Subject: Re: Serial speed.
To: Matthew Reilly <mjreilly@flashcom.net>
From: Bill Studenmund <wrstuden@nas.nasa.gov>
List: port-mac68k
Date: 07/08/1999 13:29:44
On Wed, 7 Jul 1999, Matthew Reilly wrote:

> Bill Studenmund wrote:
> 
> [snip]
> 
> > The reason I haven't merged in the change is, as Frederick mentioned, only
> > a few machines support receiving lots of characters per second. If I turn
> > on the baud rate, people will expect to receive at that rate, they won't,
> > and they'll be pissed. :-)
> 
> Thanks Bill, is this the patch? It's the only one I found searching the archive.

Nope. That one fixes a problem where someone copied & pasted the
initialization table from another port when they shoudn't have.

That fix should already be in there.

You want to find the line:

		xcs->cs_clocks[0].flags = ZSC_RTXBRG;

which is line 317 in version 1.31. Make it

		xcs->cs_clocks[0].flags = ZSC_RTXBRG | ZSC_RTXDIV;

And you'll be able to do 230400.

As I said before, you can TRANSMIT at this rate fine. You just can't
receive at faster than about 5KB/sec (both burst & sustained unless the
bursts are 3 char or smaller).

Remember we have a 3 byte input FIFO. Over flow it, and you'll see
messages about silo overflows. (More for other readers thinking of doing
this) If you're using PPP or slip, loosing one byte requires the ENTIRE
packat to be re-transmitted, lowering your throughput.

If the printer is as chatty as an HP (one byte per two seconds), the above
patch is fine.

Take care,

Bill