffaf87b9000) libc.so.6 => /lib64/libc.so.6 (0x00007ffaf840e000) libdl.so.2 => /lib64/libdl.so.2 (0x00007ffaf820a000) librt.so.1 => /lib64/librt.so.1 (0x00007ffaf8002000) libm.so.6 => /lib64/libm.so.6 (0x00007ffaf7cfe000) libresolv.so.2 => /lib64/libresolv.so.2 (0x00007ffaf7ae5000) libdw.so.1 => /usr/lib64/libdw.so.1 (0x00007ffaf789a000) /lib64/ld-linux-x86-64.so.2 (0x00007ffaf8e14000) libelf.so.1 => /usr/lib64/libelf.so.1 (0x00007ffaf7683000) libbz2.so.1 => /lib64/libbz2.so.1 (0x00007ffaf7473000) libz.so.1 => /lib64/libz.so.1 (0x00007ffaf725b000)
I suspect this was introduced by Kay's re-org of the static helper libs. https://github.com/systemd/systemd/commit/a095315b3c31f7a419baceac82c26c3c5ac0cd12 Anyway, easy enough to fix. commit c608ad85aa8470dcd01143bcea69e5efed7a8aba Author: Mike Gilbert <floppym@gentoo.org> Date: Sat Aug 15 08:51:00 2015 -0400 sys-fs/udev: Pass --disable-elfutils to configure This prevents an automagic dependency on dev-libs/elfutils. Bug: https://bugs.gentoo.org/557622 Package-Manager: portage-2.2.20_p134 sys-fs/udev/{udev-9999.ebuild => udev-224-r1.ebuild} | 1 + sys-fs/udev/udev-9999.ebuild | 1 + 2 files changed, 2 insertions(+)
Are we sure that hard-disabling elfutils is the answer to this? If udev has some optional need for it, I would rather have udev-224 have the elfutils use flag added than revbump and hard disable them.
(In reply to William Hubbs from comment #2) Its only use is for obtaining stack traces from coredumps that are written to the journal. No relevance to udev.