Debugging udev

[James Westby <jw+debian@jameswestby.net>]

Hi,

I have a problem which looks like it is udev (0.093-1) hanging at boot time. 
The messages it look something like

[grub prompt]
INIT version 2.86 booting
Starting the hotplug events dispatcher...udevd.
Synthesising initial hotplug events...done.
Waiting until /dev is fully populated...
[cursor blinks for a couple of seconds]

Then boot will either continue normally or will hang before printing
anything else. It actually stops rather than waiting forever as the
cursor stops flashing. The only thing I can do is restart with the power
button. Occaisonaly there will be one of four messages (from memory, in
decreasing order of frequency)

hdb: not ready for command.
hda: not ready for command.
hdb: DMA channel in use.
hdb: timeout waiting for DMA.

The intermittency leads me to think that it is some sort of race
condition, and these messages suggest that it is perhaps involving the
hard drives (hdb has my /).

I think this has been the case since installing udev, and I have been
meaning to write this message for a while, but a couple of days ago it
happened about 50 times before I booted sucessfully.

I saw something suggesting that user rules caused the problems in most
cases, so I removed all of mine, but this had no effect. 

So my question, is there a way to debug udev at such an early stage of
the boot process? Is there, for example, a boot parameter that would
cause udev to output what it is doing during this time so I can find out
which rule/device is causing the problem?

I don't want to submit a report with such little information, especially
on udev, but at least that would let me find out how the maintainer
would go about debugging the problem. Hopefully though the readers of
d-u can give me some good tips. 

Thanks,

James


-- 
  James Westby
  jw+debian@jameswestby.net
  http://jameswestby.net/