Message-ID: <20210119160033.GB611676@xps15>
Date:   Tue, 19 Jan 2021 09:00:33 -0700
From:   Mathieu Poirier <mathieu.poirier@...aro.org>
To:     Arnaud POULIQUEN <arnaud.pouliquen@...com>
Cc:     Arnaud POULIQUEN - foss <arnaud.pouliquen@...s.st.com>,
        Ohad Ben-Cohen <ohad@...ery.com>,
        "linux-stm32@...md-mailman.stormreply.com" 
        <linux-stm32@...md-mailman.stormreply.com>,
        "linux-remoteproc@...r.kernel.org" <linux-remoteproc@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        Bjorn Andersson <bjorn.andersson@...aro.org>
Subject: Re: [Linux-stm32] [PATCH] rpmsg: char: return an error if device
 already open

On Fri, Jan 15, 2021 at 10:13:35AM +0100, Arnaud POULIQUEN wrote:
> Hi Mathieu,
> 
> 
> On 1/14/21 8:05 PM, Mathieu Poirier wrote:
> > On Wed, Jan 06, 2021 at 02:37:14PM +0100, Arnaud Pouliquen wrote:
> >> The rpmsg_create_ept function is invoked when the device is opened.
> >> As only one endpoint must be created per device. It is not
> >> possible to open the same device twice.
> >> The fix consists in returning -EBUSY when device is already
> >> opened.
> >>
> >> Fixes: c0cdc19f84a4 ("rpmsg: Driver for user space endpoint interface")
> >> Signed-off-by: Arnaud Pouliquen <arnaud.pouliquen@...s.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 4bbbacdbf3bb..360a1ab0a9c4 100644
> >> --- a/drivers/rpmsg/rpmsg_char.c
> >> +++ b/drivers/rpmsg/rpmsg_char.c
> >> @@ -127,6 +127,9 @@ 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;
> >> +
> > 
> > I rarely had to work so hard to review a 2 line patch...
> 
> That means that my commit description was not enough explicit...
> 
> > 
> > As far as I can tell the actual code is doing the right thing.  If user space is
> > trying to open the same eptdev more than once function rpmsg_create_ept() should
> > complain and the operation denied, wich is what the current code is doing.  
> > 
> > There is currently two customers for this API - SMD and GLINK.  The SMD code is
> > quite clear that if the channel is already open, the operation will be
> > denied [1].  The GLINK code isn't as clear but the fact that it returns NULL on
> > error conditions [2] is a good indication that things are working the same way.
> > 
> > What kind of use case are you looking to address?  Is there any way you can use
> > rpdev->ops->create_ept() as it is currently done?
> 
> This patch was part of the IOCTL rpmsg series. I sent it separately at Bjorn's
> request [1].
>

I am looking at [1] later today - I will get back to this patch when I have more
context.
 
> I detect the issue using the RPMSG_ADDR_ANY for the source address when tested
> it with the rpmsf_virtio bus. In this case at each sys open of the device, a new
> endpoint is created because a new source address is allocated.
> 
> [1]https://patchwork.kernel.org/project/linux-remoteproc/patch/20201222105726.16906-11-arnaud.pouliquen@foss.st.com/
> 
> Thanks,
> Arnaud
> 
> > 
> > Thanks,
> > Mathieu
> > 
> > [1]. https://elixir.bootlin.com/linux/v5.11-rc3/source/drivers/rpmsg/qcom_smd.c#L920
> > [2]. https://elixir.bootlin.com/linux/v5.11-rc3/source/drivers/rpmsg/qcom_glink_native.c#L1149
> > 
> >>  	get_device(dev);
> >>  
> >>  	ept = rpmsg_create_ept(rpdev, rpmsg_ept_cb, eptdev, eptdev->chinfo);
> >> -- 
> >> 2.17.1
> >>
> > _______________________________________________
> > Linux-stm32 mailing list
> > Linux-stm32@...md-mailman.stormreply.com
> > https://st-md-mailman.stormreply.com/mailman/listinfo/linux-stm32
> > 

Powered by blists - more mailing lists