cpu/native/periph/uart: uart_poweroff() closes the file descriptor + extended baud-rate support on Linux

[derMihai]

Contribution description

uart_poweroff() now closes the underlying file descriptor

This is needed if for re-initializing the UART device. An example use-case is baud-rate probing.

On Linux, extended baud-rate support

On Linux, the extended baud-rates are defined in bits/termios-baud.h as follows:

/* Extra output baud rates (not in POSIX).  */
#define  B57600    0010001
#define  B115200   0010002
#define  B230400   0010003
#define  B460800   0010004
...

As such, I'm not sure if they are available on other platforms too, so I enabled them for Linux only.

Testing procedure

The following should be a valid use-case:

   static uint32_t const baud_rates[] = {
            230400,
            460800,
            921600,
            3000000,
    };
    int res;
    for (unsigned i = 0; i < ARRAY_SIZE(baud_rates); i++) {
        uart_poweroff(UART_DEV(0));
        res = uart_init(UART_DEV(0), baud_rates[i], device_rx_handler, dev);
        assert(res == 0);

        res = device_probing_procedure(UART_DEV(0));
        if (res == 0) {
            /* device responded */
            break;
        }
    }

[riot-ci]

Murdock results

✔️ PASSED

795415d cpu/native/periph/uart: uart_poweroff() closes the file descriptor

Success	Failures	Total	Runtime
10014	0	10016	09m:38s

Artifacts

• Documentation preview

[derMihai]

On a second glance, it looks that some baudrates already included (e.g. 115200) are not POSIX either, so I think I should enable them all uncoditionally.

[fabian18]

Should that be ~O_ASYNC?

[maribu]

@derMihai and @benpicco ⬆️

[derMihai]

Ouch, yes, ~O_ASYNC. Why was this merged?

[benpicco]

I didn't notice it 🙈

[derMihai]

Haha ok I'll fix is

[maribu]

Don't worry, shit happens 😉