Where are block@... database entries?

Where are block@... database entries?

[Eugene Crosser]

I run slackware-current, which keeps pretty much up to date with udev
(currently comes with udev 0.64).  Since some time ago, maybe in Spring,
rc.udev/udevstart stopped creating block@... entries in the
/dev/.udev... database.  Now, only class@... files are created by udevstart.

As a result of that (I think), HAL does not recognize CDROM drive as
such, and do not poll it.  Consequently, gnome-volume-manager do not
automount an inserved disk.

On a system where there are old /dev/.udev.tdb/block@* entries lingering
since April, automount works.

Any advice/ideas how to make it work?

And BTW, I *think* that "udev_log" in udev.conf (and maybe other
entries?) should be uppercased.  This file is sourced into shell
scripts, and then the binaries such as udevstart check for enviroment
variable UDEV_LOG, in uppercase.  Do I miss something?

Thanks
Eugene


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: Where are block@... database entries?

[Marco d'Itri]

On Sep 07, Eugene Crosser <crosser@average.org> wrote:

> Any advice/ideas how to make it work?
You need to upgrade HAL.

> And BTW, I *think* that "udev_log" in udev.conf (and maybe other
> entries?) should be uppercased.  This file is sourced into shell
> scripts, and then the binaries such as udevstart check for enviroment
> variable UDEV_LOG, in uppercase.  Do I miss something?
Yes. The file is also a real configuration file of the programs.

-- 
ciao,
Marco


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: Where are block@... database entries?

[Kay Sievers]

On Wed, Sep 07, 2005 at 11:11:35AM +0200, Marco d'Itri wrote:
> On Sep 07, Eugene Crosser <crosser@average.org> wrote:
> 
> > Any advice/ideas how to make it work?
> You need to upgrade HAL.

Yes, see the RELEASE-NOTES for 061:
  http://www.kernel.org/git/gitweb.cgi?p=linux/hotplug/udev.git;a=blob;hb=HEAD;f=RELEASE-NOTES

At the same time HAL was changed and released to work with the new udev.

Note:
Randomly upgrading critical parts of the system, without resolving all
package requirements for the  kernel, udev, hal, hal-susbsytem-managers
like NetworkManager, dbus, desktop-integration like gnome-volume-manager,
will probably break your setup.

We have complex interdependencies today and we will get even more of this
while moving forward providing a better hardware integration into the
whole system. We just started to move in that direction, so be prepared ... :)

Distros will need to have a well defined release cycle and package maintainers
which keep the package requirements up-to-date and make coordinated efforts
to synchronize the package versions, otherwise they may not be able to provide
a working setup.
HAL depends on a certain udev version and udev depends on a specific
kernel version. Upgrading only a subset of the packages may render
a part or the whole system unusable.

> > And BTW, I *think* that "udev_log" in udev.conf (and maybe other
> > entries?) should be uppercased.  This file is sourced into shell
> > scripts, and then the binaries such as udevstart check for enviroment
> > variable UDEV_LOG, in uppercase.  Do I miss something?
> Yes. The file is also a real configuration file of the programs.

Right! All the binary udev tools parse this file. This is the primary
use of this file, while sourcing it in with a shell works too.
(UDEV_LOG is exported from the udev tools after reading it from the
config file, but with the log priority in its numerial representation
not the text used in the config.)

Kay


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: Where are block@... database entries?

[Eugene Crosser]

Kay Sievers wrote:

>>>Any advice/ideas how to make it work?
>>
>>You need to upgrade HAL.
> 
> Yes, see the RELEASE-NOTES for 061:
>   http://www.kernel.org/git/gitweb.cgi?p=linux/hotplug/udev.git;a=blob;hb=HEAD;f=RELEASE-NOTES
> 
> At the same time HAL was changed and released to work with the new udev.

There is no indication in the release notes on which version of HAL is
appropriate.  I am running 0.5.4, and it does not work, do I need to
upgrade?

> Note:
> Randomly upgrading critical parts of the system, without resolving all
> package requirements for the  kernel, udev, hal, hal-susbsytem-managers
> like NetworkManager, dbus, desktop-integration like gnome-volume-manager,
> will probably break your setup.

It's hard to break what does not work anyway...  The problem is that
udev is a part of slackware distribution, but hal is not, and it has to
come with gnome (FRG in my case).  Now I am trying to figure a working
setup, to be able to advice FRG folks after that.  Existing documents on
the web are hopelessly out of date, so please bear with me.

>>>And BTW, I *think* that "udev_log" in udev.conf (and maybe other
>>>entries?) should be uppercased.  This file is sourced into shell
>>>scripts, and then the binaries such as udevstart check for enviroment
>>>variable UDEV_LOG, in uppercase.  Do I miss something?
>>
>>Yes. The file is also a real configuration file of the programs.
> 
> Right! All the binary udev tools parse this file. This is the primary
> use of this file, while sourcing it in with a shell works too.
> (UDEV_LOG is exported from the udev tools after reading it from the
> config file, but with the log priority in its numerial representation
> not the text used in the config.)

I just noticed that when I change udev_log in udev.conf, udevstart does
*not* become more verbose as I expected.  When I set UDEV_LOG
environment variable and then run udevstart, it *does* become verbose.
I figured that this looks like a bug...

Eugene


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: Where are block@... database entries?

[Kay Sievers]

On Wed, Sep 07, 2005 at 05:26:10PM +0400, Eugene Crosser wrote:
> Kay Sievers wrote:
> 
> >>>Any advice/ideas how to make it work?
> >>
> >>You need to upgrade HAL.
> > 
> > Yes, see the RELEASE-NOTES for 061:
> >   http://www.kernel.org/git/gitweb.cgi?p=linux/hotplug/udev.git;a=blob;hb=HEAD;f=RELEASE-NOTES
> > 
> > At the same time HAL was changed and released to work with the new udev.
> 
> There is no indication in the release notes on which version of HAL is
> appropriate.  I am running 0.5.4, and it does not work, do I need to
> upgrade?

No, 0.5.4 should work. Here is the HAL ChangeLog:
  http://cvs.freedesktop.org/hal/hal/ChangeLog?rev=1.546&view=markup

> > Note:
> > Randomly upgrading critical parts of the system, without resolving all
> > package requirements for the  kernel, udev, hal, hal-susbsytem-managers
> > like NetworkManager, dbus, desktop-integration like gnome-volume-manager,
> > will probably break your setup.
> 
> It's hard to break what does not work anyway...  The problem is that
> udev is a part of slackware distribution, but hal is not, and it has to
> come with gnome (FRG in my case).  Now I am trying to figure a working
> setup, to be able to advice FRG folks after that.  Existing documents on
> the web are hopelessly out of date, so please bear with me.

Looking at the HAL debug output may show something:
  http://hal.freedesktop.org/wiki/Software_2fHalTraces

> >>>And BTW, I *think* that "udev_log" in udev.conf (and maybe other
> >>>entries?) should be uppercased.  This file is sourced into shell
> >>>scripts, and then the binaries such as udevstart check for enviroment
> >>>variable UDEV_LOG, in uppercase.  Do I miss something?
> >>
> >>Yes. The file is also a real configuration file of the programs.
> > 
> > Right! All the binary udev tools parse this file. This is the primary
> > use of this file, while sourcing it in with a shell works too.
> > (UDEV_LOG is exported from the udev tools after reading it from the
> > config file, but with the log priority in its numerial representation
> > not the text used in the config.)
> 
> I just noticed that when I change udev_log in udev.conf, udevstart does
> *not* become more verbose as I expected.  When I set UDEV_LOG
> environment variable and then run udevstart, it *does* become verbose.
> I figured that this looks like a bug...

Yes, it's a special case for udevstart. Enabling debugging output in
udevstart may cause udevstart to take several minutes to finish on
systems with "sync" syslog configurations. That may lead to a system
unable to bootup correctly. udevstart ignores the setting from the config
file, you can only debug it on the commandline. All other udev tools don't
override the setting.

See line 351:
  http://www.kernel.org/git/?p=linux/hotplug/udev.git;a=blob;hÎ96f38879186c81d185f520393a2b18c4428166;hb=HEAD;f=udevstart.c#l351

Kay


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: Where are block@... database entries?

[Eugene Crosser]

Kay Sievers wrote:

> No, 0.5.4 should work. Here is the HAL ChangeLog:
>   http://cvs.freedesktop.org/hal/hal/ChangeLog?rev=1.546&view=markup
[...]
> Looking at the HAL debug output may show something:
>   http://hal.freedesktop.org/wiki/Software_2fHalTraces

Ok, at the very least, now I know where to look next.  Thanks for that! :-)

>  udevstart ignores the setting from the config
> file, you can only debug it on the commandline.

Ah, I undestand now.

Eugene


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: Where are block@... database entries?

[Kay Sievers]

On Wed, Sep 07, 2005 at 06:07:04PM +0400, Eugene Crosser wrote:
> Kay Sievers wrote:
> 
> > No, 0.5.4 should work. Here is the HAL ChangeLog:
> >   http://cvs.freedesktop.org/hal/hal/ChangeLog?rev=1.546&view=markup
> [...]
> > Looking at the HAL debug output may show something:
> >   http://hal.freedesktop.org/wiki/Software_2fHalTraces
> 
> Ok, at the very least, now I know where to look next.  Thanks for that! :-)

Look with "lshal" if your cdrom is recognized by HAL. If not, but
udev has created a device node for it, post the debug HAL prints to the
console to:
  hal@lists.freedesktop.org

I will, or somebody else will catch it there.

Good luck,
Kay


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: Where are block@... database entries?

[Marco d'Itri]

[-- Attachment #1: Type: text/plain, Size: 355 bytes --]

On Sep 07, Kay Sievers <kay.sievers@vrfy.org> wrote:

> unable to bootup correctly. udevstart ignores the setting from the config
> file, you can only debug it on the commandline. All other udev tools don't
> override the setting.
I wondered about this myself... it looks like something which should be
prominently documented.

-- 
ciao,
Marco

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: Where are block@... database entries?

[Eugene Crosser]

[About CD/VDV not being mounted on insert]

Kay Sievers wrote:

>>>No, 0.5.4 should work. Here is the HAL ChangeLog:
>>>  http://cvs.freedesktop.org/hal/hal/ChangeLog?rev=1.546&view=markup
>>
>>[...]
>>
>>>Looking at the HAL debug output may show something:
>>>  http://hal.freedesktop.org/wiki/Software_2fHalTraces
>>
>>Ok, at the very least, now I know where to look next.  Thanks for that! :-)
> 
> 
> Look with "lshal" if your cdrom is recognized by HAL. If not, but
> udev has created a device node for it, post the debug HAL prints to the
> console to:
>   hal@lists.freedesktop.org

I figured the cause of HAL not realizing that /dev/hdX is a CDROM: it
was running as 'haldaemon' user in the group 'haldeamon', whereas
/dev/hd* belong to root:disk and root:cdrom with mode 660.  HAL was
trying to open devices, but failed.  When I specified in /etc/group that
haldaemon is a member of 'disk' and 'cdrom' groups, automount magically
began to work!

I wonder what would be the proper setup?  I don't seem to find any
relevant directions on the web...

...now to figure why USB disk is not automounted either...

Eugene


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: Where are block@... database entries?

[Seth W. Klein]

[-- Attachment #1: Type: text/plain, Size: 528 bytes --]

On Wed, Sep 07, 2005 at 06:30:26PM +0200, Marco d'Itri wrote:
> On Sep 07, Kay Sievers <kay.sievers@vrfy.org> wrote:
> 
> > unable to bootup correctly. udevstart ignores the setting from the config
> > file, you can only debug it on the commandline. All other udev tools don't
> > override the setting.
>
> I wondered about this myself... it looks like something which should be
> prominently documented.

I agree.  Patch attached.

Cheers,
Seth W. Klein
-- 
sk@sethwklein.net     AIM: sethwklein     http://www.sethwklein.net/

[-- Attachment #2: udev-068-udevstart-man-log.patch --]
[-- Type: text/plain, Size: 707 bytes --]

--- udevstart.8.orig	2005-09-11 16:11:34.000000000 -0400
+++ udevstart.8	2005-09-11 16:47:11.000000000 -0400
@@ -10,6 +10,19 @@
 to create the nodes for every valid device found. It can be used to fill a
 empty device directory with nodes for all devices currently available on
 the system.
+.SH "ENVIRONMENT"
+.TP
+.B UDEV_LOG
+Sets the log level.
+.B udevstart
+ignores the setting in /etc/udev/udev.conf
+because enabling debugging output
+on systems where syslog syncs the disk after writing each line
+may cause it to take several minutes to run.
+.P
+The complete list of environment variables used by
+.B udevstart
+is "documented" only by the source code.
 .SH "SEE ALSO"
 .BR udev (8)
 .SH AUTHORS