Re: MiNTlib (I think I'm on pl43): Bug in sprintf().

[Stephen Usher <Stephen.Usher@earth.ox.ac.uk>]

>> Well I've had a fun weekend trying to get the NET/2 BSD version of
>> telnet/in.telnetd running and... I've got telnet working.
>
>Hmm, I'm sorry not having mentioned it before you started to port....
>I've already done working ports of telnet and telnetd and have not up-
>loaded them because I wanted to do some more tests. :(
>
>But I'll do it today...

Yeah, so Torsten Scherer (spelling?) told me, he sent me a copy.

>[sprintf stuff deleted]
>
>> PPS. in.telnetd isn't working.. well it is, partially, it doesn't echo
>> characters and it loses the first character of every packet it sends! I've
>> tested telnet by calling my Sun 3/80 and in.telnetd by calling it from the
>> Sun.
>
>This is due to PTY packet mode, which is not supported by Mint. Look for
>ioctl ( , TIOCPKT, ) and the main select loop.

Aha! This was causing me much scratching of my head.

I'll fix my version and send you a copy to see what you think of it.

By the way, I'll also send you a copy of pty.c from portlib which I've
modified to automatically generate a symbolic link in /dev in the form
/dev/tty[pqrs]n and returns this as the slave's name, this allows such
programs as finger etc. to work correctly.

>> PPPS. MiNTnet's PPP device works well, except for when there's a data
>> overrun when it gets hopelessly confused and has to be restarted (after
>> clearing the serial port input buffer using kermit).
>
>Hmm, do you use HSmodem or some such serial routine replacement (or even
>modm0dev)? And what buffers do overflow, the BIOS recv and/or send buffers?

I'm using the standard BIOS devices. It seems (I have no way of checking) as
though if more data than sld can handle arrives then it fails to read any
more and stops, the only way to get things going again is to kill slattach,
clear out the buffers using some terminal program then start slattach again.
If the buffers aren't cleared with a terminal program first then the second
slattach will also not work.

>There might be a bug in the circular buffer handling in MintNet, which I
>don't believe (SLIP uses nearly the same code), have to look at it.

I don't think this is the case as you can start any number of slattach
processes and they still won't work until the serial buffers have been
flushed.

>What seems more likely is a problem with the wakeselect on the BIOS devices.
>PPP input uses a deamon to receive and send data from/to the serial port
>selecting the devices. What I have noticed is that when using the Mint
>builtin drivers with Howards select patches applied then input data comes
>in rather largish hunks, ie the wakeup seems not to occur always immediatly
>after data has arrived. But I never noticed complete hangs...
>
>Migh this be a bug in checkbttys or is this 'normal', Howard?
>
>I have HSmodem running with 4k receive buffers at 19200.

Does HSmodem work on all the serial ports, ie. the SCC based ones on the TT?

>> Kay, could you get your
>> slattach to set the baud rate of the serial port using a command line option
>> please so I can place the PPP startup in an rc file and not use kermit to
>> set the port speed.
>
>Well, I usually do an
>stty speed 19200 < /dev/ttya
>for setting the terminals speed. If you find it better to have only one
>program I might add it when I get more time (not until 15th of July).

Aha! Now, why didn't I think of that?

>
>Kay.
>

Steve

-- 
---------------------------------------------------------------------------
Computer Systems Administrator, Dept. of Earth Sciences, Oxford University.
E-Mail: steve@uk.ac.ox.earth (JANET) steve@earth.ox.ac.uk (Internet).
Tel:- Oxford (0865) 282110 (UK) or +44 865 282110 (International).