Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Fri, 8 Oct 2021 16:42:53 -0700
From:   Bjorn Andersson <bjorn.andersson@linaro.org>
To:     Arnaud Pouliquen <arnaud.pouliquen@foss.st.com>
Cc:     Ohad Ben-Cohen <ohad@wizery.com>,
        Mathieu Poirier <mathieu.poirier@linaro.org>,
        linux-remoteproc@vger.kernel.org, linux-kernel@vger.kernel.org,
        linux-stm32@st-md-mailman.stormreply.com, julien.massot@iot.bzh
Subject: Re: [PATCH v4 3/4] rpmsg: char: Add possibility to use default
 endpoint of the rpmsg device.
Message-ID: <YWDXfeR79u5QQvk3@ripper>
References: <20210712131900.24752-1-arnaud.pouliquen@foss.st.com>
 <20210712131900.24752-4-arnaud.pouliquen@foss.st.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20210712131900.24752-4-arnaud.pouliquen@foss.st.com>
Precedence: bulk

On Mon 12 Jul 06:18 PDT 2021, Arnaud Pouliquen wrote:

> Current implementation create/destroy a new endpoint on each
> rpmsg_eptdev_open/rpmsg_eptdev_release calls.
> 
> For a rpmsg device created by the NS announcement mechanism we need to
> use a unique static endpoint that is the default rpmsg device endpoint
> associated to the channel.
> 

Why do you need this endpoint associated with the channel? Afaict the
read/write operations still operate on eptdev->ept, so who does use the
default endpoint for the device?

> This patch prepares the introduction of a rpmsg channel device for the
> char device. The rpmsg channel device will require a default endpoint to
> communicate to the remote processor.
> 
> Add the static_ept field in rpmsg_eptdev structure. This boolean
> determines the behavior on rpmsg_eptdev_open and rpmsg_eptdev_release call.
> 
> - If static_ept == false:
>   Use the legacy behavior by creating a new endpoint each time
>   rpmsg_eptdev_open is called and release it when rpmsg_eptdev_release
>   is called on /dev/rpmsgX device open/close.
> 
> - If static_ept == true:
>   use the rpmsg device default endpoint for the communication.
> - Address the update of _rpmsg_chrdev_eptdev_create in e separate patch for readability.
> 
> Add protection in rpmsg_eptdev_ioctl to prevent to destroy a default endpoint.
> 
> Signed-off-by: Arnaud Pouliquen <arnaud.pouliquen@foss.st.com>
> Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org>
> Tested-by: Julien Massot <julien.massot@iot.bzh>
> ---
>  drivers/rpmsg/rpmsg_char.c | 21 +++++++++++++++++++--
>  1 file changed, 19 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/rpmsg/rpmsg_char.c b/drivers/rpmsg/rpmsg_char.c
> index 50b7d4b00175..bd728d90ba4c 100644
> --- a/drivers/rpmsg/rpmsg_char.c
> +++ b/drivers/rpmsg/rpmsg_char.c
> @@ -45,6 +45,8 @@ static DEFINE_IDA(rpmsg_minor_ida);
>   * @queue_lock:	synchronization of @queue operations
>   * @queue:	incoming message queue
>   * @readq:	wait object for incoming queue
> + * @static_ept: specify if the endpoint has to be created at each device opening or
> + *              if the default endpoint should be used.
>   */
>  struct rpmsg_eptdev {
>  	struct device dev;
> @@ -59,6 +61,8 @@ struct rpmsg_eptdev {
>  	spinlock_t queue_lock;
>  	struct sk_buff_head queue;
>  	wait_queue_head_t readq;
> +
> +	bool static_ept;

I think you can skip rpmsg_create_default_ept() if you just make this
struct rpmsg_endpoint *.

>  };
>  
>  int rpmsg_chrdev_eptdev_destroy(struct device *dev, void *data)
> @@ -116,7 +120,15 @@ static int rpmsg_eptdev_open(struct inode *inode, struct file *filp)
>  
>  	get_device(dev);
>  
> -	ept = rpmsg_create_ept(rpdev, rpmsg_ept_cb, eptdev, eptdev->chinfo);
> +	/*
> +	 * If the static_ept is set to true, the rpmsg device default endpoint is used.
> +	 * Else a new endpoint is created on open that will be destroyed on release.
> +	 */
> +	if (eptdev->static_ept)
> +		ept = rpdev->ept;

This would be:
	if (eptdev->static_ept)
		ept = eptdev->static_ept;

> +	else
> +		ept = rpmsg_create_ept(rpdev, rpmsg_ept_cb, eptdev, eptdev->chinfo);
> +
>  	if (!ept) {
>  		dev_err(dev, "failed to open %s\n", eptdev->chinfo.name);
>  		put_device(dev);
> @@ -137,7 +149,8 @@ static int rpmsg_eptdev_release(struct inode *inode, struct file *filp)
>  	/* Close the endpoint, if it's not already destroyed by the parent */
>  	mutex_lock(&eptdev->ept_lock);
>  	if (eptdev->ept) {
> -		rpmsg_destroy_ept(eptdev->ept);
> +		if (!eptdev->static_ept)
> +			rpmsg_destroy_ept(eptdev->ept);
>  		eptdev->ept = NULL;
>  	}
>  	mutex_unlock(&eptdev->ept_lock);
> @@ -264,6 +277,10 @@ static long rpmsg_eptdev_ioctl(struct file *fp, unsigned int cmd,
>  	if (cmd != RPMSG_DESTROY_EPT_IOCTL)
>  		return -EINVAL;
>  
> +	/* Don't allow to destroy a default endpoint. */
> +	if (eptdev->ept == eptdev->rpdev->ept)

And this would be if:
	if (eptdev->static_ept)
		return -EPERM;

Wouldn't -EINVAL or something be better than -EPERM when you try to
destroy one of these endpoints?

It's not that we don't have permission, it's that it's not a valid
operation on this object.

Regards,
Bjorn

> +		return -EPERM;
> +
>  	return rpmsg_chrdev_eptdev_destroy(&eptdev->dev, NULL);
>  }
>  
> -- 
> 2.17.1
>