[libcamera-devel] [PATCH] libcamera: V4L2Device: Remove the controls order assumption in updateControls()

[Laurent Pinchart]

Hi Hiro,

Thank you for the patch.

On Wed, May 26, 2021 at 03:43:11PM +0900, Hirokazu Honda wrote:
> The original updateControls() has the assumption that ctrls and
> v4l2Ctrls lists are in the same order. It is dependent on the
> caller implementation though. This changes updateControls()
> implementation so that it works without the assumption.
> 
> Signed-off-by: Hirokazu Honda <hiroh at chromium.org>
> ---
>  src/libcamera/v4l2_device.cpp | 25 ++++++++-----------------
>  1 file changed, 8 insertions(+), 17 deletions(-)
> 
> diff --git a/src/libcamera/v4l2_device.cpp b/src/libcamera/v4l2_device.cpp
> index caafbc2d..aaca7171 100644
> --- a/src/libcamera/v4l2_device.cpp
> +++ b/src/libcamera/v4l2_device.cpp
> @@ -244,12 +244,6 @@ ControlList V4L2Device::getControls(const std::vector<uint32_t> &ids)
>  
>  	ControlList ctrls{ controls_ };
>  
> -	/*
> -	 * Start by filling the ControlList. This can't be combined with filling
> -	 * v4l2Ctrls, as updateControls() relies on both containers having the
> -	 * same order, and the control list is based on a map, which is not
> -	 * sorted by insertion order.
> -	 */
>  	for (uint32_t id : ids) {
>  		const auto iter = controls_.find(id);
>  		if (iter == controls_.end()) {
> @@ -623,19 +617,16 @@ void V4L2Device::updateControlInfo()
>  void V4L2Device::updateControls(ControlList *ctrls,
>  				Span<const v4l2_ext_control> v4l2Ctrls)
>  {
> -	unsigned int i = 0;
> -	for (auto &ctrl : *ctrls) {
> -		if (i == v4l2Ctrls.size())
> -			break;
> +	for (const v4l2_ext_control &v4l2Ctrl : v4l2Ctrls) {
> +		const unsigned int id = v4l2Ctrl.id;
>  
> -		const struct v4l2_ext_control *v4l2Ctrl = &v4l2Ctrls[i];
> -		unsigned int id = ctrl.first;
> -		ControlValue &value = ctrl.second;
> +		ControlValue value = ctrls->get(id);
>  
> -		const auto iter = controls_.find(id);
> +		const auto &iter = controls_.find(id);

find() returns an iterator, not a reference to an iterator. Storing it
in a reference here will work as the lifetime of rvalues is extended
when a const lvalue reference is bound to them, but it's misleading, and
doesn't save any memory or CPU time. I think you should thus keep the
original code.

> +		ASSERT(iter != controls_.end());

I'd add a blank line here.

With these changes,

Reviewed-by: Laurent Pinchart <laurent.pinchart at ideasonboard.com>

>  		switch (iter->first->type()) {
>  		case ControlTypeInteger64:
> -			value.set<int64_t>(v4l2Ctrl->value64);
> +			value.set<int64_t>(v4l2Ctrl.value64);
>  			break;
>  
>  		case ControlTypeByte:
> @@ -650,11 +641,11 @@ void V4L2Device::updateControls(ControlList *ctrls,
>  			 * \todo To be changed when support for string controls
>  			 * will be added.
>  			 */
> -			value.set<int32_t>(v4l2Ctrl->value);
> +			value.set<int32_t>(v4l2Ctrl.value);
>  			break;
>  		}
>  
> -		i++;
> +		ctrls->set(id, value);
>  	}
>  }
>  

-- 
Regards,

Laurent Pinchart