Re: [ECOS] Redboot for Assabet goes silent

[Gary Thomas <gthomas at redhat dot com>]

On 23-Nov-2000 Martin van Veen wrote:
> Hi guys,
> 
> "Jesper" == Jesper Skov <jskov@redhat.com>
> "Gary" == Gary Thomas <gthomas@redhat.com> 
> 
>>>>> And it did fix it for me too. Downloading over a serial works
>>>>> now, but during download a NAK for each packet is returned. sx
>>>>> resends it and then the packet is accepted. That makes the
>>>>> download very slow, but having a slow download is much better
>>>>> than having no download at all :-)
>>>>
>>>> Gary> I'm glad it fixed it for you. 
>>>> Gary> As for the NAK problem, what is your host computer?
>>>> Gary> Have you tried using Ymodem?  I've tried this with both
>>>> Gary> Linux and Windows hosts and I get pretty good performance,
>>>> Gary> without any extra NAKs.
>>>
>>> Jesper> Is there only one flavor of assabet boards? I ask because
>>> Jesper> I also had a NAK-for-every-packet problem, and it was
>>> Jesper> basically due to the target CPU being too slow.
>>
>> Gary> Yes, however they can be set up to run at different
>> Gary> (software selectable) clock frequencies. Is this truly
>> Gary> an Assabet, or some other SA1110 based system?
> 
> As far as I know I've got a true Assabet board (including the Neponset)
> There is a sticker on the backside of the board saying that my board is
> a 8A1110DEVDBD. If I look at the pictures on the ARM Linux on Assabet
> page (www.cs.cmu.edu/~wearable/software/assabet.html) I've got the same
> board.
> 
> Triggered by Gary's statement that he has no problems on Linux and
> Windows hosts, I've done some testing. I've got a Linux host, but I
> also tested using a Windows NT host with HyperTerminal. All tests
> showed that the data is sent (getting loads of NAKs) and at the end
> is rejected (which I did not notice at the time I wrote the piece of
> text above). In other words: I can download, but the data does not
> arrive :-(
> 
> First, the serial port I use is configured as:
> 
>   $ stty < /dev/ttyS0
>   speed 38400 baud; line = 0;
>   -brkint -imaxbel
> 
> I try to upload a file containing a romfs, so I have to upload it
> in raw mode. I use minicom -o /dev/ttyS0 to make contact with the
> board. Then I type the Redboot load command and then exit minicom
> to enter the sx or sb command on the command line. Below you can
> find the output of my tests...
> 
> ----
> 
>   Redboot> load -r -v -m xMODEM -b 0x1000000 romfs.img
> 
>   $ sx -k romfs.img < /dev/ttyS0 > /dev/ttyS0

Why break out of minicom at all?  I always just use ^A-S (send file) which
sends directly without messing around with the line, etc.
>  
> I performed the same tests on a Windows NT machine with HyperTerminal.
> Having contact with the board is no problem, but uploading using
> different protocols (XMODEM/XMODEM 1K/YMODEM/YMODEM-G) gives the same
> result as on the Linux host... (HyperTerminal gives me no error
> messages, but tells that I can wait forever before the data is
> uploaded).
> 
> Strange enough though, is that I can upload over the serial line when
> I am sending over an application for debugging with gdb, so I guess
> that the serial port configuration is not the problem... (unless gdb
> is setting up the serial port itself).

Also, do you have the latest patches in place?  I'm not sure that they
have made it to anonCVS yet.  There were serious problems with the
X/Y modem stuff which kept it from working at all using Windows.

I can send them to you [privately, no need to clutter the mailing list].