[gentoo-user] udev and external harddisk, some more info

[gentoo-user] udev and external harddisk, some more info

[Willie Wong]

(This is not a duplicate of the mail I sent yesterday, though,
curiously I never got that one back from the list [It did make gmane])

To recap: when I plugged in my harddrive, the kernel recognized the
device, but udev failed to create the appropriate entries in /dev. 

I know that udev did receive the events, because udevmonitor shows so
and because the appropriate device appeared under /sys/block/sda (and
other places it is supposed to go).

One curious thing though: because I remembered that usb-drives used to
work about a year ago with my mp3-player, I tried downgrading udev,
and it works!

So somewhere between udev-068-r1 and udev-087-r1 something changed so
that the device nodes are not created. 

I am running the 2.6.11-hardened-r15 kernel. 

(I am assuming that using a usb mass storage device is a common enough
thing that, the fact that I don't see anything on bugzilla about this
means something is misconfigured on my end.)

Any ideas?

-----------------quote dmesg-----------------
usb 1-3: new high speed USB device using ehci_hcd and address 8
scsi3 : SCSI emulation for USB Mass Storage devices
usb-storage: device found at 8
usb-storage: waiting for device to settle before scanning
  Vendor: ST332062  Model: 0A                Rev: 0000
  Type:   Direct-Access                      ANSI SCSI revision: 00
SCSI device sda: 625142448 512-byte hdwr sectors (320073 MB)
sda: assuming drive cache: write through
SCSI device sda: 625142448 512-byte hdwr sectors (320073 MB)
sda: assuming drive cache: write through
 sda: unknown partition table
Attached scsi disk sda at scsi3, channel 0, id 0, lun 0
Attached scsi generic sg0 at scsi3, channel 0, id 0, lun 0,  type 0
usb-storage: device scan complete
usb 1-3: USB disconnect, address 8
----------------end quote------------------------     

-----------------quote console output----------------
sep / # udevmonitor 
udevmonitor prints the received event from the kernel [UEVENT]
and the event which udev sends out after rule processing [UDEV]

UEVENT[1155758086.517805] add <at> /devices/pci0000:00/0000:00:1d.7/usb1/1-3
UEVENT[1155758086.517857] add <at> /devices/pci0000:00/0000:00:1d.7/usb1/1-3/1-3:1.0
UEVENT[1155758086.521381] add <at> /class/scsi_host/host1
UDEV  [1155758089.825906] add <at> /devices/pci0000:00/0000:00:1d.7/usb1/1-3
UEVENT[1155758091.523830] add <at> /devices/pci0000:00/0000:00:1d.7/usb1/1-3/1-3:1.0/host1/target1:0:0/1:0:0:0
UEVENT[1155758091.526080] add <at> /block/sda
UEVENT[1155758091.552249] add <at> /class/scsi_device/1:0:0:0
UEVENT[1155758091.552297] add <at> /class/scsi_generic/sg0
UDEV  [1155758093.153366] add <at> /devices/pci0000:00/0000:00:1d.7/usb1/1-3/1-3:1.0
UDEV  [1155758093.170993] add <at> /class/scsi_host/host1
UDEV  [1155758099.763624] add <at> /devices/pci0000:00/0000:00:1d.7/usb1/1-3/1-3:1.0/host1/target1:0:0/1:0:0:0
UDEV  [1155758099.787844] add <at> /block/sda
UDEV  [1155758099.817401] add <at> /class/scsi_generic/sg0
UDEV  [1155758099.825812] add <at> /class/scsi_device/1:0:0:0
------------------end quote-----------------------

W

-- 
"`You ARE Zaphod Beeblebrox?'
`Yeah,' said Zaphod, `but don't shout it out or they'll all 
want one.'
`THE Zaphod Beeblebrox?'
`No, just A Zaphod Bebblebrox, didn't you hear I come in 
six packs?'
`But sir,' it squealed, `I just heard on the sub-ether 
radio report. It said you were dead...'
`Yeah, that's right, I just haven't stopped moving yet.'"

- Zaphod and the Guide's receptionist. 
Sortir en Pantoufles: up 21 days, 21:37
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev and external harddisk, some more info

[Richard Fish]

On 8/17/06, Willie Wong <wwong@princeton.edu> wrote:
> Any ideas?

Since udev is seeing the events, I'm guessing something in your udev
rules.  Have you done an etc-update?  Anything relevant in
/etc/udev/rules.d/10-local.rules?

-Richard
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev and external harddisk, some more info

[Willie Wong]

On Thu, Aug 17, 2006 at 11:48:59PM -0700, Penguin Lover Richard Fish squawked:
> On 8/17/06, Willie Wong <wwong@princeton.edu> wrote:
> >Any ideas?
> 
> Since udev is seeing the events, I'm guessing something in your udev
> rules.  Have you done an etc-update?  Anything relevant in
> /etc/udev/rules.d/10-local.rules?
> 

Yes, I've always performed the etc-update after upgrading udev. The
only line in 10-local is the following:

-------/etc/udev/rules.d/10-local.rules-----------
KERNEL=="js*",          NAME="input/%k", MODE="664", SYMLINK+="%k"
--------end file----------------------------------

which is to make joysticks work under wine. 

The thing is, I compared the rules (50-udev.rules) between three
versions of udev, and to my eyes, the rules for sd* and sg* are
virtually identical for these versions:
  1) udev-068-r1, which works on my desktop.
  2) udev-087-r1, the one that was giving me woes
  3) udev-094, one that is running on my ~x86 laptop and also works

So I am not quite sure whether it is a rules problem or a problem on
how udev handles the events. Is there a way to check that?

W
-- 
"`Right,' said Ford, `I'm going to have a look.'
He glanced round at the others.
`Is no one going to say, "No you can't possibly, let me go 
instead"?'
They all shook their heads.
`Oh well.'"

- Ford attempting to be heroic whilst being seiged by 
Shooty and Bangbang. 
Sortir en Pantoufles: up 22 days, 17:19
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev and external harddisk, some more info

[Richard Fish]

On 8/18/06, Willie Wong <wwong@princeton.edu> wrote:
> So I am not quite sure whether it is a rules problem or a problem on
> how udev handles the events. Is there a way to check that?

Well you can start with "udevcontrol log_priority=debug".  That should
cause udev to output debug information to /var/log/messages.  What you
are looking for is a sequence of messages in the form of:

udev_done: seq 3110, pid [8009] exit with 1, 0 seconds old
udev_event_run: seq 3111 forked, pid [8024], 'add' 'scsi_disk', 0 seconds old
udev_event_run: seq 3112 forked, pid [8025], 'add' 'block', 0 seconds old
udev_event_run: seq 3115 forked, pid [8026], 'add' 'scsi_device', 0 seconds old
udev_event_run: seq 3116 forked, pid [8027], 'add' 'scsi_generic', 0 seconds old
run_program: '/sbin/modprobe '
udev_rules_get_name: reset symlink list
udev_rules_get_name: add symlink 'sdb'
udev_rules_get_name: rule applied, 'sdb' becomes 'usbkey'
...

The "udev_rules_get_name" parts would be the most interesting.

Other than that, the only other thing I might suggest is try a newer
kernel version.

-Richard
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev and external harddisk, some more info

[Willie Wong]

On Fri, Aug 18, 2006 at 10:02:12AM -0700, Penguin Lover Richard Fish squawked:
> On 8/18/06, Willie Wong <wwong@princeton.edu> wrote:
> >So I am not quite sure whether it is a rules problem or a problem on
> >how udev handles the events. Is there a way to check that?
> 
> Other than that, the only other thing I might suggest is try a newer
> kernel version.

What do you know, it worked! Upgrading to 2.6.16-hardened-r10 fixed
the issue. And I just wrote some custom rules for symlinks. 

You have my heart-felt thanks, Richard. 

W
-- 
Pintsize: Curses! HOURS in there and I STILL don't have mutant ice powers!
Pintsize: Sorry waffles, you can't be my sidekick until I have some 
    superhero powers to fight crime with... What? "Waffle powers"?
    Somehow I don't see "soaking up syrup" or "browning in a toaster"
    getting us a lot of hot supervillain ladies. 
Sortir en Pantoufles: up 32 min
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev and external harddisk, some more info

[Richard Fish]

On 8/19/06, Willie Wong <wwong@princeton.edu> wrote:
> What do you know, it worked! Upgrading to 2.6.16-hardened-r10 fixed
> the issue. And I just wrote some custom rules for symlinks.

Hrm, we can only hope that the API between the kernel and udev
stabilizes in the future.  It really should _not_ be necessary to
change kernel versions when upgrading udev or vice-versa.

-Richard
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] udev and external harddisk, some more info

[Willie Wong]

On Sat, Aug 19, 2006 at 01:59:36PM -0700, Penguin Lover Richard Fish squawked:
> Hrm, we can only hope that the API between the kernel and udev
> stabilizes in the future.  It really should _not_ be necessary to
> change kernel versions when upgrading udev or vice-versa.

Now that you mentioned it, it would've been nice if there were a notice
to that effect when I upggraded udev. And I *do* read the portage
emerge logs. 

But now I am just basking in the happiness of everything working like
it's supposed to: udev, kernel, even xorg7. 

W
-- 
Proof that "Suki is good":
according to spaceballs: "[now we know evil will
    always triumph, because] good is dumb"
according to S: "ok, now I feel really dumb"
the proof follows.
~~~~~~~
(on W's attempt to prove the phrase "suki is good" by syllogism)
S: it's okay, you don't need to proove [sic] it: it's a definition
W: nice
S: i spelled prove wrong
S: geez
W: don't worry
W: that doesn't make you evil.
Sortir en Pantoufles: up  4:06
-- 
gentoo-user@gentoo.org mailing list