Re: [RFCv4 08/21] [WAR] v4l2-ctrls: do not clone non-standard controls

[Hans Verkuil <hverkuil@xs4all.nl>]

On 02/20/18 05:44, Alexandre Courbot wrote:
> Only standard controls can be successfully cloned: handler_new_ref, used
> by v4l2_ctrl_request_clone(), forcibly calls v4l2_ctrl_new_std() which
> fails to find custom controls names, and we eventually hit the condition
> that name == NULL in v4l2_ctrl_new().

Hmm, the core reason is that handler_new_ref tries to automatically create
a new control class if it didn't exist yet. Which is OK for standard control
classes but not for non-standard control classes such as is used in vivid.

I will have to think about this.

Regards,

	Hans

> 
> This prevents us from using non-standard controls with requests, but
> that is enough for testing purposes.
> 
> Signed-off-by: Alexandre Courbot <acourbot@chromium.org>
> ---
>  drivers/media/v4l2-core/v4l2-ctrls.c | 5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/drivers/media/v4l2-core/v4l2-ctrls.c b/drivers/media/v4l2-core/v4l2-ctrls.c
> index 166647817efb..7a81aa5959c3 100644
> --- a/drivers/media/v4l2-core/v4l2-ctrls.c
> +++ b/drivers/media/v4l2-core/v4l2-ctrls.c
> @@ -2772,6 +2772,11 @@ int v4l2_ctrl_request_clone(struct v4l2_ctrl_handler *hdl,
>  		if (filter && !filter(ctrl))
>  			continue;
>  		err = handler_new_ref(hdl, ctrl, &new_ref, false);
> +		if (err) {
> +			printk("%s: handler_new_ref on control %x (%s) returned %d\n", __func__, ctrl->id, ctrl->name, err);
> +			err = 0;
> +			continue;
> +		}
>  		if (err)
>  			break;
>  		if (from->is_request)
>