Re: [PATCH v2] serial: 8250: add gpio support to exar

[Sudip Mukherjee <sudipm.mukherjee@gmail.com>]

On Sat, Dec 19, 2015 at 04:43:20PM +0000, One Thousand Gnomes wrote:
> On Sat, 19 Dec 2015 20:53:57 +0530
> Sudip Mukherjee <sudipm.mukherjee@gmail.com> wrote:

Hi Alan,

> 
> > Exar XR17V352/354/358 chips have 16 multi-purpose inputs/outputs which
> > can be controlled using gpio interface.
> > Add support to use these pins and select GPIO_SYSFS also so that these
> > pins can be used from the userspace through sysfs.
> 
> I don't actually see why this is a separate module the way you've done
> it. As you've got runtime requirements for it in the serial driver based
> on the Kconfig entries it'll either always be needed or never, so having
> it as a separate module just wastes memory, better just to build the
> extra file into the driver in this case.

Just a bit confused here. Are you asking to make it bool instead of
tristate in the Kconfig

config SERIAL_8250_EXAR_GPIO
	bool "Support for GPIO pins on XR17V352/354/358"

or are you asking to include all these gpio code as part of 8250_pci.c ?
I thought module will be better as it will not be needed by anyone who
is not using Exar chip, and even those who are using, some of them may
not need it. This is optional for only those who wants to use the gpio
capability of that chip.

> 
> >
> > --- a/drivers/tty/serial/8250/8250_pci.c
> > +++ b/drivers/tty/serial/8250/8250_pci.c
> > @@ -1764,12 +1764,29 @@ xr17v35x_has_slave(struct serial_private *priv)
> >  	        (dev_id == PCI_DEVICE_ID_EXAR_XR17V8358));
> >  }
> >  
> > +extern void xr17v35x_gpio_exit(struct uart_8250_port *);
> > +extern int xr17v35x_gpio_init(struct pci_dev *, struct uart_8250_port *);
> > +
> > +static void pci_xr17v35x_exit(struct pci_dev *dev)
> > +{
> > +#if defined(CONFIG_SERIAL_8250_EXAR_GPIO)
> > +	struct serial_private *priv = pci_get_drvdata(dev);
> > +	struct uart_8250_port *port = NULL;
> > +
> > +	if (priv)
> > +		port = serial8250_get_port(priv->line[0]);
> > +
> > +	xr17v35x_gpio_exit(port);
> 
> This looks odd - how will priv be NULL if the init succeeded ? And if you
> can tolerate NULL isn't the whole thing
> 
> 	xr17v35x_gpio_exit(serial8250_get_port(priv->line[0]));
> 
> which conventiently means you can make xr17v35x_gpio_exit a null function
> in the header and keep the ifdefs there

True, It can not be NULL here. In my v1 I kept the ifdefs in the header
file but if I try with CONFIG_SERIAL_8250_EXAR_GPIO=m then
#ifdef CONFIG_SERIAL_8250_EXAR_GPIO is becoming false. But if I keep it
in the c file it is working.

> 
> > +#endif
> > +}
> > +
> >  static int
> >  pci_xr17v35x_setup(struct serial_private *priv,
> >  		  const struct pciserial_board *board,
> >  		  struct uart_8250_port *port, int idx)
> >  {
> >  	u8 __iomem *p;
> > +	int ret;
> >  
> >  	p = pci_ioremap_bar(priv->dev, 0);
> >  	if (p == NULL)
> > @@ -1807,7 +1824,16 @@ pci_xr17v35x_setup(struct serial_private *priv,
> >  	writeb(128, p + UART_EXAR_RXTRG);
> >  	iounmap(p);
> >  
> > -	return pci_default_setup(priv, board, port, idx);
> > +	ret = pci_default_setup(priv, board, port, idx);
> > +	if (ret)
> > +		return ret;
> > +
> > +#if defined(CONFIG_SERIAL_8250_EXAR_GPIO)
> > +	if (idx == 0 && priv->dev->vendor == PCI_VENDOR_ID_EXAR)
> > +		ret = xr17v35x_gpio_init(priv->dev, port);
> > +#endif
> > +
> > +	return ret;
> >  }
> 
> I would suggest
> 
> 1. Put all the checks on idx == 0 and the like into the gpio driver code
> so it makes less mess here. It's gpio knowledge not serial knowledge.
> 
> 2. Can you not just remove the VENDOR_ID_EXAR check ?

pci_xr17v35x_setup() is also used by PCI_VENDOR_ID_COMMTECH. I donot
know if they also have the same gpio capability or not. So gpio code
should only be used by VENDOR_ID_EXAR.

> 
> 3. If you tidy the code up so the gpio call is
> 
> 	ret = pci_default_setup(blah)
> 	if (ret == 0)
> 		ret = xr17v35x_gpio_init(priv->dev, port);
> 	return ret;

If i have to pass idx to the gpio code then it should be:

	ret = xr17v35x_gpio_init(priv->dev, port, idx);

> 
> Then you can just ifdef the gpio_init/exit methods to return 0 in the
> header and avoid ifdef mess anywhere else
> 
> 
> The gpio driver itself I'd suggest fixing
> 
> +	exar_gpio = devm_kzalloc(&dev->dev, sizeof(*exar_gpio), GFP_KERNEL);
> +	if (!exar_gpio) {
> +		ret = -ENOMEM;
> +		goto err_unmap;
> +	}
> +
> +	/* assuming we will never have more than 99 boards */
> +	buf = devm_kzalloc(&dev->dev, strlen("exar_gpio") + 3, GFP_KERNEL);
> +	if (!buf) {
> +		ret = -ENOMEM;
> +		goto err_unmap;
> +	}
> 
> 
> It's not worth making buf a separate allocation. Just declare a 16 byte
> buffer or whatever at the end of the struct and stop worrying about the
> number of boards. Not only will it actually be smaller than all the code
> to handle the case, it'll be more reliable.
> 
> The other thing that looks a bit wrong is exar_get(). You've got it
> returning negative error codes on a NULL. Firstly as far as I can see the
> NULL case can't happen, secondly if it does you take negative values from
> it and treat them as valid then use them, which is nonsensical.

Yes, that is a mess. Return type is unsigned and I am returning negative
error code. :(
I will fix it. But I will first wait for your reply on my mentioned
confusion first.

regards
sudip