linux-kernel - Re: BUG: unable to handle kernel paging request at 6b6b6bcb (v4l2_device

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:	Tue, 6 Sep 2011 10:20:00 +0200
From:	Laurent Pinchart <laurent.pinchart@...asonboard.com>
To:	Sitsofe Wheeler <sitsofe@...oo.com>
Cc:	Hans Verkuil <hverkuil@...all.nl>,
	Dave Young <hidave.darkstar@...il.com>,
	linux-media@...r.kernel.org, linux-kernel@...r.kernel.org,
	Guennadi Liakhovetski <g.liakhovetski@....de>,
	Mauro Carvalho Chehab <mchehab@...radead.org>
Subject: Re: BUG: unable to handle kernel paging request at 6b6b6bcb (v4l2_device_disconnect+0x11/0x30)

On Tuesday 06 September 2011 00:31:03 Sitsofe Wheeler wrote:
> On Mon, Sep 05, 2011 at 12:16:42PM +0200, Hans Verkuil wrote:
> > On Monday, September 05, 2011 12:13:26 Hans Verkuil wrote:
> > > The original order is correct, but what I missed is that for drivers
> > > that release (free) everything in the videodev release callback the
> > > v4l2_device struct is also freed and v4l2_device_put will fail.
> > > 
> > > To fix this, add this code just before the vdev->release call:
> > > 	/* Do not call v4l2_device_put if there is no release callback set. */
> > > 	if (v4l2_dev->release == NULL)
> > > 	
> > > 		v4l2_dev = NULL;
> > > 
> > > If there is no release callback, then the refcounting is pointless
> > > anyway.
> > > 
> > > This should work.
> > 
> > Note that in the long run using the v4l2_device release callback
> > instead of the videodev release is better. But it's a lot of work to
> > convert everything so that's long term. I'm quite surprised BTW that
> > this bug wasn't found much earlier.
> 
> This inline patch fixes the second "poison overwritten" problem so:
> Tested-by: Sitsofe Wheeler <sitsofe@...oo.com>
> 
> However, it does not prevent the original oops that was reported in the
> original message. Yang Ruirui's patch in
> https://lkml.org/lkml/2011/9/1/74 seems to be required to resolve
> that initial problem - can it be ACK'd? Yang's patch is reproduced
> inline below:
> 
> For uvc device, dev->vdev.dev is the &intf->dev,
> uvc_delete code is as below:
> 	usb_put_intf(dev->intf);
> 	usb_put_dev(dev->udev);
> 
> 	uvc_status_cleanup(dev);
> 	uvc_ctrl_cleanup_device(dev);
> 
> ## the intf dev is released above, so below code will oops.
> 
> 	if (dev->vdev.dev)
> 		v4l2_device_unregister(&dev->vdev);
> Fix it by get_device in v4l2_device_register and put_device in
> v4l2_device_disconnect

Acked-by: Laurent Pinchart <laurent.pinchart@...asonboard.com>

> ---
>  drivers/media/video/v4l2-device.c |    2 ++
>  1 file changed, 2 insertions(+)
> diff --git a/drivers/media/video/v4l2-device.c
> b/drivers/media/video/v4l2-device.c index c72856c..e6a2c3b 100644
> --- a/drivers/media/video/v4l2-device.c
> +++ b/drivers/media/video/v4l2-device.c
> @@ -38,6 +38,7 @@ int v4l2_device_register(struct device *dev, struct
> v4l2_device *v4l2_dev) mutex_init(&v4l2_dev->ioctl_lock);
>  	v4l2_prio_init(&v4l2_dev->prio);
>  	kref_init(&v4l2_dev->ref);
> +	get_device(dev);
>  	v4l2_dev->dev = dev;
>  	if (dev == NULL) {
>  		/* If dev == NULL, then name must be filled in by the caller */
> @@ -93,6 +94,7 @@ void v4l2_device_disconnect(struct v4l2_device *v4l2_dev)
> 
>  	if (dev_get_drvdata(v4l2_dev->dev) == v4l2_dev)
>  		dev_set_drvdata(v4l2_dev->dev, NULL);
> +	put_device(v4l2_dev->dev);
>  	v4l2_dev->dev = NULL;
>  }
>  EXPORT_SYMBOL_GPL(v4l2_device_disconnect);

-- 
Regards,

Laurent Pinchart
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/