diff mbox series

[v2,10/16] rpmsg: char: allow only one endpoint per device

Message ID 20201222105726.16906-11-arnaud.pouliquen@foss.st.com (mailing list archive)
State New, archived
Headers show
Series introduce generic IOCTL interface for RPMsg channels management | expand

Commit Message

Arnaud Pouliquen Dec. 22, 2020, 10:57 a.m. UTC
Only one endpoint can be created per device, prevent from multi open.

Signed-off-by: Arnaud Pouliquen <arnaud.pouliquen@foss.st.com>
---
 drivers/rpmsg/rpmsg_char.c | 3 +++
 1 file changed, 3 insertions(+)

Comments

Bjorn Andersson Jan. 5, 2021, 12:59 a.m. UTC | #1
On Tue 22 Dec 04:57 CST 2020, Arnaud Pouliquen wrote:

> Only one endpoint can be created per device, prevent from multi open.
> 

Having multiple invocations of rpmsg_create_ept() with the same chinfo
sounds like a bad idea. I think in the SMD and GLINK case the underlying
transport would complain that the related chinfo is already "busy", but
this seems like an appropriate fix regardless.

Please add a proper Fixes: tag and send this outside of this patch
series.

Thanks,
Bjorn

> Signed-off-by: Arnaud Pouliquen <arnaud.pouliquen@foss.st.com>
> ---
>  drivers/rpmsg/rpmsg_char.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/drivers/rpmsg/rpmsg_char.c b/drivers/rpmsg/rpmsg_char.c
> index 66e01b979e72..4b0674a2e3e9 100644
> --- a/drivers/rpmsg/rpmsg_char.c
> +++ b/drivers/rpmsg/rpmsg_char.c
> @@ -122,6 +122,9 @@ static int rpmsg_eptdev_open(struct inode *inode, struct file *filp)
>  	struct rpmsg_device *rpdev = eptdev->rpdev;
>  	struct device *dev = &eptdev->dev;
>  
> +	if (eptdev->ept)
> +		return -EBUSY;
> +
>  	get_device(dev);
>  
>  	ept = rpmsg_create_ept(rpdev, rpmsg_ept_cb, eptdev, eptdev->chinfo);
> -- 
> 2.17.1
>
Arnaud Pouliquen Jan. 5, 2021, 5:05 p.m. UTC | #2
On 1/5/21 1:59 AM, Bjorn Andersson wrote:
> On Tue 22 Dec 04:57 CST 2020, Arnaud Pouliquen wrote:
> 
>> Only one endpoint can be created per device, prevent from multi open.
>>
> 
> Having multiple invocations of rpmsg_create_ept() with the same chinfo
> sounds like a bad idea. I think in the SMD and GLINK case the underlying
> transport would complain that the related chinfo is already "busy", but
> this seems like an appropriate fix regardless.
> 
> Please add a proper Fixes: tag and send this outside of this patch
> series.

I will send it in a separate patch.

Regards,
Arnaud

> 
> Thanks,
> Bjorn
> 
>> Signed-off-by: Arnaud Pouliquen <arnaud.pouliquen@foss.st.com>
>> ---
>>  drivers/rpmsg/rpmsg_char.c | 3 +++
>>  1 file changed, 3 insertions(+)
>>
>> diff --git a/drivers/rpmsg/rpmsg_char.c b/drivers/rpmsg/rpmsg_char.c
>> index 66e01b979e72..4b0674a2e3e9 100644
>> --- a/drivers/rpmsg/rpmsg_char.c
>> +++ b/drivers/rpmsg/rpmsg_char.c
>> @@ -122,6 +122,9 @@ static int rpmsg_eptdev_open(struct inode *inode, struct file *filp)
>>  	struct rpmsg_device *rpdev = eptdev->rpdev;
>>  	struct device *dev = &eptdev->dev;
>>  
>> +	if (eptdev->ept)
>> +		return -EBUSY;
>> +
>>  	get_device(dev);
>>  
>>  	ept = rpmsg_create_ept(rpdev, rpmsg_ept_cb, eptdev, eptdev->chinfo);
>> -- 
>> 2.17.1
>>
diff mbox series

Patch

diff --git a/drivers/rpmsg/rpmsg_char.c b/drivers/rpmsg/rpmsg_char.c
index 66e01b979e72..4b0674a2e3e9 100644
--- a/drivers/rpmsg/rpmsg_char.c
+++ b/drivers/rpmsg/rpmsg_char.c
@@ -122,6 +122,9 @@  static int rpmsg_eptdev_open(struct inode *inode, struct file *filp)
 	struct rpmsg_device *rpdev = eptdev->rpdev;
 	struct device *dev = &eptdev->dev;
 
+	if (eptdev->ept)
+		return -EBUSY;
+
 	get_device(dev);
 
 	ept = rpmsg_create_ept(rpdev, rpmsg_ept_cb, eptdev, eptdev->chinfo);