Did you consider a combined value for all options?

E.g.,

typedef uint16_t uart_mode_t
#define UART_PARITY_OFFSET 0
#define UART_PARITY_MASK 0x3
#define UART_DATABITS_OFFSET 3
#define UART_DATABITS_MASK 0x3

//...
int parity = mode & UART_PARITY_MASK >> UART_PARITY_OFFSET
int databits = mode & UART_DATABITS_MASK >> UART_DATABITS_OFFSET

In addition, define the most used ones as speaking defines:

#define UART_MODE_8N1 (UART_DATABITS_8 << UART_DATABITS_OFFSET | UART_PARITY_N << UART_PARITY_OFFSET ...)

My main reason would be that this way, uart users actually using modes don't need all the fields:

typedef struct  {
 //...
 uint8_t uart_parity;
 uint8_t uart_databits;
 uint8_t uart_stopbits;
} random_driver_params_t;

// vs
typedef struct  {
 //...
 uart_mode_t uart_mode;
} random_driver_params_t;

@kaspar030 we already had this discussion here and so I've made this PR according to it.

v2 changes:

• add support for partial 6-bit and full 7-bit modes to the devices defining USART_CR1_M1
• add tests for even and odd parity against PHiLIP hardware

TODO: convert uart_mode() to a feature

v3 changes:

• convert uart_mode to a feature periph_uart_modecfg
• move tests to a separate commit

@bluephoton could you test this PR with your setup and report the results here? Thanks.

@yegorich , I gave it a try but it didn't work! I know that I need to invert RX for my scenario so I added the following code (at the very end before you enable uart):

MODIFY_REG(dev(uart)->CR2, USART_CR2_RXINV_Msk, USART_CR2_RXINV);

but this didn't help!

I'll look bit more tomorrow, in case its something with my environment.

Here is how I initialize (working case):

uart_init_ex(uart, baudrate, UartReceiveCallback, this, false);
uart_configure(uart, UART_9BITS, UART_PARITY_EVEN, UART_STOPBITS_TWO, UART_RX_INVERT);
uart_enable(uart);

Here is with new uart_mode (not working case):

uart_init(uart, baudrate, UartReceiveCallback, this);
uart_mode(uart, UART_DATABITS_8, UART_PARITY_EVEN, UART_STOPBITS_2);

Couple notes on implementation:

• I'm not an expert in this, but not a fan of starting up device, stopping it to configure, then starting it again.
• Not sure why we have UART_DATABITS_9 if we are setting this internally (code could use some comment around this as it's not intuitive)

@yegorich I found the culprit! Two issues:

• uart_stop() line 171 & 178 should be

cr1 &= ~USART_CR1_UE;   // note the ~ should be there in both cases

• stopbits seems to be not set correctly, when I change it as follow it works:

// cr2 |= stopbits;
cr2 |= USART_CR2_STOP_1;

• And of course inverting RX bin was required and you should add while you are at it.

I also recommend replacing all numeric constants with actual macro as defined in CMSIS header file. Also not use _M (mask) and use actual bit value (even if it happen to be the same) like:

if (databits == UART_DATABITS_8) {
            //cr1 |= USART_CR1_M;
            cr1 |= USART_CR1_M0;

Edit: Added comments on the exact code location with changes required

@MrKevinWeiss could you give this a final test-run? Thx. If you succeed, feel free to merge!

Ok hold for retesting

Seems good to me. GO!
Nice job @yegorich!