Am Mittwoch, 9. Juli 2003 02:09 schrieb Greg KH:
> On Wed, Jul 09, 2003 at 12:14:00AM +0200, Oliver Neukum wrote:
> > Am Dienstag, 8. Juli 2003 23:39 schrieb Greg KH:
> > > On Tue, Jul 08, 2003 at 04:08:41PM -0400, Alan Stern wrote:
> > > > .txt. But do you really need all these device
> > > > numbers? Consider using devfs, which does dynamic device number
> > > > assignment.
> > >
> > > It does? Through what calls?
> >
> > devfs_register():
> >
> > if ( ( S_ISCHR (mode) || S_ISBLK (mode) ) &&
> > (flags & DEVFS_FL_AUTO_DEVNUM) )
>
> Not present in 2.5, and only 1 driver used it in 2.4.
>
> Well, the ia64 port had some support for some devices that used this,
> but it is minimal.
>
> So no, using devfs for dynamic number assignment is not a valid
> solution.
You could certainly argue that devfs should not generate numbers.
It should just export them nicely. By the same argument, what USB
currently does, is wrong as well.
The right place to generate numbers is in the driver core IMHO.
Generally device numbers are just a way to address devices
from user space. Several more ways could be thought of.
Having such information outside the generic layer is a layering
violation.
Regards
Oliver
-------------------------------------------------------
This SF.Net email sponsored by: Parasoft
Error proof Web apps, automate testing & more.
Download & eval WebKing and get a free book.
www.parasoft.com/bulletproofapps
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
