Andy Shevchenko wrote: > Which version of POSIX and SUS standards are in consideration? I admit my ignorance - but I've always been told that the behaviour of always unconditionally asserting both DTR & RTS on serial port open is mandated by both POSIX and SUS, which is why all Unix-style OSes have been following this behaviour up until 2021-04, when FreeBSD 13.0 became the world's first Unix-style OS to provide an *option* for users to opt out (on a per-serial-device basis) of this standards- mandated behaviour. > Is it only the USB class of devices that are affected or do we have > examples on other buses? In my own use case, it is only USB: my hw device is one where a USB-serial chip (FT2232D in my case) and the circuit that repurposes DTR & RTS outputs from one of the UARTs (FT2232D Channel B in my case) are inseparably integrated on the same PCB, with a custom USB VID:PID identifying the device as a whole. However, I have been told that in order to be acceptable into Linux mainline, the proposed solution has to work for all similarly affected parties and not just my device, hence I am also considering a "generic" case where a custom hw device would have an old-fashioned RS-232 electrical interface and could be connected to "any" serial port. > Logically I would put them otherwise, first to check a custom flag and > then the existence of the callback. I can make this change in the next version of my patch series, once I get a clarification from Greg as to the correct way to denote the chain of authorship and revision. M~
