Re: [PATCH v7 09/16] media: ti: j721e-csi2rx: add support for processing virtual channels

[Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>]

Hi,

On 11/09/2025 13:28, Rishikesh Donadkar wrote:
> From: Jai Luthra <j-luthra@ti.com>
> 
> Use get_frame_desc() to get the frame desc from the connected source,
> and use the provided virtual channel instead of hardcoded one.
> 
> get_frame_desc() returns the same information when called on each stream
> start, so instead get the VCs for all the routed stream at first
> stream start and cache this information in the driver.
> 
> get_frame_desc() works per stream, but as we don't support multiple
> streams yet, we will just always use stream 0. If the source doesn't
> support get_frame_desc(), fall back to the previous method of always
> capturing virtual channel 0.
> 
> Reviewed-by: Yemike Abhilash Chandra <y-abhilashchandra@ti.com>
> Co-developed-by: Pratyush Yadav <p.yadav@ti.com>
> Signed-off-by: Pratyush Yadav <p.yadav@ti.com>
> Signed-off-by: Jai Luthra <j-luthra@ti.com>
> Signed-off-by: Rishikesh Donadkar <r-donadkar@ti.com>
> ---
>  .../platform/ti/j721e-csi2rx/j721e-csi2rx.c   | 107 ++++++++++++++++++
>  1 file changed, 107 insertions(+)
> 
> diff --git a/drivers/media/platform/ti/j721e-csi2rx/j721e-csi2rx.c b/drivers/media/platform/ti/j721e-csi2rx/j721e-csi2rx.c
> index f3443c3ab01a..903722b5aea7 100644
> --- a/drivers/media/platform/ti/j721e-csi2rx/j721e-csi2rx.c
> +++ b/drivers/media/platform/ti/j721e-csi2rx/j721e-csi2rx.c
> @@ -32,6 +32,7 @@
>  #define SHIM_DMACNTX_YUV422		GENMASK(27, 26)
>  #define SHIM_DMACNTX_DUAL_PCK_CFG	BIT(24)
>  #define SHIM_DMACNTX_SIZE		GENMASK(21, 20)
> +#define SHIM_DMACNTX_VC			GENMASK(9, 6)
>  #define SHIM_DMACNTX_FMT		GENMASK(5, 0)
>  #define SHIM_DMACNTX_YUV422_MODE_11	3
>  #define SHIM_DMACNTX_SIZE_8		0
> @@ -103,6 +104,7 @@ struct ti_csi2rx_dev;
>  
>  struct ti_csi2rx_ctx {
>  	struct ti_csi2rx_dev		*csi;
> +	struct v4l2_subdev_route	*route;
>  	struct video_device		vdev;
>  	struct vb2_queue		vidq;
>  	struct mutex			mutex; /* To serialize ioctls. */
> @@ -111,6 +113,8 @@ struct ti_csi2rx_ctx {
>  	struct media_pad		pad;
>  	u32				sequence;
>  	u32				idx;
> +	u32				vc;
> +	u32				stream;
>  };
>  
>  struct ti_csi2rx_dev {
> @@ -134,6 +138,7 @@ struct ti_csi2rx_dev {
>  		dma_addr_t		paddr;
>  		size_t			len;
>  	} drain;
> +	bool				vc_cached;
>  };
>  
>  static inline struct ti_csi2rx_dev *to_csi2rx_dev(struct v4l2_subdev *sd)
> @@ -610,6 +615,7 @@ static void ti_csi2rx_setup_shim(struct ti_csi2rx_ctx *ctx)
>  	}
>  
>  	reg |= FIELD_PREP(SHIM_DMACNTX_SIZE, fmt->size);
> +	reg |= FIELD_PREP(SHIM_DMACNTX_VC, ctx->vc);
>  
>  	writel(reg, csi->shim + SHIM_DMACNTX(ctx->idx));
>  
> @@ -884,6 +890,82 @@ static void ti_csi2rx_buffer_queue(struct vb2_buffer *vb)
>  	}
>  }
>  
> +static int ti_csi2rx_get_route(struct ti_csi2rx_ctx *ctx)
> +{
> +	struct ti_csi2rx_dev *csi = ctx->csi;
> +	struct media_pad *pad;
> +	struct v4l2_subdev_state *state;
> +	struct v4l2_subdev_route *r;
> +
> +	/* Get the source pad connected to this ctx */
> +	pad = media_entity_remote_source_pad_unique(ctx->pad.entity);
> +	if (!pad) {
> +		dev_err(csi->dev, "No pad connected to ctx %d\n", ctx->idx);
> +		return -ENODEV;
> +	}
> +
> +	state = v4l2_subdev_lock_and_get_active_state(&csi->subdev);
> +
> +	for_each_active_route(&state->routing, r) {
> +		if (!(r->flags & V4L2_SUBDEV_ROUTE_FL_ACTIVE))
> +			continue;
> +		if (r->source_pad != pad->index)
> +			continue;
> +
> +		ctx->route = r;
> +	}
> +
> +	v4l2_subdev_unlock_state(state);

I'm still learning the structure for this driver, but the pattern here
does not look right. If you unlock the state, the state can change. You
can't store the route and expect it to stay the same after unlock.
Here you are even storing a pointer to the state, which can be dangerous.

Our current v4l2 state locking is not perfect and we need to do some
manual locking/unlocking in the drivers for now, and, indeed, we just
have to expect that it doesn't change. But that has to happen inside the
enable/disable path, with such locking that it's "as safe as possible",
which means locking early, as much as possible and keeping the lock
during the processing. In many cases the framework deals with that (and
that's our future target generally).

In the subdev enable path, .enable_streams already has the state locked
and the driver doesn't have to deal with it. If using s_stream (but why
would you?) you need to lock it yourself, and it usually makes sense to
lock almost as the first thing, and pass the state forward (or use
v4l2_subdev_get_locked_active_state()).

Here all this processing is happening from ti_csi2rx_start_streaming(),
so not even from the subdev's ops. So I'm not sure yet how to clean this
up, but, as I said, it just doesn't look right.

 Tomi

> +
> +	if (!ctx->route)
> +		return -ENODEV;
> +
> +	return 0;
> +}
> +
> +static int ti_csi2rx_get_vc(struct ti_csi2rx_ctx *ctx)
> +{
> +	struct ti_csi2rx_dev *csi = ctx->csi;
> +	struct ti_csi2rx_ctx *curr_ctx;
> +	struct v4l2_mbus_frame_desc fd;
> +	struct media_pad *source_pad;
> +	struct v4l2_subdev_route *curr_route;
> +	int ret;
> +	unsigned int i, j;
> +
> +	/* Get the frame desc form source */
> +	source_pad = media_entity_remote_pad_unique(&csi->subdev.entity, MEDIA_PAD_FL_SOURCE);
> +	if (!source_pad)
> +		return -ENODEV;
> +
> +	ret = v4l2_subdev_call(csi->source, pad, get_frame_desc, source_pad->index, &fd);
> +	if (ret)
> +		return ret;
> +
> +	if (fd.type != V4L2_MBUS_FRAME_DESC_TYPE_CSI2)
> +		return -EINVAL;
> +
> +	for (i = 0; i < csi->num_ctx; i++) {
> +		curr_ctx = &csi->ctx[i];
> +
> +		/* Capture VC 0 by default */
> +		curr_ctx->vc = 0;
> +
> +		ret = ti_csi2rx_get_route(curr_ctx);
> +		if (ret)
> +			continue;
> +
> +		curr_route = curr_ctx->route;
> +		curr_ctx->stream = curr_route->sink_stream;
> +
> +		for (j = 0; j < fd.num_entries; j++)
> +			if (curr_ctx->stream == fd.entry[j].stream)
> +				curr_ctx->vc = fd.entry[j].bus.csi2.vc;
> +	}
> +
> +	return 0;
> +}
> +
>  static int ti_csi2rx_start_streaming(struct vb2_queue *vq, unsigned int count)
>  {
>  	struct ti_csi2rx_ctx *ctx = vb2_get_drv_priv(vq);
> @@ -904,6 +986,25 @@ static int ti_csi2rx_start_streaming(struct vb2_queue *vq, unsigned int count)
>  	if (ret)
>  		goto err;
>  
> +	/* If no stream is routed to this ctx, exit early */
> +	ret = ti_csi2rx_get_route(ctx);
> +	if (ret)
> +		goto err;
> +
> +	/* Get the VC for all enabled ctx on first stream start */
> +	mutex_lock(&csi->mutex);
> +	if (!csi->vc_cached) {
> +		ret = ti_csi2rx_get_vc(ctx);
> +		if (ret == -ENOIOCTLCMD) {
> +			ctx->vc = 0;
> +		} else if (ret < 0) {
> +			mutex_unlock(&csi->mutex);
> +			goto err;
> +		}
> +		csi->vc_cached = true;
> +	}
> +	mutex_unlock(&csi->mutex);
> +
>  	ti_csi2rx_setup_shim(ctx);
>  
>  	ctx->sequence = 0;
> @@ -950,6 +1051,10 @@ static void ti_csi2rx_stop_streaming(struct vb2_queue *vq)
>  	writel(0, csi->shim + SHIM_CNTL);
>  	writel(0, csi->shim + SHIM_DMACNTX(ctx->idx));
>  
> +	mutex_lock(&csi->mutex);
> +	csi->vc_cached = false;
> +	mutex_unlock(&csi->mutex);
> +
>  	ret = v4l2_subdev_call(&csi->subdev, video, s_stream, 0);
>  	if (ret)
>  		dev_err(csi->dev, "Failed to stop subdev stream\n");
> @@ -1303,6 +1408,8 @@ static int ti_csi2rx_init_ctx(struct ti_csi2rx_ctx *ctx)
>  
>  	ti_csi2rx_fill_fmt(fmt, &ctx->v_fmt);
>  
> +	ctx->route = NULL;
> +
>  	ctx->pad.flags = MEDIA_PAD_FL_SINK;
>  	vdev->entity.ops = &ti_csi2rx_video_entity_ops;
>  	ret = media_entity_pads_init(&ctx->vdev.entity, 1, &ctx->pad);