Andrew Benton wrote:
> Matthew Burgess wrote:
>
>>> # Populate /dev with all the devices that are already
available,
>>> # and save it's status so we can report failures.
>>> udevstart || failed=1
>>
>>
>> According to the latest (076) version I don't think the above
>> invocation of `udevstart' is required anymore.
>
>
> I'm using udev-076 and if I don't run udevstart with the bootscript then
> /dev doesn't get populated and the other bootscripts fail as they can't
> mount the partitions. At least I think that's what happens.
Of course, I wasn't remembering the RELEASE-NOTES properly. This is
what I've currently got in my notes (though it's untested at the moment):
1. sed -i '[EMAIL PROTECTED] /sbin/udevsend @@' /etc/rc.d/init.d/udev
2. sed -i '/udevstart/ a\
udevd --daemon' /etc/rc.d/init.d/udev
3. mkdir /lib/udev/devices &&
ln -s /proc/self/fd /lib/udev/devices/fd &&
ln -s /proc/self/fd/0 /lib/udev/devices/stdin &&
ln -s /proc/self/fd/1 /lib/udev/devices/stdout &&
ln -s /proc/self/fd/2 /lib/udev/devices/stderr &&
ln -s /proc/kcore /lib/udev/devices/dev/core &&
mkdir pts &&
mkdir shm
4. sed -i -e '18,29 d' -e '72,75 d' /etc/rc.d/init.d/udev
5. find /etc/rc.d -name \*hotplug -delete
3 and 4 assume that the new /lib/udev/devices feature can handle
symlinks (and directories) correctly and that udev and/or the kernel
don't already create those devices correctly without having entries
present in /lib/udev/devices.
This also doesn't address any of the changes necessary for the rules
file to play ball.
Regards,
Matt.
--
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page