[etherlab-users] system_time_base

[xiaoliang.li at scbd.com.cn]

When I run the /example/rtai_rtdm_dc.  Why the system_time_base is less than
zero for sometimes. Than the master will report error message
""Synchronization error"" ,the AL status code is 0x001A.

 

 

The code is following:

 

void update_master_clock(void)

{

#if SYNC_MASTER_TO_REF

    // calc drift (via un-normalised time diff)

    int32_t delta = dc_diff_ns - prev_dc_diff_ns;

    prev_dc_diff_ns = dc_diff_ns;

 

    // normalise the time diff

    dc_diff_ns =

        ((dc_diff_ns + (cycle_ns / 2)) % cycle_ns) - (cycle_ns / 2);

 

    // only update if primary master

    if (dc_started) {

 

        // add to totals

        dc_diff_total_ns += dc_diff_ns;

        dc_delta_total_ns += delta;

        dc_filter_idx++;

 

        if (dc_filter_idx >= DC_FILTER_CNT) {

            // add rounded delta average

            dc_adjust_ns +=

                ((dc_delta_total_ns + (DC_FILTER_CNT / 2)) / DC_FILTER_CNT);

 

            // and add adjustment for general diff (to pull in drift)

            dc_adjust_ns += sign(dc_diff_total_ns / DC_FILTER_CNT);

 

            // limit crazy numbers (0.1% of std cycle time)

            if (dc_adjust_ns < -1000) {

                dc_adjust_ns = -1000;

            }

            if (dc_adjust_ns > 1000) {

                dc_adjust_ns =  1000;

            }

 

            // reset

            dc_diff_total_ns = 0LL;

            dc_delta_total_ns = 0LL;

            dc_filter_idx = 0;

        }

 

        // add cycles adjustment to time base (including a spot adjustment)

        system_time_base += dc_adjust_ns + sign(dc_diff_ns);

    }

    else {

        dc_started = (dc_diff_ns != 0);

 

        if (dc_started) {

            // output first diff

            rt_printk("First master diff: %d.\n", dc_diff_ns);

 

            // record the time of this initial cycle

            dc_start_time_ns = dc_time_ns;

        }

}

 

         When system_time_base is less than zero ,the master can not be sync
to the ref clock slave.

But I do not know ,why the system_time_base can be less than zero?

 

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.etherlab.org/pipermail/etherlab-users/attachments/20160420/937a55bb/attachment-0002.htm>