Re: [PATCH 4/5] USB: core: Add API to change the wireless_status

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Thu, 2023-02-23 at 11:25 -0500, Alan Stern wrote:
> On Thu, Feb 23, 2023 at 05:17:13PM +0100, Bastien Nocera wrote:
> > On Thu, 2023-02-23 at 10:41 -0500, Alan Stern wrote:
> > > On Thu, Feb 23, 2023 at 02:24:51PM +0100, Bastien Nocera wrote:
> > > > Allow device specific drivers to change the wireless status of
> > > > a
> > > > device.
> > > > This will allow user-space to know whether the device is
> > > > available,
> > > > whether or not specific USB interfaces can detect it.
> > > > 
> > > > This can be used by wireless headsets with USB receivers to
> > > > propagate to
> > > > user-space whether or not the headset is turned on, so as to
> > > > consider it
> > > > as unavailable, and not switch to it just because the receiver
> > > > is
> > > > plugged in.
> > > > 
> > > > Signed-off-by: Bastien Nocera <hadess@xxxxxxxxxx>
> > > > ---
> > > >  drivers/usb/core/message.c | 13 +++++++++++++
> > > >  drivers/usb/core/usb.c     | 24 ++++++++++++++++++++++++
> > > >  include/linux/usb.h        |  4 ++++
> > > >  3 files changed, 41 insertions(+)
> > > > 
> > > > diff --git a/drivers/usb/core/message.c
> > > > b/drivers/usb/core/message.c
> > > > index 127fac1af676..d5c7749d515e 100644
> > > > --- a/drivers/usb/core/message.c
> > > > +++ b/drivers/usb/core/message.c
> > > > @@ -1908,6 +1908,18 @@ static void
> > > > __usb_queue_reset_device(struct
> > > > work_struct *ws)
> > > >         usb_put_intf(iface);    /* Undo _get_ in
> > > > usb_queue_reset_device() */
> > > >  }
> > > >  
> > > > +/*
> > > > + * Internal function to set the wireless_status sysfs
> > > > attribute
> > > > + * See usb_set_wireless_status() for more details
> > > > + */
> > > > +static void __usb_wireless_status_intf(struct work_struct *ws)
> > > > +{
> > > > +       struct usb_interface *iface =
> > > > +               container_of(ws, struct usb_interface,
> > > > wireless_status_work);
> > > > +
> > > > +       usb_update_wireless_status_attr(iface);
> > > > +       usb_put_intf(iface);    /* Undo _get_ in
> > > > usb_set_wireless_status() */
> > > > +}
> > > 
> > > Have you thought about what will happen if this routine ends up
> > > running 
> > > after the interface has been deleted?
> > 
> > I believe that usb_release_interface() will only be called once the
> > last reference to the interface is dropped, so bar any refcounting
> > bugs, the interface should always exist when this function is
> > called.
> 
> Yes, but what about the calls made by
> usb_update_wireless_status_attr(): 
> sysfs_update_group(), sysfs_notify(), and kobject_uevent()?  Will
> they 
> work properly when called for an object that has been unregistered
> from 
> sysfs?

Those calls are made before the last reference to the interface can be
dropped by our own call to usb_put_intf(). So, in effect, the interface
is kept alive until we're done with our work so we can't ever be poking
at objects that disappeared.

Am I missing something about the object model, or the refcounting here?



[Index of Archives]     [Linux Media]     [Linux Input]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [Old Linux USB Devel Archive]

  Powered by Linux