Am 22.10.2011 10:25, schrieb Greg KH:
On Fri, Oct 21, 2011 at 11:04:09PM +0200, Josua Dietze wrote:
When adding the ID of a composite device dynamically to a driver,
all hitherto unbound interfaces are bound to this driver which is
often not intended.
It isn't? What happens wrong with this? Can't you just unbind the ones
you don't want?
Yes, I could, but if the dynamic adding is done automatically, say, by an udev
rule upon device discovery, the process is becoming time-sensitive.
For instance, it may happen that e.g. storage interfaces are mistakenly bound
before the storage driver is able to bind. Unbinding specific interfaces would
be even more time-sensitive in that context, wouldn't it?
Example:
$ echo "1234 2a2a ff">/sys/bus/usb-serial/drivers/generic/new_id
will bind only vendor-specific interfaces to usbserial.
You really shouldn't be using the generic usb serial driver for anything
"real", so that's not a good reason to accept this patch :)
It's literally only a "generic" example. My real-world use revolves around the
"option" driver (and I keep telling everyone *not* to use "usbserial" BTW :)).
The current 3G/4G sticks usually expose one or two storage interfaces in
addition to their serial/vendor-specific interfaces. To support brand-new models
that are not known to the 3G driver yet, the adding of dynamic IDs to the driver
is very useful.
And your patch is linewrapped :(
Oops, I'm sorry. Will repost if acceptance is likely.
Thank you,
Josua Dietze
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
