[ECOS] EB 40A, Serial driver stops after 10 minutes

[ECOS] EB 40A, Serial driver stops after 10 minutes

[Shiv Subramaniam]

Hello
   We have a variant of EB40A with only the clock
speed differing - 32.768 Mhz (instead of 66Mhz). 
   After making the change for Clock speed
CYGNUM_HAL_ARM_AT91_CLOCK_SPEED to 3276800. We
compiled the code with original serial driver - The
result was that a lot of packets were getting dropped
(38400 baud).

   We then changed the driver to "Improved AT91 serial
driver" by John Recker from the contrib site
(http://ecos.sourceware.org/contrib.html). This driver
did not drop packets for the rate - 760 bytes per
second (baud: 38400). After 10 minutes - the whole
system comes to a halt - The system does not respond
to the packets sent to the serial port.

The code consists of two threads -
==================================
1. Thread-1 receives packets from serial port and
posts it to the mailbox of thread-2
2. Thread-2 - reads message from the mailbox and sends
the data out through the same serial port.



Another interesting behavior was that when we have a
simple single thread loopback solution. The system was
observed to be stable for 30 minutes.

Can anyone tell me what could be the problem?

Thanks,
Shiv


__________________________________
Do you Yahoo!?
Find out what made the Top Yahoo! Searches of 2003
http://search.yahoo.com/top2003

-- 
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss

Re: [ECOS] EB 40A, Serial driver stops after 10 minutes

[Thomas Koeller]

On Monday 05 January 2004 21:43, Shiv Subramaniam wrote:
> Hello
>    We have a variant of EB40A with only the clock
> speed differing - 32.768 Mhz (instead of 66Mhz).
>    After making the change for Clock speed
> CYGNUM_HAL_ARM_AT91_CLOCK_SPEED to 3276800. We
> compiled the code with original serial driver - The
> result was that a lot of packets were getting dropped
> (38400 baud).

Use ecos from CVS, if you can. The serial driver that
comes with ecos releases 1.3 and 2.0 is known to loose
data at higher speeds. The one in current CVS is
largely reworked and does not suffer from this problem.

tk
-- 
Thomas Koeller, Software Development

Basler Vision Technologies
An der Strusbek 60-62
22926 Ahrensburg
Germany

Tel +49 (4102) 463-162
Fax +49 (4102) 463-239

mailto:thomas dot koeller at baslerweb dot com
http://www.baslerweb.com


-- 
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss

[ECOS] Re: EB 40A, Serial driver stops after 10 minutes

[Daniel Néri]

Shiv Subramaniam <smaniam@yahoo.com> writes:

>    We then changed the driver to "Improved AT91 serial driver" by
> John Recker from the contrib site
> (http://ecos.sourceware.org/contrib.html). This driver did not drop
> packets for the rate - 760 bytes per second (baud: 38400). After 10
> minutes - the whole system comes to a halt - The system does not
> respond to the packets sent to the serial port.

This is not surprising as this code violates some basic eCos
principles, e.g. locking/unlocking a mutex in DSR context, and worse,
operating on a semaphore in ISR context.

I suggest that the maintainers remove the link to this code from the
contributions page, now that a better serial driver exists in official
sources.


Best regards,
-- 
Daniel Neri
dne@mayonnaise.net


-- 
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss

Re: [ECOS] EB 40A, Serial driver stops after 10 minutes

[Shiv Subramaniam]

Thanks Thomas,
   With the latest code from CVS the driver code is
stable for about 100 minutes. The system still comes
to halt after that.

Thanks,
Shiv

--- Thomas Koeller <thomas.koeller@baslerweb.com>
wrote:
> On Monday 05 January 2004 21:43, Shiv Subramaniam
> wrote:
> > Hello
> >    We have a variant of EB40A with only the clock
> > speed differing - 32.768 Mhz (instead of 66Mhz).
> >    After making the change for Clock speed
> > CYGNUM_HAL_ARM_AT91_CLOCK_SPEED to 3276800. We
> > compiled the code with original serial driver -
> The
> > result was that a lot of packets were getting
> dropped
> > (38400 baud).
> 
> Use ecos from CVS, if you can. The serial driver
> that
> comes with ecos releases 1.3 and 2.0 is known to
> loose
> data at higher speeds. The one in current CVS is
> largely reworked and does not suffer from this
> problem.
> 
> tk
> -- 
> Thomas Koeller, Software Development
> 
> Basler Vision Technologies
> An der Strusbek 60-62
> 22926 Ahrensburg
> Germany
> 
> Tel +49 (4102) 463-162
> Fax +49 (4102) 463-239
> 
> mailto:thomas dot koeller at baslerweb dot com
> http://www.baslerweb.com
> 


__________________________________
Do you Yahoo!?
Yahoo! Hotjobs: Enter the "Signing Bonus" Sweepstakes
http://hotjobs.sweepstakes.yahoo.com/signingbonus

-- 
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss