RE: [PATCH] tty: serdev: serdev-ttyport: add devt for ctrl->dev

[Sherry Sun]

> -----Original Message-----
> From: Greg KH <gregkh@xxxxxxxxxxxxxxxxxxx>
> Sent: 2023年3月15日 19:10
> To: Sherry Sun <sherry.sun@xxxxxxx>
> Cc: jirislaby@xxxxxxxxxx; robh@xxxxxxxxxx; linux-serial@xxxxxxxxxxxxxxx;
> linux-kernel@xxxxxxxxxxxxxxx; dl-linux-imx <linux-imx@xxxxxxx>
> Subject: Re: [PATCH] tty: serdev: serdev-ttyport: add devt for ctrl->dev
> 
> On Wed, Mar 15, 2023 at 06:54:00PM +0800, Sherry Sun wrote:
> > For serdev framework, the serdev_controller device is the tty device,
> > which is also the child device of the uart_port device. If we don't
> > set devt property for ctrl->dev, device_find_child(uport->dev, ...)
> > may always return NULL in uart_suspend_port() function, which prevents
> > us from properly handling uart port suspend, so fix it here.
> >
> > Fixes: bed35c6dfa6a ("serdev: add a tty port controller driver")
> > Signed-off-by: Sherry Sun <sherry.sun@xxxxxxx>
> > ---
> >  drivers/tty/serdev/serdev-ttyport.c | 2 ++
> >  1 file changed, 2 insertions(+)
> >
> > diff --git a/drivers/tty/serdev/serdev-ttyport.c
> > b/drivers/tty/serdev/serdev-ttyport.c
> > index bba37ab90215..c58af8141380 100644
> > --- a/drivers/tty/serdev/serdev-ttyport.c
> > +++ b/drivers/tty/serdev/serdev-ttyport.c
> > @@ -268,6 +268,7 @@ struct device *serdev_tty_port_register(struct
> > tty_port *port,  {
> >  	struct serdev_controller *ctrl;
> >  	struct serport *serport;
> > +	dev_t devt = MKDEV(drv->major, drv->minor_start) + idx;
> >  	int ret;
> >
> >  	if (!port || !drv || !parent)
> > @@ -282,6 +283,7 @@ struct device *serdev_tty_port_register(struct
> tty_port *port,
> >  	serport->tty_idx = idx;
> >  	serport->tty_drv = drv;
> >
> > +	ctrl->dev.devt = devt;
> 
> This feels wrong as you can't just create a magic dev_t out of no where and
> expect it to be handled properly.  Where now is this dev_t exposed?
> 
Hi Greg,
Now dev_t is the key point to get the tty->dev in uart_suspend_port()/uart_resume_port() and alloc_tty_struct(), it is set in tty_register_device_attr() but not in serdev_tty_port_register().

To be frank, I am also not sure if this is the right way to fix the issue, maybe we can have more discussion here regarding how to get the correct tty_dev in uart_suspend_port().
At least with the code logic in serdev framework we are using now, we will always get NULL tty_dev from device_find_child(). I believe this is not what we want here.

static int serial_match_port(struct device *dev, void *data)
{
    struct uart_match *match = data;
    struct tty_driver *tty_drv = match->driver->tty_driver;
    dev_t devt = MKDEV(tty_drv->major, tty_drv->minor_start) +
        match->port->line;

    return dev->devt == devt; /* Actually, only one tty per port */
}
int uart_suspend_port(struct uart_driver *drv, struct uart_port *uport)
{
    struct uart_state *state = drv->state + uport->line;
    struct tty_port *port = &state->port;
    struct device *tty_dev;
    struct uart_match match = {uport, drv};

    mutex_lock(&port->mutex);

    tty_dev = device_find_child(uport->dev, &match, serial_match_port);
    if (uport->line == 2)
    if (tty_dev && device_may_wakeup(tty_dev)) {
        enable_irq_wake(uport->irq);
        put_device(tty_dev);
        mutex_unlock(&port->mutex);
        return 0;
    }
    put_device(tty_dev);
...

Best Regards
Sherry


> Something else feels wrong here, sorry, I do not think this is correct.
> 
> thanks,
> 
> greg k-h