I went a different route since I couldn't figure out if the TUN/TAP feature 
could support the need.

I hacked the code to make an array of slaves.  I've done up to 500 simulated 
slaves so far.  This is on a single IBM ThinkPad T43 (Centrino) running Ubuntu 
10.04.4LTS.

There is a problem with topology detection because the downstream link appears 
on port 3, but I only enable ports 0 and 1.  This doesn't seem to affect the 
linear topology I'm interested in.

There is also a problem with LRW command, with the size of the packet and/or 
the number of slaves so the packet size is wrong and an incorrect area in the 
packet is written.

I also changed the code to use the bin file of my particular slave of interest. 
 This is not quite complete or generic at this time.

All the above is sufficient for my internal testing needs.  I would be willing 
to hand this code off after some cleaning with the understanding there would be 
de-hack-ification and debugging provided by others and re-released.  I've also 
not been doing git updates, so the code may be weeks out of date.

Thanks for your understanding,
Bruno

 

 

 

-----Original Message-----
From: Shahbaz Youssefi <[email protected]>
To: rbenyehuda <[email protected]>
Cc: babbittner <[email protected]>; etherlab-users 
<[email protected]>
Sent: Tue, Jan 15, 2013 4:05 am
Subject: Re: [etherlab-users] Open source virtual Linux Ethercat Slave



Hi Raz,


Is it possible with your software to fake networks with large number of slaves? 
The way I understand, you should have `ethX` interfaces for your slaves to hook 
on to. Perhaps you could try  [email protected]'s idea of named pipes and if 
any of the given parameters (to the user-space one) couldn't be opened, you 
automatically hook on to a named pipe to transfer the data?


This would be very useful to test software targeted to work with large number 
of slaves, without actually having to buy that many slaves (or network cards!)

Shahbaz




On Mon, Nov 26, 2012 at 8:39 AM, Raz Ben Yehuda <[email protected]> wrote:

On Mon, 2012-11-26 at 02:35 +0100, [email protected] wrote:
> I'm attempting to run multiple virtual slaves on one host.  I setup a
> TUN/TAP interface via tunctl named tap0.  Then I issued the commands:
> ./ecslave tap0 &
> ./ecslave eth0 tap0 &
> but I get a message the tap0 interface is LINK DOWN.
> Possibly I'm misusing the TUN/TAP virtual interfaces?

user space ecslave checks that /sys/class/net/ethX/operstate is "up".
Do you have this path for tap0 ?


> Alternatively I'd like to use something like named pipes, but I
> believe that would involve modifying the code.  If I were to modify
> the code

This is what open source is all about. i will be happy to adopt your
code.

>   I think I'd add an option to specify the number of slaves to
> virtualize and make an array of slaves internally.
>
> Any other ideas?
>
> Thanks,
> Bruno




>
>
>
> -----Original Message-----
> From: Ben Yehuda, Raz <[email protected]>
> To: etherlab-users <[email protected]>
> Sent: Wed, Nov 21, 2012 2:25 pm
> Subject: [etherlab-users] Open source virtual Linux Ethercat Slave
>
> Hello etherlab users
>
> I wrote an open source linux virtual ethercat slave.
>
> $ git clone git://git.code.sf.net/p/ecslave/code ecslave
>
> Ecslave is "tailored" to etherlab and it provides a virtual
> and cheap solution to set up and debug etherlab master application.
>
> At the moment ec_slave is user space application, but i am going to add a
> version to kernel mode in the near future.
>
> wiki :
> https://sourceforge.net/p/ecslave/wiki/Howto/
>
> I will be happy for debug and input.
>
> raz
>
> _______________________________________________
> etherlab-users mailing list
> [email protected]
> http://lists.etherlab.org/mailman/listinfo/etherlab-users


--
https://sites.google.com/site/ironspeedlinux/


_______________________________________________
etherlab-users mailing list
[email protected]
http://lists.etherlab.org/mailman/listinfo/etherlab-users




 
_______________________________________________
etherlab-users mailing list
[email protected]
http://lists.etherlab.org/mailman/listinfo/etherlab-users

Reply via email to