Re: [PATCH v21 3/9] media: test-drivers: Set REQBUFS minimum number of buffers

From: Hans Verkuil <hverkuil-cisco@xs4all.nl>
To: Benjamin Gaignard <benjamin.gaignard@collabora.com>,
	mchehab@kernel.org, tfiga@chromium.org, m.szyprowski@samsung.com,
	ezequiel@vanguardiasur.com.ar, p.zabel@pengutronix.de,
	nicolas@ndufresne.ca
Cc: linux-media@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-rockchip@lists.infradead.org, kernel@collabora.com
Subject: Re: [PATCH v21 3/9] media: test-drivers: Set REQBUFS minimum number of buffers
Date: Mon, 18 Mar 2024 11:06:09 +0100	[thread overview]
Message-ID: <43d27ed7-94ff-4916-b1f3-07204b2f54ab@xs4all.nl> (raw)
In-Reply-To: <20240314153226.197445-4-benjamin.gaignard@collabora.com>

Hi Benjamin,

On 14/03/2024 4:32 pm, Benjamin Gaignard wrote:
> Instead of using 'min_queued_buffers' field to specify the
> minimum number of buffers to be allocated when calling REQBUF
> use 'min_reqbufs_allocation' field which is dedicated to this
> purpose.
> 
> While at it rename vivid_create_queue() parameter.
> 
> Signed-off-by: Benjamin Gaignard <benjamin.gaignard@collabora.com>
> ---
>  drivers/media/test-drivers/vimc/vimc-capture.c | 2 +-
>  drivers/media/test-drivers/vivid/vivid-core.c  | 4 ++--
>  2 files changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/media/test-drivers/vimc/vimc-capture.c b/drivers/media/test-drivers/vimc/vimc-capture.c
> index 2a2d19d23bab..97693561f1e4 100644
> --- a/drivers/media/test-drivers/vimc/vimc-capture.c
> +++ b/drivers/media/test-drivers/vimc/vimc-capture.c
> @@ -432,7 +432,7 @@ static struct vimc_ent_device *vimc_capture_add(struct vimc_device *vimc,
>  	q->mem_ops = vimc_allocator == VIMC_ALLOCATOR_DMA_CONTIG
>  		   ? &vb2_dma_contig_memops : &vb2_vmalloc_memops;
>  	q->timestamp_flags = V4L2_BUF_FLAG_TIMESTAMP_MONOTONIC;
> -	q->min_queued_buffers = 2;
> +	q->min_reqbufs_allocation = 2;
>  	q->lock = &vcapture->lock;
>  	q->dev = v4l2_dev->dev;
>  
> diff --git a/drivers/media/test-drivers/vivid/vivid-core.c b/drivers/media/test-drivers/vivid/vivid-core.c
> index 159c72cbb5bf..11b8520d9f57 100644
> --- a/drivers/media/test-drivers/vivid/vivid-core.c
> +++ b/drivers/media/test-drivers/vivid/vivid-core.c
> @@ -861,7 +861,7 @@ static const struct media_device_ops vivid_media_ops = {
>  static int vivid_create_queue(struct vivid_dev *dev,
>  			      struct vb2_queue *q,
>  			      u32 buf_type,
> -			      unsigned int min_queued_buffers,
> +			      unsigned int min_reqbufs_allocation,
>  			      const struct vb2_ops *ops)
>  {
>  	if (buf_type == V4L2_BUF_TYPE_VIDEO_CAPTURE && dev->multiplanar)
> @@ -898,7 +898,7 @@ static int vivid_create_queue(struct vivid_dev *dev,
>  	q->mem_ops = allocators[dev->inst] == 1 ? &vb2_dma_contig_memops :
>  						  &vb2_vmalloc_memops;
>  	q->timestamp_flags = V4L2_BUF_FLAG_TIMESTAMP_MONOTONIC;
> -	q->min_queued_buffers = supports_requests[dev->inst] ? 0 : min_queued_buffers;
> +	q->min_reqbufs_allocation = min_reqbufs_allocation;
>  	q->lock = &dev->mutex;
>  	q->dev = dev->v4l2_dev.dev;
>  	q->supports_requests = supports_requests[dev->inst];

How did you test this series? If I run the test-media script using the patch
v4l2-compliance (the v9 series) I get two failures. Both are traced to code
in vivid: meta_out_queue_setup() and touch_cap_queue_setup():

Both have this code:

        if (*nplanes) {
                if (sizes[0] < size)
                        return -EINVAL;
        } else {
                sizes[0] = size;
        }

        if (q_num_bufs + *nbuffers < 2)
                *nbuffers = 2 - q_num_bufs;

        *nplanes = 1;
        return 0;

It's the *nbuffers assignment that is wrong. Looking at vivid-core.c I see
that vivid_create_queue() is called with min_reqbufs_allocation set to 1 for
these two devices. I think that should be 2, and then the code above can
be dropped.

It is probably best to post a v21.1 3/9 patch only.

Regards,

	Hans