Subject:
|
Re: LNPd keeps sending TX_FAILURE
|
Newsgroups:
|
lugnet.robotics.rcx.legos
|
Date:
|
Fri, 7 Jan 2005 03:26:50 GMT
|
Viewed:
|
7146 times
|
| |
|
|
I don't use LNPd or BrickOS so I don't know if this will work. It makes
all sorts of problems go away when using the RCX ROM message handling
routines, I don't know how much of the ROM routines are used by LNPd or
whether, if it doesn't, it has replicated the same problems.
Try inserting a 35 millisecond delay between receipt of message from RCX
and transmission of next message from PC. See if that works.
Every character transmitted by RCX is also echoed on the RCX receiver. I
think some of the code you were wading through may be related to
confirming that the echo-ed character matchs the transmitted character;
if it doesn't then I guess this is an error or a collision.
"Tyler Mandry" <tmandry@gmail.com> wrote in message
news:I9x7Do.qxM@lugnet.com...
> I'm running LNPd on Gentoo Linux 2.6.7. Everything seems to work fine
> except
> transmitting messages to the RCX. The |lnptest| application that comes
> with LNPd
> keeps reporting a |Collision| every time it tries to transmit
> something. It does
> however correctly report messages that it receives (from the
> corresponding
> program for the RCX). I tried going in |transceiver.c| of the LNPd
> source and
> commenting out sections that called |tx_error()|, but then another
> section would
> call it. In fact, all of the places that could call |tx_error()|
> called it when
> the one before it was commented out: the {frame error} check in
> |rcx_read()| (I
> don't pretend to know what a frame error is), the {transmit collision}
> check
> that compares |rcv_buffer| with |tx_verify|, and
> |check_interbyte_timeout()|.
>
> Here's the information I've been able to glean about each of these
> errors. It's
> not that long of a read. The frame error seems to be triggered if a
> call to
> |ioctl()| requesting |TIOCGICOUNT|, which apparently ["reads serial
> port inline
> interrupt counts"], returns a different value than the previous value
> in
> |rcx_read()|. Doesn't exactly clear things up for me, but hopefully it
> will help
> determine the problem. |tx_verify| is a global variable and is set by
> |rcx_write()| during a write. It is compared to |rcv_buffer|, which is
> read from
> the transceiver by |rcx_read()| (which does the checking). If they
> differ,
> |tx_error()| is called which causes LNPd to send |TX_FAILURE|. Both of
> these
> error checks only call |tx_error()| if the global |active| varible is
> true,
> which is set in |rcx_write()|. So, it seems that after a write
> (|rcx_write()|
> called), |rcx_read()| reads back the data that was written (an echo?)
> and
> compares it to the data that should have been written. It also checks
> to make
> sure that no "inline interrupts" have been called since the last
> write, I'm not
> sure exactly what this means, possibly that something was sent to the
> transceiver (but don't take my word for it). As for
> |check_interbyte_timeout()|,
> it's called by |run_transceiver()| when there isn't anything to read
> and checks
> if the "interbyte timeout" has been reached, which is updated every
> |rcx_read()|. I don't know how correct my inferences are, but
> hopefully it'll
> help those who haven't gone as in-depth into the source as I have to
> figure it
> out.
>
> Can anyone explain why this might be happening? Transmitting to the
> RCX is vital
> to the project I'm working on, so any help is greatly appreciated.
>
> Tyler Mandry
|
|
Message is in Reply To:
 | | LNPd keeps sending TX_FAILURE
|
| I'm running LNPd on Gentoo Linux 2.6.7. Everything seems to work fine except transmitting messages to the RCX. The lnptest application that comes with LNPd keeps reporting a Collision every time it tries to transmit something. It does however (...) (20 years ago, 7-Jan-05, to lugnet.robotics.rcx.legos, FTX)
|
6 Messages in This Thread:
 
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
