Hi all, This might be one of those "obvious in hindsight" things, but: what is the limit for the maximum supported mailbox size? Either in EtherCAT in general or in EtherLab in particular?
What prompted this is that I had a slave experimentally set to a 2048 byte mailbox size in BOOT mode; performing foe_read or foe_write with this setup resulted in a livelock in the depths of Etherlab immediately following posting the request packet. (Ctrl-C etc stopped working, and the kernel periodically printed "blocked task" warnings; the only way I could get it out of this state was to unplug the network cable.) With everything else identical, configuring the slave with a 1024 byte mailbox size results in correct behaviour. Given the values, I assume the limit will be somewhere near the typical 1500 byte max Ethernet payload size. So: 1. Is this supposed to work (perhaps with jumbo frames or segmentation)? 2. If it is supposed to work, is it the master or the slave at fault? 3. If it's not supposed to work, shouldn't the master just select a smaller size if it doesn't support large sizes? While it's supposed to respect the slave's preference, ultimately the master is in control of mailbox sizes. Regards, Gavin Lambert _______________________________________________ etherlab-users mailing list [email protected] http://lists.etherlab.org/mailman/listinfo/etherlab-users
