>On Wed, Dec 13, 2000 at 03:34:07PM -0800, Adam J. Richter wrote:
Tim Waugh wrote:
>If we have user-space matching, why have kernel-space matching as
The user space matching scheme that I described will only
determine what modules to load. Once the modules are loaded, it is
up to them to figure out which devices to bind to. Imagine that
I have a system with three parallel ports, one of which has a
zip drive attached to it. Once the module is loaded, how does
it figure out which port to bind to, if it cannot check the
ieee1284 device string?
While it would be possible to write more user level code so
that the modules would be loaded with additional kernel arguments
telling it what to bind to, there would still need to be a kernel
based mechanism for the cases where we do want to support hot plugging
even if we cannot reliably detect it in all cases. I suppose that
too could be communicated through some new ioctl's, but that raises
other issues about how a user level program would communicate this
information to the kernel and how it would be run (e.g., spawned by
/sbin/hotplug, a persistent daemon, a cron job). At the moment, I
think the simplest approach is probably to be consistent with the way
the clients of MODULE_DEVICE_TABLE.
>I'd say ditch it and deprecate lp=auto.
This facility would eventually cause lp=auto to be rewritten
>> Do any devices actually say "CLASS:" in their ID information?
OK. Thanks for the information. Also, are you sure that there
is a device where a keyword is given in lower case (for example "cls:"
is used instead of "CLS:")?
Adam J. Richter __ ______________ 4880 Stevens Creek Blvd, Suite 104
email@example.com \ / San Jose, California 95129-1034
+1 408 261-6630 | g g d r a s i l United States of America
fax +1 408 261-6631 "Free Software For The Rest Of Us."
-- To unsubscribe, send mail to: firstname.lastname@example.org --
-- with the single word "unsubscribe" in the body of the message. --
This archive was generated by hypermail 2b29 : Wed Dec 13 2000 - 19:28:27 EST