2023-03-24 17:32:21

by Wei Chen

Subject: [PATCH] media: hi846: Fix memleak in hi846_init_controls()

hi846_init_controls doesn't clean the allocated ctrl_hdlr
in case there is a failure, which causes memleak. Add
v4l2_ctrl_handler_free to free the resource properly.

Signed-off-by: Wei Chen <[email protected]>
---
drivers/media/i2c/hi846.c | 1 +
1 file changed, 1 insertion(+)

diff --git a/drivers/media/i2c/hi846.c b/drivers/media/i2c/hi846.c
index 7c61873b7198..c45a6511d2c1 100644
--- a/drivers/media/i2c/hi846.c
+++ b/drivers/media/i2c/hi846.c
@@ -1472,6 +1472,7 @@ static int hi846_init_controls(struct hi846 *hi846)
if (ctrl_hdlr->error) {
dev_err(&client->dev, "v4l ctrl handler error: %d\n",
ctrl_hdlr->error);
+ v4l2_ctrl_handler_free(ctrl_hdlr);
return ctrl_hdlr->error;
}

--
2.25.1

2023-03-26 08:56:58

by Martin Kepplinger

Subject: Re: [PATCH] media: hi846: Fix memleak in hi846_init_controls()

Am Freitag, dem 24.03.2023 um 17:24 +0000 schrieb Wei Chen:
> hi846_init_controls doesn't clean the allocated ctrl_hdlr
> in case there is a failure, which causes memleak. Add
> v4l2_ctrl_handler_free to free the resource properly.
>
> Signed-off-by: Wei Chen <[email protected]>
> ---
>  drivers/media/i2c/hi846.c | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/drivers/media/i2c/hi846.c b/drivers/media/i2c/hi846.c
> index 7c61873b7198..c45a6511d2c1 100644
> --- a/drivers/media/i2c/hi846.c
> +++ b/drivers/media/i2c/hi846.c
> @@ -1472,6 +1472,7 @@ static int hi846_init_controls(struct hi846
> *hi846)
>         if (ctrl_hdlr->error) {
>                 dev_err(&client->dev, "v4l ctrl handler error: %d\n",
>                         ctrl_hdlr->error);
> +               v4l2_ctrl_handler_free(ctrl_hdlr);
>                 return ctrl_hdlr->error;
>         }
>  

hi Wei, thanks for the patch. It looks like I forgot that indeed, but
to me it looks like the subsequent error paths of
v4l2_fwnode_device_parse() and v4l2_ctrl_new_fwnode_properties() would
leak the same thing.

And since we only assign it to the subdev's ctrl_handler in the success
case later, you could simply add an error label:


@@ -1663,21 +1663,26 @@ static int hi846_init_controls(struct hi846
*hi846)
if (ctrl_hdlr->error) {
dev_err(&client->dev, "v4l ctrl handler error: %d\n",
ctrl_hdlr->error);
- return ctrl_hdlr->error;
+ ret = ctrl_hdlr->error;
+ goto error;
}

ret = v4l2_fwnode_device_parse(&client->dev, &props);
if (ret)
- return ret;
+ goto error;

ret = v4l2_ctrl_new_fwnode_properties(ctrl_hdlr,
&hi846_ctrl_ops,
&props);
if (ret)
- return ret;
+ goto error;

hi846->sd.ctrl_handler = ctrl_hdlr;

return 0;
+
+error:
+ v4l2_ctrl_handler_free(ctrl_hdlr);
+ return ret;
}


what do you think? thanks!

martin

2023-03-26 09:22:06

by Wei Chen

Subject: Re: [PATCH] media: hi846: Fix memleak in hi846_init_controls()

Dear Martin,

Thanks for pointing out my mistake.

I cross-checked other implementations of drivers under media/i2c. I
agree with you. Let me send the second version of my patch.

Thanks,
Wei

On Sun, 26 Mar 2023 at 16:50, Martin Kepplinger <[email protected]> wrote:
>
> Am Freitag, dem 24.03.2023 um 17:24 +0000 schrieb Wei Chen:
> > hi846_init_controls doesn't clean the allocated ctrl_hdlr
> > in case there is a failure, which causes memleak. Add
> > v4l2_ctrl_handler_free to free the resource properly.
> >
> > Signed-off-by: Wei Chen <[email protected]>
> > ---
> > drivers/media/i2c/hi846.c | 1 +
> > 1 file changed, 1 insertion(+)
> >
> > diff --git a/drivers/media/i2c/hi846.c b/drivers/media/i2c/hi846.c
> > index 7c61873b7198..c45a6511d2c1 100644
> > --- a/drivers/media/i2c/hi846.c
> > +++ b/drivers/media/i2c/hi846.c
> > @@ -1472,6 +1472,7 @@ static int hi846_init_controls(struct hi846
> > *hi846)
> > if (ctrl_hdlr->error) {
> > dev_err(&client->dev, "v4l ctrl handler error: %d\n",
> > ctrl_hdlr->error);
> > + v4l2_ctrl_handler_free(ctrl_hdlr);
> > return ctrl_hdlr->error;
> > }
> >
>
> hi Wei, thanks for the patch. It looks like I forgot that indeed, but
> to me it looks like the subsequent error paths of
> v4l2_fwnode_device_parse() and v4l2_ctrl_new_fwnode_properties() would
> leak the same thing.
>
> And since we only assign it to the subdev's ctrl_handler in the success
> case later, you could simply add an error label:
>
>
> @@ -1663,21 +1663,26 @@ static int hi846_init_controls(struct hi846
> *hi846)
> if (ctrl_hdlr->error) {
> dev_err(&client->dev, "v4l ctrl handler error: %d\n",
> ctrl_hdlr->error);
> - return ctrl_hdlr->error;
> + ret = ctrl_hdlr->error;
> + goto error;
> }
>
> ret = v4l2_fwnode_device_parse(&client->dev, &props);
> if (ret)
> - return ret;
> + goto error;
>
> ret = v4l2_ctrl_new_fwnode_properties(ctrl_hdlr,
> &hi846_ctrl_ops,
> &props);
> if (ret)
> - return ret;
> + goto error;
>
> hi846->sd.ctrl_handler = ctrl_hdlr;
>
> return 0;
> +
> +error:
> + v4l2_ctrl_handler_free(ctrl_hdlr);
> + return ret;
> }
>
>
> what do you think? thanks!
>
> martin
>
>