Package: udev
Version: 0.114-2

Because of this bug, udev does not create the /dev entries (sda and
sda1) for my USB key (a Creative Muvo 128Mo). I solved it (see the end
of the mail) for my configuration. But here is the description of what
happen.

I log in a console to see kernel messages and also run udev with log in
'debug' mode instead of 'err'. After the key has been plugged, the
kernel send the usual messages about a new usb peripheral and report
that it will be named sda and has one partition sda1. All udev stuff is
working and the genuine entries are created in /dev (the entry by-path,
by-label...) until it comes to run 'vol_id --export /dev/sda'. Here the
kernel raise a problem of I/O access on block 256000 and the key is
removed by the kernel because of I/O errors. Finally, udev being aware
of this removal, delete the sda entry of /dev.

I stopped udev (/etc/init.d/udev stop), unplugged my key and replug it.
I create the nodes by hand ('mknod sda b 8 0' and 'mknod sda1 b 8 1')
and was able to mount the sda1 partition and read it correctly. Then, I
try to run '/lib/udev/vol_id --export /dev/sda' and it fails as in the
udev scripts, making the kernel rejecting my key. By the way, it works
perfectly on sda1.

I looked in the /etc/udev/rules.d/z25_persistent.rules and vol_id is
called once at the end. I added then:
KERNEL=="sd*[!0-9]", SYSFS{removable}=="1",
GOTO="persistent_storage_end"
just after the line:
KERNEL=="hd*[!0-9]", SYSFS{removable}=="1",
GOTO="persistent_storage_end"
May I suggest to include this line in the official udev script if
vol_id is indeed made for scanning partition only?

So with the additional line, it works for me. But you may be interested
in finding why vol_id makes the kernel reject my key. If so please send
me instructions on what to do or what information you may need. I test
it for different kernels (2.6.18 and 2.6.21) and the behaviour was the
same, always related to vol_id from udev.

Best regards,

Damien.



-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Reply via email to