When i boot, the rc init script say "please do not use this script with baselayout 1" "the udev init script is written for baselayout 2" so, i have baselayout 2 installed on my system for long time ago, and he's here. lamusic ~ % eix -ceI baselayout [I] sys-apps/baselayout (2.0.0@25/12/2008): Filesystem baselayout and init scripts This bug is reproductible on kernel 2.6.28 and 2.6.26, probably on all. On my hardened mixed system ~x86 on kvm run correctly, and on my work with an ~amd64 multilib system on kernel 2.6.27 run correctly. It's impossible for me give a screenshot or a log file, all services crashes because udev not start. For time on this machine i downgrade on openrc 0.3* . Reproducible: Always
Created attachment 176366 [details] This is my emerge --info
Created attachment 176367 [details] This is my emerge --info
Did you run etc-update after your last upgrade? Also, please paste your "emerge -pv openrc udev sysvinit" output. It's weird, cause I have a same box(same profile, 2.6.28 kernel, openrc-0.4.1, udev-135-r2 and sysvinit-2.86-r12) and I don't experience any problems like yours. I will leave the bug just ASSIGNED (which means accepted), until we are sure whether it's something of your side, or ours.
Yes i have etc-update, and my output is here : lamusic ~ % sudo emerge -pv openrc udev sysvinit These are the packages that would be merged, in order: Calculating dependencies... done! [ebuild R ] sys-apps/sysvinit-2.86-r12 USE="(-ibm) (-selinux) -static" 101 kB [ebuild R ] sys-fs/udev-135-r2 USE="(-selinux)" 437 kB [ebuild U ] sys-apps/openrc-0.4.1 [0.3.0-r1] USE="ncurses pam unicode -debug" 0 kB Total: 3 packages (1 upgrade, 2 reinstalls), Size of downloads: 537 kB I have for time on openrc-0.3.0-r1. And this run correctly.
Since I _might_ be the cause for this, do you mind trying with openrc 0.4 and pam USE flag disabled?
I'm test for time with =openrc-0.4.0 and =openrc-0.4.1 with USE flag pam disable, udev crash with the same errors : "the udev init script is written for baselayout 2" "please do not use this script with baselayout 1"
I assign the bug to the proper maintainers for deeper investigation.
Sorry but i think, i say the cause of problem, it's not a bug for openrc, i'm in nomultilib profile for no long time ago and one symlink is broken my /lib64 not link with /lib and of course this udev script : "" if [ ! -e /lib/librc.so -a -f /etc/init.d/sysfs ]; then "" don't like my broken symlink, i'm recreate symlinks and i test this.
Yes openrc-0.4.1 run correctly, but i'm a tiny visual problem, this time is very openrc the cause; i'm sure; i'm test this, with zsh, bash, and other shell. I put a screenshot, but with copy it's not visible.
Created attachment 176424 [details] my screenshot with rxvt-unicode and zsh here.
(In reply to comment #10) > Created an attachment (id=176424) [edit] > my screenshot with rxvt-unicode and zsh here. > i think this attachnement isn't visible, so my screenshot : http://omploader.org/vMTF4dQ
If udev should be fixed to work with broken /lib -> /lib64 symlink, someone who is on a multilib system can please check if any files udev installs access /lib instead of /lib64 and should be changed at emerge time.
I know there was an issue with hardened amd64 stages a long time with regard to /lib not pointing to /lib64. It was up to the user to manually fix these issues when using that stage. The issue you're having is exactly that. You need to get a list of packages that install stuff in /lib and remove /lib and symlink it over then re-emerge all those packages.
udev-135-r3 no longer relies on checking /lib/librc.so to get info about openrc installed or not, but just uses /etc/init.d/sysfs existance.
As already written, it should work fine with for example udev-135-r4 (that should go stable soon).