Hardware Flow Control bug on Suns

[raymond r rankins]

The following is a problem we are having using hardware flow control on
the Sun workstation serial ports.  If anyone has experienced similar
problems, and hopefully resolved it in some way, please let us know.  You
can email to either myself (rankins at argentina.crd.ge.com) or to
cannon at bahamas.crd.ge.com.

Thanks.

--------------------------------------------------------------------

In an application we are attempting to use hardware flow control (RTS/CTS)
between a Sun and a PC.  On the PC side we are using a commercial library
of routines which appear to work properly in managing the line.  On the
Sun side, however, we have discovered two conditions which are causing us
some grief.

1) Sun has not implemented HW flow control on the receiving side in their
   serial line drivers for ttya and ttyb: RTS is not lowered when the buffer
   becomes nearly full.  We had heard that flow control had been implemented
   in the driver on the first 4 ports of the ALM board.  We have an ALM board
   and have tried to test this.  In doing so, however, we ran into problem
   number two below.

2) When we configure the device for HW flow control, the C read behaves
   improperly.  It appears that if there is a full buf (as defined in the man
   page on read) of data in the input buffer then the read will do fine.  If
   there is not a full buffer (i.e. if it can't read nbytes of data), it will
   block on the read.  With this (probably unrelated) bug we couldn't test
   the ALM and the driver for flow control.

In an attempt to diagnose problem #2 above we tried software flow control
using the ALM and that didn't work either.  The C read command would still
block if buf was not full.

Ray Rankins          |(518) 387-7340 | INTERNET: rankins at argentina.crd.ge.com
2 Moonglow Rd.       |(518) 583-3320 | COMPUSERVE: 71131,3236
Gansevoort, NY 12831 |               | AmericaOnline: RayRankins