Hi, On 19/01/15 13:31, Arnd Bergmann wrote: > On Friday 16 January 2015 18:37:17 Mark Rutland wrote: >> >>> Makes some sense, so what about: >>> ----------------------- >>> Required properties: >>> - compatible: must be "arm,sbsa-uart" >>> - reg: exactly one register range >>> - interrupts: exactly one interrupt specifier >>> >>> Optional properties: >>> - current-speed : the current active speed of the UART >>> - fifo-size : always 32 as per the SBSA specification >>> - word-size : the number of payload bits per word >>> - parity : used parity method, can be: >>> "n": no parity >>> "e": even parity >>> "o": odd parity >>> "m": always mark (logical 1) >>> "s": always space (logical 0) >>> - stop-bits : the number of stop bits after the payload >> >> Generally those look fine, though someone more familiar with serial >> should take a look. >> >> We can drop fifo-size given it's fixed. Anything that we know outright >> doesn't need to be described. >> >> We might want to make them mandatory. I don't see any value in not >> knowing. > > I think making them optional is useless, so either make them mandatory or > drop them if we can avoid it. Unfortunately, LCR is not part of SBSA, > that would let us drop the properties and still report the correct settings. > > Would we be better off documenting that 8-n-1 is the only allowed settting? > I can't think of a reason why anyone would use something else. Reading the spec again I see that 8-bit word len is mandated: "The generic UART uses 8-bit words, equivalent to UARTLCR_H.WLEN == b11." So given this I guess anything other than (8)n1 is pretty useless and wouldn't probably be used anyways. This means that we can get rid of the optional properties altogether and just add "current-speed" to the list of required ones. I am about to implement this. Cheers, Andre. -- To unsubscribe from this list: send the line "unsubscribe linux-serial" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html