Re: [PATCH v4] tty: serial: Add RS422 flag to struct serial_rs485

[Lino Sanfilippo]

Hi,

On 13.11.23 10:41, Crescent CY Hsieh wrote:
> Add "SER_RS485_MODE_RS422" flag to struct serial_rs485, so that serial
> port can switch interface into RS422 if supported by using ioctl command
> "TIOCSRS485".
>
> By treating RS422 as a mode of RS485, which means while enabling RS422
> there are two flags need to be set (SER_RS485_ENABLED and
> SER_RS485_MODE_RS422), it would make things much easier. For example
> some places that checks for "SER_RS485_ENABLED" won't need to be rewritten.
>
> There are only two things need to be noticed:
>
> - While enabling RS422, other RS485 flags should not be set.
> - RS422 doesn't need to deal with termination, so while disabling RS485
>   or enabling RS422, uart_set_rs485_termination() shall return.

Commit messages should use the imperative form.

>
> Signed-off-by: Crescent CY Hsieh <crescentcy.hsieh@xxxxxxxx>
>
> ---
> Changes from v3 to v4:
> - Include 'linux/const.h' header in '/include/uapi/linux/serial.h'
> - Replace BIT() with _BITUL() which defined in
>   '/include/uapi/linux/const.h'
>
> Changes from v2 to v3:
> - Remove "SER_RS422_ENABLED" flag from legacy flags.
> - Revise "SER_RS422_ENABLED" into "SER_RS485_MODE_RS422".
> - Remove the code which checks the conflicts between SER_RS485_ENABLED
>   and SER_RS422_ENABLED.
> - Add return check in uart_set_rs485_termination().
>
> Changes from v1 to v2:
> - Revise the logic that checks whether RS422/RS485 are enabled
>   simultaneously.
>
> v3: https://lore.kernel.org/all/20231108060719.11775-1-crescentcy.hsieh@xxxxxxxx/
> v2: https://lore.kernel.org/all/20231101064404.45711-1-crescentcy.hsieh@xxxxxxxx/
> v1: https://lore.kernel.org/all/20231030053632.5109-1-crescentcy.hsieh@xxxxxxxx/
>
> ---
>  drivers/tty/serial/serial_core.c | 11 ++++++++++-
>  include/uapi/linux/serial.h      | 19 +++++++++++--------
>  2 files changed, 21 insertions(+), 9 deletions(-)
>
> diff --git a/drivers/tty/serial/serial_core.c b/drivers/tty/serial/serial_core.c
> index 831d03361..777f091a4 100644
> --- a/drivers/tty/serial/serial_core.c
> +++ b/drivers/tty/serial/serial_core.c
> @@ -1376,6 +1376,13 @@ static void uart_sanitize_serial_rs485(struct uart_port *port, struct serial_rs4
>  		return;
>  	}
>
> +	/* Clear other RS485 flags and return if enabling RS422 */
> +	if (rs485->flags & SER_RS485_MODE_RS422) {
> +		memset(rs485, 0, sizeof(*rs485));
> +		rs485->flags |= (SER_RS485_ENABLED | SER_RS485_MODE_RS422)> +		return;
> +	}
> +
>  	/* Pick sane settings if the user hasn't */
>  	if ((supported_flags & (SER_RS485_RTS_ON_SEND|SER_RS485_RTS_AFTER_SEND)) &&
>  	    !(rs485->flags & SER_RS485_RTS_ON_SEND) ==
> @@ -1400,7 +1407,9 @@ static void uart_sanitize_serial_rs485(struct uart_port *port, struct serial_rs4
>  static void uart_set_rs485_termination(struct uart_port *port,
>  				       const struct serial_rs485 *rs485)
>  {
> -	if (!(rs485->flags & SER_RS485_ENABLED))
> +	/* Return while disabling RS485 or enabling RS422 */
> +	if (!(rs485->flags & SER_RS485_ENABLED) ||
> +	    (rs485->flags & SER_RS485_ENABLED && rs485->flags & SER_RS485_MODE_RS422))

Is this check needed at all? If no termination GPIO is specified, gpiod_set_value_cansleep()
does nothing. If a termination GPIO is specified it will be deasserted in case of RS422
(since for the RS422 case the SER_RS485_TERMINATE_BUS flag is deleted in
uart_sanitize_serial_rs485()). This seems like a good behaviour to me, so IMO the above check
is not needed.

>
>  	gpiod_set_value_cansleep(port->rs485_term_gpio,
> diff --git a/include/uapi/linux/serial.h b/include/uapi/linux/serial.h
> index 53bc1af67..9086367db 100644
> --- a/include/uapi/linux/serial.h
> +++ b/include/uapi/linux/serial.h
> @@ -11,6 +11,7 @@
>  #ifndef _UAPI_LINUX_SERIAL_H
>  #define _UAPI_LINUX_SERIAL_H
>
> +#include <linux/const.h>
>  #include <linux/types.h>
>
>  #include <linux/tty_flags.h>
> @@ -137,17 +138,19 @@ struct serial_icounter_struct {
>   * * %SER_RS485_ADDRB		- Enable RS485 addressing mode.
>   * * %SER_RS485_ADDR_RECV - Receive address filter (enables @addr_recv). Requires %SER_RS485_ADDRB.
>   * * %SER_RS485_ADDR_DEST - Destination address (enables @addr_dest). Requires %SER_RS485_ADDRB.
> + * * %SER_RS485_MODE_RS422	- Enable RS422. Requires %SER_RS485_ENABLED.

All of the above flags are only effective if RS485_ENABLED is set, so no need
to mention this explicitly for the MODE_RS422 flag.

>   */
>  struct serial_rs485 {
>  	__u32	flags;
> -#define SER_RS485_ENABLED		(1 << 0)
> -#define SER_RS485_RTS_ON_SEND		(1 << 1)
> -#define SER_RS485_RTS_AFTER_SEND	(1 << 2)
> -#define SER_RS485_RX_DURING_TX		(1 << 4)
> -#define SER_RS485_TERMINATE_BUS		(1 << 5)
> -#define SER_RS485_ADDRB			(1 << 6)
> -#define SER_RS485_ADDR_RECV		(1 << 7)
> -#define SER_RS485_ADDR_DEST		(1 << 8)
> +#define SER_RS485_ENABLED		_BITUL(0)
> +#define SER_RS485_RTS_ON_SEND		_BITUL(1)
> +#define SER_RS485_RTS_AFTER_SEND	_BITUL(2)
> +#define SER_RS485_RX_DURING_TX		_BITUL(3)
> +#define SER_RS485_TERMINATE_BUS		_BITUL(4)
> +#define SER_RS485_ADDRB			_BITUL(5)
> +#define SER_RS485_ADDR_RECV		_BITUL(6)
> +#define SER_RS485_ADDR_DEST		_BITUL(7)
> +#define SER_RS485_MODE_RS422		_BITUL(8)
>
>  	__u32	delay_rts_before_send;
>  	__u32	delay_rts_after_send;


Regards,
Lino