In article <20040930023324.GG15736@bcd.geek.com.au>,
	dan@geek.com.au (Daniel Carosone) writes:
> 
> I was playing with a CDMA modem PCMCIA care we have for testing.  It
> probes as a completely standard com port, 16550a with fifo. =20
> FWIW, CIS probe says <CMOTECH, 5100P, 01, >
> 
> I can open the com port with cu, and the card seems to come to life,
> but I get garbage that looks like a baud mismatch.
> 
> The windows driver reports connections at 230400 bps.  Perhaps the
> card's pseudo-com port only works at this rate, but I can't get com(4)
> to go there.
> 
> I dimly remember some discussion in the distant past about people
> playing with rate multipliers for higher speeds, but can't see
> anything obvious in com.c to suggest how this might be accessible.
> 
> I'll build a kernel with COM_HAYESP in case it's emulating one of
> those; any other suggestions?

Try building a current kernel with options COM_16650 (COM16650 on 2.0).

See if you can get docs from the manufacturer, or at least see what
they claim it to be in the literature. Some pucs are 16550a, but with
a hack to set the clock multiplier, some really are 16650 (which we
support with COM_16650). There is another chip, the TI 16550c, that
we don't support in high speed mode at all.

-- 
Frederick