Re: [PATCH 2/2] usb: typec: tipd: Support wakeup

[Jun Nie <jun.nie@linaro.org>]

Bryan O'Donoghue <bryan.odonoghue@linaro.org> 于2023年1月5日周四 19:05写道：
>
> On 05/01/2023 07:50, Jun Nie wrote:
> > Enable wakeup when pluging or unpluging USB cable. It is up to other
> > components to hold system in active mode, such as display, so that
> > user can receive the notification.
> >
> > Signed-off-by: Jun Nie <jun.nie@linaro.org>
> > ---
> >   drivers/usb/typec/tipd/core.c | 38 +++++++++++++++++++++++++++++++++++
> >   1 file changed, 38 insertions(+)
> >
> > diff --git a/drivers/usb/typec/tipd/core.c b/drivers/usb/typec/tipd/core.c
> > index 46a4d8b128f0..485b90c13078 100644
> > --- a/drivers/usb/typec/tipd/core.c
> > +++ b/drivers/usb/typec/tipd/core.c
> > @@ -95,6 +95,7 @@ struct tps6598x {
> >       struct power_supply_desc psy_desc;
> >       enum power_supply_usb_type usb_type;
> >
> > +     int wakeup;
> >       u16 pwr_status;
> >   };
> >
> > @@ -846,6 +847,12 @@ static int tps6598x_probe(struct i2c_client *client)
> >       i2c_set_clientdata(client, tps);
> >       fwnode_handle_put(fwnode);
> >
> > +     tps->wakeup = device_property_read_bool(tps->dev, "wakeup-source");
> > +     if (tps->wakeup) {
> > +             device_init_wakeup(&client->dev, true);
> > +             enable_irq_wake(client->irq);
> > +     }
>
> Does the ordering of device_init_wakeup() and enable_irq_wake() matter ?
>
> The sequence in drivers/usb/typec/tcpm/tcpci_maxim.c is
> enable_irq_wake() and then device_init_wakeup() ?

With reading related code, I believe it is better to put device_init_wakeup()
before enable_irq_wake() logically. Though it shall not matter in real world.

device_init_wakeup() register the wakeup source and setup sysfs etc, which
is puerly software side infrastructure. enable_irq_wake() setup interrupt
configuration, including software and hardware sides. I assume there is no
suspend/resume process involves wakeup event before probe function finish
in real world. If there is, device_init_wakeup() should come before before
enable_irq_wake() strictly.

- Jun

>
> Reviewed-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>