Received-SPF: pass (google.com: domain of linux-kernel+bounces-195264-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Message-ID: <a962e94f-a669-40e1-8b28-9ef077cf69be@xs4all.nl>
Date: Thu, 30 May 2024 12:31:59 +0200
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: [PATCH v2] media: uapi: v4l: Change V4L2_TYPE_IS_CAPTURE
 condition
To: Nas Chung <nas.chung@chipsnmedia.com>, mchehab@kernel.org,
 linux-media@vger.kernel.org, sebastian.fricke@collabora.com,
 m.tretter@pengutronix.de
Cc: linux-kernel@vger.kernel.org
References: <20240528020425.4994-1-nas.chung@chipsnmedia.com>
Content-Language: en-US, nl
From: Hans Verkuil <hverkuil@xs4all.nl>
Autocrypt: addr=hverkuil@xs4all.nl; keydata=
 xsFNBFQ84W0BEAC7EF1iL4s3tY8cRTVkJT/297h0Hz0ypA+ByVM4CdU9sN6ua/YoFlr9k0K4
 BFUlg7JzJoUuRbKxkYb8mmqOe722j7N3HO8+ofnio5cAP5W0WwDpM0kM84BeHU0aPSTsWiGR
 yw55SOK2JBSq7hueotWLfJLobMWhQii0Zd83hGT9SIt9uHaHjgwmtTH7MSTIiaY6N14nw2Ud
 C6Uykc1va0Wqqc2ov5ihgk/2k2SKa02ookQI3e79laOrbZl5BOXNKR9LguuOZdX4XYR3Zi6/
 BsJ7pVCK9xkiVf8svlEl94IHb+sa1KrlgGv3fn5xgzDw8Z222TfFceDL/2EzUyTdWc4GaPMC
 E/c1B4UOle6ZHg02+I8tZicjzj5+yffv1lB5A1btG+AmoZrgf0X2O1B96fqgHx8w9PIpVERN
 YsmkfxvhfP3MO3oHh8UY1OLKdlKamMneCLk2up1Zlli347KMjHAVjBAiy8qOguKF9k7HOjif
 JCLYTkggrRiEiE1xg4tblBNj8WGyKH+u/hwwwBqCd/Px2HvhAsJQ7DwuuB3vBAp845BJYUU3
 06kRihFqbO0vEt4QmcQDcbWINeZ2zX5TK7QQ91ldHdqJn6MhXulPKcM8tCkdD8YNXXKyKqNl
 UVqXnarz8m2JCbHgjEkUlAJCNd6m3pfESLZwSWsLYL49R5yxIwARAQABzSFIYW5zIFZlcmt1
 aWwgPGh2ZXJrdWlsQHhzNGFsbC5ubD7CwZUEEwECACgFAlQ84W0CGwMFCRLMAwAGCwkIBwMC
 BhUIAgkKCwQWAgMBAh4BAheAACEJEL0tYUhmFDtMFiEEBSzee8IVBTtonxvKvS1hSGYUO0wT
 7w//frEmPBAwu3OdvAk9VDkH7X+7RcFpiuUcJxs3Xl6jpaA+SdwtZra6W1uMrs2RW8eXXiq/
 80HXJtYnal1Y8MKUBoUVhT/+5+KcMyfVQK3VFRHnNxCmC9HZV+qdyxAGwIscUd4hSlweuU6L
 6tI7Dls6NzKRSTFbbGNZCRgl8OrF01TBH+CZrcFIoDgpcJA5Pw84mxo+wd2BZjPA4TNyq1od
 +slSRbDqFug1EqQaMVtUOdgaUgdlmjV0+GfBHoyCGedDE0knv+tRb8v5gNgv7M3hJO3Nrl+O
 OJVoiW0G6OWVyq92NNCKJeDy8XCB1yHCKpBd4evO2bkJNV9xcgHtLrVqozqxZAiCRKN1elWF
 1fyG8KNquqItYedUr+wZZacqW+uzpVr9pZmUqpVCk9s92fzTzDZcGAxnyqkaO2QTgdhPJT2m
 wpG2UwIKzzi13tmwakY7OAbXm76bGWVZCO3QTHVnNV8ku9wgeMc/ZGSLUT8hMDZlwEsW7u/D
 qt+NlTKiOIQsSW7u7h3SFm7sMQo03X/taK9PJhS2BhhgnXg8mOa6U+yNaJy+eU0Lf5hEUiDC
 vDOI5x++LD3pdrJVr/6ZB0Qg3/YzZ0dk+phQ+KlP6HyeO4LG662toMbFbeLcBjcC/ceEclII
 90QNEFSZKM6NVloM+NaZRYVO3ApxWkFu+1mrVTXOwU0EVDzhbQEQANzLiI6gHkIhBQKeQaYs
 p2SSqF9c++9LOy5x6nbQ4s0X3oTKaMGfBZuiKkkU6NnHCSa0Az5ScRWLaRGu1PzjgcVwzl5O
 sDawR1BtOG/XoPRNB2351PRp++W8TWo2viYYY0uJHKFHML+ku9q0P+NkdTzFGJLP+hn7x0RT
 DMbhKTHO3H2xJz5TXNE9zTJuIfGAz3ShDpijvzYieY330BzZYfpgvCllDVM5E4XgfF4F/N90
 wWKu50fMA01ufwu+99GEwTFVG2az5T9SXd7vfSgRSkzXy7hcnxj4IhOfM6Ts85/BjMeIpeqy
 TDdsuetBgX9DMMWxMWl7BLeiMzMGrfkJ4tvlof0sVjurXibTibZyfyGR2ricg8iTbHyFaAzX
 2uFVoZaPxrp7udDfQ96sfz0hesF9Zi8d7NnNnMYbUmUtaS083L/l2EDKvCIkhSjd48XF+aO8
 VhrCfbXWpGRaLcY/gxi2TXRYG9xCa7PINgz9SyO34sL6TeFPSZn4bPQV5O1j85Dj4jBecB1k
 z2arzwlWWKMZUbR04HTeAuuvYvCKEMnfW3ABzdonh70QdqJbpQGfAF2p4/iCETKWuqefiOYn
 pR8PqoQA1DYv3t7y9DIN5Jw/8Oj5wOeEybw6vTMB0rrnx+JaXvxeHSlFzHiD6il/ChDDkJ9J
 /ejCHUQIl40wLSDRABEBAAHCwXwEGAECAA8FAlQ84W0CGwwFCRLMAwAAIQkQvS1hSGYUO0wW
 IQQFLN57whUFO2ifG8q9LWFIZhQ7TA1WD/9yxJvQrpf6LcNrr8uMlQWCg2iz2q1LGt1Itkuu
 KaavEF9nqHmoqhSfZeAIKAPn6xuYbGxXDrpN7dXCOH92fscLodZqZtK5FtbLvO572EPfxneY
 UT7JzDc/5LT9cFFugTMOhq1BG62vUm/F6V91+unyp4dRlyryAeqEuISykhvjZCVHk/woaMZv
 c1Dm4Uvkv0Ilelt3Pb9J7zhcx6sm5T7v16VceF96jG61bnJ2GFS+QZerZp3PY27XgtPxRxYj
 AmFUeF486PHx/2Yi4u1rQpIpC5inPxIgR1+ZFvQrAV36SvLFfuMhyCAxV6WBlQc85ArOiQZB
 Wm7L0repwr7zEJFEkdy8C81WRhMdPvHkAIh3RoY1SGcdB7rB3wCzfYkAuCBqaF7Zgfw8xkad
 KEiQTexRbM1sc/I8ACpla3N26SfQwrfg6V7TIoweP0RwDrcf5PVvwSWsRQp2LxFCkwnCXOra
 gYmkrmv0duG1FStpY+IIQn1TOkuXrciTVfZY1cZD0aVxwlxXBnUNZZNslldvXFtndxR0SFat
 sflovhDxKyhFwXOP0Rv8H378/+14TaykknRBIKEc0+lcr+EMOSUR5eg4aURb8Gc3Uc7fgQ6q
 UssTXzHPyj1hAyDpfu8DzAwlh4kKFTodxSsKAjI45SLjadSc94/5Gy8645Y1KgBzBPTH7Q==
In-Reply-To: <20240528020425.4994-1-nas.chung@chipsnmedia.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit

On 28/05/2024 04:04, Nas Chung wrote:
> Explicitly compare a buffer type only with valid buffer types,
> to avoid matching the buffer type outside of valid buffer
> type set.
> 
> Signed-off-by: Nas Chung <nas.chung@chipsnmedia.com>
> ---
>  include/uapi/linux/videodev2.h | 7 ++++++-
>  1 file changed, 6 insertions(+), 1 deletion(-)
> 
> diff --git a/include/uapi/linux/videodev2.h b/include/uapi/linux/videodev2.h
> index fe6b67e83751..fa2b7086e480 100644
> --- a/include/uapi/linux/videodev2.h
> +++ b/include/uapi/linux/videodev2.h
> @@ -157,6 +157,10 @@ enum v4l2_buf_type {
>  	V4L2_BUF_TYPE_PRIVATE              = 0x80,
>  };
>  
> +#define V4L2_TYPE_IS_VALID(type)		\
> +	((type) >= V4L2_BUF_TYPE_VIDEO_CAPTURE	\
> +	 && (type) <= V4L2_BUF_TYPE_META_OUTPUT)
> +
>  #define V4L2_TYPE_IS_MULTIPLANAR(type)			\
>  	((type) == V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE	\
>  	 || (type) == V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE)
> @@ -171,7 +175,8 @@ enum v4l2_buf_type {
>  	 || (type) == V4L2_BUF_TYPE_SDR_OUTPUT			\
>  	 || (type) == V4L2_BUF_TYPE_META_OUTPUT)
>  
> -#define V4L2_TYPE_IS_CAPTURE(type) (!V4L2_TYPE_IS_OUTPUT(type))
> +#define V4L2_TYPE_IS_CAPTURE(type)	\
> +	(V4L2_TYPE_IS_VALID(type) && !V4L2_TYPE_IS_OUTPUT(type))
>  
>  enum v4l2_tuner_type {
>  	V4L2_TUNER_RADIO	     = 1,

This patch introduced this warning:

In function 'find_format_by_index',
    inlined from 'vdec_enum_fmt' at drivers/media/platform/qcom/venus/vdec.c:452:8:
drivers/media/platform/qcom/venus/vdec.c:172:32: warning: 'valid' may be used uninitialized [-Wmaybe-uninitialized]
  172 |                 if (k == index && valid)
      |                     ~~~~~~~~~~~^~~~~~~~
drivers/media/platform/qcom/venus/vdec.c: In function 'vdec_enum_fmt':
drivers/media/platform/qcom/venus/vdec.c:157:22: note: 'valid' was declared here
  157 |                 bool valid;
      |                      ^~~~~

This driver does this:

	bool valid;

        if (V4L2_TYPE_IS_OUTPUT(type)) {
                valid = venus_helper_check_codec(inst, fmt[i].pixfmt);
        } else if (V4L2_TYPE_IS_CAPTURE(type)) {
                valid = venus_helper_check_format(inst, fmt[i].pixfmt);

With your patch both V4L2_TYPE_IS_OUTPUT(type) and V4L2_TYPE_IS_CAPTURE(type)
can return false, something that wasn't possible without this patch.

In this driver the fix would just be to drop the second 'if' altogether, so just
'} else {'.

Since these defines are part of the public API, this change introduces a subtle
behavior difference that can affect applications.

That said, I do consider this an improvement.

I would like some changes, though:

1) Just after "V4L2_BUF_TYPE_META_OUTPUT          = 14," in enum v4l2_buf_type,
   add a comment saying that V4L2_TYPE_IS_VALID and (for output types)
   V4L2_TYPE_IS_OUTPUT must be updated if a new type is added. It's all
   too easy to forget that otherwise.
2) Add a patch fixing the venus/vdec.c code.
3) Something else I noticed, but I think this change should be in a separate patch:
   V4L2_TYPE_IS_OUTPUT() returns true for V4L2_BUF_TYPE_VIDEO_OVERLAY, but that
   definitely belongs to CAPTURE. Nobody really uses that type anymore, but still,
   it should be fixed.

Regards,

	Hans