Re: [PATCH 5/8] usb: typec: tcpm: validate VDO count before reading Attention status VDO

[Greg Kroah-Hartman <gregkh@linuxfoundation.org>]

On Thu, May 14, 2026 at 02:55:31PM -0700, Badhri Jagan Sridharan wrote:
> On Wed, May 13, 2026 at 8:53 AM Greg Kroah-Hartman
> <gregkh@linuxfoundation.org> wrote:
> >
> > A broken/malicious device can send the incorrect count for an attention
> > status VDO, which will cause the kernel to read uninitialized stack data
> > and send it off elsewhere.
> >
> > Fix this up by correctly verifying the count for the attention object.
> >
> > Assisted-by: gkh_clanker_t1000
> > Cc: Badhri Jagan Sridharan <badhri@google.com>
> > Cc: Heikki Krogerus <heikki.krogerus@linux.intel.com>
> > Cc: stable <stable@kernel.org>
> > Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> > ---
> >  drivers/usb/typec/tcpm/tcpm.c | 2 ++
> >  1 file changed, 2 insertions(+)
> >
> > diff --git a/drivers/usb/typec/tcpm/tcpm.c b/drivers/usb/typec/tcpm/tcpm.c
> > index 44dab6c32c33..2cab74ed71a7 100644
> > --- a/drivers/usb/typec/tcpm/tcpm.c
> > +++ b/drivers/usb/typec/tcpm/tcpm.c
> > @@ -2639,6 +2639,8 @@ static void tcpm_handle_vdm_request(struct tcpm_port *port,
> >                         }
> >                         break;
> >                 case ADEV_ATTENTION:
> > +                       if (cnt < 2)
> > +                               break;
> >                         if (typec_altmode_attention(adev, p[1]))
> >                                 tcpm_log(port, "typec_altmode_attention no port partner altmode");
> >                         break;
> > --
> 
> Hi Greg,
> 
> While the `cnt` check helps fix this specific case, I do see multiple
> other instances in the same function where the validity of p[0] and
> p[1] seems to be assumed (for instance calls to
> typec_altmode_vdm(adev, p[0], &p[1], cnt)). To make the patch more
> complete and robust, shouldn't p also be initialized to 0? This will
> prevent leaking uninitialized stack data in those instances too.

Good point, we should be checking this for all of these, not just this
one instance.  I'll fix this up and do a new version, thanks.

greg k-h