[PATCH 1/1] rpmsg: virtio_rpmsg_bus: Fix rpmsg_create_channel memory leak

[PATCH 1/1] rpmsg: virtio_rpmsg_bus: Fix rpmsg_create_channel memory leak

[Loic Pallardy]

In case of rpmsg_register_device() failure, vch previously
allocated must be free.

fixes: 6eed598a0491 ("rpmsg: Split off generic tail of create_channel()")

Signed-off-by: Loic Pallardy <loic.pallardy@st.com>
---
 drivers/rpmsg/virtio_rpmsg_bus.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/rpmsg/virtio_rpmsg_bus.c b/drivers/rpmsg/virtio_rpmsg_bus.c
index 664f957..10f927e 100644
--- a/drivers/rpmsg/virtio_rpmsg_bus.c
+++ b/drivers/rpmsg/virtio_rpmsg_bus.c
@@ -431,8 +431,10 @@ static struct rpmsg_device *rpmsg_create_channel(struct virtproc_info *vrp,
 	rpdev->dev.parent = &vrp->vdev->dev;
 	rpdev->dev.release = virtio_rpmsg_release_device;
 	ret = rpmsg_register_device(rpdev);
-	if (ret)
+	if (ret) {
+		kfree(vch);
 		return NULL;
+	}
 
 	return rpdev;
 }
-- 
1.9.1

Re: [PATCH 1/1] rpmsg: virtio_rpmsg_bus: Fix rpmsg_create_channel memory leak

[Suman Anna]

Hi Loic,

On 07/25/2018 04:08 PM, Loic Pallardy wrote:
> In case of rpmsg_register_device() failure, vch previously
> allocated must be free.
> 

Isn't this already handled through the virtio_rpmsg_release_device()
callback as part of the put_device() in rpmsg_register_device() failure.

regards
Suman

> fixes: 6eed598a0491 ("rpmsg: Split off generic tail of create_channel()")
> 
> Signed-off-by: Loic Pallardy <loic.pallardy@st.com>
> ---
>  drivers/rpmsg/virtio_rpmsg_bus.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/rpmsg/virtio_rpmsg_bus.c b/drivers/rpmsg/virtio_rpmsg_bus.c
> index 664f957..10f927e 100644
> --- a/drivers/rpmsg/virtio_rpmsg_bus.c
> +++ b/drivers/rpmsg/virtio_rpmsg_bus.c
> @@ -431,8 +431,10 @@ static struct rpmsg_device *rpmsg_create_channel(struct virtproc_info *vrp,
>  	rpdev->dev.parent = &vrp->vdev->dev;
>  	rpdev->dev.release = virtio_rpmsg_release_device;
>  	ret = rpmsg_register_device(rpdev);
> -	if (ret)
> +	if (ret) {
> +		kfree(vch);
>  		return NULL;
> +	}
>  
>  	return rpdev;
>  }
> 

RE: [PATCH 1/1] rpmsg: virtio_rpmsg_bus: Fix rpmsg_create_channel memory leak

[Loic PALLARDY]

Hi Suman,

> -----Original Message-----
> From: Suman Anna <s-anna@ti.com>
> Sent: Thursday, July 26, 2018 12:37 AM
> To: Loic PALLARDY <loic.pallardy@st.com>; bjorn.andersson@linaro.org;
> ohad@wizery.com
> Cc: linux-remoteproc@vger.kernel.org; linux-kernel@vger.kernel.org;
> Arnaud POULIQUEN <arnaud.pouliquen@st.com>;
> benjamin.gaignard@linaro.org
> Subject: Re: [PATCH 1/1] rpmsg: virtio_rpmsg_bus: Fix
> rpmsg_create_channel memory leak
> 
> Hi Loic,
> 
> On 07/25/2018 04:08 PM, Loic Pallardy wrote:
> > In case of rpmsg_register_device() failure, vch previously
> > allocated must be free.
> >
> 
> Isn't this already handled through the virtio_rpmsg_release_device()
> callback as part of the put_device() in rpmsg_register_device() failure.

Yes you're right, put_device() is doing the job.
Please forget this patch.

Regards,
Loic 

> 
> regards
> Suman
> 
> > fixes: 6eed598a0491 ("rpmsg: Split off generic tail of create_channel()")
> >
> > Signed-off-by: Loic Pallardy <loic.pallardy@st.com>
> > ---
> >  drivers/rpmsg/virtio_rpmsg_bus.c | 4 +++-
> >  1 file changed, 3 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/rpmsg/virtio_rpmsg_bus.c
> b/drivers/rpmsg/virtio_rpmsg_bus.c
> > index 664f957..10f927e 100644
> > --- a/drivers/rpmsg/virtio_rpmsg_bus.c
> > +++ b/drivers/rpmsg/virtio_rpmsg_bus.c
> > @@ -431,8 +431,10 @@ static struct rpmsg_device
> *rpmsg_create_channel(struct virtproc_info *vrp,
> >  	rpdev->dev.parent = &vrp->vdev->dev;
> >  	rpdev->dev.release = virtio_rpmsg_release_device;
> >  	ret = rpmsg_register_device(rpdev);
> > -	if (ret)
> > +	if (ret) {
> > +		kfree(vch);
> >  		return NULL;
> > +	}
> >
> >  	return rpdev;
> >  }
> >