Trying to boot the system results in a continuous loop of two error messages. "udevd[2108]: get_netlink_msg: unable to receive kernel netlink message: socket..." is all I managed to write down o the error messages before it rebooted on its own because my hard drive was now corrupt... sys-fs/udev-100-r2 Steps to recreate 1. install gentoo 2. build kernel with genkernel 3. boot from initramfs 4. watch udev kill your hard drive
Created attachment 99702 [details] Emerge --info Emerge --info from chroot off livecd
Created attachment 99703 [details] Kernel config hardened-sources-2.6.17 config
Created attachment 99704 [details] dmesg output dmesg output from within the livecd
Please attach your kernel configuration file, I'm guessing that you don't have a needed config option enabled.
oh nevermind, you did attach it, will look at it now...
Ah, change CONFIG_NETFILTER_NETLINK to be Y not M and see if that fixes this issue.
(In reply to comment #6) > Ah, change CONFIG_NETFILTER_NETLINK to be Y not M and see if that fixes this > issue. > Same issue even with the above compiled into the kernel, rather than as a modules. it did give me a chance to grab the rest of the error messages, again requiring a good 'ole --rebuild-tree... "udevd[2108]: get_netlink_msg: unable to receive kernel netlink message: socket operation on non-socket udevd[2108]: get_ctrl_msg: unable to receive udev user message: socket operation on non-socket"
(In reply to comment #7) > > "udevd[2108]: get_netlink_msg: unable to receive kernel netlink message: socket > operation on non-socket > udevd[2108]: get_ctrl_msg: unable to receive udev user message: socket > operation on non-socket" > Got the same problem here. Hardware I use: Toshiba Satellite P100. This is weird because I can boot using livecd's but have problems when trying to boot it from hdd. I've tried to use couple of different versions of udev but with no success. Any ideas?
This bug is still present in udev-103, but interestingly it appears only on the kernel running from the desktop but not when it runs from a CD. And doesn't appear on every machine either.
I just ran into this bug when trying to move a Gentoo installation from the first hard disk on a desktop system to the second hard disk, via cd / ; tar cf - --one-file-system . | (cd /mnt/sdb4 ; tar xpf - ) (The installation was all on one partition.) I wondered whether it might have anything to do with the empty /dev in the new filesystem, and indeed it seems that was part of the cause: populating that /dev with the contents of the old /dev fixed the problem.
*** Bug 159376 has been marked as a duplicate of this bug. ***
(In reply to comment #10) > I just ran into this bug when trying to move a Gentoo installation from the > first hard disk on a desktop system to the second hard disk, via > > cd / ; tar cf - --one-file-system . | (cd /mnt/sdb4 ; tar xpf - ) > > (The installation was all on one partition.) I wondered whether it might have > anything to do with the empty /dev in the new filesystem, and indeed it seems > that was part of the cause: populating that /dev with the contents of the old > /dev fixed the problem. could someone in the know please comment on whether this is a recommended fix? regards, /iaw
no, those directions are not recommended the simpler fix is: mkdir /mnt/fixit mount --bind / /mnt/fixit cp -a /dev/* /mnt/fixit/dev/ umount /mnt/fixit rmdir /mnt/fixit i can reproduce this error when /dev/console is missing ... so something about udev sucks
With udev-104-r5 it should finally work again to boot with empty /dev directory.
I can confirm that this fixes my problem.
Upgrading to Udev-104-r5 and copying /dev from a livecd fixed the problem.
In udev-107 there is finally a patch against this, which removes the need for workarounds in udev-start.sh
