media: i2c: rdacm2: Remove an incorrect fwnode_handle_put() call

  The commit in Fixes has removed an fwnode_handle_put() call in the error
handling path of the probe.

Remove the same call from the remove function.

Fixes: 1029939b3782 ("media: v4l: async: Simplify async sub-device fwnode matching")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
---
/!\   This patch is highly speculative. Review with care.   /!\

If it is correct, it is likely that other similar issue lurk in commit
1029939b3782. I've not looked in detail and my cocci script did not
trigger on them but drivers/media/i2c/max9286.c also looks like a
similar candidate.
---
 drivers/media/i2c/rdacm21.c | 1 -
 1 file changed, 1 deletion(-)
  

Hi Christophe

On Sat, Aug 26, 2023 at 10:05:06AM +0200, Christophe JAILLET wrote:
> The commit in Fixes has removed an fwnode_handle_put() call in the error
> handling path of the probe.
>
> Remove the same call from the remove function.
>
> Fixes: 1029939b3782 ("media: v4l: async: Simplify async sub-device fwnode matching")
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>

Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>

> ---
> /!\   This patch is highly speculative. Review with care.   /!\
>
> If it is correct, it is likely that other similar issue lurk in commit
> 1029939b3782. I've not looked in detail and my cocci script did not
> trigger on them but drivers/media/i2c/max9286.c also looks like a
> similar candidate.

I think the call to  fwnode_handle_put(priv->sd.fwnode) in
max9286_v4l2_unregister() can indeed be removed, yes!

Thanks
  j

> ---
>  drivers/media/i2c/rdacm21.c | 1 -
>  1 file changed, 1 deletion(-)
>
> diff --git a/drivers/media/i2c/rdacm21.c b/drivers/media/i2c/rdacm21.c
> index a36a709243fd..3e22df36354f 100644
> --- a/drivers/media/i2c/rdacm21.c
> +++ b/drivers/media/i2c/rdacm21.c
> @@ -608,7 +608,6 @@ static void rdacm21_remove(struct i2c_client *client)
>  	v4l2_async_unregister_subdev(&dev->sd);
>  	v4l2_ctrl_handler_free(&dev->ctrls);
>  	i2c_unregister_device(dev->isp);
> -	fwnode_handle_put(dev->sd.fwnode);
>  }
>
>  static const struct of_device_id rdacm21_of_ids[] = {
> --
> 2.34.1
>
  

On Mon, Aug 28, 2023 at 09:48:10AM +0200, Jacopo Mondi wrote:
> Hi Christophe
> 
> On Sat, Aug 26, 2023 at 10:05:06AM +0200, Christophe JAILLET wrote:
> > The commit in Fixes has removed an fwnode_handle_put() call in the error
> > handling path of the probe.
> >
> > Remove the same call from the remove function.
> >
> > Fixes: 1029939b3782 ("media: v4l: async: Simplify async sub-device fwnode matching")
> > Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> 
> Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>

The subject line should read "rdacm21", not "rdacm2". with that fixed,

Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>

> > ---
> > /!\   This patch is highly speculative. Review with care.   /!\
> >
> > If it is correct, it is likely that other similar issue lurk in commit
> > 1029939b3782. I've not looked in detail and my cocci script did not
> > trigger on them but drivers/media/i2c/max9286.c also looks like a
> > similar candidate.
> 
> I think the call to  fwnode_handle_put(priv->sd.fwnode) in
> max9286_v4l2_unregister() can indeed be removed, yes!

I agree.

> > ---
> >  drivers/media/i2c/rdacm21.c | 1 -
> >  1 file changed, 1 deletion(-)
> >
> > diff --git a/drivers/media/i2c/rdacm21.c b/drivers/media/i2c/rdacm21.c
> > index a36a709243fd..3e22df36354f 100644
> > --- a/drivers/media/i2c/rdacm21.c
> > +++ b/drivers/media/i2c/rdacm21.c
> > @@ -608,7 +608,6 @@ static void rdacm21_remove(struct i2c_client *client)
> >  	v4l2_async_unregister_subdev(&dev->sd);
> >  	v4l2_ctrl_handler_free(&dev->ctrls);
> >  	i2c_unregister_device(dev->isp);
> > -	fwnode_handle_put(dev->sd.fwnode);
> >  }
> >
> >  static const struct of_device_id rdacm21_of_ids[] = {
  

On Mon, Aug 28, 2023 at 01:57:24PM +0300, Laurent Pinchart wrote:
> On Mon, Aug 28, 2023 at 09:48:10AM +0200, Jacopo Mondi wrote:
> > Hi Christophe
> > 
> > On Sat, Aug 26, 2023 at 10:05:06AM +0200, Christophe JAILLET wrote:
> > > The commit in Fixes has removed an fwnode_handle_put() call in the error
> > > handling path of the probe.
> > >
> > > Remove the same call from the remove function.
> > >
> > > Fixes: 1029939b3782 ("media: v4l: async: Simplify async sub-device fwnode matching")
> > > Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
> > 
> > Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
> 
> The subject line should read "rdacm21", not "rdacm2". with that fixed,
> 
> Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
> 
> > > ---
> > > /!\   This patch is highly speculative. Review with care.   /!\
> > >
> > > If it is correct, it is likely that other similar issue lurk in commit
> > > 1029939b3782. I've not looked in detail and my cocci script did not
> > > trigger on them but drivers/media/i2c/max9286.c also looks like a
> > > similar candidate.
> > 
> > I think the call to  fwnode_handle_put(priv->sd.fwnode) in
> > max9286_v4l2_unregister() can indeed be removed, yes!
> 
> I agree.

drivers/media/platform/nxp/imx-mipi-csis.c also needs a fix.

Christophe, do you plan to send patches for those ? If not, I can handle
it.

> > > ---
> > >  drivers/media/i2c/rdacm21.c | 1 -
> > >  1 file changed, 1 deletion(-)
> > >
> > > diff --git a/drivers/media/i2c/rdacm21.c b/drivers/media/i2c/rdacm21.c
> > > index a36a709243fd..3e22df36354f 100644
> > > --- a/drivers/media/i2c/rdacm21.c
> > > +++ b/drivers/media/i2c/rdacm21.c
> > > @@ -608,7 +608,6 @@ static void rdacm21_remove(struct i2c_client *client)
> > >  	v4l2_async_unregister_subdev(&dev->sd);
> > >  	v4l2_ctrl_handler_free(&dev->ctrls);
> > >  	i2c_unregister_device(dev->isp);
> > > -	fwnode_handle_put(dev->sd.fwnode);
> > >  }
> > >
> > >  static const struct of_device_id rdacm21_of_ids[] = {
  

Le 28/08/2023 à 13:16, Laurent Pinchart a écrit :
> On Mon, Aug 28, 2023 at 01:57:24PM +0300, Laurent Pinchart wrote:
>> On Mon, Aug 28, 2023 at 09:48:10AM +0200, Jacopo Mondi wrote:
>>> Hi Christophe
>>>
>>> On Sat, Aug 26, 2023 at 10:05:06AM +0200, Christophe JAILLET wrote:
>>>> The commit in Fixes has removed an fwnode_handle_put() call in the error
>>>> handling path of the probe.
>>>>
>>>> Remove the same call from the remove function.
>>>>
>>>> Fixes: 1029939b3782 ("media: v4l: async: Simplify async sub-device fwnode matching")
>>>> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
>>>
>>> Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
>>
>> The subject line should read "rdacm21", not "rdacm2". with that fixed,
>>
>> Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
>>
>>>> ---
>>>> /!\   This patch is highly speculative. Review with care.   /!\
>>>>
>>>> If it is correct, it is likely that other similar issue lurk in commit
>>>> 1029939b3782. I've not looked in detail and my cocci script did not
>>>> trigger on them but drivers/media/i2c/max9286.c also looks like a
>>>> similar candidate.
>>>
>>> I think the call to  fwnode_handle_put(priv->sd.fwnode) in
>>> max9286_v4l2_unregister() can indeed be removed, yes!
>>
>> I agree.
> 
> drivers/media/platform/nxp/imx-mipi-csis.c also needs a fix.
> 
> Christophe, do you plan to send patches for those ? If not, I can handle
> it.

I'll propose patches for both in the coming days.

CJ