Re: [PATCH RESEND net-next v7 05/16] net: pcs: add Renesas MII converter driver

["Clément Léger" <clement.leger@bootlin.com>]

Le Tue, 14 Jun 2022 22:31:09 -0700,
Jakub Kicinski <kuba@kernel.org> a écrit :

> On Fri, 10 Jun 2022 12:37:01 +0200 Clément Léger wrote:
> > Subject: [PATCH RESEND net-next v7 05/16] net: pcs: add Renesas MII converter driver
> > 
> > Add a PCS driver for the MII converter that is present on the Renesas
> > RZ/N1 SoC. This MII converter is reponsible for converting MII to
> > RMII/RGMII or act as a MII pass-trough. Exposing it as a PCS allows to
> > reuse it in both the switch driver and the stmmac driver. Currently,
> > this driver only allows the PCS to be used by the dual Cortex-A7
> > subsystem since the register locking system is not used.  
> 
> Could someone with MII &| PCS knowledge cast an eye over this code?
> All I can do is point out error path issues...
> 
> > +struct phylink_pcs *miic_create(struct device *dev, struct device_node *np)
> > +{
> > +	struct platform_device *pdev;
> > +	struct miic_port *miic_port;
> > +	struct device_node *pcs_np;
> > +	struct miic *miic;
> > +	u32 port;
> > +
> > +	if (!of_device_is_available(np))
> > +		return ERR_PTR(-ENODEV);
> > +
> > +	if (of_property_read_u32(np, "reg", &port))
> > +		return ERR_PTR(-EINVAL);
> > +
> > +	if (port > MIIC_MAX_NR_PORTS || port < 1)
> > +		return ERR_PTR(-EINVAL);
> > +
> > +	/* The PCS pdev is attached to the parent node */
> > +	pcs_np = of_get_parent(np);  
> 
> of_get_parent()? ..
> 
> > +	if (!pcs_np)
> > +		return ERR_PTR(-ENODEV);
> > +
> > +	if (!of_device_is_available(pcs_np))
> > +		return ERR_PTR(-ENODEV);  
> 
> .. more like of_leak_parent()

Indeed, I'll fix that.

> 
> > +	pdev = of_find_device_by_node(pcs_np);
> > +	if (!pdev || !platform_get_drvdata(pdev))
> > +		return ERR_PTR(-EPROBE_DEFER);
> > +
> > +	miic_port = kzalloc(sizeof(*miic_port), GFP_KERNEL);
> > +	if (!miic_port)
> > +		return ERR_PTR(-ENOMEM);
> > +
> > +	miic = platform_get_drvdata(pdev);
> > +	device_link_add(dev, miic->dev, DL_FLAG_AUTOREMOVE_CONSUMER);
> > +
> > +	miic_port->miic = miic;
> > +	miic_port->port = port - 1;
> > +	miic_port->pcs.ops = &miic_phylink_ops;
> > +
> > +	return &miic_port->pcs;
> > +}
> > +EXPORT_SYMBOL(miic_create);  
> 
> > +static int miic_parse_dt(struct device *dev, u32 *mode_cfg)
> > +{
> > +	s8 dt_val[MIIC_MODCTRL_CONF_CONV_NUM];
> > +	struct device_node *np = dev->of_node;
> > +	struct device_node *conv;
> > +	u32 conf;
> > +	int port;
> > +
> > +	memset(dt_val, MIIC_MODCTRL_CONF_NONE, sizeof(dt_val));
> > +
> > +	of_property_read_u32(np, "renesas,miic-switch-portin", &conf);
> > +	dt_val[0] = conf;
> > +
> > +	for_each_child_of_node(np, conv) {
> > +		if (of_property_read_u32(conv, "reg", &port))
> > +			continue;
> > +
> > +		if (!of_device_is_available(conv))
> > +			continue;
> > +
> > +		if (of_property_read_u32(conv, "renesas,miic-input", &conf) == 0)
> > +			dt_val[port] = conf;
> > +
> > +		of_node_put(conv);  
> 
> Don't these iteration functions put() the current before taking the
> next one all by themselves? Or is there supposed to be a "break" here?

Yes, of_node_put() should actually only be called in case of early exit.
I'll fix that.


-- 
Clément Léger,
Embedded Linux and Kernel engineer at Bootlin
https://bootlin.com