2.6.12 udev hangs at boot

2.6.12 udev hangs at boot

[Nick Warne]

Hi All,

2.6.12, GCC 3.4.4 on Slackware 10 base

New 2.6.12 build hangs at initialising udev dynamic device directory on boot.  
I used make oldconfig from 2.6.11.12, and all the new changes I selected N, 
all bar nvidia FB support (I built several times, as I have a GeForce4 card, 
so suspected the nvidia FB support at first and turned off).

2.6.11.12 works perfect.

I have just spent an hour or so investigating, but I am none the wiser.

Ideas what could be causing this?

Nick
-- 
"When you're chewing on life's gristle,
Don't grumble, Give a whistle..."

Re: 2.6.12 udev hangs at boot

[Alistair John Strachan]

On Saturday 18 Jun 2005 13:32, you wrote:
> Hi All,
>
> 2.6.12, GCC 3.4.4 on Slackware 10 base
>
> New 2.6.12 build hangs at initialising udev dynamic device directory on
> boot. I used make oldconfig from 2.6.11.12, and all the new changes I
> selected N, all bar nvidia FB support (I built several times, as I have a
> GeForce4 card, so suspected the nvidia FB support at first and turned off).
>
> 2.6.11.12 works perfect.
>
> I have just spent an hour or so investigating, but I am none the wiser.
>
> Ideas what could be causing this?

I had this problem because I was running an ancient version of udev (0.34, 
versus 0.58, at the time..). Try upgrading udev if it's out of date.

-- 
Cheers,
Alistair.

personal:   alistair()devzero!co!uk
university: s0348365()sms!ed!ac!uk
student:    CS/CSim Undergraduate
contact:    1F2 55 South Clerk Street,
            Edinburgh. EH8 9PP.

Re: 2.6.12 udev hangs at boot

[Nick Warne]

On Saturday 18 June 2005 14:03, Alistair John Strachan wrote:
> > New 2.6.12 build hangs at initialising udev dynamic device directory on
> > boot. I used make oldconfig from 2.6.11.12, and all the new changes I
> > selected N, all bar nvidia FB support (I built several times, as I have a
> > GeForce4 card, so suspected the nvidia FB support at first and turned
> > off).
> >
> > 2.6.11.12 works perfect.
> >
> > I have just spent an hour or so investigating, but I am none the wiser.
> >
> > Ideas what could be causing this?
>
> I had this problem because I was running an ancient version of udev (0.34,
> versus 0.58, at the time..). Try upgrading udev if it's out of date.

Thanks, that worked :-)

Strange how the mind works.  I thought about that, but only for a nanosecond.  
After reading the changelog, I was convinced (or didn't think no more about) 
as the udev code stuff hadn't changed from 2.6.11.12 -> 2.6.12.

Nick
-- 
"When you're chewing on life's gristle,
Don't grumble, Give a whistle..."

Re: 2.6.12 udev hangs at boot

[Andrew Walrond]

On Saturday 18 June 2005 14:44, Nick Warne wrote:
> >
> > I had this problem because I was running an ancient version of udev
> > (0.34, versus 0.58, at the time..). Try upgrading udev if it's out of
> > date.
>
> Thanks, that worked :-)
>

FWIW In the udev 058 announcement, Greg said:

"Note, if you are running a kernel newer than 2.6.12-rc4 (including the
-mm releases) and you have any custom udev rules, you MUST upgrade to
the latest version to allow udev to work properly.  This change happened
because of a previously-unrealized reliance in libsysfs on the presence
of a useless sysfs file that has recently been removed.  Hopefully the
libsysfs people will be releasing a new version shortly with this change
in it for those packages who rely on it."

Just a reminder because I bet many people will get caught out by this!

Andrew Walrond

Re: 2.6.12 udev hangs at boot

[Jeff Garzik]

Nick Warne wrote:
> New 2.6.12 build hangs at initialising udev dynamic device directory on boot.  


Did you try simply waiting a while?

udev took a long time to initialize (30-40 seconds) for me, then 
everything worked and the machine booted just fine.

I've seen this on both new and old udev.  Some patience will fix things :)

	Jeff

Re: 2.6.12 udev hangs at boot

[Nick Warne]

On Saturday 18 June 2005 17:53, Jeff Garzik wrote:
> Nick Warne wrote:
> > New 2.6.12 build hangs at initialising udev dynamic device directory on
> > boot.
>
> Did you try simply waiting a while?
>
> udev took a long time to initialize (30-40 seconds) for me, then
> everything worked and the machine booted just fine.
>
> I've seen this on both new and old udev.  Some patience will fix things :)

Yes, I waited a while first time.  No drive activity, no nothing.  Keyboard 
was still awake though, so it wasn't a 'crash' as such. The boot just stopped 
there twiddling it's thumbs...

Installing the latest udev though makes the machine boot so fast I can't see 
the 'initialising udev devices' message unless I scroll back to see in 
console.  I thought at first I broke it, and udev wasn't working at all now 
and was being ignored, but it is working just fine :-)

But remember, what got me was it boots fine on 2.6.11.12, insomuch I never 
really saw the udev message away and never had to investigate it before.

Nick
-- 
"When you're chewing on life's gristle,
Don't grumble, Give a whistle..."

Re: 2.6.12 udev hangs at boot

[Andrew Haninger]

Thanks for pointing this out.

I also have a Slackware 10 machine here on which I was trying to get
2.6.12 working. I was noticing odd things changing when I switched
between the default (2.4.26) kernel and 2.6.12. 'less' wouldn't work,
my /dev/sd* devices weren't there so I couldn't mount my external hard
disk or cdrom drive, and there was that hang at boot time.

I've since downloaded a new udev and installed it and all above
problems are resolved. I did have to do a 'pkgremove udev", though, to
get the /dev/ devices back. (Except that I did it through pkgtool so
'udev' might not be the right name to give to pkgremove.)

I had even checked the Documentation/Changes file to see if Slackware
10 came with some outdated package, and udev wasn't mentioned. I'm not
sure where else I should have looked for information like that,
though.

Anyway, just a heads-up to anyone else experiencing a breaking of
'less' and missing /dev files.

-Andy

Re: 2.6.12 udev hangs at boot

[Denis Vlasenko]

On Saturday 18 June 2005 20:06, Nick Warne wrote:
> On Saturday 18 June 2005 17:53, Jeff Garzik wrote:
> > Nick Warne wrote:
> > > New 2.6.12 build hangs at initialising udev dynamic device directory on
> > > boot.
> >
> > Did you try simply waiting a while?
> >
> > udev took a long time to initialize (30-40 seconds) for me, then
> > everything worked and the machine booted just fine.
> >
> > I've seen this on both new and old udev.  Some patience will fix things :)
> 
> Yes, I waited a while first time.  No drive activity, no nothing.  Keyboard 
> was still awake though, so it wasn't a 'crash' as such. The boot just stopped 
> there twiddling it's thumbs...
> 
> Installing the latest udev though makes the machine boot so fast I can't see 
> the 'initialising udev devices' message unless I scroll back to see in 
> console.  I thought at first I broke it, and udev wasn't working at all now 
> and was being ignored, but it is working just fine :-)
> 
> But remember, what got me was it boots fine on 2.6.11.12, insomuch I never 
> really saw the udev message away and never had to investigate it before.

udev-030 is looking for detached_state somewhere inside /sys tree,
but somewhere between 2.6.12-rc3 and -rc5 detached_state is gone.

I installed udev-058 and it fixed things.

It took me more than hour to figure out, from 02:00 to after 03:00.

Greg, any plans to distribute udev and hotplug within kernel tarballs
so that people do not need to track such changes continuously?

After all, udev is tied to /sys layout which changes with kernel
and also udev is vital for properly functioning boot process
(I mean: if udev breaks, you cannot easily debug it: your box
is unusable since lots of things gone avry like gettys not having ttys
to let you log in...).
--
vda

Re: 2.6.12 udev hangs at boot

[Stefan Smietanowski]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi Denis.

> After all, udev is tied to /sys layout which changes with kernel
> and also udev is vital for properly functioning boot process

Not if you use a static /dev.

// Stefan
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (MingW32)

iD8DBQFCtpe0Brn2kJu9P78RAhrmAJ9EhCY1JQu0bkU3Tx8xP+oGc6/eHQCbBybu
hjZMEAlLc9gPc7vl238LtEE=
=9FdR
-----END PGP SIGNATURE-----

Re: 2.6.12 udev hangs at boot

[Denis Vlasenko]

On Monday 20 June 2005 13:17, Stefan Smietanowski wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Hi Denis.
> 
> > After all, udev is tied to /sys layout which changes with kernel
> > and also udev is vital for properly functioning boot process
> 
> Not if you use a static /dev.

Static /dev kind of defeats the purpose of udev.
I do not want to go back to the days when I had tons of
/dev/{h,s}d{a,b,c,d,e,f}{0,1,2,3,4,5,6,7,8,9} for every
IDE and SCSI block device possible. Same for /dev/tty*.
I want them appear on the fly, if/when hardware is present.
--
vda

Re: 2.6.12 udev hangs at boot

[Stefan Smietanowski]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi Denis.

>>>After all, udev is tied to /sys layout which changes with kernel
>>>and also udev is vital for properly functioning boot process
>>
>>Not if you use a static /dev.
> 
> 
> Static /dev kind of defeats the purpose of udev.
> I do not want to go back to the days when I had tons of
> /dev/{h,s}d{a,b,c,d,e,f}{0,1,2,3,4,5,6,7,8,9} for every
> IDE and SCSI block device possible. Same for /dev/tty*.
> I want them appear on the fly, if/when hardware is present.

Of course it defeats the purpose of udev. That's not what I
was answering however. It was the blanket statement that
"udev is vital for properly functioning boot process" as
if udev was necessary. udev (for me) is a very fine tool
that serves very many purposes, however it is not vital.

// Stefan
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (MingW32)

iD8DBQFCtu3jBrn2kJu9P78RAghtAJwN9Nv5jYuVqchP66I2fM3khizUNACfWfjI
bwUjbisvDWk6qCCaC13u1no=
=z8Tr
-----END PGP SIGNATURE-----

Re: 2.6.12 udev hangs at boot

[Bill Davidsen]

Denis Vlasenko wrote:
> On Monday 20 June 2005 13:17, Stefan Smietanowski wrote:
> 
>>-----BEGIN PGP SIGNED MESSAGE-----
>>Hash: SHA1
>>
>>Hi Denis.
>>
>>
>>>After all, udev is tied to /sys layout which changes with kernel
>>>and also udev is vital for properly functioning boot process
>>
>>Not if you use a static /dev.
> 
> 
> Static /dev kind of defeats the purpose of udev.

udev is a great thing is you are putting out a distribution which will 
be used on many configurations. That doesn't include most of us.

> I do not want to go back to the days when I had tons of
> /dev/{h,s}d{a,b,c,d,e,f}{0,1,2,3,4,5,6,7,8,9} for every
> IDE and SCSI block device possible. Same for /dev/tty*.
> I want them appear on the fly, if/when hardware is present.

One size doesn't fit all with that, it's nice to have hardware that the 
kernel doesn't try to use, particularly if it's hardware that the kernel 
almost but not quite knows HOW to use...

In other words, there are reasons for running static /dev, although I 
suggest trimming it to something resembling what you need.

RELATED TOPIC: It would be really nice of the changelog started with a 
"REQUIRED UPDATES" section, even though the vast majority of cases the 
description would be "none." Seeing that git didn't even provide a full 
Changelog in the previous tradition, I guess that's not likely.


-- 
    -bill davidsen (davidsen@tmr.com)
"The secret to procrastination is to put things off until the
  last possible moment - but no longer"  -me

Re: 2.6.12 udev hangs at boot

[Greg KH]

On Mon, Jun 20, 2005 at 01:04:10PM +0300, Denis Vlasenko wrote:
> 
> Greg, any plans to distribute udev and hotplug within kernel tarballs
> so that people do not need to track such changes continuously?

Nope.  But if you use udev, you should read the announcements for new
releases, as I did say this was required for 2.6.12, and gave everyone a
number of weeks notice :)

thanks,

greg k-h

Re: 2.6.12 udev hangs at boot

[Jeff Garzik]

Greg KH wrote:
> On Mon, Jun 20, 2005 at 01:04:10PM +0300, Denis Vlasenko wrote:
> 
>>Greg, any plans to distribute udev and hotplug within kernel tarballs
>>so that people do not need to track such changes continuously?
> 
> 
> Nope.  But if you use udev, you should read the announcements for new
> releases, as I did say this was required for 2.6.12, and gave everyone a
> number of weeks notice :)

Since udev is required for booting, it sounds like you're putting people 
in an upgrade-or-no-boot situation.

That's lame.  The kernel should support udev's out in the field, on 
people's boxes (RHEL, SLES?, Fedora, ...).

	Jeff

Re: 2.6.12 udev hangs at boot

[Greg KH]

On Mon, Jun 20, 2005 at 01:24:23PM -0400, Jeff Garzik wrote:
> Greg KH wrote:
> >On Mon, Jun 20, 2005 at 01:04:10PM +0300, Denis Vlasenko wrote:
> >
> >>Greg, any plans to distribute udev and hotplug within kernel tarballs
> >>so that people do not need to track such changes continuously?
> >
> >
> >Nope.  But if you use udev, you should read the announcements for new
> >releases, as I did say this was required for 2.6.12, and gave everyone a
> >number of weeks notice :)
> 
> Since udev is required for booting, it sounds like you're putting people 
> in an upgrade-or-no-boot situation.

Well, they don't _have_ to upgrade their kernel :)

> That's lame.  The kernel should support udev's out in the field, on 
> people's boxes (RHEL, SLES?, Fedora, ...).

This was caused by an unfortunate assumption in older versions of udev
about what was contained in the sysfs tree.  udev is now fixed to not
make that assumption.  So this was not a kernel bug, but a udev/libsysfs
one (and I wasn't going to keep the old kernel behavior for this minor
issue.)

As for working with people's boxes, only the very oldest versions of
udev (like the reported 030 version which is a year old and I do not
think shipped by any distro) would have the "lockup" issue.  On all of
the other ones, only custom rules written by users would have issues
(meaning, not work).  I do not know of any shipping, supported distro
that currently has a boot lockup issue (if so, please let me know.)

thanks,

greg k-h

Re: 2.6.12 udev hangs at boot

[Nick Warne]

On Monday 20 June 2005 18:34, you wrote:

> As for working with people's boxes, only the very oldest versions of
> udev (like the reported 030 version which is a year old and I do not
> think shipped by any distro) would have the "lockup" issue.  On all of
> the other ones, only custom rules written by users would have issues
> (meaning, not work).  I do not know of any shipping, supported distro
> that currently has a boot lockup issue (if so, please let me know.)

It appears the issue people are seeing is with Slack 10, which shipped with 
udev 0.26 - and I presume there was 'custom' rules Patrick had built in.

Nick

-- 
"When you're chewing on life's gristle,
Don't grumble, Give a whistle..."

Re: 2.6.12 udev hangs at boot

[Greg KH]

On Mon, Jun 20, 2005 at 08:00:08PM +0100, Nick Warne wrote:
> On Monday 20 June 2005 18:34, you wrote:
> 
> > As for working with people's boxes, only the very oldest versions of
> > udev (like the reported 030 version which is a year old and I do not
> > think shipped by any distro) would have the "lockup" issue.  On all of
> > the other ones, only custom rules written by users would have issues
> > (meaning, not work).  I do not know of any shipping, supported distro
> > that currently has a boot lockup issue (if so, please let me know.)
> 
> It appears the issue people are seeing is with Slack 10, which shipped with 
> udev 0.26 - and I presume there was 'custom' rules Patrick had built in.

Ick.  Hm, there's not been any updates for slack since then? (note,
there was no 0.26 release, there are no '.' in udev releases.)

Any Slackware users want to pester them for updates?

thanks,

greg k-h

Re: 2.6.12 udev hangs at boot

[Nick Warne]

On Monday 20 June 2005 20:21, Greg KH wrote:

> > It appears the issue people are seeing is with Slack 10, which shipped
> > with udev 0.26 - and I presume there was 'custom' rules Patrick had built
> > in.
>
> Ick.  Hm, there's not been any updates for slack since then? (note,
> there was no 0.26 release, there are no '.' in udev releases.)
>
> Any Slackware users want to pester them for updates?

Remember this is Slackware 10 here I am talking about - Slackware 10.1 has 
been released since, that uses as stock udev 50.  Slackware current uses udev 
54.  Trouble is here also, GLIBC has been updated in latest Slackware[s], so 
there is no real upgrade path for Slack 10 users other than the whole 
caboodle - which breaks a lot if you have all the latest 'other stuff' built 
from source anyway.

I guess many users don't upgrade all the system like I do to find these 
problems.  This appears to be just a gotcha for old Slackware 10 users like 
me.  Sometimes you read stuff about doing an upgrade, and unless it pokes yer 
eye out with a big stick you miss it... so it is isn't a big deal as long as 
people know about it - it's an easy fix.

Nick
-- 
"When you're chewing on life's gristle,
Don't grumble, Give a whistle..."

Re: 2.6.12 udev hangs at boot

[David Lang]

I ran into the same issue last week on fedora core 3 so it's not _just_ a 
slackware problem.

David Lang

On Mon, 20 Jun 2005, Nick Warne wrote:

> Date: Mon, 20 Jun 2005 20:32:30 +0100
> From: Nick Warne <nick@linicks.net>
> To: Greg KH <gregkh@suse.de>
> Cc: linux-kernel@vger.kernel.org
> Subject: Re: 2.6.12 udev hangs at boot
> 
> On Monday 20 June 2005 20:21, Greg KH wrote:
>
>>> It appears the issue people are seeing is with Slack 10, which shipped
>>> with udev 0.26 - and I presume there was 'custom' rules Patrick had built
>>> in.
>>
>> Ick.  Hm, there's not been any updates for slack since then? (note,
>> there was no 0.26 release, there are no '.' in udev releases.)
>>
>> Any Slackware users want to pester them for updates?
>
> Remember this is Slackware 10 here I am talking about - Slackware 10.1 has
> been released since, that uses as stock udev 50.  Slackware current uses udev
> 54.  Trouble is here also, GLIBC has been updated in latest Slackware[s], so
> there is no real upgrade path for Slack 10 users other than the whole
> caboodle - which breaks a lot if you have all the latest 'other stuff' built
> from source anyway.
>
> I guess many users don't upgrade all the system like I do to find these
> problems.  This appears to be just a gotcha for old Slackware 10 users like
> me.  Sometimes you read stuff about doing an upgrade, and unless it pokes yer
> eye out with a big stick you miss it... so it is isn't a big deal as long as
> people know about it - it's an easy fix.
>
> Nick
> -- 
> "When you're chewing on life's gristle,
> Don't grumble, Give a whistle..."
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>

-- 
There are two ways of constructing a software design. One way is to make it so simple that there are obviously no deficiencies. And the other way is to make it so complicated that there are no obvious deficiencies.
  -- C.A.R. Hoare

Re: 2.6.12 udev hangs at boot

[Greg KH]

On Mon, Jun 20, 2005 at 12:42:39PM -0700, David Lang wrote:
> I ran into the same issue last week on fedora core 3 so it's not _just_ a 
> slackware problem.

FC 3, with default udev install will not boot with 2.6.12?  What version
of udev is the latest for FC3?

thanks,

greg k-h

Re: 2.6.12 udev hangs at boot

[David Lang]

I'll check tomorrow, I was useing what was on the FC3 iso images, it would 
eventually boot, but it would hang on udev with a 2.6.11.x or 2.6.12-pre6 
kernel for 2-3 min before continueing through the boot.

David Lang

  On Mon, 20 Jun 
2005, Greg KH wrote:

> Date: Mon, 20 Jun 2005 23:28:43 -0700
> From: Greg KH <gregkh@suse.de>
> To: David Lang <david.lang@digitalinsight.com>
> Cc: Nick Warne <nick@linicks.net>, linux-kernel@vger.kernel.org
> Subject: Re: 2.6.12 udev hangs at boot
> 
> On Mon, Jun 20, 2005 at 12:42:39PM -0700, David Lang wrote:
>> I ran into the same issue last week on fedora core 3 so it's not _just_ a
>> slackware problem.
>
> FC 3, with default udev install will not boot with 2.6.12?  What version
> of udev is the latest for FC3?
>
> thanks,
>
> greg k-h
>

-- 
There are two ways of constructing a software design. One way is to make it so simple that there are obviously no deficiencies. And the other way is to make it so complicated that there are no obvious deficiencies.
  -- C.A.R. Hoare

Re: 2.6.12 udev hangs at boot

[Greg KH]

On Mon, Jun 20, 2005 at 11:31:15PM -0700, David Lang wrote:
> I'll check tomorrow, I was useing what was on the FC3 iso images, it would 
> eventually boot, but it would hang on udev with a 2.6.11.x or 2.6.12-pre6 
> kernel for 2-3 min before continueing through the boot.

Ok, if you saw this on 2.6.11.x, then it's not the same udev issue that
was caused in the 2.6.12-preX tree.  That must be something different.

thanks,

greg k-h

Re: 2.6.12 udev hangs at boot

[Jesper Juhl]

On 6/20/05, Greg KH <gregkh@suse.de> wrote:
> On Mon, Jun 20, 2005 at 08:00:08PM +0100, Nick Warne wrote:
> > On Monday 20 June 2005 18:34, you wrote:
> >
> > > As for working with people's boxes, only the very oldest versions of
> > > udev (like the reported 030 version which is a year old and I do not
> > > think shipped by any distro) would have the "lockup" issue.  On all of
> > > the other ones, only custom rules written by users would have issues
> > > (meaning, not work).  I do not know of any shipping, supported distro
> > > that currently has a boot lockup issue (if so, please let me know.)
> >
> > It appears the issue people are seeing is with Slack 10, which shipped with
> > udev 0.26 - and I presume there was 'custom' rules Patrick had built in.
> 
> Ick.  Hm, there's not been any updates for slack since then? (note,
> there was no 0.26 release, there are no '.' in udev releases.)
> 
There has been updates since then.
There's been a newer stable Slackware release since 10.0. Slackware
10.1 was released on  2005-02-06 and ships udev 050.
Slackware-current (the unstable/development branch) is currently using udev 054.
Both the udev 050 and udev 054 Slackware packages should install fine
on a Slackware 10.0 box as far as I can see. So people just need to
upgrade their distro, or at least the udev bit of it :)
I'm running Slackware-current here with udev 054 and everything is just fine.

-- 
Jesper Juhl <jesper.juhl@gmail.com>
Don't top-post  http://www.catb.org/~esr/jargon/html/T/top-post.html
Plain text mails only, please      http://www.expita.com/nomime.html

Re: 2.6.12 udev hangs at boot

[Andrew Haninger]

On 6/20/05, Nick Warne <nick@linicks.net> wrote:
> It appears the issue people are seeing is with Slack 10, which shipped with
> udev 0.26 - and I presume there was 'custom' rules Patrick had built in.
I was going to point this out, but Slack 10.1 seems to have shipped
with v0.50. While 10.1 would be thought to be an upgrade to 10.0, any
experienced Slackware user knows that upgrades are discourages. So,
you'd be very likely to find many Slackware 10 machines out there
(actually, I still have v9.1 on one of my laptops).

Anyway, something someplace obvious (e.g. a note in Changes in the
area where it tells what minimum package versions are needed) would be
helpful.

-Andy

Re: 2.6.12 udev hangs at boot

[Josh Boyer]

On Mon, 2005-06-20 at 10:34 -0700, Greg KH wrote:
> On Mon, Jun 20, 2005 at 01:24:23PM -0400, Jeff Garzik wrote:
> > Greg KH wrote:
> > >On Mon, Jun 20, 2005 at 01:04:10PM +0300, Denis Vlasenko wrote:
> > >
> > >>Greg, any plans to distribute udev and hotplug within kernel tarballs
> > >>so that people do not need to track such changes continuously?
> > >
> > >
> > >Nope.  But if you use udev, you should read the announcements for new
> > >releases, as I did say this was required for 2.6.12, and gave everyone a
> > >number of weeks notice :)
> > 
> > Since udev is required for booting, it sounds like you're putting people 
> > in an upgrade-or-no-boot situation.
> 
> Well, they don't _have_ to upgrade their kernel :)
> 
> > That's lame.  The kernel should support udev's out in the field, on 
> > people's boxes (RHEL, SLES?, Fedora, ...).
> 
> This was caused by an unfortunate assumption in older versions of udev
> about what was contained in the sysfs tree.  udev is now fixed to not
> make that assumption.  So this was not a kernel bug, but a udev/libsysfs
> one (and I wasn't going to keep the old kernel behavior for this minor
> issue.)
> 
> As for working with people's boxes, only the very oldest versions of
> udev (like the reported 030 version which is a year old and I do not
> think shipped by any distro) would have the "lockup" issue.  On all of
> the other ones, only custom rules written by users would have issues
> (meaning, not work).  I do not know of any shipping, supported distro
> that currently has a boot lockup issue (if so, please let me know.)

SLES 9 shipped with udev-021.

http://www.novell.com/products/linuxpackages/enterpriseserver/i386/udev.html

Is that effected by this?

josh

Re: 2.6.12 udev hangs at boot

[Greg KH]

On Mon, Jun 20, 2005 at 02:49:28PM -0500, Josh Boyer wrote:
> 
> SLES 9 shipped with udev-021.
> 
> http://www.novell.com/products/linuxpackages/enterpriseserver/i386/udev.html
> 
> Is that effected by this?

First off, why would you be running a kernel.org kernel on an
"enterprise" distro?  I wouldn't recommend that at all, and you probably
loose your support contract if you try that out.

Secondly, no, it is not, see the previous post in this thread.

thanks,

greg k-h

Re: 2.6.12 udev hangs at boot

[Olaf Hering]

 On Mon, Jun 20, Jeff Garzik wrote:

> That's lame.  The kernel should support udev's out in the field, on 
> people's boxes (RHEL, SLES?, Fedora, ...).

The udev package as shipped with SLES9 works just fine with every kernel
version. Modulo the scsi targetN changes in sysfs layout, but that
doesnt prevent booting.

Re: 2.6.12 udev hangs at boot

[Daniel Barkalow]

On Mon, 20 Jun 2005, Greg KH wrote:

> On Mon, Jun 20, 2005 at 01:04:10PM +0300, Denis Vlasenko wrote:
> > 
> > Greg, any plans to distribute udev and hotplug within kernel tarballs
> > so that people do not need to track such changes continuously?
> 
> Nope.  But if you use udev, you should read the announcements for new
> releases, as I did say this was required for 2.6.12, and gave everyone a
> number of weeks notice :)

Shouldn't this be listed in Changes? It looks like Changes only mentions
the existance of udev, but doesn't specify a required version, despite
there being a version requirement. (Not that I really think too many
people would think to look, but it would at least have a chance of
reaching people who look carefully at kernels but don't read the mailing
list)

	-Daniel
*This .sig left intentionally blank*

Re: 2.6.12 udev hangs at boot

[Jesper Juhl]

On Mon, 20 Jun 2005, Daniel Barkalow wrote:

> On Mon, 20 Jun 2005, Greg KH wrote:
> 
> > On Mon, Jun 20, 2005 at 01:04:10PM +0300, Denis Vlasenko wrote:
> > > 
> > > Greg, any plans to distribute udev and hotplug within kernel tarballs
> > > so that people do not need to track such changes continuously?
> > 
> > Nope.  But if you use udev, you should read the announcements for new
> > releases, as I did say this was required for 2.6.12, and gave everyone a
> > number of weeks notice :)
> 
> Shouldn't this be listed in Changes? It looks like Changes only mentions
> the existance of udev, but doesn't specify a required version, despite
> there being a version requirement. (Not that I really think too many
> people would think to look, but it would at least have a chance of
> reaching people who look carefully at kernels but don't read the mailing
> list)
> 
Sounds like a good idea to me. Here's a patch. Does this look OK to you 
Greg?  
I was tempted to write udev 031 or so, since Documentation/Changes is 
supposed to list absolute minimum versions, and as far as I could read 
from your previous posts 030 would have been the last one with the lock-up 
bug, but then I read this little bit in the udev 058 announcement:
	"Note, if you are running a kernel newer than 2.6.12-rc4 
	 (including the -mm releases) and you have any custom udev rules, 
	 you MUST upgrade to the latest version to allow udev to work 
	 properly."
And given that the latest release at that time was 058 (since this is in 
the announcement of that release), I figured that the minimum requirement 
for flawless operation actually is udev 058 for the 2.6.12 and future 
kernels - so that's what I ended up writing.


Add required udev version to Documentation/Changes

Signed-off-by: Jesper Juhl <juhl-lkml@dif.dk>
---

 Documentation/Changes |    1 +
 1 files changed, 1 insertion(+)

--- linux-2.6.12-orig/Documentation/Changes	2005-06-17 21:48:29.000000000 +0200
+++ linux-2.6.12/Documentation/Changes	2005-06-21 00:07:17.000000000 +0200
@@ -64,6 +64,7 @@
 o  nfs-utils              1.0.5                   # showmount --version
 o  procps                 3.2.0                   # ps --version
 o  oprofile               0.5.3                   # oprofiled --version
+o  udev                   058                     # udevinfo -V
 
 Kernel compilation
 ==================

Re: 2.6.12 udev hangs at boot

[Tomasz Torcz]

On Mon, Jun 20, 2005 at 05:26:46PM -0400, Daniel Barkalow wrote:
> On Mon, 20 Jun 2005, Greg KH wrote:
> 
> > On Mon, Jun 20, 2005 at 01:04:10PM +0300, Denis Vlasenko wrote:
> > > 
> > > Greg, any plans to distribute udev and hotplug within kernel tarballs
> > > so that people do not need to track such changes continuously?
> > 
> > Nope.  But if you use udev, you should read the announcements for new
> > releases, as I did say this was required for 2.6.12, and gave everyone a
> > number of weeks notice :)
> 
> Shouldn't this be listed in Changes? It looks like Changes only mentions
> the existance of udev, but doesn't specify a required version, despite
> there being a version requirement.

 Probably.

 Add udev version requirement to Changes files.

Signed-off-by: Tomasz Torcz <zdzichu@irc.pl>

--- linux-2.6.12/Documentation/Changes  2005-05-07 10:56:52.000000000 +0200
+++ linux-fixed/Documentation/Changes   2005-06-21 00:19:58.000000000 +0200
@@ -64,6 +64,7 @@ o  isdn4k-utils           3.1pre1       
 o  nfs-utils              1.0.5                   # showmount --version
 o  procps                 3.2.0                   # ps --version
 o  oprofile               0.5.3                   # oprofiled --version
+o  udev                   058                     # udevinfo -V
 
 Kernel compilation
 ==================


-- 
Tomasz Torcz                 Morality must always be based on practicality.
zdzichu@irc.-nie.spam-.pl                -- Baron Vladimir Harkonnen

Re: 2.6.12 udev hangs at boot

[Nick Warne]

Andrew Walrond wrote:

> On Saturday 18 June 2005 14:44, Nick Warne wrote:
>> >
>> > I had this problem because I was running an ancient version of udev
>> > (0.34, versus 0.58, at the time..). Try upgrading udev if it's out of
>> > date.
>>
>> Thanks, that worked :-)
>>
> 
> FWIW In the udev 058 announcement, Greg said:
> 
> "Note, if you are running a kernel newer than 2.6.12-rc4 (including the
> -mm releases) and you have any custom udev rules, you MUST upgrade to
> the latest version to allow udev to work properly.  This change happened
> because of a previously-unrealized reliance in libsysfs on the presence
> of a useless sysfs file that has recently been removed.  Hopefully the
> libsysfs people will be releasing a new version shortly with this change
> in it for those packages who rely on it."
> 
> Just a reminder because I bet many people will get caught out by this!

Yes - and also the download link on the main udev page is broke :-/

Here is where to get the latest udev build:

http://www.kernel.org/pub/linux/utils/kernel/hotplug/

Nick
-- 
"When you're chewing on life's gristle,
Don't grumble, Give a whistle..."

Re: 2.6.12 udev hangs at boot

[Greg KH]

On Sat, Jun 18, 2005 at 04:32:44PM +0100, Nick Warne wrote:
> 
> Yes - and also the download link on the main udev page is broke :-/

How is it broken?  Works for me.  What do you get?

thanks,

greg k-h

Re: 2.6.12 udev hangs at boot

[Nick Warne]

Andrew Haninger wrote:

> Anyway, just a heads-up to anyone else experiencing a breaking of
> 'less' and missing /dev files.

Yep... I had 'less' break too (you will find 'man' is broke also, rolling on 
from that).

It turns out to be a problem (typo?) in  /etc/udev/rules.d/udev.rules

Try changing:

# pty devices
KERNEL="pty[p-za-e][0-9a-f]*", NAME="pty/m%n", SYMLINK="%k"
KERNEL="tty[p-za-e][0-9a-f]*", NAME="tty/s%n", SYMLINK="%k"

to:

# pty devices
KERNEL="pty[p-za-e][0-9a-f]*", NAME="pty/m%n", SYMLINK="%k"
KERNEL="tty[p-za-e][0-9a-f]*", NAME="pty/s%n", SYMLINK="%k"

(change is in second line tty -> pty)

As to the missing /dev/ entries - remember you are using udev now - they 
appear 'on the fly' as and when you plug something in - ensure you have set 
'hotplug' to start.

Nick
-- 
"When you're chewing on life's gristle,
Don't grumble, Give a whistle..."

Re: 2.6.12 udev hangs at boot

[Andrew Haninger]

On 6/19/05, Nick Warne <nick@linicks.net> wrote:
> As to the missing /dev/ entries - remember you are using udev now - they
> appear 'on the fly' as and when you plug something in - ensure you have set
> 'hotplug' to start.
I have set hotplug to start on Slackware 10 (chmod a+x
/etc/rc.d/rc.hotplug) and it seems to start fine. When running the
default udev (v0.26) it loads all that I need for my external hard
disk except for sd_mod. If I manually modprobed sd_mod, /dev/sda and
/dev/sda1 would be automatically created, which was nice. With the
latest version of udev (v0.58) it loads sd_mod when I plug in the
drive, but /dev/sd** are not created dynamically (they are there all
the time). Having them created dynamically is nifty, but not necessary
for me. It would be nice if it was a little more brainless (e.g. /dev
entries are created on the fly by default) since I don't know how to
turn on/off dynamic creation of /dev entries.

Also, starting hotplug on boot gives a (slight) delay on my system.
This might be normal, but I'm not sure.

I'm probably not understanding something correctly, but this is what I'm seeing.

Thanks.

Re: 2.6.12 udev hangs at boot

[Greg KH]

On Sun, Jun 19, 2005 at 04:39:27PM +0100, Nick Warne wrote:
> Andrew Haninger wrote:
> 
> > Anyway, just a heads-up to anyone else experiencing a breaking of
> > 'less' and missing /dev files.
> 
> Yep... I had 'less' break too (you will find 'man' is broke also, rolling on 
> from that).
> 
> It turns out to be a problem (typo?) in  /etc/udev/rules.d/udev.rules
> 
> Try changing:
> 
> # pty devices
> KERNEL="pty[p-za-e][0-9a-f]*", NAME="pty/m%n", SYMLINK="%k"
> KERNEL="tty[p-za-e][0-9a-f]*", NAME="tty/s%n", SYMLINK="%k"
> 
> to:
> 
> # pty devices
> KERNEL="pty[p-za-e][0-9a-f]*", NAME="pty/m%n", SYMLINK="%k"
> KERNEL="tty[p-za-e][0-9a-f]*", NAME="pty/s%n", SYMLINK="%k"
> 
> (change is in second line tty -> pty)

Hm, that's what already ships with the udev tarball in the gentoo rule
set (which is usually the most up-to-date rule set in the tarball.)

Which one are you looking at?

> As to the missing /dev/ entries - remember you are using udev now - they 
> appear 'on the fly' as and when you plug something in - ensure you have set 
> 'hotplug' to start.

Not necessarily, people are getting udev working and full hotplug
support by setting /sbin/hotplug to NULL.  Ah, the magic of netlink...

thanks,

greg k-h

Re: 2.6.12 udev hangs at boot

[Nick Warne]

On Tuesday 21 June 2005 07:45, you wrote:

> > It turns out to be a problem (typo?) in  /etc/udev/rules.d/udev.rules
> >
> > Try changing:
> >
> > # pty devices
> > KERNEL="pty[p-za-e][0-9a-f]*", NAME="pty/m%n", SYMLINK="%k"
> > KERNEL="tty[p-za-e][0-9a-f]*", NAME="tty/s%n", SYMLINK="%k"
> >
> > to:
> >
> > # pty devices
> > KERNEL="pty[p-za-e][0-9a-f]*", NAME="pty/m%n", SYMLINK="%k"
> > KERNEL="tty[p-za-e][0-9a-f]*", NAME="pty/s%n", SYMLINK="%k"
> >
> > (change is in second line tty -> pty)
>
> Hm, that's what already ships with the udev tarball in the gentoo rule
> set (which is usually the most up-to-date rule set in the tarball.)
>
> Which one are you looking at?

This wasn't my 'solution' - I googled:

http://kerneltrap.org/node/4474

I don't understand why 'less' broke at all...

Nick
-- 
"When you're chewing on life's gristle,
Don't grumble, Give a whistle..."