Hello!

Well... I am pretty new to udev - forced to use the good new udev
system as I need to migrate to 2.6.13. ;-)
I try to work my way through all the things google gave me to understand
udev but I got stuck now somewhere in the middle:

After installing "udev from scratch" on an embedded system
(not being able to follow Decibels UDEV Primer directly as I don't use
any package management tools.) I still cannot get my mtd
(memory technology device) node created dynamically by udev when I
insert my module to the kernel.

Basically, my installation was like this:

- compiled Linux-2.6.13-rc2 w/ CONFIG_UNIX98_PTYS and CONFIG_DEVPTS_FS
and CONFIG_HOTPLUG and friends.

- manually created /dev nodes on local filesystem to be able to boot
and login via ssh at least:
cd /dev
mknod -m 660 console c 5 1
mknod -m 666 null c 1 3
mknod -m 666 hda b 3 0
mknod -m 666 hda1 b 3 1
mknod -m 666 hda2 b 3 2
(I guess I will have to move those manually created devices to
/etc/udev/devices, right? Where is documentation for that and who
is copying that to /dev and when?)

- to be able to login via ssh:
mknod /dev/ptmx c 5 2
chmod 666 /dev/ptmx
mkdir /dev/pts            

- Add to /etc/fstab:
  none            /dev/pts        devpts        gid=5,mode=620    0 0
	(Note:  gid=5 is applicable for RedHat systems for which group "tty" has
	gid 5.  Adjust according to your distribution.  Use mode=600 if you want
	"mesg n" to be default.)

- mount /dev/pts

- installed hotplug-2004_09_23

- installed udev-062 with udevdir=/dev

- added mtd sections to /etc/udev/rules.d/50-udev.rules:
#KERNEL="mtd[0-9]*", NAME="mtd/%k", MODE="0666"
KERNEL="mtd*"            NAME="%k", OWNER="root", GROUP="disk", MODE="0640"
(I tested both lines)

(BTW: in which order are the files in /etc/udev/rules.d processed if
there are more than one? Are there any naming conventions which apply?
Where is the documentation for that?)

- I've checked /udev/uevent_listen which works pretty fine while
loading my mtd modules:

$ ./uevent_listen
[1121084998] received 'remove' from '/module/mminfo'
ACTION=remove
DEVPATH=/module/mminfo
SUBSYSTEM=module
SEQNUM=181

[1121085000] received 'add' from '/module/mminfo'
ACTION=add
DEVPATH=/module/mminfo
SUBSYSTEM=module
SEQNUM=182

while doing
$ rmmod mminfo
$ insmod mminfo.ko

Now I ran into the following problem while loading my module
which appeared as /dev/mtd/0 and/or /dev/mtd/0ro in devfs:

Jul 11 13:33:29 ecam udevd[14647]: udevd.c: seq 177 forked, pid 14653, 2 seconds old
Jul 11 13:33:29 ecam udev[14653]: udev_rules.c: reset run list
Jul 11 13:33:29 ecam udev[14654]: udev_utils.c: exec of program '/sbin/udev_run_devd' failed
Jul 11 13:33:29 ecam udevd[14647]: udevd.c: seq 177 exit, 2 seconds old
Jul 11 13:42:32 ecam udevd[14647]: udevd.c: seq 178 queued, devpath '/module/mminfo'
Jul 11 13:42:32 ecam udevd[14647]: udevd.c: seq 178 forked, pid 14662, 0 seconds old

I patched udev_utils.c (appended, whitespaces fixed, too) to get the clue
that the call to /sbin/udev_run_devd failed.

Well, there is no valuable documentation about udev_run_devd neither in
writing_udev_rules nor in Decibels UDEV primer. So I decided to remove
that line from the 50-udev.rules:
# emulate dev.d/
#RUN="/sbin/udev_run_devd"
without really knowing what that is...
(What is a directory multiplexer doing? Do I need that? For what?)

The error messge about failed exec of udev_run_devd doesn't appear anymore.
But I still cannot get my /dev/mtd/0 created...

What did I miss? udev seems to work as udevstart gives me all the ttys and
friends... but I cannot get my /dev/mtd.
Is there a step by step installation for the not-mainstream distros?

Best greets,

Clemens Koller
_______________________________
R&D Imaging Devices
Anagramm GmbH
Rupert-Mayer-Str. 45/1
81379 Muenchen
Germany

http://www.anagramm.de
Phone: +49-89-741518-50
Fax: +49-89-741518-19