Messages in this thread | | | Date | Tue, 20 Oct 2020 14:36:09 +0200 | From | Pavel Machek <> | Subject | Re: [PATCH v9 3/3] leds: trigger: implement a tty trigger |
| |
Hi!
> Usage is as follows: > > myled=ledname > tty=ttyS0 > > echo tty > /sys/class/leds/$myled/trigger > echo $tty > /sys/class/leds/$myled/ttyname > > . When this new trigger is active it periodically checks the tty's > statistics and when it changed since the last check the led is flashed > once. > > Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> > ---
> +static ssize_t ttyname_show(struct device *dev, > + struct device_attribute *attr, char *buf) > +{ > + struct ledtrig_tty_data *trigger_data = led_trigger_get_drvdata(dev); > + ssize_t len = 0;
Unused value. Not sure if some checker will complain.
> +static ssize_t ttyname_store(struct device *dev, > + struct device_attribute *attr, const char *buf, > + size_t size) > +{ > + struct ledtrig_tty_data *trigger_data = led_trigger_get_drvdata(dev); > + char *ttyname; > + ssize_t ret = size;
> + ledtrig_tty_halt(trigger_data); > + > + mutex_lock(&trigger_data->mutex); > + > + if (size > 0 && buf[size - 1] == '\n') > + size -= 1; > + > + if (size) { > + ttyname = kmemdup_nul(buf, size, GFP_KERNEL); > + if (!ttyname) { > + ret = -ENOMEM; > + goto out_unlock;
If this happens, you return error to the user and trigger_data->ttyname remains with the old value, but trigger is now stopped. That is not exactly consistent.
> +static void ledtrig_tty_work(struct work_struct *work) > +{ > + struct ledtrig_tty_data *trigger_data = > + container_of(work, struct ledtrig_tty_data, dwork.work); > + struct serial_icounter_struct icount; > + int ret; > + > + mutex_lock(&trigger_data->mutex); > + > + BUG_ON(!trigger_data->ttyname); > + > + /* try to get the tty corresponding to $ttyname */ > + if (!trigger_data->tty) { > + dev_t devno; > + struct tty_struct *tty; > + int ret; > + > + ret = tty_dev_name_to_number(trigger_data->ttyname, &devno); > + if (ret < 0) > + /* > + * A device with this name might appear later, so keep > + * retrying. > + */ > + goto out;
Poll every 100 msec... Hmm.... Okay, I guess?
> + ret = tty_get_icount(trigger_data->tty, &icount); > + if (ret) { > + mutex_unlock(&trigger_data->mutex); > + dev_info(trigger_data->tty->dev, "Failed to get icount, stopped polling\n"); > + mutex_unlock(&trigger_data->mutex);
Eh?
> + return; > + } > + > + if (icount.rx != trigger_data->rx || > + icount.tx != trigger_data->tx) { > + led_set_brightness(trigger_data->led_cdev, LED_ON);
Are you sure about LED_ON here? It should use current brightness selected by brightness file...
> +static int ledtrig_tty_activate(struct led_classdev *led_cdev) > +{ > + struct ledtrig_tty_data *trigger_data; > + > + trigger_data = kzalloc(sizeof(*trigger_data), GFP_KERNEL); > + if (!trigger_data) > + return -ENOMEM; > + > + led_set_trigger_data(led_cdev, trigger_data); > + > + INIT_DELAYED_WORK(&trigger_data->dwork, ledtrig_tty_work); > + trigger_data->led_cdev = led_cdev; > + mutex_init(&trigger_data->mutex);
How is this protected from concurrent access from sysfs?
Best regards, Pavel
-- http://www.livejournal.com/~pavelmachek [unhandled content-type:application/pgp-signature] | |