Re: [Patch v2 6/6] tty: serial: 8250: Add support for MOXA PCIe boards to switch interface between RS422/RS485

[Jiri Slaby <jirislaby@kernel.org>]

On 16. 10. 23, 5:37, Crescent CY Hsieh wrote:
> MOXA PCIe boards have 4 serial interfaces and don't require additional
> stuff to switch between interfaces:
> 
> - RS232
> - RS422
> - RS485_2W (half-duplex)
> - RS485_4W (full-duplex)
> 
> By using ioctl command "TIOCRS485", it can switch between default
> interface and RS485 if supported.
> 
> That means, for RS422/RS485 board, it can switch between RS422 and
> RS485 by setting the flags within struct serial_rs485.
> 
> However, for the RS232/RS422/RS485 board, it can only switch between
> RS232 and RS485, there's no flag for switching interface into RS422.
> 
> This patch uses "SER_RS485_TERMINATE_BUS" to represent RS422 as a
> workaround solution.
> 
> Signed-off-by: Crescent CY Hsieh <crescentcy.hsieh@moxa.com>
> ---
>   drivers/tty/serial/8250/8250_pci.c | 52 ++++++++++++++++++++++++++++++
>   1 file changed, 52 insertions(+)
> 
> diff --git a/drivers/tty/serial/8250/8250_pci.c b/drivers/tty/serial/8250/8250_pci.c
> index 72dd27141801..e2e8a28242bc 100644
> --- a/drivers/tty/serial/8250/8250_pci.c
> +++ b/drivers/tty/serial/8250/8250_pci.c
> @@ -1976,6 +1976,10 @@ pci_sunix_setup(struct serial_private *priv,
>   #define MOXA_RS485_2W	0x0F
>   #define MOXA_UIR_OFFSET	0x04
>   
> +static const struct serial_rs485 pci_moxa_rs485_supported = {
> +	.flags = SER_RS485_ENABLED | SER_RS485_RX_DURING_TX | SER_RS485_TERMINATE_BUS,
> +};
> +
>   static bool pci_moxa_is_mini_pcie(unsigned short device)
>   {
>   	if (device == PCI_DEVICE_ID_MOXA_CP102N	||
> @@ -1989,6 +1993,11 @@ static bool pci_moxa_is_mini_pcie(unsigned short device)
>   	return false;
>   }
>   
> +static bool pci_moxa_match_x1xx(unsigned short device)
> +{
> +	return (device & 0x0F00) == 0x0100;
> +}
> +
>   static bool pci_moxa_match_x3xx(unsigned short device)

I wonder if it made more sense to have only a single function like:

static bool pci_moxa_match(unsigned short device, unsigned short matching)
{
	return (device & 0x0F00) == matching;
}

and call:
pci_moxa_match(device, 0x0100)
pci_moxa_match(device, 0x0300)
?

Perhaps name the function better. According to what the second digit in 
the device ID in fact means.

>   {
>   	return (device & 0x0F00) == 0x0300;
> @@ -2016,6 +2025,36 @@ static int pci_moxa_set_interface(const struct pci_dev *dev,
>   	return 0;
>   }
>   
> +static int pci_moxa_rs485_config(struct uart_port *port,
> +				 struct ktermios *termios,
> +				 struct serial_rs485 *rs485)
> +{
> +	struct pci_dev *dev = to_pci_dev(port->dev);
> +	unsigned short device = dev->device;
> +	u8 mode = MOXA_RS232;
> +
> +	if (rs485->flags & SER_RS485_ENABLED) {
> +		/* Use SER_RS485_TERMINATE_BUS to represent RS422 as a workaround. */
> +		if (rs485->flags & SER_RS485_TERMINATE_BUS) {
> +			mode = MOXA_RS422;
> +		} else {
> +			if (rs485->flags & SER_RS485_RX_DURING_TX)
> +				mode = MOXA_RS485_4W;
> +			else
> +				mode = MOXA_RS485_2W;
> +		}
> +	} else {
> +		/*
> +		 * RS232 is not supported for those device IDs of
> +		 * MOXA PCIe boards match the pattern 0x*3**.

I am not native, but I can hardly parse this.

Maybe:
RS232 is not supported on MOXA PCIe boards with device IDs matching the 
pattern 0x*3**.
?

> +		 */
> +		if (pci_moxa_match_x3xx(device))
> +			return -EINVAL;

Perhaps EOPNOTSUPP?

> +	}
> +
> +	return pci_moxa_set_interface(dev, port->port_id, mode);
> +}
> +
>   static int pci_moxa_init(struct pci_dev *dev)
>   {
>   	unsigned short device = dev->device;

thanks,
-- 
js
suse labs