Re: CAN and MPC855T/860T

In message [off-list ref] you wrote:

I know that TQ Components integrated two intel 82527 on their TQMs /
STK8xxL.
But I couldn't find any sources supporting them by now.

You didn't ask before...

Could you please give me some pointers to driver source code.

Actually, the driver source code is just one part of the issue.  What
are you going to do with it?

IMHO there are two Open Source projects that look interesting:

   1. http://home.wanadoo.nl/arnaud/
   2. http://www.esstin.uhp-nancy.fr/~tissered/CanFestival/

The first provides generic CAN support  for  a  couple  of  controler
boards  and chips. We have a (preliminary version of) a device driver
for it running on the TQ modules.

The second project implements (parts of)  the  CANopen  protocol  and
integrates  with  RTLinux. However, it is a non-trivial job to get it
running on the TQ hardware, and I'm not sure if the  effort  is  well
invested.


See also:

	http://www.microcontrol.net/CANpie/

Wolfgang Denk

--
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd@denx.de
Of course there's no reason for it, it's just our policy.

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Wolfgang Denk wrote:

IMHO there are two Open Source projects that look interesting:

   1. http://home.wanadoo.nl/arnaud/

Two days ago I got this code working using a SJA1000 chip on a PowerPC
system. It's very "interesting" code, for the following reasons:

1) Its initialisation is back to front. Most Linux subsystems you load a
hardware driver which calls a function to register itself. With this you
load the CAN subsystem which takes module arguments telling it which
hardware drivers to use.


2) When the interrupt handlers call worker functions, they pass the
parameters through global variables rather than as function parameters.

3) It has two different ways of keeping track of its circular buffers,
by an index or by pointers. Different routines use different methods,
with hilarious consequences. (This may particularly cause problems for
users of Pelican mode.)

4) It has several loops with udelays() in them. One of these gets called
by an interrupt handler.

I'm not entirely sure what to do with my changes, as I also integrated
it into the kernel tree (the target system doesn't use modules).

--
Adrian Cox   http://www.humboldt.co.uk/


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/