usbcore vs. udev

usbcore vs. udev

[Bill Nottingham]

usbcore seems to only generate udev events on removal; it never
generates add events. Any particular reason?

(2.6.10-acX, if it matters.)

Bill


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
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: usbcore vs. udev

[Kay Sievers]

On Tue, 2005-01-25 at 13:47 -0500, Bill Nottingham wrote:
> usbcore seems to only generate udev events on removal; it never
> generates add events. Any particular reason?

What kind of event do you expect? And how is that related to udev?

Thanks,
Kay



-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
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: usbcore vs. udev

[Greg KH]

On Tue, Jan 25, 2005 at 01:47:27PM -0500, Bill Nottingham wrote:
> usbcore seems to only generate udev events on removal; it never
> generates add events. Any particular reason?
> 
> (2.6.10-acX, if it matters.)

What type of udev events?  Those for the usb major number?  So things
like the USB printer driver don't work anymore?

What version of udev?  And does this happen in a "clean" 2.6.10?

thanks,

greg k-h


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
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: usbcore vs. udev

[Bill Nottingham]

Greg KH (greg@kroah.com) said: 
> On Tue, Jan 25, 2005 at 01:47:27PM -0500, Bill Nottingham wrote:
> > usbcore seems to only generate udev events on removal; it never
> > generates add events. Any particular reason?
> > 
> > (2.6.10-acX, if it matters.)
> 
> What type of udev events?  Those for the usb major number?  So things
> like the USB printer driver don't work anymore?

modprobe usbcore yields no events to udev, as far as I can tell
by putting a catchall script in /etc/dev.d/default.

modprobe -r usbcore yields:

ACTION=remove
DEVNAME=/dev/usb
DEVPATH=/class/usb

and

ACTION=remove
DEVNAME=/dev/usb_host
DEVPATH=/class/usb_host

I'd expect to get 'add' for the same devname/devpath on load
of usbcore, or no events on usbcore removal.

> What version of udev?

039.

> And does this happen in a "clean" 2.6.10?

Will try this and more recent udev.

Bill


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
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: usbcore vs. udev

[Bill Nottingham]

Bill Nottingham (notting@redhat.com) said: 
> > And does this happen in a "clean" 2.6.10?
> 
> Will try this and more recent udev.

Stock 2.6.10 behaves the same. Oddly, updating to udev-050 makes
the remove events on usbcore removal disappear.

Bill


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
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: usbcore vs. udev

[Greg KH]

On Tue, Jan 25, 2005 at 08:33:41PM -0500, Bill Nottingham wrote:
> Greg KH (greg@kroah.com) said: 
> > On Tue, Jan 25, 2005 at 01:47:27PM -0500, Bill Nottingham wrote:
> > > usbcore seems to only generate udev events on removal; it never
> > > generates add events. Any particular reason?
> > > 
> > > (2.6.10-acX, if it matters.)
> > 
> > What type of udev events?  Those for the usb major number?  So things
> > like the USB printer driver don't work anymore?
> 
> modprobe usbcore yields no events to udev, as far as I can tell
> by putting a catchall script in /etc/dev.d/default.

Nor should it.  Why would you think it should?  No /dev nodes are
owned by usbcore.

> modprobe -r usbcore yields:
> 
> ACTION=remove
> DEVNAME=/dev/usb
> DEVPATH=/class/usb
> 
> and
> 
> ACTION=remove
> DEVNAME=/dev/usb_host
> DEVPATH=/class/usb_host
> 
> I'd expect to get 'add' for the same devname/devpath on load
> of usbcore, or no events on usbcore removal.

Ah, you mean the class add and remove.  Hm, I think we don't emit those
because those are classes (not class_dev), and at remove time we don't
know to not emit them.  I'm considering always emiting them at add also,
but it might mess up some userspace stuff, so I'm trying to be cautious
and test it out a lot first.

But again, udev doesn't care about these events, nor should it.

thanks,

greg k-h


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
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: usbcore vs. udev

[Bill Nottingham]

Greg KH (greg@kroah.com) said: 
> > modprobe usbcore yields no events to udev, as far as I can tell
> > by putting a catchall script in /etc/dev.d/default.
> 
> Nor should it.  Why would you think it should?

Only because it emitted events on remove. :)

> Ah, you mean the class add and remove.  Hm, I think we don't emit those
> because those are classes (not class_dev), and at remove time we don't
> know to not emit them.  I'm considering always emiting them at add also,
> but it might mess up some userspace stuff, so I'm trying to be cautious
> and test it out a lot first.
> 
> But again, udev doesn't care about these events, nor should it.

Probably not. Perhaps hooking into hotplug for the class is better;
what I was originally looking at is hooking into the class/device
plug for the mounting of usbfs.

Bill


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
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: usbcore vs. udev

[Kay Sievers]

On Tue, 2005-01-25 at 23:12 -0500, Bill Nottingham wrote:
> Greg KH (greg@kroah.com) said: 
> > > modprobe usbcore yields no events to udev, as far as I can tell
> > > by putting a catchall script in /etc/dev.d/default.
> > 
> > Nor should it.  Why would you think it should?
> 
> Only because it emitted events on remove. :)

I can't reproduce this here. I expect, with 050 you don't get any event
in dev.d/, not for "remove" nor for the "add" event of usbcore.
The events for class creation are just recently introduced by a change
in the driver core and I blacklisted these for udev with version 049.

> > Ah, you mean the class add and remove.  Hm, I think we don't emit those
> > because those are classes (not class_dev), and at remove time we don't
> > know to not emit them.  I'm considering always emiting them at add also,
> > but it might mess up some userspace stuff, so I'm trying to be cautious
> > and test it out a lot first.
> > 
> > But again, udev doesn't care about these events, nor should it.
> 
> Probably not. Perhaps hooking into hotplug for the class is better;
> what I was originally looking at is hooking into the class/device
> plug for the mounting of usbfs.

Yes, hotplug.d/ should work fine for this. With the hotplug events
managed by udevd (Fedora devel already swiched to it) the only
difference between hotplug.d/ and dev.d/ is that /dev.d fires only on
node creation/removal. Everything else, the environment, the
availability of the node or wait_for_sysfs is exactly the same.

Thanks,
Kay



-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
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