linux-kernel - Re: [PATCH v2] media: videobuf2: forbid remove

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]

Message-ID: <138b53bd-8b89-4260-bfa1-ff88ae2c1050@kernel.org>
Date: Thu, 23 Oct 2025 12:54:33 +0200
From: Hans Verkuil <hverkuil+cisco@...nel.org>
To: Marek Szyprowski <m.szyprowski@...sung.com>, linux-media@...r.kernel.org,
 linux-kernel@...r.kernel.org
Cc: Tomasz Figa <tfiga@...omium.org>,
 Mauro Carvalho Chehab <mchehab@...nel.org>,
 Guennadi Liakhovetski <g.liakhovetski@....de>,
 Benjamin Gaignard <benjamin.gaignard@...labora.com>,
 Hans Verkuil <hverkuil@...nel.org>, stable@...r.kernel.org,
 Shuangpeng Bai <SJB7183@....edu>
Subject: Re: [PATCH v2] media: videobuf2: forbid remove_bufs when legacy
 fileio is active

On 23/10/2025 11:01, Hans Verkuil wrote:
> On 21/10/2025 11:56, Hans Verkuil wrote:
>> Hi Marek,
>>
>> On 20/10/2025 18:01, Marek Szyprowski wrote:
>>> vb2_ioctl_remove_bufs() call manipulates queue internal buffer list,
>>> potentially overwriting some pointers used by the legacy fileio access
>>> mode. Add a vb2_verify_memory_type() check symmetrical to
>>> vb2_ioctl_create_bufs() to forbid that ioctl when fileio is active to
>>> protect internal queue state between subsequent read/write calls.
>>>
>>> CC: stable@...r.kernel.org
>>> Fixes: a3293a85381e ("media: v4l2: Add REMOVE_BUFS ioctl")
>>> Reported-by: Shuangpeng Bai<SJB7183@....edu>
>>> Suggested-by: Benjamin Gaignard <benjamin.gaignard@...labora.com>
>>> Signed-off-by: Marek Szyprowski <m.szyprowski@...sung.com>
>>
>> I'll pick this up as a fix for v6.18. I think this is important enough to
>> not wait for v6.19.
> 
> Hmm, it's failing on v4l2-compliance. I'm debugging to see whether it is a
> kernel or v4l2-compliance problem.
> 
> Regards,
> 
> 	Hans
> 
>>
>> Regards,
>>
>> 	Hans
>>
>>> ---
>>> v2:
>>> - dropped a change to vb2_ioctl_create_bufs(), as it is already handled
>>>   by the vb2_verify_memory_type() call
>>> - replaced queue->type check in vb2_ioctl_remove_bufs() by a call to
>>>   vb2_verify_memory_type() which covers all cases
>>>
>>> v1: https://lore.kernel.org/all/20251016111154.993949-1-m.szyprowski@samsung.com/
>>> ---
>>>  drivers/media/common/videobuf2/videobuf2-v4l2.c | 6 ++++--
>>>  1 file changed, 4 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/drivers/media/common/videobuf2/videobuf2-v4l2.c b/drivers/media/common/videobuf2/videobuf2-v4l2.c
>>> index d911021c1bb0..0de7490292fe 100644
>>> --- a/drivers/media/common/videobuf2/videobuf2-v4l2.c
>>> +++ b/drivers/media/common/videobuf2/videobuf2-v4l2.c
>>> @@ -1000,9 +1000,11 @@ int vb2_ioctl_remove_bufs(struct file *file, void *priv,
>>>  			  struct v4l2_remove_buffers *d)
>>>  {
>>>  	struct video_device *vdev = video_devdata(file);
>>> +	int res;
>>>  
>>> -	if (vdev->queue->type != d->type)
>>> -		return -EINVAL;
>>> +	res = vb2_verify_memory_type(vdev->queue, vdev->queue->memory, d->type);
>>> +	if (res)
>>> +		return res;
>>>  
>>>  	if (d->count == 0)
>>>  		return 0;

This is the problem. For the corner case where d->count == 0 it can be that
vdev->queue->memory is VB2_MEMORY_UNKNOWN (that happens if no buffers were ever
queued). But it should still return 0 in that case. Also the fileio test doesn't
apply in that case, but that's not tested in v4l2-compliance.

I suggest this:

	if (d->count == 0)
		return d->type == vdev->queue->type ? 0 : -EINVAL;

	res = vb2_verify_memory_type(vdev->queue, vdev->queue->memory, d->type);
	if (res)
		return res;

I tested this and it passes v4l2-compliance.

Marek, can you post a v3?

Thank you,

	Hans

>>
>>
> 
>