Re: [PATCH v8 05/14] media: rkisp1: add Rockchip ISP1 subdev driver

[Dafna Hirschfeld]

Hi,
I (Dafna Hirschfeld) will work in following months with Helen Koike to fix the issues
in the TODO file of this driver: drivers/staging/media/rkisp1/TODO

On 15.08.19 15:17, Sakari Ailus wrote:
> On Thu, Aug 15, 2019 at 11:24:22AM +0300, Sakari Ailus wrote:
> > Hi Helen,
> >
> > On Wed, Aug 14, 2019 at 09:58:05PM -0300, Helen Koike wrote:
> >
> > ...
> >
> > > > > +static int rkisp1_isp_sd_set_fmt(struct v4l2_subdev *sd,
> > > > > +				 struct v4l2_subdev_pad_config *cfg,
> > > > > +				 struct v4l2_subdev_format *fmt)
> > > > > +{
> > > > > +	struct rkisp1_device *isp_dev = sd_to_isp_dev(sd);
> > > > > +	struct rkisp1_isp_subdev *isp_sd = &isp_dev->isp_sdev;
> > > > > +	struct v4l2_mbus_framefmt *mf = &fmt->format;
> > > > > +
> > > >
> > > > Note that for sub-device nodes, the driver is itself responsible for
> > > > serialising the access to its data structures.
> > >
> > > But looking at subdev_do_ioctl_lock(), it seems that it serializes the
> > > ioctl calls for subdevs, no? Or I'm misunderstanding something (which is
> > > most probably) ?
> >
> > Good question. I had missed this change --- subdev_do_ioctl_lock() is
> > relatively new. But setting that lock is still not possible as the struct

'the struct' - do you mean the 'vdev' struct allocated in
'v4l2_device_register_subdev_nodes' ?

> > is allocated in the framework and the device is registered before the

> > driver gets hold of it. It's a good idea to provide the same serialisation
> > for subdevs as well.
> >
> > I'll get back to this later.
>
> The main reason is actually that these ops are also called through the
> sub-device kAPI, not only through the uAPI, and the locks are only taken
> through the calls via uAPI.

actually it seems that although 'subdev_do_ioctl_lock' exit, I wonder if
any subdevice uses that vdev->lock in  subdev_do_ioctl_lock.
It is not initialized in v4l2_device_register_subdev_nodes where the vdev is allocated
and I wonder if any subdevice actually initialize it somewhere else. For example it is null in this
driver and in vimc.

> So adding the locks to uAPI calls alone would not address the issue.

What I can do is add a mutex to every struct of a subdevice and lock it
at the beginning of each subdevice operation.
Is this an acceptable solution?

Thanks,
Dafna
>