[etherlab-dev] EtherCAT master changes for version 1.4

Florian Pose fp at igh-essen.com
Fri Feb 1 12:35:09 CET 2008


Hi etherlab-dev's,

I'm planning to implement a few changes for the version 1.4.0 of the
EtherCAT master, that I'd like to discuss with you. I'm currently
gathering ideas for the new interface. Here's an (commented) excerpt of
my TODO file:

* Realtime interface changes:

    - PDO configuration.
    
      Configuration of mapped PDO entries.

    - SDO access.
    
      SDO reading/writing via realtime interface.

    - Replace process data pointers with offset.
    
      PDO registration functions will return an offset in the domain's
      process data image, rather than later setting setting a process
      data pointer. See next point.

    - External memory for domains
    
      This is important, if you want to use the process data in a shared
      memory block, for example.

* 'Real' Mailbox handler

    The current mailbox handler allows only one protocol access, at a
    time. It is necessary to serialise accesses to avoid confusion.

* Read PDO mapping via CoE during bus scan.

    A current problem is, that the mapping information stored in the SDO
    dictionary are not present at PDO configuration time, if the
    application is loaded directly after the master. This will be fixed.

* SDO write access in sysfs.

    When the SDO writing is anebled via the RT interface, this will be
    enabled, too.

* Remove get_cycles() calls and references to cpu_khz to increase
  portability.

    It's hard to port the master to architectures other than IA-32, if
    these symbols are used.

* Remove ecdb.h and let lsec output PDO information 'cut-and-pastable' for
  applications.

    The ecdb.h file has the drive to get wrong, and anyway never was
    meant to be complete. lsec shall have the ability to output the
    C-code PDO registrations for a given bus structure.

* Update documentation.

    ... ;-)

If you have any wishes, ideas, comments, please don't hesitate to post
them.

-- 
Best regards,
Florian Pose

http://etherlab.org



More information about the Etherlab-dev mailing list