Re: [PATCH 1/8] bus: mhi: core: Validate channel ID when processing command completions

From: Greg KH
Date: Thu Jun 24 2021 - 11:27:25 EST


On Thu, Jun 24, 2021 at 08:17:52PM +0530, Manivannan Sadhasivam wrote:
> On Thu, Jun 24, 2021 at 04:39:51PM +0200, Greg KH wrote:
> > On Thu, Jun 24, 2021 at 08:02:48PM +0530, Manivannan Sadhasivam wrote:
> > > On Thu, Jun 24, 2021 at 03:50:45PM +0200, Greg KH wrote:
> > > > On Mon, Jun 21, 2021 at 09:46:09PM +0530, Manivannan Sadhasivam wrote:
> > > > > From: Bhaumik Bhatt <bbhatt@xxxxxxxxxxxxxx>
> > > > >
> > > > > MHI reads the channel ID from the event ring element sent by the
> > > > > device which can be any value between 0 and 255. In order to
> > > > > prevent any out of bound accesses, add a check against the maximum
> > > > > number of channels supported by the controller and those channels
> > > > > not configured yet so as to skip processing of that event ring
> > > > > element.
> > > > >
> > > > > Cc: stable@xxxxxxxxxxxxxxx
> > > > > Fixes: 1d3173a3bae7 ("bus: mhi: core: Add support for processing events from client device")
> > > > > Signed-off-by: Bhaumik Bhatt <bbhatt@xxxxxxxxxxxxxx>
> > > > > Reviewed-by: Hemant Kumar <hemantk@xxxxxxxxxxxxxx>
> > > > > Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@xxxxxxxxxx>
> > > > > Reviewed-by: Jeffrey Hugo <quic_jhugo@xxxxxxxxxxx>
> > > > > Link: https://lore.kernel.org/r/1619481538-4435-1-git-send-email-bbhatt@xxxxxxxxxxxxxx
> > > > > Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@xxxxxxxxxx>
> > > > > ---
> > > > > drivers/bus/mhi/core/main.c | 15 ++++++++++-----
> > > > > 1 file changed, 10 insertions(+), 5 deletions(-)
> > > > >
> > > > > diff --git a/drivers/bus/mhi/core/main.c b/drivers/bus/mhi/core/main.c
> > > > > index 22acde118bc3..ed07421c4870 100644
> > > > > --- a/drivers/bus/mhi/core/main.c
> > > > > +++ b/drivers/bus/mhi/core/main.c
> > > > > @@ -773,11 +773,16 @@ static void mhi_process_cmd_completion(struct mhi_controller *mhi_cntrl,
> > > > > cmd_pkt = mhi_to_virtual(mhi_ring, ptr);
> > > > >
> > > > > chan = MHI_TRE_GET_CMD_CHID(cmd_pkt);
> > > > > - mhi_chan = &mhi_cntrl->mhi_chan[chan];
> > > > > - write_lock_bh(&mhi_chan->lock);
> > > > > - mhi_chan->ccs = MHI_TRE_GET_EV_CODE(tre);
> > > > > - complete(&mhi_chan->completion);
> > > > > - write_unlock_bh(&mhi_chan->lock);
> > > > > + WARN_ON(chan >= mhi_cntrl->max_chan);
> > > >
> > > > What can ever trigger this WARN_ON()? Do you mean to reboot a machine
> > > > if panic-on-warn is set?
> > > >
> > > > If this can actually happen, then check for it and recover properly,
> > > > don't just blindly warn and then keep on going.
> > > >
> > >
> > > We can't do much here other than warning the user and dropping the
> > > command.
> >
> > But you didn't warn anyone. Well, you rebooted the machine, is that ok?
> > If this can be triggered by a user, this should never happen.
> >
> > Do not use WARN_ON() ever please.
> >
> > > There is no recovery possible because, the device has sent the command
> > > completion event on a wrong channel. It can't happen usually unless a
> > > malcious device sits on the other end.
> >
> > Then just eat the message and move on, please do not crash the box.
> >
>
> Okay. We'll spit an error message and drop the event.

If this can be triggered by a user, don't provide a way to DoS the
kernel error log.

thanks,

greg k-h