2020-12-28 12:19:07

by Ezequiel Garcia

Subject: [PATCH] media: v4l2-async: Put fwnode after last access

fwnode_handle_put() should be called after the fwnode
is last accessed. Fix it.

Fixes: b98158d837ef ("media: v4l2-async: Accept endpoints and devices for fwnode matching")
Signed-off-by: Ezequiel Garcia <[email protected]>
---
drivers/media/v4l2-core/v4l2-async.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/media/v4l2-core/v4l2-async.c b/drivers/media/v4l2-core/v4l2-async.c
index e3ab003a6c85..1303c9b83138 100644
--- a/drivers/media/v4l2-core/v4l2-async.c
+++ b/drivers/media/v4l2-core/v4l2-async.c
@@ -78,6 +78,7 @@ static bool match_fwnode(struct v4l2_async_notifier *notifier,
bool asd_fwnode_is_ep;
bool sd_fwnode_is_ep;
struct device *dev;
+ bool match;

/*
* Both the subdev and the async subdev can provide either an endpoint
@@ -113,9 +114,10 @@ static bool match_fwnode(struct v4l2_async_notifier *notifier,
other_fwnode = sd->fwnode;
}

- fwnode_handle_put(dev_fwnode);
+ match = (dev_fwnode == other_fwnode);

- if (dev_fwnode != other_fwnode)
+ fwnode_handle_put(dev_fwnode);
+ if (!match)
return false;

/*
--
2.29.2

2020-12-28 13:20:00

by Laurent Pinchart

Subject: Re: [PATCH] media: v4l2-async: Put fwnode after last access

Hi Ezequiel,

Thank you for the patch.

On Mon, Dec 28, 2020 at 09:17:25AM -0300, Ezequiel Garcia wrote:
> fwnode_handle_put() should be called after the fwnode
> is last accessed. Fix it.
>
> Fixes: b98158d837ef ("media: v4l2-async: Accept endpoints and devices for fwnode matching")
> Signed-off-by: Ezequiel Garcia <[email protected]>
> ---
> drivers/media/v4l2-core/v4l2-async.c | 6 ++++--
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/media/v4l2-core/v4l2-async.c b/drivers/media/v4l2-core/v4l2-async.c
> index e3ab003a6c85..1303c9b83138 100644
> --- a/drivers/media/v4l2-core/v4l2-async.c
> +++ b/drivers/media/v4l2-core/v4l2-async.c
> @@ -78,6 +78,7 @@ static bool match_fwnode(struct v4l2_async_notifier *notifier,
> bool asd_fwnode_is_ep;
> bool sd_fwnode_is_ep;
> struct device *dev;
> + bool match;
>
> /*
> * Both the subdev and the async subdev can provide either an endpoint
> @@ -113,9 +114,10 @@ static bool match_fwnode(struct v4l2_async_notifier *notifier,
> other_fwnode = sd->fwnode;
> }
>
> - fwnode_handle_put(dev_fwnode);
> + match = (dev_fwnode == other_fwnode);
>
> - if (dev_fwnode != other_fwnode)

This only performs a pointer comparison, it doesn't access dev_fwnode. I
don't think the change is necessary.

> + fwnode_handle_put(dev_fwnode);
> + if (!match)
> return false;
>
> /*

--
Regards,

Laurent Pinchart

2020-12-28 16:58:02

by Ezequiel Garcia

Subject: Re: [PATCH] media: v4l2-async: Put fwnode after last access

On Mon, 2020-12-28 at 15:16 +0200, Laurent Pinchart wrote:
> Hi Ezequiel,
>
> Thank you for the patch.
>
> On Mon, Dec 28, 2020 at 09:17:25AM -0300, Ezequiel Garcia wrote:
> > fwnode_handle_put() should be called after the fwnode
> > is last accessed. Fix it.
> >
> > Fixes: b98158d837ef ("media: v4l2-async: Accept endpoints and devices for fwnode matching")
> > Signed-off-by: Ezequiel Garcia <[email protected]>
> > ---
> >  drivers/media/v4l2-core/v4l2-async.c | 6 ++++--
> >  1 file changed, 4 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/media/v4l2-core/v4l2-async.c b/drivers/media/v4l2-core/v4l2-async.c
> > index e3ab003a6c85..1303c9b83138 100644
> > --- a/drivers/media/v4l2-core/v4l2-async.c
> > +++ b/drivers/media/v4l2-core/v4l2-async.c
> > @@ -78,6 +78,7 @@ static bool match_fwnode(struct v4l2_async_notifier *notifier,
> >         bool asd_fwnode_is_ep;
> >         bool sd_fwnode_is_ep;
> >         struct device *dev;
> > +       bool match;
> >  
> >         /*
> >          * Both the subdev and the async subdev can provide either an endpoint
> > @@ -113,9 +114,10 @@ static bool match_fwnode(struct v4l2_async_notifier *notifier,
> >                 other_fwnode = sd->fwnode;
> >         }
> >  
> > -       fwnode_handle_put(dev_fwnode);
> > +       match = (dev_fwnode == other_fwnode);
> >  
> > -       if (dev_fwnode != other_fwnode)
>
> This only performs a pointer comparison, it doesn't access dev_fwnode. I
> don't think the change is necessary.
>

Ah, of course. I'm always confused by this.

Thanks,
Ezequiel