linux-kernel - Re: [PATCH] media: videobuf2-core: copy vb planes unconditionally

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 for Android: free password hash cracker in your pocket

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

Message-ID: <94f7c775-10bd-46d5-bc89-a18533dcab68@xs4all.nl>
Date: Thu, 7 Nov 2024 10:17:16 +0100
From: Hans Verkuil <hverkuil@...all.nl>
To: Tomasz Figa <tfiga@...omium.org>, Tudor Ambarus <tudor.ambarus@...aro.org>
Cc: m.szyprowski@...sung.com, mchehab@...nel.org, yunkec@...omium.org,
 linux-media@...r.kernel.org, linux-kernel@...r.kernel.org,
 linux-samsung-soc@...r.kernel.org, andre.draszik@...aro.org,
 kernel-team@...roid.com, willmcvicker@...gle.com, stable@...r.kernel.org
Subject: Re: [PATCH] media: videobuf2-core: copy vb planes unconditionally

On 07/11/2024 00:41, Tomasz Figa wrote:
> On Wed, Nov 6, 2024 at 9:18 PM Tudor Ambarus <tudor.ambarus@...aro.org> wrote:
>>
>> Copy the relevant data from userspace to the vb->planes unconditionally
>> as it's possible some of the fields may have changed after the buffer
>> has been validated.
>>
>> Keep the dma_buf_put(planes[plane].dbuf) calls in the first
>> `if (!reacquired)` case, in order to be close to the plane validation code
>> where the buffers were got in the first place.
>>
>> Cc: stable@...r.kernel.org
>> Fixes: 95af7c00f35b ("media: videobuf2-core: release all planes first in __prepare_dmabuf()")
>> Signed-off-by: Tudor Ambarus <tudor.ambarus@...aro.org>
>> ---
>>  .../media/common/videobuf2/videobuf2-core.c   | 28 ++++++++++---------
>>  1 file changed, 15 insertions(+), 13 deletions(-)
>>
> 
> Thanks for the fix.
> 
> Acked-by: Tomasz Figa <tfiga@...omium.org>
> 
> (We probably need some tests to verify this behavior... It seems like
> the way v4l2-compliance is implemented [1] would only trigger the
> !reacquired case on most drivers.)
> 
> [1] https://git.linuxtv.org/v4l-utils.git/tree/utils/v4l2-compliance/v4l2-test-buffers.cpp#n2071
> (just queuing all imported buffers in order and re-queuing them
> exactly as they are dequeued [2])
> [2] https://git.linuxtv.org/v4l-utils.git/tree/utils/v4l2-compliance/v4l2-test-buffers.cpp#n1299

I'll see if I can improve that test.

Regards,

	Hans

> 
> Best regards,
> Tomasz
> 
>> diff --git a/drivers/media/common/videobuf2/videobuf2-core.c b/drivers/media/common/videobuf2/videobuf2-core.c
>> index f07dc53a9d06..c0cc441b5164 100644
>> --- a/drivers/media/common/videobuf2/videobuf2-core.c
>> +++ b/drivers/media/common/videobuf2/videobuf2-core.c
>> @@ -1482,18 +1482,23 @@ static int __prepare_dmabuf(struct vb2_buffer *vb)
>>                         }
>>                         vb->planes[plane].dbuf_mapped = 1;
>>                 }
>> +       } else {
>> +               for (plane = 0; plane < vb->num_planes; ++plane)
>> +                       dma_buf_put(planes[plane].dbuf);
>> +       }
>>
>> -               /*
>> -                * Now that everything is in order, copy relevant information
>> -                * provided by userspace.
>> -                */
>> -               for (plane = 0; plane < vb->num_planes; ++plane) {
>> -                       vb->planes[plane].bytesused = planes[plane].bytesused;
>> -                       vb->planes[plane].length = planes[plane].length;
>> -                       vb->planes[plane].m.fd = planes[plane].m.fd;
>> -                       vb->planes[plane].data_offset = planes[plane].data_offset;
>> -               }
>> +       /*
>> +        * Now that everything is in order, copy relevant information
>> +        * provided by userspace.
>> +        */
>> +       for (plane = 0; plane < vb->num_planes; ++plane) {
>> +               vb->planes[plane].bytesused = planes[plane].bytesused;
>> +               vb->planes[plane].length = planes[plane].length;
>> +               vb->planes[plane].m.fd = planes[plane].m.fd;
>> +               vb->planes[plane].data_offset = planes[plane].data_offset;
>> +       }
>>
>> +       if (reacquired) {
>>                 /*
>>                  * Call driver-specific initialization on the newly acquired buffer,
>>                  * if provided.
>> @@ -1503,9 +1508,6 @@ static int __prepare_dmabuf(struct vb2_buffer *vb)
>>                         dprintk(q, 1, "buffer initialization failed\n");
>>                         goto err_put_vb2_buf;
>>                 }
>> -       } else {
>> -               for (plane = 0; plane < vb->num_planes; ++plane)
>> -                       dma_buf_put(planes[plane].dbuf);
>>         }
>>
>>         ret = call_vb_qop(vb, buf_prepare, vb);
>> --
>> 2.47.0.199.ga7371fff76-goog
>>