Linux-Hams archive - July 1998: Re: Baycom problem - comment and question

Re: Baycom problem - comment and question

Thomas Sailer (vmhv.pgbwis@mail.dy.fi)
Thu, 02 Jul 1998 10:47:49 +0200


> I just looked up the data sheet on the 16550 and as you say it DOES NOT
> support synchronous communication. The driver just use the UART as an IO
> port.

To give you some expert insight from the guy who did the Linux
driver :-)

The newer drivers (i.e. baycom_ser_fdx in the 2.1.x series) use
the Delta-CTS interrupt, that is, if the CTS line (which is connected
to the receive data output of the modem chip) changes, you get
an interrupt. The interrupt then reads a time stamp counter and
updates its PLL and feeds the received bits into the HDLC decoder.
So worst case you get one interrupt per bit, on average it's less.

Although I wouldn't recommend this, people used the driver successfully
up to 9600 bauds.

> I am still kind of leery of this approach, If you can recompiling the
> kernel while using the modem I would say its OK. I guess I am just
> skeptical. I do plan on buying one so I guess I will find out sone
> enough.

You don't even need to buy one to find out :-) The new driver
(baycom_ser_fdx) is full duplex capable, as its name implies.
Tie the DTR and the CTS pins together and use it as a loopback
interface.

BTW if you're using IDE disks, you should hdparm -u1 all your disks,
otherwise the IDE driver will disable interrupts for disgustingly
long periods of time, which will make the baycom driver fail.

Tom