Subject: Re: 16550 Serail Controllers
To: Gerry Simmons Jr. <darykon!simmons@nwnexus.wa.com>
From: None <winans@aps.anl.gov>
List: netbsd-users
Date: 01/25/1995 11:50:08
>Thanks to everyone on the feedback. One thing to note is that most of the
>I/O boards use ChipSets and the 16XXX serial controller is inside a single
>100 pin Flat Package along with the Parallel Port and usually a game port.
>Both VLB I/O boards I have are this way.
>
>The reason I was asking is in regards to someone else's inquiry on the
>"silo overflow error." I too get this error in SLIP, but don't think my
>serial ports have FIFO's. Is is possible to get this error even without
>a FIFO? Is this simply a serial channel overflow message?

I get those ALOT too.  I have FIFOs and hardware flow control enabled
and so on... however, after looking at the com.c source, it appears
that the on-chip flow control is not being used... it is the calls from
the tty code to stop and restart the data that seem to be operating the
CTS line.  This, IMHO, sucks and might cause me to try my hand at a new
driver some day.  If my understanding is accurate on this, you will
ALWAYS be able to get overflows on in-bound characters.

Can anyone confirm this minfeature of the com driver?

The good news is that at least in the outbound direction from your box,
things will work well because modems have large buffers and operate
their hardware flow control properly.  So the only occasional character
loss should be from the modem to your box.

--John


! John Winans                     Advanced Photon Source  (Controls)    !
! winans@aps.anl.gov              Argonne National Laboratory, Illinois !
!                                                                       !
!"The large print giveth, and the small print taketh away." - Tom Waits !