Re: KASAN: use-after-free Read in usbhid_power

[Alan Stern <stern@rowland.harvard.edu>]

On Thu, 8 Aug 2019, Andrey Konovalov wrote:

> On Thu, Jul 25, 2019 at 5:09 PM Alan Stern <stern@rowland.harvard.edu> wrote:
> >
> > On Thu, 25 Jul 2019, Oliver Neukum wrote:
> >
> > > Am Mittwoch, den 24.07.2019, 17:02 -0400 schrieb Alan Stern:
> > > > On Wed, 24 Jul 2019, Oliver Neukum wrote:
> > > >
> > > > >  drivers/hid/usbhid/hid-core.c | 13 +++++++++++++
> > > > >  1 file changed, 13 insertions(+)
> > > > >
> > > > > diff --git a/drivers/hid/usbhid/hid-core.c b/drivers/hid/usbhid/hid-core.c
> > > > > index c7bc9db5b192..98b996ecf4d3 100644
> > > > > --- a/drivers/hid/usbhid/hid-core.c
> > > > > +++ b/drivers/hid/usbhid/hid-core.c
> > > > > @@ -1229,6 +1229,17 @@ static int usbhid_power(struct hid_device *hid, int lvl)
> > > > >   struct usbhid_device *usbhid = hid->driver_data;
> > > > >   int r = 0;
> > > > >
> > > > > + spin_lock_irq(&usbhid->lock);
> > > > > + if (test_bit(HID_DISCONNECTED, &usbhid->iofl)) {
> > > > > +         r = -ENODEV;
> > > > > +         spin_unlock_irq(&usbhid->lock);
> > > > > +         goto bail_out;
> > > > > + } else {
> > > > > +         /* protect against disconnect */
> > > > > +         usb_get_dev(interface_to_usbdev(usbhid->intf));
> > > > > +         spin_unlock_irq(&usbhid->lock);
> > > > > + }
> > > > > +
> > > > >   switch (lvl) {
> > > > >   case PM_HINT_FULLON:
> > > > >           r = usb_autopm_get_interface(usbhid->intf);
> > > > > @@ -1238,7 +1249,9 @@ static int usbhid_power(struct hid_device *hid, int lvl)
> > > > >           usb_autopm_put_interface(usbhid->intf);
> > > > >           break;
> > > > >   }
> > > > > + usb_put_dev(interface_to_usbdev(usbhid->intf));
> > > > >
> > > > > +bail_out:
> > > > >   return r;
> > > > >  }

> This report looks like very similar to these two:
> 
> https://syzkaller.appspot.com/bug?extid=b156665cf4d1b5e00c76
> https://syzkaller.appspot.com/bug?extid=3cbe5cd105d2ad56a1df

It also seems to resemble extids a7a6b9c609b9457c62c6, 
62a1e04fd3ec2abf099e, and 75e6910bf03e266a277f, although this may be an 
illusion.

> Maybe we should mark those two as duplicates.
> 
> Hillf suggested a fix on one of them, but it looks different from what
> you propose:
> 
> https://groups.google.com/d/msg/syzkaller-bugs/xW7LvKfpyn0/SpKbs5ZLEAAJ

Go ahead and try it out on all of them.  I don't have a clear feeling 
about it, not having worked on usbhid in quite a while.

Alan Stern