Re: SUMMARY: Hardware Flow Control ... WHERE ARE YOU SUN !!!!!!

From: Jim Prescott (
Date: Fri Oct 02 1992 - 23:41:18 CDT

In article <> (David Clunie) writes:
>My question now is this ... surely some one from Sun reads this group,
>and surely there must be thousands of frustrated sun users who cannot
>believe that Sun's cannot perform incoming hardware flow control.
>Why the hell don't you guys fix the bloody driver !@#$

Probably because it isn't broken. Sun ports implement RTS/CTS handshaking
exactly as specified by the rs232c standard. The desired behavior (which
is actually in the most recent version of the rs232 standard, now called
ANSI/EIA/TIA-232-E) could potentially upset some stupid device that depends
upon RTS acting like another DTR.

The missing modem control on the second SLC port seems to be another
example of Sun following an existing standard to the letter instead of
doing what 99.999% of their customers would have preferred. They only had
room on the machine for a single db25 connector but they wanted to provide
2 serial ports. Someone decided the best approach was to put all the pins
for both ports into a single db-25. Fortunately the rs232 spec includes
a secondary data channel with its own TD, RD, RTS, CTS, RTS; unfortunately
it leaves out DSR, which Suns ignore, and DTR, which is at least strongly
desirable on a modem line. Sun could have just put the second DTR on some
vacant pin (perhaps pin 11 which rs232 leaves unassigned) but didn't.

>the Zilog SCC is capable of de-asserting the >appropriate line.

The situation isn't really symmetric. It can be told to drop RTS. It can
also be told to automatically stop sending data whenever CTS drops. The
kernel has more to do in RTS case.

A smarter chip might have a mode where it automatically drops RTS whenever
its silo is about to fill up.

In any event, I believe they wanted it to work the way it does; it wasn't
an oversight.

>so where >is the patch guys !

Possibly in the Sun Consulting HSPEED special (or possibly not, all it says
is that it allows 38.4k input).

>At the very least where is something clearly stated in the documentation

The operation of the crtscts flag is described in termio(4). The zs,
mpc and spc man pages indicate that they support the crtscts flag
(although only on the first 4 mcp ports). The mti driver ignores crtscts
and requires CTS to be high.

Solaris 2.0's termio can request bi-directional RTS/CTS flow control so
perhaps Sun will implement it in their drivers.

Jim Prescott		{rit,tropix,ur-valhalla}!moscom!jgp

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:50 CDT