Re: [PATCH] Input: Added check for grabbed device

[Dmitry Torokhov <dmitry.torokhov@gmail.com>]

Hi Anshul,

On Sun, Jan 04, 2015 at 12:25:55AM -0800, Anshul Garg wrote:
> From: Anshul Garg <aksgarg1989@gmail.com>
> 
> If input device is grabbed then client which has grabbed the device can
> flush or write to the device so for other clients -EINVAL should be returned.
> 
> Signed-off-by: Anshul Garg <anshul.g@samsung.com>
> ---
>  drivers/input/evdev.c |    7 +++++++
>  1 file changed, 7 insertions(+)
> 
> diff --git a/drivers/input/evdev.c b/drivers/input/evdev.c
> index b1a52ab..105e489 100644
> --- a/drivers/input/evdev.c
> +++ b/drivers/input/evdev.c
> @@ -277,6 +277,8 @@ static int evdev_flush(struct file *file, fl_owner_t id)
>  
>  	if (!evdev->exist || client->revoked)
>  		retval = -ENODEV;
> +	else if (evdev->grab && evdev->grab != client)
> +		retval = -EINVAL;
>  	else
>  		retval = input_flush_device(&evdev->handle, file);
>  
> @@ -475,6 +477,11 @@ static ssize_t evdev_write(struct file *file, const char __user *buffer,
>  		goto out;
>  	}
>  
> +
> +	if (evdev->grab && evdev->grab != client) {
> +		retval = -EINVAL;
> +		goto out;
> +	}

Grab is supposed to be transparent to the other users, so returning
error here is not desirable. Also, input_inject_event() will already
ignore events if device is grabbed, so everything is good as far as
evdev_write() concerned.

Flush is a bit trickier: the driver should only "flush" data that
belongs to the user that issued flush request (and the owner is
identified by file handle passed to flush() method). I think it is OK to
allow owner to flush its own data even while device is grabbed.

Thanks.

-- 
Dmitry