Oliver Neukum <oliver@xxxxxxxxxx> writes:
> Am Dienstag, 24. Januar 2012, 00:52:02 schrieb Bjørn Mork:
>> /* --- method tables --- */
>>
>> @@ -111,13 +112,45 @@ struct wdm_device {
>> struct work_struct rxwork;
>> int werr;
>> int rerr;
>> +
>> + struct list_head device_list;
>> };
>>
>> static struct usb_driver wdm_driver;
>>
>> +/* return intfdata if we own the interface, else look up intf in the list */
>> static struct wdm_device *wdm_get_device(struct usb_interface *intf)
>> {
>> - struct wdm_device *desc = usb_get_intfdata(intf);
>> + struct wdm_device *desc = NULL;
>> +
>> + pr_debug("%s: intf->dev.driver=%p, &wdm_driver=%p\n", __func__, to_usb_driver(intf->dev.driver), &wdm_driver);
>> +
>> + mutex_lock(&wdm_mutex);
>> + if (to_usb_driver(intf->dev.driver) == &wdm_driver)
>> + desc = usb_get_intfdata(intf);
>> + else
>> + list_for_each_entry(desc, &wdm_device_list, device_list)
>> + if (desc->intf == intf)
>> + break;
>> + mutex_unlock(&wdm_mutex);
>> +
>> + return desc;
>> +}
>
> Well, this is conceptually inelegant. If we need to maintain our own list,
> we can put every instance on it. We should have as few code paths as
> possible.
Agreed. The above is a result of coding-as-you-go without actually
knowing where you're going...
Given that I ended up putting every device on the list, then everthying
should use the list lookup. Which makes it possible to drop the
usb_set_intfdata() and then also combine even more of the probe/register
by moving the now common usb_register_dev part to create.
Bjørn
--
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
