Re: [MAT-devel] modbus slave function 0x03 and 0x04

[Jiri B. <ji...@ba...>]

> Jiri:
> > It might be useful to allow the 03 and 04 maps to be configured
> > separately, so that the MatPLC can be substituted into an
> > already-existing system.
...

Andrey:
> I am not sure it is a good option.. And the standard provides for a
> situation when all functions access the same chunk of memory. This is
> what GMM is in MatPLC. Why would it be better to have different maps?

The only situation would be when the MatPLC is to replace/augment an
existing system, and the previous system has separate register banks. In
that situation one would want to keep the existing addresses.

For new systems, making them identical would generally be better.

> > When coils are implemented, they'll slot right into this system.

> Coils or  registers are the same stuff. The only difference is the
> access granularity (1 bit or 16 bits).

Yeah, but again it can be useful to have them in a separate bank.

> > (Obviously, since you're doing the implementing and I'm just
> > talking, feel free to ignore my suggestions; I guess put them in the
> > wishlist where they belong.)

> Actually I appreciate your comments very much! And I agree with most
> of them! :)

No worries; just so you don't feel pressured :-)

> And while we are at it, what do you think would be the best way to
> implement access control?

I agree with Mario, leave per-point access control up to the MatPLC
core.

Global access control (for TCP) is a different matter, but
 - I don't think Modbus has much of a provision for that, and
 - there's already a modbus-aware-firewall project.

About the maximum you could do is make an option to listen on a
particular interface, but I'm not sure it's worthwhile given the above.


Jiri
-- 
Jiri Baum <ji...@ba...>           http://www.csse.monash.edu.au/~jirib
  MAT LinuxPLC project --- http://mat.sf.net --- Machine Automation Tools