Re: [PATCH] : ir257_usb_disconnect_atomic-2.diff

[Jean Tourrilhes]

On Tue, Apr 02, 2002 at 07:00:38PM -0800, Greg KH wrote:
> On Tue, Apr 02, 2002 at 06:24:13PM -0800, Jean Tourrilhes wrote:
> > @@ -1519,33 +1544,47 @@ static void *irda_usb_probe(struct usb_d
> >  /*
> >   * The current irda-usb device is removed, the USB layer tell us
> >   * to shut it down...
> > + * One of the constraints is that when we exit this function,
> > + * we cannot use the usb_device no more. Gone. Destroyed. kfree().
> > + * Most other subsystem allow you to destroy the instance at a time
> > + * when it's convenient to you, to postpone it to a later date, but
> > + * not the USB subsystem.
> > + * So, we must make bloody sure that everything gets deactivated.
> > + * Jean II
> 
> That's one of the next things I'm going to be working on fixing :)

	By the time you will "fix" that, all the USB driver will be
fixed to workaround this issue. Actually, it force to be a bit more
careful about the disconnect, and avoid zombies instances all over the
place, so is not such a bad thing after all.

> The patch looks good.  Thanks for setting the proper GFP_* flag.

	That was Martin... Don't need to thank, because the impact is
limited to the IrDA driver...

> greg k-h

	Have fun...

	Jean