Problems w/ serial flow control

[Philip Prindeville]

Rick Stevens wrote:
> On Wed, 2006-05-31 at 12:14 -0600, Philip Prindeville wrote:
>   
>> Tim wrote:
>>     
>>> On Tue, 2006-05-30 at 21:25 -0600, Philip Prindeville wrote:
>>>   
>>>       
>>>> I'm running FC3 on an MSI-7142 (K8MM-V motherboard), and using
>>>> the two serial ports on the m/b.  I used the stock
>>>> out-of-the-shrinkwrap settings for most hardware (except for the sound
>>>> board and X drivers), and things have worked fine.
>>>>
>>>> But... when connecting to a router's console at 115200 with either
>>>> RTS/CTS or XON/XOFF flow control, I'm seeing dropped characters.
>>>>     
>>>>         
>>> Looking at a friend's router, his one's serial port connection *ONLY*
>>> had the three-wire type of serial connection.  So his couldn't possibly
>>> use hardware handshaking (RTS/CTS).  Perhaps yours can't too?
>>>
>>> Also, check what your router says about its maximum port speeds and
>>> handshaking protocols.
>>>
>>>   
>>>       
>> It is an 8-wire connection (over RJ-45 to DB-9), and can do 115200 (max 
>> rate)
>> with either XON/XOFF or RTS/CTS.
>>
>> I've tried both, and they're both lossy.
>>
>> The issue is on the PC as the receiver...
>>     
>
> Remember that both ends have to agree on the flow control.  If one is
> running  RTS/CTS and the other is XON/XOFF, you don't have flow control.
> At 115K, I'd always opt for RTS/CTS...XON/XOFF requires CPU intervention
> unless you have fairly sophisticated hardware.  All you have to do is
> be a bit pokey at responding to an interrupt and you'll drop characters.
> RTS/CTS is usually done at the hardware level and is a bit more
> reliable.
>
>   

I started with RTS/CTS...  And had it set on both ends.

I'm using the FC5 version of kermit.  My .kermrc looks like:

set modem type none
set line /dev/ttyS0
set carrier-watch off
set speed 115200
set flow rts/cts
set parity none
set stop-bits 1
connect

Not sure what the difference is between "/dev/tty0" and
"/dev/ttyS0"...

-Philip