DRIVER vs. PHYSDEVDRIVER for block susbsystem

DRIVER vs. PHYSDEVDRIVER for block susbsystem

[Andrey Borzenkov]

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

The example persistent rules distributed with udev use DRIVER matching for 
SUBSYSTEM=block. Unfortunately, it appears that DRIVER is simply not defined 
in this case - it is fetched only for the case when devpath is /devices/* 
not /block/*.

It effectively means that for IDE CD-ROMs persistent rules are never executed, 
see also http://qa.mandriva.com/show_bug.cgi?id!720.

Is DRIVER expected to be set for SUBSYSTEM=block (I will submit patch)? 
Otherwise it would be nice to fix examples in rules.

Thank you

- -andrey
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (GNU/Linux)

iD8DBQFEJYr8R6LMutpd94wRAnMcAJ0RrAN2rXMMY7BcemqyT8t30Biz4gCgxeMD
b5KrUSOujKE0FAcFkaUD5Go=wsdk
-----END PGP SIGNATURE-----


-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid\x110944&bid$1720&dat\x121642
_______________________________________________
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: DRIVER vs. PHYSDEVDRIVER for block susbsystem

[Kay Sievers]

On Sat, Mar 25, 2006 at 09:24:30PM +0300, Andrey Borzenkov wrote:
> The example persistent rules distributed with udev use DRIVER matching for 
> SUBSYSTEM=block. Unfortunately, it appears that DRIVER is simply not defined 
> in this case - it is fetched only for the case when devpath is /devices/* 
> not /block/*.

No DRIVER matches up the whole chain of devices. But the driver value of
block is "" which matches successfully the non-match of the values it is
compared against.

The rules have changed and should work that way:
  http://www.kernel.org/git/?p=linux/hotplug/udev.git;a=blob;hb=HEAD;f=etc/udev/60-persistent-storage.rules

> It effectively means that for IDE CD-ROMs persistent rules are never executed, 
> see also http://qa.mandriva.com/show_bug.cgi?id!720.

We should not introduce any new users of the "device" link or the
PHYSDEV* values, they will go away some day along with the odd idea
to distinguish between class and physical devices.

> Is DRIVER expected to be set for SUBSYSTEM=block (I will submit patch)? 
> Otherwise it would be nice to fix examples in rules.

No, class devices don't have a driver for now.

Thanks,
Kay


-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid\x110944&bid$1720&dat\x121642
_______________________________________________
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: DRIVER vs. PHYSDEVDRIVER for block susbsystem

[Andrey Borzenkov]

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

On Saturday 25 March 2006 21:50, Kay Sievers wrote:
> On Sat, Mar 25, 2006 at 09:24:30PM +0300, Andrey Borzenkov wrote:
> > The example persistent rules distributed with udev use DRIVER matching
> > for SUBSYSTEM=block. Unfortunately, it appears that DRIVER is simply not
> > defined in this case - it is fetched only for the case when devpath is
> > /devices/* not /block/*.
>
> No DRIVER matches up the whole chain of devices. But the driver value of
> block is "" which matches successfully the non-match of the values it is
> compared against.
>
> The rules have changed and should work that way:
>  
> http://www.kernel.org/git/?p=linux/hotplug/udev.git;a=blob;hb=HEAD;f=etc/ud
>ev/60-persistent-storage.rules
>

Those rules will never match for ide-cs or ide-floppy as intended (while this 
"fixes" it for ide-cdrom, sure :)

> We should not introduce any new users of the "device" link or the
> PHYSDEV* values, they will go away some day along with the odd idea
> to distinguish between class and physical devices.
>
> > Is DRIVER expected to be set for SUBSYSTEM=block (I will submit patch)?
> > Otherwise it would be nice to fix examples in rules.
>
> No, class devices don't have a driver for now.
>

In this case adding driver for block devices looks like the right thing, does 
not it? Strictly speaking, it could be generalized to "if device symlink 
exists, fetch driver from it"; this will go away when sysfs layout will have 
changed.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (GNU/Linux)

iD8DBQFEJZWvR6LMutpd94wRAjEYAKDQ/WZVjsAOy8ag7K5Pya4F07+qHQCbBnOM
zv1UWpLC8ax5HoZSCS0RbQw=XkPH
-----END PGP SIGNATURE-----


-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid\x110944&bid$1720&dat\x121642
_______________________________________________
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: DRIVER vs. PHYSDEVDRIVER for block susbsystem

[Kay Sievers]

On Sat, Mar 25, 2006 at 10:10:38PM +0300, Andrey Borzenkov wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On Saturday 25 March 2006 21:50, Kay Sievers wrote:
> > On Sat, Mar 25, 2006 at 09:24:30PM +0300, Andrey Borzenkov wrote:
> > > The example persistent rules distributed with udev use DRIVER matching
> > > for SUBSYSTEM=block. Unfortunately, it appears that DRIVER is simply not
> > > defined in this case - it is fetched only for the case when devpath is
> > > /devices/* not /block/*.
> >
> > No DRIVER matches up the whole chain of devices. But the driver value of
> > block is "" which matches successfully the non-match of the values it is
> > compared against.
> >
> > The rules have changed and should work that way:
> >  
> > http://www.kernel.org/git/?p=linux/hotplug/udev.git;a=blob;hb=HEAD;f=etc/ud
> >ev/60-persistent-storage.rules
> >
> 
> Those rules will never match for ide-cs or ide-floppy as intended (while this 
> "fixes" it for ide-cdrom, sure :)

Care to explain?

> > We should not introduce any new users of the "device" link or the
> > PHYSDEV* values, they will go away some day along with the odd idea
> > to distinguish between class and physical devices.
> >
> > > Is DRIVER expected to be set for SUBSYSTEM=block (I will submit patch)?
> > > Otherwise it would be nice to fix examples in rules.
> >
> > No, class devices don't have a driver for now.
> >
> 
> In this case adding driver for block devices looks like the right thing, does 
> not it? Strictly speaking, it could be generalized to "if device symlink 
> exists, fetch driver from it"; this will go away when sysfs layout will have 
> changed.

No, the block devices don't have a driver.

Thanks,
Kay


-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid\x110944&bid$1720&dat\x121642
_______________________________________________
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: DRIVER vs. PHYSDEVDRIVER for block susbsystem

[Andrey Borzenkov]

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

On Saturday 25 March 2006 22:24, Kay Sievers wrote:

> > > No DRIVER matches up the whole chain of devices. But the driver value
> > > of block is "" which matches successfully the non-match of the values
> > > it is compared against.
> > >
> > > The rules have changed and should work that way:
> > >
> > > http://www.kernel.org/git/?p=linux/hotplug/udev.git;a=blob;hb=HEAD;f=et
> > >c/ud ev/60-persistent-storage.rules
> >
> > Those rules will never match for ide-cs or ide-floppy as intended (while
> > this "fixes" it for ide-cdrom, sure :)
>
> Care to explain?
>

Sure. As you stated before, 'driver value of block is ""'. The "" can never 
match "ide-cd|ide-floppy". And non-block are filtered away two lines before.

> > > We should not introduce any new users of the "device" link or the
> > > PHYSDEV* values, they will go away some day along with the odd idea
> > > to distinguish between class and physical devices.
[...]
> > In this case adding driver for block devices looks like the right thing,
> > does not it? Strictly speaking, it could be generalized to "if device
> > symlink exists, fetch driver from it"; this will go away when sysfs
> > layout will have changed.
>
> No, the block devices don't have a driver.
>

Is not this the "odd idea to distinguish between class and physical 
device"? :)

Regards

- -andrey
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (GNU/Linux)

iD8DBQFEJZs+R6LMutpd94wRAhlhAKCBf5wJk8YF7mAKGasYGKk4QooaBACeMz6M
3JKACK6OUVHaepBjRDuGaEUj1H
-----END PGP SIGNATURE-----


-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid\x110944&bid$1720&dat\x121642
_______________________________________________
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: DRIVER vs. PHYSDEVDRIVER for block susbsystem

[Kay Sievers]

On Sat, Mar 25, 2006 at 10:34:21PM +0300, Andrey Borzenkov wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On Saturday 25 March 2006 22:24, Kay Sievers wrote:
> 
> > > > No DRIVER matches up the whole chain of devices. But the driver value
> > > > of block is "" which matches successfully the non-match of the values
> > > > it is compared against.
> > > >
> > > > The rules have changed and should work that way:
> > > >
> > > > http://www.kernel.org/git/?p=linux/hotplug/udev.git;a=blob;hb=HEAD;f=et
> > > >c/ud ev/60-persistent-storage.rules
> > >
> > > Those rules will never match for ide-cs or ide-floppy as intended (while
> > > this "fixes" it for ide-cdrom, sure :)
> >
> > Care to explain?
> >
> 
> Sure. As you stated before, 'driver value of block is ""'. The "" can never 
> match "ide-cd|ide-floppy". And non-block are filtered away two lines before.

No, DRIVER matches up the chain of all devices, it has nothing to do
with the event a non-block device. Did you test it?

> > > > We should not introduce any new users of the "device" link or the
> > > > PHYSDEV* values, they will go away some day along with the odd idea
> > > > to distinguish between class and physical devices.
> [...]
> > > In this case adding driver for block devices looks like the right thing,
> > > does not it? Strictly speaking, it could be generalized to "if device
> > > symlink exists, fetch driver from it"; this will go away when sysfs
> > > layout will have changed.
> >
> > No, the block devices don't have a driver.
> >
> 
> Is not this the "odd idea to distinguish between class and physical 
> device"? :)

No, it's not. The correct value for the driver of a blockdev is "", as
there is no driver. Copying the value from the parent would be wrong.

Kay


-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid\x110944&bid$1720&dat\x121642
_______________________________________________
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: DRIVER vs. PHYSDEVDRIVER for block susbsystem

[Andrey Borzenkov]

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

On Saturday 25 March 2006 22:43, Kay Sievers wrote:
> No, DRIVER matches up the chain of all devices, it has nothing to do
> with the event a non-block device. Did you test it?
>

you are right, sorry. Thank you for your patience.

- -andrey
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (GNU/Linux)

iD8DBQFEJaMCR6LMutpd94wRArgIAJ0U5nQC6e7j3HYUB6LtHzmMgWTZeACg0SIQ
YENMLY3ysD/H7UtGtUeWV9s=RWA8
-----END PGP SIGNATURE-----


-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid\x110944&bid$1720&dat\x121642
_______________________________________________
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