Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 12 Jul 2004 20:22:15 +0100
From:      Andy Holyer <andyh@hhbb.co.uk>
To:        "Alex Mitchell" <alex@vestigocorp.com>
Cc:        freebsd-questions@FreeBSD.org
Subject:   Re: LRZSZ
Message-ID:  <C894EC53-D438-11D8-A8CC-000D93511A6A@hhbb.co.uk>
In-Reply-To: <FKEJLIBBEDCPHJIAGPPBMEOBGDAA.alex@vestigocorp.com>
References:  <FKEJLIBBEDCPHJIAGPPBMEOBGDAA.alex@vestigocorp.com>

next in thread | previous in thread | raw e-mail | index | archive | help

On 9 Jul 2004, at 21:55, Alex Mitchell wrote:
>
> A very large automotive supplier (our client) uses HHP Dolphin 7200 
> scanners to scan their automotive crates.  The scans are batched in 
> the scanners.  When they are ready to upload the data to our server, 
> they dock the scanner in the cradle.  The cradle is connected to a 56K 
> US Robotics modem.  We wrote the scanner application using 16bit C 
> compiler (because the scanners are DOS based).  We're using a Y-Modem 
> protocol to upload the data files.  The scanner dials into a modem 
> pool to upload the data.  When a connection is established, LRZ is 
> used to receive the data files from the scanner.  With smaller files 
> (7K and smaller), this works fine.  I've tried uploading a 12K file.  
> After uploading 8K, the scanner returns an NAK error.  This error 
> keeps repeating until the scanner gives up and the upload fails.  I've 
> tried a 24K file, it fails after 16K.  I've tried a 56K file and it 
> fails at 16K.  NAK error each time.  I've tried the above scenario on 
> FreeBSD 4.7 and FreeBSD 5.2.1 about 30 times for each file size, with 
> the exact same result each time.

I had a very similar problem about 10 years ago. Macs trying to send 
mail just hung when they had more than x lines of text.

It turned out in the end that the then MacOs TCP/IP stack was ignoring 
the stated MTU, and just tried to send the whole thing in one huge 
packet. On dial-up, this overflowed the kernel IO buffer size,  and  
hung the process.

In the end I looked at the serial port handling code in the kernel, and 
the buffer size was sensible for a 1990-era machine with 64MB of ram, 
being about 1K each. Since we had plenty of RAM on the boxes, I 
sacrificed a bit by upping the buffer size define to something big (I 
think 64K), rebuilt the kernel and the problem went away.

I wonder if something similar is happening to you?
---
Andy Holyer, Technical stuff
Hedgehog Broadband, 11 Marlborough Place Brighton BN1 1UB
08451 260895 x 241



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?C894EC53-D438-11D8-A8CC-000D93511A6A>