Hello, All three routines are always called from the (Xenomai) cyclic task. I have tried periods of 250 usec, 500 usec, and 5 msec.
ecrt_master_application_time() and ecrt_master_sync_slave_clocks() are always called at every cyclic cycle. I initially called ecrt_master_sync_reference_clock() every cycle but when that wasn't performing well, I tried skipping one or more cycles but that didn't seem to make a difference. Performance is better when using smaller periods. In Xenomai, I call rt_timer_read() which should return nanoseconds since 1970. (The configuration for Xenomai has CONFIG_XENO_OPT_TIMING_PERIODIC=n aka "tickless mode"). This is supplied to _application_time(). master_sync_reference() is called immediately after, and ecrt_domain_queue() and ecrt_master_send() very soon after that. Currently, I stop calling master_sync_reference_clock() shortly after the slave configuration state machine has put all of the slaves into their OP states. I continue to call ecrt_master_sync_slave_clocks() for the life of the app. Best regards, --George Broz Moog, Inc. Industrial Group -----<[email protected]> wrote: ----- To: <[email protected]> From: Florian Pose Sent by: Date: 01/30/2012 07:27AM Subject: Re: [etherlab-users] failing dc-sync-check (and rtdm patch) Hi, do you call the ecrt_master_sync_reference_clock() and ecrt_master_sync_slave_clocks() cyclically? Do you call ecrt_master_application_time() cyclically? What kind of time dou you pass to the latter? -- Best regards, Florian Pose _______________________________________________ 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
