hiya

I was using 2.6.11.12 for a long time until i decided to finally go for it and make the jump to a newer udev for 2.6.14. I had compiled a 2.6.14 kernel from before but it paused for a long time initializing udev and I learnt that it was necessary to update udev.

SO, today, I downloaded udev 0.76 and compiled and installed it while using my 2.6.11.12 then rebooted into my 2.6.14 and got this unfortunate error:

udevsend[XXXX]: main: error sending message: connection refused

hundreds of times on the screen (with the XXXX replaced by a different number every time). Is this something to do with hotplugging? I have an /etc/hotplug and an /etc/hotplug.d directory. I tried recompiling my 2.6.14 but this made little difference. The machine just about boots but doesn't go any further than GDM. The prompt is fully functional though and it has mounted my disks.

By the way, I use Slackware (10.1 I think). What was all this talk about udev rules? Do I need to change anything? I just left it to use its defaults.




...

help!

make sure you have
CONFIG_KOBJECT_UEVENT
enabled in the kernel
http://www.ussg.iu.edu/hypermail/lin...06.2/1088.html
seems the older hotplug hook doesn't work anymore in favor of this

yea, it was built in the last time I built my kernel. I catted the kernel source's .config and found it:

CONFIG_KOBJECT_UEVENT=y

what else should I do?

ok this is as far as i understand it
i think the problem is in the bootscripts
old school was udevsend (called by hotplug) automatically started udevd
this is from the kernel today hotplug readme
so i think your init script needs to start udevd --daemon. It no longer happens magically
through udevsend.

what i would do is find a newer system and look at the udev boot or possibly someone will come along here and show.

ok, i think (i THINK ) it's all ok now! I finally ended up putting that line at the very beginning of rc.S and my system seems to be fairly content. It does throw up a one line error on boot like this:

udevd[XXXX]: add_to_rules: error parsing SYSFS attribute in 'BUS'

but other than that things seem to be fine. Do you know what that is? My m8's pc has done the same actually under similar conditions.

Thanks for your help though m8 - I couldn't have got it working without you! I'm now updating my entire Slackware installation to 10.2 with slapt-get and getting ready to install Gnome (at the moment I use Flux).

Later

ah, one thing. Just before it switches to runlevel 4 I get this:

none on /mnt/ramfs type tmpfs (rw)
mount: none already mounted or /sys busy
mount: according to mtab, sysfs is already mounted on /sys

and that's all it says about the whole ordeal ... I notice no other problems, errors or drops in performance.

seeing as my understanding of /proc and /sys and similar things is fairly limited I wondered if anyone knew what this was about?