[PATCH 00/25] media: v4l2-mem2mem: Reduce cargo-cult

[PATCH 00/25] media: v4l2-mem2mem: Reduce cargo-cult

[Laurent Pinchart]

Hello,

The v4l2_m2m_get_vq() function never returns NULL, but many mem2mem
drivers still check its return value and consider NULL as an error. This
may have originated a long time ago from valid checks when
v4l2_m2m_get_vq() could return NULL, with drivers then just copying the
checks. This series attempts to stop the cargo-cult behaviour.

Patch 01/25 starts by explicitly stating in kerneldoc that the
v4l2_m2m_get_vq() function never returns NULL. All the other patches
drop NULL checks from drivers.

I have carefully checked all patched locations in all drivers. They fall
in 3 categories:

- Checks in the VIDIOC_G_FMT, VIDIOC_TRY_FMT and VIDIOC_S_FMT handlers:
  Those may have been added to ensure that the format type has a valid
  value, but that is ensured by the V4L2 ioctl core before calling the
  handlers. The checks can be dropped without a need to replace them
  with proper type checks.

- Checks in the VIDIOC_S_SELECTION handler: The only location where this
  is performed has an explicit type check, so the NULL check can also be
  dropped.

- Checks in other locations where the type parameter to the
  v4l2_m2m_get_vq() function is hardcoded: The hardcoded type is valid,
  so the NULL check can't have been meant to check the type. It can also
  be removed.

There's no dependency between any of those patches so they can be merged
in any order.

Laurent Pinchart (25):
  media: v4l2-mem2mem: Document that v4l2_m2m_get_vq() never returns
    NULL
  media: allgro-dvt: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: meson-g2d: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: amphion: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: coda: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: imagination: e5010: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: m2m-deinterlace: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: mediatek: jpeg: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: mediatek: vcodec: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: dw100: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: imx-jpeg: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: imx-pxp: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: nxp: imx8-isi: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: mx2_emmaprp: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: qcom: iris: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: qcom: venus: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: renesas: fdp1: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: rcar_jpu: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: platform: rga: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: samsung: s5p-g2d: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: samsung: s5p-jpeg: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: stm32: dma2d: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: ti: vpe: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: vicodec: Drop unneeded v4l2_m2m_get_vq() NULL check
  media: vim2m: Drop unneeded v4l2_m2m_get_vq() NULL check

 drivers/media/platform/allegro-dvt/allegro-core.c    |  2 --
 drivers/media/platform/amlogic/meson-ge2d/ge2d.c     |  5 -----
 drivers/media/platform/amphion/vdec.c                |  2 --
 drivers/media/platform/amphion/venc.c                |  2 --
 .../media/platform/chips-media/coda/coda-common.c    |  4 ----
 drivers/media/platform/imagination/e5010-jpeg-enc.c  |  4 ----
 drivers/media/platform/m2m-deinterlace.c             |  7 -------
 drivers/media/platform/mediatek/jpeg/mtk_jpeg_core.c |  7 -------
 .../mediatek/vcodec/decoder/mtk_vcodec_dec.c         |  7 -------
 .../vcodec/decoder/vdec/vdec_av1_req_lat_if.c        |  2 --
 .../vcodec/decoder/vdec/vdec_vp9_req_lat_if.c        |  2 --
 .../mediatek/vcodec/encoder/mtk_vcodec_enc.c         |  8 --------
 drivers/media/platform/nxp/dw100/dw100.c             |  7 -------
 drivers/media/platform/nxp/imx-jpeg/mxc-jpeg.c       |  4 ----
 drivers/media/platform/nxp/imx-pxp.c                 |  7 -------
 drivers/media/platform/nxp/imx8-isi/imx8-isi-m2m.c   |  2 --
 drivers/media/platform/nxp/mx2_emmaprp.c             |  7 -------
 drivers/media/platform/qcom/iris/iris_vdec.c         |  2 --
 drivers/media/platform/qcom/venus/vdec.c             |  2 --
 drivers/media/platform/qcom/venus/venc.c             |  2 --
 drivers/media/platform/renesas/rcar_fdp1.c           |  3 ---
 drivers/media/platform/renesas/rcar_jpu.c            |  8 --------
 drivers/media/platform/rockchip/rga/rga.c            |  4 ----
 drivers/media/platform/samsung/s5p-g2d/g2d.c         |  4 ----
 drivers/media/platform/samsung/s5p-jpeg/jpeg-core.c  |  7 -------
 drivers/media/platform/st/stm32/dma2d/dma2d.c        |  5 -----
 drivers/media/platform/ti/vpe/vpe.c                  |  7 -------
 drivers/media/test-drivers/vicodec/vicodec-core.c    |  7 -------
 drivers/media/test-drivers/vim2m.c                   | 12 ------------
 drivers/media/v4l2-core/v4l2-mem2mem.c               | 12 +-----------
 include/media/v4l2-mem2mem.h                         |  3 +++
 31 files changed, 4 insertions(+), 153 deletions(-)


base-commit: e5f0a698b34ed76002dc5cff3804a61c80233a7a
-- 
Regards,

Laurent Pinchart

[PATCH 15/25] media: qcom: iris: Drop unneeded v4l2_m2m_get_vq() NULL check

[Laurent Pinchart]

The v4l2_m2m_get_vq() function never returns NULL. The check may have
been intended to catch invalid format types, but that's not needed as
the V4L2 core picks the appropriate VIDIOC_S_FMT ioctl handler based on
the format type, so the type can't be incorrect. Drop the unneeded
return value check.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
---
 drivers/media/platform/qcom/iris/iris_vdec.c | 2 --
 1 file changed, 2 deletions(-)

diff --git a/drivers/media/platform/qcom/iris/iris_vdec.c b/drivers/media/platform/qcom/iris/iris_vdec.c
index d670b51c5839..1e9ffdbb6e18 100644
--- a/drivers/media/platform/qcom/iris/iris_vdec.c
+++ b/drivers/media/platform/qcom/iris/iris_vdec.c
@@ -191,8 +191,6 @@ int iris_vdec_s_fmt(struct iris_inst *inst, struct v4l2_format *f)
 	u32 codec_align;
 
 	q = v4l2_m2m_get_vq(inst->m2m_ctx, f->type);
-	if (!q)
-		return -EINVAL;
 
 	if (vb2_is_busy(q))
 		return -EBUSY;
-- 
Regards,

Laurent Pinchart

Re: [PATCH 15/25] media: qcom: iris: Drop unneeded v4l2_m2m_get_vq() NULL check

[Bryan O'Donoghue]

On 08/10/2025 18:50, Laurent Pinchart wrote:
> The v4l2_m2m_get_vq() function never returns NULL. The check may have
> been intended to catch invalid format types, but that's not needed as
> the V4L2 core picks the appropriate VIDIOC_S_FMT ioctl handler based on
> the format type, so the type can't be incorrect. Drop the unneeded
> return value check.
> 
> Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> ---
>   drivers/media/platform/qcom/iris/iris_vdec.c | 2 --
>   1 file changed, 2 deletions(-)
> 
> diff --git a/drivers/media/platform/qcom/iris/iris_vdec.c b/drivers/media/platform/qcom/iris/iris_vdec.c
> index d670b51c5839..1e9ffdbb6e18 100644
> --- a/drivers/media/platform/qcom/iris/iris_vdec.c
> +++ b/drivers/media/platform/qcom/iris/iris_vdec.c
> @@ -191,8 +191,6 @@ int iris_vdec_s_fmt(struct iris_inst *inst, struct v4l2_format *f)
>   	u32 codec_align;
>   
>   	q = v4l2_m2m_get_vq(inst->m2m_ctx, f->type);
> -	if (!q)
> -		return -EINVAL;
>   
>   	if (vb2_is_busy(q))
>   		return -EBUSY;
Reviewed-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>

Re: [PATCH 15/25] media: qcom: iris: Drop unneeded v4l2_m2m_get_vq() NULL check

[Dikshita Agarwal]

On 10/8/2025 11:20 PM, Laurent Pinchart wrote:
> The v4l2_m2m_get_vq() function never returns NULL. The check may have
> been intended to catch invalid format types, but that's not needed as
> the V4L2 core picks the appropriate VIDIOC_S_FMT ioctl handler based on
> the format type, so the type can't be incorrect. Drop the unneeded
> return value check.
> 
> Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> ---
>  drivers/media/platform/qcom/iris/iris_vdec.c | 2 --
>  1 file changed, 2 deletions(-)
> 
> diff --git a/drivers/media/platform/qcom/iris/iris_vdec.c b/drivers/media/platform/qcom/iris/iris_vdec.c
> index d670b51c5839..1e9ffdbb6e18 100644
> --- a/drivers/media/platform/qcom/iris/iris_vdec.c
> +++ b/drivers/media/platform/qcom/iris/iris_vdec.c
> @@ -191,8 +191,6 @@ int iris_vdec_s_fmt(struct iris_inst *inst, struct v4l2_format *f)
>  	u32 codec_align;
>  
>  	q = v4l2_m2m_get_vq(inst->m2m_ctx, f->type);
> -	if (!q)
> -		return -EINVAL;
>  
>  	if (vb2_is_busy(q))
>  		return -EBUSY;

The same change would be required for iris_venc.c as well which is part of
linux-next[1]

[1]:
https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/tree/drivers/media/platform/qcom/iris/iris_venc.c#n271

Thanks,
Dikshita

Re: [PATCH 15/25] media: qcom: iris: Drop unneeded v4l2_m2m_get_vq() NULL check

[Laurent Pinchart]

Hi Dikshita,

On Thu, Oct 09, 2025 at 07:23:20PM +0530, Dikshita Agarwal wrote:
> On 10/8/2025 11:20 PM, Laurent Pinchart wrote:
> > The v4l2_m2m_get_vq() function never returns NULL. The check may have
> > been intended to catch invalid format types, but that's not needed as
> > the V4L2 core picks the appropriate VIDIOC_S_FMT ioctl handler based on
> > the format type, so the type can't be incorrect. Drop the unneeded
> > return value check.
> > 
> > Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> > ---
> >  drivers/media/platform/qcom/iris/iris_vdec.c | 2 --
> >  1 file changed, 2 deletions(-)
> > 
> > diff --git a/drivers/media/platform/qcom/iris/iris_vdec.c b/drivers/media/platform/qcom/iris/iris_vdec.c
> > index d670b51c5839..1e9ffdbb6e18 100644
> > --- a/drivers/media/platform/qcom/iris/iris_vdec.c
> > +++ b/drivers/media/platform/qcom/iris/iris_vdec.c
> > @@ -191,8 +191,6 @@ int iris_vdec_s_fmt(struct iris_inst *inst, struct v4l2_format *f)
> >  	u32 codec_align;
> >  
> >  	q = v4l2_m2m_get_vq(inst->m2m_ctx, f->type);
> > -	if (!q)
> > -		return -EINVAL;
> >  
> >  	if (vb2_is_busy(q))
> >  		return -EBUSY;
> 
> The same change would be required for iris_venc.c as well which is part of
> linux-next[1]

Thank you for noticing. I'll rebase this series on v6.18-rc1 once it
gets released and will make sure iris_venc.c is addressed.

> [1]: https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/tree/drivers/media/platform/qcom/iris/iris_venc.c#n271

-- 
Regards,

Laurent Pinchart

[PATCH 16/25] media: qcom: venus: Drop unneeded v4l2_m2m_get_vq() NULL check

[Laurent Pinchart]

The v4l2_m2m_get_vq() function never returns NULL. The check may have
been intended to catch invalid format types, but that's not needed as
the V4L2 core picks the appropriate VIDIOC_S_FMT ioctl handler based on
the format type, so the type can't be incorrect. Drop the unneeded
return value check.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
---
 drivers/media/platform/qcom/venus/vdec.c | 2 --
 drivers/media/platform/qcom/venus/venc.c | 2 --
 2 files changed, 4 deletions(-)

diff --git a/drivers/media/platform/qcom/venus/vdec.c b/drivers/media/platform/qcom/venus/vdec.c
index 29b0d6a5303d..8c77db0f6e76 100644
--- a/drivers/media/platform/qcom/venus/vdec.c
+++ b/drivers/media/platform/qcom/venus/vdec.c
@@ -329,8 +329,6 @@ static int vdec_s_fmt(struct file *file, void *fh, struct v4l2_format *f)
 	struct vb2_queue *q;
 
 	q = v4l2_m2m_get_vq(inst->m2m_ctx, f->type);
-	if (!q)
-		return -EINVAL;
 
 	if (vb2_is_busy(q))
 		return -EBUSY;
diff --git a/drivers/media/platform/qcom/venus/venc.c b/drivers/media/platform/qcom/venus/venc.c
index c0a0ccdded80..0fe4cc37118b 100644
--- a/drivers/media/platform/qcom/venus/venc.c
+++ b/drivers/media/platform/qcom/venus/venc.c
@@ -241,8 +241,6 @@ static int venc_s_fmt(struct file *file, void *fh, struct v4l2_format *f)
 	struct vb2_queue *q;
 
 	q = v4l2_m2m_get_vq(inst->m2m_ctx, f->type);
-	if (!q)
-		return -EINVAL;
 
 	if (vb2_is_busy(q))
 		return -EBUSY;
-- 
Regards,

Laurent Pinchart

Re: [PATCH 16/25] media: qcom: venus: Drop unneeded v4l2_m2m_get_vq() NULL check

[Bryan O'Donoghue]

On 08/10/2025 18:50, Laurent Pinchart wrote:
> The v4l2_m2m_get_vq() function never returns NULL. The check may have
> been intended to catch invalid format types, but that's not needed as
> the V4L2 core picks the appropriate VIDIOC_S_FMT ioctl handler based on
> the format type, so the type can't be incorrect. Drop the unneeded
> return value check.
> 
> Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> ---
>   drivers/media/platform/qcom/venus/vdec.c | 2 --
>   drivers/media/platform/qcom/venus/venc.c | 2 --
>   2 files changed, 4 deletions(-)
> 
> diff --git a/drivers/media/platform/qcom/venus/vdec.c b/drivers/media/platform/qcom/venus/vdec.c
> index 29b0d6a5303d..8c77db0f6e76 100644
> --- a/drivers/media/platform/qcom/venus/vdec.c
> +++ b/drivers/media/platform/qcom/venus/vdec.c
> @@ -329,8 +329,6 @@ static int vdec_s_fmt(struct file *file, void *fh, struct v4l2_format *f)
>   	struct vb2_queue *q;
>   
>   	q = v4l2_m2m_get_vq(inst->m2m_ctx, f->type);
> -	if (!q)
> -		return -EINVAL;
>   
>   	if (vb2_is_busy(q))
>   		return -EBUSY;
> diff --git a/drivers/media/platform/qcom/venus/venc.c b/drivers/media/platform/qcom/venus/venc.c
> index c0a0ccdded80..0fe4cc37118b 100644
> --- a/drivers/media/platform/qcom/venus/venc.c
> +++ b/drivers/media/platform/qcom/venus/venc.c
> @@ -241,8 +241,6 @@ static int venc_s_fmt(struct file *file, void *fh, struct v4l2_format *f)
>   	struct vb2_queue *q;
>   
>   	q = v4l2_m2m_get_vq(inst->m2m_ctx, f->type);
> -	if (!q)
> -		return -EINVAL;
>   
>   	if (vb2_is_busy(q))
>   		return -EBUSY;
Reviewed-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>

Re: [PATCH 16/25] media: qcom: venus: Drop unneeded v4l2_m2m_get_vq() NULL check

[Dikshita Agarwal]

On 10/8/2025 11:20 PM, Laurent Pinchart wrote:
> The v4l2_m2m_get_vq() function never returns NULL. The check may have
> been intended to catch invalid format types, but that's not needed as
> the V4L2 core picks the appropriate VIDIOC_S_FMT ioctl handler based on
> the format type, so the type can't be incorrect. Drop the unneeded
> return value check.
> 
> Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> ---
>  drivers/media/platform/qcom/venus/vdec.c | 2 --
>  drivers/media/platform/qcom/venus/venc.c | 2 --
>  2 files changed, 4 deletions(-)
> 
> diff --git a/drivers/media/platform/qcom/venus/vdec.c b/drivers/media/platform/qcom/venus/vdec.c
> index 29b0d6a5303d..8c77db0f6e76 100644
> --- a/drivers/media/platform/qcom/venus/vdec.c
> +++ b/drivers/media/platform/qcom/venus/vdec.c
> @@ -329,8 +329,6 @@ static int vdec_s_fmt(struct file *file, void *fh, struct v4l2_format *f)
>  	struct vb2_queue *q;
>  
>  	q = v4l2_m2m_get_vq(inst->m2m_ctx, f->type);
> -	if (!q)
> -		return -EINVAL;
>  
>  	if (vb2_is_busy(q))
>  		return -EBUSY;
> diff --git a/drivers/media/platform/qcom/venus/venc.c b/drivers/media/platform/qcom/venus/venc.c
> index c0a0ccdded80..0fe4cc37118b 100644
> --- a/drivers/media/platform/qcom/venus/venc.c
> +++ b/drivers/media/platform/qcom/venus/venc.c
> @@ -241,8 +241,6 @@ static int venc_s_fmt(struct file *file, void *fh, struct v4l2_format *f)
>  	struct vb2_queue *q;
>  
>  	q = v4l2_m2m_get_vq(inst->m2m_ctx, f->type);
> -	if (!q)
> -		return -EINVAL;
>  
>  	if (vb2_is_busy(q))
>  		return -EBUSY;

Reviewed-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>

Thanks,
Dikshita

Re: [PATCH 00/25] media: v4l2-mem2mem: Reduce cargo-cult

[Nicolas Dufresne]

[-- Attachment #1: Type: text/plain, Size: 6036 bytes --]

Hi,

Le mercredi 08 octobre 2025 à 20:50 +0300, Laurent Pinchart a écrit :
> Hello,
> 
> The v4l2_m2m_get_vq() function never returns NULL, but many mem2mem
> drivers still check its return value and consider NULL as an error. This
> may have originated a long time ago from valid checks when
> v4l2_m2m_get_vq() could return NULL, with drivers then just copying the
> checks. This series attempts to stop the cargo-cult behaviour.
> 
> Patch 01/25 starts by explicitly stating in kerneldoc that the
> v4l2_m2m_get_vq() function never returns NULL. All the other patches
> drop NULL checks from drivers.
> 
> I have carefully checked all patched locations in all drivers. They fall
> in 3 categories:
> 
> - Checks in the VIDIOC_G_FMT, VIDIOC_TRY_FMT and VIDIOC_S_FMT handlers:
>   Those may have been added to ensure that the format type has a valid
>   value, but that is ensured by the V4L2 ioctl core before calling the
>   handlers. The checks can be dropped without a need to replace them
>   with proper type checks.
> 
> - Checks in the VIDIOC_S_SELECTION handler: The only location where this
>   is performed has an explicit type check, so the NULL check can also be
>   dropped.
> 
> - Checks in other locations where the type parameter to the
>   v4l2_m2m_get_vq() function is hardcoded: The hardcoded type is valid,
>   so the NULL check can't have been meant to check the type. It can also
>   be removed.
> 
> There's no dependency between any of those patches so they can be merged
> in any order.
> 
> Laurent Pinchart (25):
>   media: v4l2-mem2mem: Document that v4l2_m2m_get_vq() never returns
>     NULL
>   media: allgro-dvt: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: meson-g2d: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: amphion: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: coda: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: imagination: e5010: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: m2m-deinterlace: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: mediatek: jpeg: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: mediatek: vcodec: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: dw100: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: imx-jpeg: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: imx-pxp: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: nxp: imx8-isi: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: mx2_emmaprp: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: qcom: iris: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: qcom: venus: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: renesas: fdp1: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: rcar_jpu: Drop unneeded v4l2_m2m_get_vq() NULL check

Why not "renesas: jpu" to match the fdp1 patch naming ?

>   media: platform: rga: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: samsung: s5p-g2d: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: samsung: s5p-jpeg: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: stm32: dma2d: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: ti: vpe: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: vicodec: Drop unneeded v4l2_m2m_get_vq() NULL check
>   media: vim2m: Drop unneeded v4l2_m2m_get_vq() NULL check

I reviewed the list and it seems complete to me.

Nicolas

> 
>  drivers/media/platform/allegro-dvt/allegro-core.c    |  2 --
>  drivers/media/platform/amlogic/meson-ge2d/ge2d.c     |  5 -----
>  drivers/media/platform/amphion/vdec.c                |  2 --
>  drivers/media/platform/amphion/venc.c                |  2 --
>  .../media/platform/chips-media/coda/coda-common.c    |  4 ----
>  drivers/media/platform/imagination/e5010-jpeg-enc.c  |  4 ----
>  drivers/media/platform/m2m-deinterlace.c             |  7 -------
>  drivers/media/platform/mediatek/jpeg/mtk_jpeg_core.c |  7 -------
>  .../mediatek/vcodec/decoder/mtk_vcodec_dec.c         |  7 -------
>  .../vcodec/decoder/vdec/vdec_av1_req_lat_if.c        |  2 --
>  .../vcodec/decoder/vdec/vdec_vp9_req_lat_if.c        |  2 --
>  .../mediatek/vcodec/encoder/mtk_vcodec_enc.c         |  8 --------
>  drivers/media/platform/nxp/dw100/dw100.c             |  7 -------
>  drivers/media/platform/nxp/imx-jpeg/mxc-jpeg.c       |  4 ----
>  drivers/media/platform/nxp/imx-pxp.c                 |  7 -------
>  drivers/media/platform/nxp/imx8-isi/imx8-isi-m2m.c   |  2 --
>  drivers/media/platform/nxp/mx2_emmaprp.c             |  7 -------
>  drivers/media/platform/qcom/iris/iris_vdec.c         |  2 --
>  drivers/media/platform/qcom/venus/vdec.c             |  2 --
>  drivers/media/platform/qcom/venus/venc.c             |  2 --
>  drivers/media/platform/renesas/rcar_fdp1.c           |  3 ---
>  drivers/media/platform/renesas/rcar_jpu.c            |  8 --------
>  drivers/media/platform/rockchip/rga/rga.c            |  4 ----
>  drivers/media/platform/samsung/s5p-g2d/g2d.c         |  4 ----
>  drivers/media/platform/samsung/s5p-jpeg/jpeg-core.c  |  7 -------
>  drivers/media/platform/st/stm32/dma2d/dma2d.c        |  5 -----
>  drivers/media/platform/ti/vpe/vpe.c                  |  7 -------
>  drivers/media/test-drivers/vicodec/vicodec-core.c    |  7 -------
>  drivers/media/test-drivers/vim2m.c                   | 12 ------------
>  drivers/media/v4l2-core/v4l2-mem2mem.c               | 12 +-----------
>  include/media/v4l2-mem2mem.h                         |  3 +++
>  31 files changed, 4 insertions(+), 153 deletions(-)
> 
> 
> base-commit: e5f0a698b34ed76002dc5cff3804a61c80233a7a

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 228 bytes --]

Re: [PATCH 00/25] media: v4l2-mem2mem: Reduce cargo-cult

[Laurent Pinchart]

On Wed, Oct 08, 2025 at 02:24:30PM -0400, Nicolas Dufresne wrote:
> Hi,
> 
> Le mercredi 08 octobre 2025 à 20:50 +0300, Laurent Pinchart a écrit :
> > Hello,
> > 
> > The v4l2_m2m_get_vq() function never returns NULL, but many mem2mem
> > drivers still check its return value and consider NULL as an error. This
> > may have originated a long time ago from valid checks when
> > v4l2_m2m_get_vq() could return NULL, with drivers then just copying the
> > checks. This series attempts to stop the cargo-cult behaviour.
> > 
> > Patch 01/25 starts by explicitly stating in kerneldoc that the
> > v4l2_m2m_get_vq() function never returns NULL. All the other patches
> > drop NULL checks from drivers.
> > 
> > I have carefully checked all patched locations in all drivers. They fall
> > in 3 categories:
> > 
> > - Checks in the VIDIOC_G_FMT, VIDIOC_TRY_FMT and VIDIOC_S_FMT handlers:
> >   Those may have been added to ensure that the format type has a valid
> >   value, but that is ensured by the V4L2 ioctl core before calling the
> >   handlers. The checks can be dropped without a need to replace them
> >   with proper type checks.
> > 
> > - Checks in the VIDIOC_S_SELECTION handler: The only location where this
> >   is performed has an explicit type check, so the NULL check can also be
> >   dropped.
> > 
> > - Checks in other locations where the type parameter to the
> >   v4l2_m2m_get_vq() function is hardcoded: The hardcoded type is valid,
> >   so the NULL check can't have been meant to check the type. It can also
> >   be removed.
> > 
> > There's no dependency between any of those patches so they can be merged
> > in any order.
> > 
> > Laurent Pinchart (25):
> >   media: v4l2-mem2mem: Document that v4l2_m2m_get_vq() never returns
> >     NULL
> >   media: allgro-dvt: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: meson-g2d: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: amphion: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: coda: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: imagination: e5010: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: m2m-deinterlace: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: mediatek: jpeg: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: mediatek: vcodec: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: dw100: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: imx-jpeg: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: imx-pxp: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: nxp: imx8-isi: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: mx2_emmaprp: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: qcom: iris: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: qcom: venus: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: renesas: fdp1: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: rcar_jpu: Drop unneeded v4l2_m2m_get_vq() NULL check
> 
> Why not "renesas: jpu" to match the fdp1 patch naming ?

I tried to go with the most common prefix as reported by git log. I
don't mind changing this, I'll wait for more reviews to see if a v2 is
needed, otherwise this can be updated when applying if desired.

> >   media: platform: rga: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: samsung: s5p-g2d: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: samsung: s5p-jpeg: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: stm32: dma2d: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: ti: vpe: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: vicodec: Drop unneeded v4l2_m2m_get_vq() NULL check
> >   media: vim2m: Drop unneeded v4l2_m2m_get_vq() NULL check
> 
> I reviewed the list and it seems complete to me.

Thank you.

> >  drivers/media/platform/allegro-dvt/allegro-core.c    |  2 --
> >  drivers/media/platform/amlogic/meson-ge2d/ge2d.c     |  5 -----
> >  drivers/media/platform/amphion/vdec.c                |  2 --
> >  drivers/media/platform/amphion/venc.c                |  2 --
> >  .../media/platform/chips-media/coda/coda-common.c    |  4 ----
> >  drivers/media/platform/imagination/e5010-jpeg-enc.c  |  4 ----
> >  drivers/media/platform/m2m-deinterlace.c             |  7 -------
> >  drivers/media/platform/mediatek/jpeg/mtk_jpeg_core.c |  7 -------
> >  .../mediatek/vcodec/decoder/mtk_vcodec_dec.c         |  7 -------
> >  .../vcodec/decoder/vdec/vdec_av1_req_lat_if.c        |  2 --
> >  .../vcodec/decoder/vdec/vdec_vp9_req_lat_if.c        |  2 --
> >  .../mediatek/vcodec/encoder/mtk_vcodec_enc.c         |  8 --------
> >  drivers/media/platform/nxp/dw100/dw100.c             |  7 -------
> >  drivers/media/platform/nxp/imx-jpeg/mxc-jpeg.c       |  4 ----
> >  drivers/media/platform/nxp/imx-pxp.c                 |  7 -------
> >  drivers/media/platform/nxp/imx8-isi/imx8-isi-m2m.c   |  2 --
> >  drivers/media/platform/nxp/mx2_emmaprp.c             |  7 -------
> >  drivers/media/platform/qcom/iris/iris_vdec.c         |  2 --
> >  drivers/media/platform/qcom/venus/vdec.c             |  2 --
> >  drivers/media/platform/qcom/venus/venc.c             |  2 --
> >  drivers/media/platform/renesas/rcar_fdp1.c           |  3 ---
> >  drivers/media/platform/renesas/rcar_jpu.c            |  8 --------
> >  drivers/media/platform/rockchip/rga/rga.c            |  4 ----
> >  drivers/media/platform/samsung/s5p-g2d/g2d.c         |  4 ----
> >  drivers/media/platform/samsung/s5p-jpeg/jpeg-core.c  |  7 -------
> >  drivers/media/platform/st/stm32/dma2d/dma2d.c        |  5 -----
> >  drivers/media/platform/ti/vpe/vpe.c                  |  7 -------
> >  drivers/media/test-drivers/vicodec/vicodec-core.c    |  7 -------
> >  drivers/media/test-drivers/vim2m.c                   | 12 ------------
> >  drivers/media/v4l2-core/v4l2-mem2mem.c               | 12 +-----------
> >  include/media/v4l2-mem2mem.h                         |  3 +++
> >  31 files changed, 4 insertions(+), 153 deletions(-)
> > 
> > 
> > base-commit: e5f0a698b34ed76002dc5cff3804a61c80233a7a

-- 
Regards,

Laurent Pinchart