Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20221212-uvc-race-v6-0-2a662f8de011@chromium.org> <20230103022540.3667-1-hdanton@sina.com>
In-Reply-To: <20230103022540.3667-1-hdanton@sina.com>
From:   Ricardo Ribalda <ribalda@chromium.org>
Date:   Tue, 3 Jan 2023 10:08:56 +0100
Message-ID: <CANiDSCuTZ17ohj=G3c_36NmnCTtwt0aBZ1Fkh-+rzwEgufK5kw@mail.gmail.com>
Subject: Re: [PATCH v6] media: uvcvideo: Fix race condition with usb_kill_urb
To:     Hillf Danton <hdanton@sina.com>
Cc:     Laurent Pinchart <laurent.pinchart@ideasonboard.com>,
        Mauro Carvalho Chehab <mchehab@kernel.org>,
        Max Staudt <mstaudt@google.com>,
        Sergey Senozhatsky <senozhatsky@chromium.org>,
        Yunke Cao <yunkec@chromium.org>, linux-kernel@vger.kernel.org,
        linux-media@vger.kernel.org, stable@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Hi Hillf

Thanks for the heads up

On Tue, 3 Jan 2023 at 03:25, Hillf Danton <hdanton@sina.com> wrote:
>
> On 02 Jan 2023 15:48:01 +0100 Ricardo Ribalda <ribalda@chromium.org>
> > --- a/drivers/media/usb/uvc/uvc_ctrl.c
> > +++ b/drivers/media/usb/uvc/uvc_ctrl.c
> > @@ -1442,6 +1442,9 @@ static void uvc_ctrl_status_event_work(struct work_struct *work)
> >
> >       uvc_ctrl_status_event(w->chain, w->ctrl, w->data);
> >
> > +     if (dev->flush_status)
> > +             return;
> > +
> >       /* Resubmit the URB. */
> >       w->urb->interval = dev->int_ep->desc.bInterval;
> >       ret = usb_submit_urb(w->urb, GFP_KERNEL);
> > diff --git a/drivers/media/usb/uvc/uvc_status.c b/drivers/media/usb/uvc/uvc_status.c
> > index 7518ffce22ed..e457889345a3 100644
> > --- a/drivers/media/usb/uvc/uvc_status.c
> > +++ b/drivers/media/usb/uvc/uvc_status.c
> > @@ -6,6 +6,7 @@
> >   *          Laurent Pinchart (laurent.pinchart@ideasonboard.com)
> >   */
> >
> > +#include <asm/barrier.h>
> >  #include <linux/kernel.h>
> >  #include <linux/input.h>
> >  #include <linux/slab.h>
> > @@ -309,5 +310,44 @@ int uvc_status_start(struct uvc_device *dev, gfp_t flags)
> >
> >  void uvc_status_stop(struct uvc_device *dev)
> >  {
> > +     struct uvc_ctrl_work *w = &dev->async_ctrl;
> > +
> > +     /* Prevent the asynchronous control handler from requeing the URB */
> > +     dev->flush_status = true;
> > +
> > +     /*
> > +      * The barrier is needed so the flush_status change is visible to other
> > +      * CPUs running the asynchronous handler before usb_kill_urb() is
> > +      * called below.
> > +      */
> > +     smp_mb();
>
> Given unpaired mb, take a look at the release/acquire memory barrier pairing
> in c5b2cbdbdac5 ("ipc/mqueue.c: update/document memory barriers")

Would it work? to replace:

dev->flush_status = true;
smp_mb();

with:
smp_store_release(&dev->flush_status, 1);

and then read it always with:

smp_load_acquire(&dev->flush_status);

Thanks!

>
> > +
> > +     /* If there is any status event on the queue, process it. */
> > +     if (cancel_work_sync(&w->work))
> > +             uvc_ctrl_status_event(w->chain, w->ctrl, w->data);
> > +
> > +     /* Kill the urb. */
> >       usb_kill_urb(dev->int_urb);
> > +
> > +     /*
> > +      * The URB completion handler may have queued asynchronous work. This
> > +      * won't resubmit the URB as flush_status is set, but it needs to be
> > +      * cancelled before returning or it could then race with a future
> > +      * uvc_status_start() call.
> > +      */
> > +     if (cancel_work_sync(&w->work))
> > +             uvc_ctrl_status_event(w->chain, w->ctrl, w->data);
> > +
> > +     /*
> > +      * From this point, there are no events on the queue and the status URB
> > +      * is dead, this is, no events will be queued until uvc_status_start()
> > +      * is called.
> > +      */
> > +     dev->flush_status = false;
> > +
> > +     /*
> > +      * Write to memory the value of flush_status before uvc_status_start()
> > +      * is called again.
> > +      */
> > +     smp_mb();
> >  }


-- 
Ricardo Ribalda