Re: Use of usb_find_interface in open is racy

[Russ Dill <russ.dill@gmail.com>]

On Wed, Nov 18, 2009 at 7:27 AM, Oliver Neukum <oliver@neukum.org> wrote:
> Am Mittwoch, 18. November 2009 11:41:25 schrieb Jiri Kosina:
>> > User space receives the hotplug event for the class device, makes the
>> > device node and notifies another program that opens the device node.
>> > The program opens the device node which calls into usb_open and then
>> > skel_open. skel_open calls usb_find_interface. usb_find_interfaces
>> > searches the klist_devices of skel_driver, finds no device associated
>> > with the minor number and returns NULL. skel_open returns -ENODEV.
>> >
>> > Control returns to really_probe and really_probe calls driver_bound
>> > which adds the device to the list of devices associated with
>> > skel_driver (klist_devices).
>> >
>> > I'm not sure what the right way to solve this is. A call to
>> > wait_for_device_probe() in the skel_open call before calling
>> > usb_find_interface fixes the problem, but it is a rather large hammer.
>>
>
> Device core code is hard to follow, but I tried.
> How about simply covering all of usb_register_dev() with minor_rwsem?
>

The race is open from the point that usb_register_dev calls
device_create (at the end of usb_register_dev) till the point that
skel_probe returns up to the point of really_probe, really_probe calls
driver_bound, and driver_bound adds the device to the driver's
klist_devices.

call stack:

device_create
usb_register_dev
skel_probe
usb_probe_device (via drv->probe())
really_probe

really_probe:

dev->driver = drv;
if (driver_sysfs_add(dev)) {
[...]
}

if (dev->bus->probe) {
[...]
} else if (drv->probe) {