Re: [PATCH v4 21/28] net: wan: Add framer framework support

[Herve Codina <herve.codina@bootlin.com>]

Hi Christophe,

On Mon, 21 Aug 2023 08:02:10 +0200
Christophe JAILLET <christophe.jaillet@wanadoo.fr> wrote:

> Le 18/08/2023 à 18:39, Christophe Leroy a écrit :
> > From: Herve Codina <herve.codina@bootlin.com>
> > 
> > A framer is a component in charge of an E1/T1 line interface.
> > Connected usually to a TDM bus, it converts TDM frames to/from E1/T1
> > frames. It also provides information related to the E1/T1 line.
> > 
> > The framer framework provides a set of APIs for the framer drivers
> > (framer provider) to create/destroy a framer and APIs for the framer
> > users (framer consumer) to obtain a reference to the framer, and
> > use the framer.
> > 
> > This basic implementation provides a framer abstraction for:
> >   - power on/off the framer
> >   - get the framer status (line state)
> >   - be notified on framer status changes
> >   - get/set the framer configuration
> > 
> > Signed-off-by: Herve Codina <herve.codina@bootlin.com>
> > Reviewed-by: Christophe Leroy <christophe.leroy@csgroup.eu>
> > Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
> > ---  
> 
> Hi,
> 
> should there be a V5, some nits below.
> 
> ...
> 
> > +int framer_power_off(struct framer *framer)
> > +{
> > +	int ret;
> > +
> > +	mutex_lock(&framer->mutex);
> > +	if (framer->power_count == 1 && framer->ops->power_off) {
> > +		ret =  framer->ops->power_off(framer);  
> 
>                       ~~
> Useless extra space

Will be remove in the next iteration.

> 
> > +		if (ret < 0) {
> > +			dev_err(&framer->dev, "framer poweroff failed --> %d\n", ret);
> > +			mutex_unlock(&framer->mutex);
> > +			return ret;
> > +		}
> > +	}
> > +	--framer->power_count;
> > +	mutex_unlock(&framer->mutex);
> > +	framer_pm_runtime_put(framer);
> > +
> > +	if (framer->pwr)
> > +		regulator_disable(framer->pwr);
> > +
> > +	return 0;
> > +}  
> 
> ...
> 
> > +struct framer *framer_create(struct device *dev, struct device_node *node,
> > +			     const struct framer_ops *ops)
> > +{
> > +	int ret;
> > +	int id;
> > +	struct framer *framer;
> > +
> > +	if (WARN_ON(!dev))
> > +		return ERR_PTR(-EINVAL);
> > +
> > +	/* get_status() is mandatory if the provider ask for polling status */
> > +	if (WARN_ON((ops->flags & FRAMER_FLAG_POLL_STATUS) && !ops->get_status))
> > +		return ERR_PTR(-EINVAL);
> > +
> > +	framer = kzalloc(sizeof(*framer), GFP_KERNEL);
> > +	if (!framer)
> > +		return ERR_PTR(-ENOMEM);
> > +
> > +	id = ida_simple_get(&framer_ida, 0, 0, GFP_KERNEL);  
> 
> ida_alloc()?
> (ida_simple_get() is deprecated)

Indeed, ida_alloc() and ida_free() will be used in the next iteration.

> 
> > +	if (id < 0) {
> > +		dev_err(dev, "unable to get id\n");
> > +		ret = id;
> > +		goto free_framer;
> > +	}
> > +
> > +	device_initialize(&framer->dev);
> > +	mutex_init(&framer->mutex);
> > +	INIT_WORK(&framer->notify_status_work, framer_notify_status_work);
> > +	INIT_DELAYED_WORK(&framer->polling_work, framer_polling_work);
> > +	BLOCKING_INIT_NOTIFIER_HEAD(&framer->notifier_list);
> > +
> > +	framer->dev.class = framer_class;
> > +	framer->dev.parent = dev;
> > +	framer->dev.of_node = node ? node : dev->of_node;
> > +	framer->id = id;
> > +	framer->ops = ops;
> > +
> > +	ret = dev_set_name(&framer->dev, "framer-%s.%d", dev_name(dev), id);
> > +	if (ret)
> > +		goto put_dev;
> > +
> > +	/* framer-supply */
> > +	framer->pwr = regulator_get_optional(&framer->dev, "framer");
> > +	if (IS_ERR(framer->pwr)) {
> > +		ret = PTR_ERR(framer->pwr);
> > +		if (ret == -EPROBE_DEFER)
> > +			goto put_dev;
> > +
> > +		framer->pwr = NULL;
> > +	}
> > +
> > +	ret = device_add(&framer->dev);
> > +	if (ret)
> > +		goto put_dev;
> > +
> > +	if (pm_runtime_enabled(dev)) {
> > +		pm_runtime_enable(&framer->dev);
> > +		pm_runtime_no_callbacks(&framer->dev);
> > +	}
> > +
> > +	return framer;
> > +
> > +put_dev:
> > +	put_device(&framer->dev);  /* calls framer_release() which frees resources */
> > +	return ERR_PTR(ret);
> > +
> > +free_framer:
> > +	kfree(framer);
> > +	return ERR_PTR(ret);
> > +}  
> 
> ...
> 
> > +void framer_provider_of_unregister(struct framer_provider *framer_provider)
> > +{
> > +	mutex_lock(&framer_provider_mutex);
> > +	list_del(&framer_provider->list);
> > +	of_node_put(framer_provider->dev->of_node);
> > +	kfree(framer_provider);
> > +	mutex_unlock(&framer_provider_mutex);  
> 
> If it make sense, of_node_put() and kfree() could maybe be out of the 
> mutex, in order to match how things are done in 
> __framer_provider_of_register().

Yes, it makes sense.
Both of_node_put() and kfree() will be moved out of the mutex.

> 
> > +}  
> 
> ...
> 
> > +static void framer_release(struct device *dev)
> > +{
> > +	struct framer *framer;
> > +
> > +	framer = dev_to_framer(dev);
> > +	regulator_put(framer->pwr);
> > +	ida_simple_remove(&framer_ida, framer->id);  
> 
> ida_free()?
> (ida_simple_remove() is deprecated)

Yes

> 
> > +	kfree(framer);
> > +}  
> 
> ...
> 

Thanks for the review,
Hervé

-- 
Hervé Codina, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com