linux-kernel - Re: [PATCH v7] media: vb2: Take queue or device lock in mmap-related vb2 ioctl handlers

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

Message-ID: <53B3A7CB.8020901@xs4all.nl>
Date:	Wed, 02 Jul 2014 08:33:47 +0200
From:	Hans Verkuil <hverkuil@...all.nl>
To:	Laurent Pinchart <laurent.pinchart@...asonboard.com>,
	Sasha Levin <sasha.levin@...cle.com>
CC:	Laurent Pinchart <laurent.pinchart+renesas@...asonboard.com>,
	linux-media@...r.kernel.org, Sakari Ailus <sakari.ailus@....fi>,
	Katsuya MATSUBARA <matsu@...l.co.jp>,
	Sylwester Nawrocki <sylvester.nawrocki@...il.com>,
	linux-sh@...r.kernel.org, LKML <linux-kernel@...r.kernel.org>,
	Dave Jones <davej@...hat.com>
Subject: Re: [PATCH v7] media: vb2: Take queue or device lock in mmap-related
 vb2 ioctl handlers

For now it is -ENOTIME for me. It's on my TODO list, but there are a number of
other things I want to finish first.

Regards,

	Hans

On 07/01/2014 11:08 PM, Laurent Pinchart wrote:
> On Wednesday 25 June 2014 12:58:03 Sasha Levin wrote:
>> Ping?
> 
> Hans, I've replied to your previous e-mail with
> 
> I'm of course fine with a different way to solve the race condition, if we can 
> find a good one. Do you already know how you would like to solve this ?
> 
>> On 05/23/2014 09:54 AM, Hans Verkuil wrote:
>>> Hi Laurent,
>>>
>>> This patch caused a circular locking dependency as reported by Sasha
>>> Levin:
>>>
>>> https://lkml.org/lkml/2014/5/5/366
>>>
>>> The reason is that copy_to/from_user is called in video_usercopy() with
>>> the
>>> core lock held. The copy functions can fault which takes the mmap_sem. If
>>> it was just video_usercopy() then it would be fairly easy to solve this,
>>> but the copy_to_/from_user functions are also called from read and write
>>> and they can be used in other unexpected places.
>>>
>>> I'm not sure if vb2_fop_get_unmapped_area() is a problem. I suspect (but
>>> I'm not sure) that when that one is called the mmap_sem isn't taken, in
>>> which case taking the lock is fine.
>>>
>>> But taking the lock in vb2_fop_mmap() does cause lockdep problems.
>>>
>>> Ideally I would like to drop taking that lock in vb2_fop_mmap and resolve
>>> the race condition that it intended to fix in a different way.
>>>
>>> Regards,
>>>
>>> 	Hans
>>>
>>> On 08/06/2013 10:10 PM, Laurent Pinchart wrote:
>>>> The vb2_fop_mmap() and vb2_fop_get_unmapped_area() functions are plug-in
>>>> implementation of the mmap() and get_unmapped_area() file operations
>>>> that calls vb2_mmap() and vb2_get_unmapped_area() on the queue
>>>> associated with the video device. Neither the
>>>> vb2_fop_mmap/vb2_fop_get_unmapped_area nor the
>>>> v4l2_mmap/vb2_get_unmapped_area functions in the V4L2 core take any
>>>> lock, leading to race conditions between mmap/get_unmapped_area and
>>>> other buffer-related ioctls such as VIDIOC_REQBUFS.
>>>>
>>>> Fix it by taking the queue or device lock around the vb2_mmap() and
>>>> vb2_get_unmapped_area() calls.
>>>>
>>>> Signed-off-by: Laurent Pinchart
>>>> <laurent.pinchart+renesas@...asonboard.com>
>>>> ---
>>>>
>>>>  drivers/media/v4l2-core/videobuf2-core.c | 18 ++++++++++++++++--
>>>>  1 file changed, 16 insertions(+), 2 deletions(-)
>>>>
>>>> diff --git a/drivers/media/v4l2-core/videobuf2-core.c
>>>> b/drivers/media/v4l2-core/videobuf2-core.c index 9fc4bab..c9b50c7 100644
>>>> --- a/drivers/media/v4l2-core/videobuf2-core.c
>>>> +++ b/drivers/media/v4l2-core/videobuf2-core.c
>>>> @@ -2578,8 +2578,15 @@ EXPORT_SYMBOL_GPL(vb2_ioctl_expbuf);
>>>>  int vb2_fop_mmap(struct file *file, struct vm_area_struct *vma)
>>>>  {
>>>>  	struct video_device *vdev = video_devdata(file);
>>>> +	struct mutex *lock = vdev->queue->lock ? vdev->queue->lock :
>>>> vdev->lock;
>>>> +	int err;
>>>>
>>>> -	return vb2_mmap(vdev->queue, vma);
>>>> +	if (lock && mutex_lock_interruptible(lock))
>>>> +		return -ERESTARTSYS;
>>>> +	err = vb2_mmap(vdev->queue, vma);
>>>> +	if (lock)
>>>> +		mutex_unlock(lock);
>>>> +	return err;
>>>>  }
>>>>  EXPORT_SYMBOL_GPL(vb2_fop_mmap);
>>>>
>>>> @@ -2685,8 +2692,15 @@ unsigned long vb2_fop_get_unmapped_area(struct
>>>> file *file, unsigned long addr,>> 
>>>>  		unsigned long len, unsigned long pgoff, unsigned long flags)
>>>>  {
>>>>  	struct video_device *vdev = video_devdata(file);
>>>> +	struct mutex *lock = vdev->queue->lock ? vdev->queue->lock :
>>>> vdev->lock;
>>>> +	int ret;
>>>>
>>>> -	return vb2_get_unmapped_area(vdev->queue, addr, len, pgoff, flags);
>>>> +	if (lock && mutex_lock_interruptible(lock))
>>>> +		return -ERESTARTSYS;
>>>> +	ret = vb2_get_unmapped_area(vdev->queue, addr, len, pgoff, flags);
>>>> +	if (lock)
>>>> +		mutex_unlock(lock);
>>>> +	return ret;
>>>>
>>>>  }
>>>>  EXPORT_SYMBOL_GPL(vb2_fop_get_unmapped_area);
>>>>  #endif
> 

--
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/