[etherlab-users] Corrupt mailbox response in FoE

Florian Pose fp at igh-essen.com
Tue Feb 5 15:03:31 CET 2013


Am 21.01.2013 16:52, schrieb Ralf Roesch:
> the current stable-1.5 branch does not work for me, when using FoE
> commands.
> I checked the FoE datagram flow with Wireshark and detected no errors.
> So I looked at the source and found a fix for my problem.
> Attached you will find a patch for further discussion.

I just applied the patch. I really tried hard to not breaking something
during the necessary changes in 9cdd7669dc0b, but FoE is hard to test,
since most slaves allow only firmware updating. Sorry for that.

> Please note: all fsm->slave by slave replacements hove nothing to do
> with the problem. (This is only code beautifying)
> The important fixes are the datagram by fsm->datagram replacements found
> in ec_slave_mbox_fetch() functions.
> BTW what are the differences between the datagram and fsm->datagram
> handles?

9cdd7669dc0b introduced a new locking-free mechanism for datagram
injection. The datagram sent in the last cycle (fsm->datagram) can now
be different from the datagram to use in the next cycle (datagram),
because slave FSM datagrams are now maintained in a ring buffer.

Best regards,
Florian Pose

More information about the Etherlab-users mailing list