[etherlab-users] Regular "working counter changes" error

[law_hit at aliyun.com]

Dear all, 

    I am running Etherlab master with Gavin's patches to drive some IO modules and servo drivers.  
All the slaves are successfully configured and switched to OP. But I got a lot of working counter 
warnings begin from it is started. Part of the dmesg log is shown below: 

[ 4764.420650] EtherCAT 0: Domain 0: 670 working counter changes - now 13/21.

[ 4765.424640] EtherCAT 0: Domain 0: 669 working counter changes - now 21/21.

[ 4766.428629] EtherCAT 0: Domain 0: 669 working counter changes - now 13/21.

[ 4767.432619] EtherCAT 0: Domain 0: 670 working counter changes - now 13/21.

[ 4768.436609] EtherCAT 0: Domain 0: 669 working counter changes - now 21/21.

[ 4769.440597] EtherCAT 0: Domain 0: 669 working counter changes - now 13/21.

[ 4770.444586] EtherCAT 0: Domain 0: 670 working counter changes - now 13/21.

[ 4771.448576] EtherCAT 0: Domain 0: 669 working counter changes - now 21/21.

    Some information of my system:  CPU: AMD ryzen5 1600x, Intel I211 network card, LMDE2(Debian8),
Kernel: 4.9.0-0.bpo.8-rt-amd64 (rt-preempt),  IgH EtherCAT master 1.5.2 d5f6daedadc3+, 
I am using the generic driver. 

    I believe this is not a timing problem of the realtime master thread. Because I have isolated serval cores (6-11)
for realtime tasks, and the max jitter of realtime thread is no more than 5us. 
    mint at lmde ~ $ sudo cyclictest -p80 --smp
    [sudo] password for mint: 
    # /dev/cpu_dma_latency set to 0us
    WARN: Running on unknown kernel version...YMMV
    policy: fifo: loadavg: 0.32 0.09 0.03 1/872 17643           
    T: 0 (17345) P:80 I:1000 C:  58319 Min:      1 Act:    2 Avg:    3 Max:      34
    T: 1 (17346) P:80 I:1500 C:  38879 Min:      1 Act:    5 Avg:    4 Max:     113
    T: 2 (17347) P:80 I:2000 C:  29159 Min:      1 Act:    3 Avg:    4 Max:      42
    T: 3 (17348) P:80 I:2500 C:  23327 Min:      1 Act:    3 Avg:    4 Max:      43
    T: 4 (17349) P:80 I:3000 C:  19439 Min:      1 Act:    3 Avg:    5 Max:      32
    T: 5 (17350) P:80 I:3500 C:  16662 Min:      1 Act:    3 Avg:    4 Max:      27
    T: 6 (17351) P:80 I:4000 C:  14579 Min:      2 Act:    3 Avg:    3 Max:       4
    T: 7 (17352) P:80 I:4500 C:  12959 Min:      2 Act:    3 Avg:    3 Max:       5
    T: 8 (17353) P:80 I:5000 C:  11663 Min:      2 Act:    3 Avg:    3 Max:       5
    T: 9 (17354) P:80 I:5500 C:  10603 Min:      2 Act:    3 Avg:    3 Max:       5
    T:10 (17355) P:80 I:6000 C:   9719 Min:      2 Act:    4 Avg:    3 Max:       4
    T:11 (17356) P:80 I:6500 C:   8972 Min:      2 Act:    3 Avg:    3 Max:       4

    I have tried to use DC but the errors still happen. Related codes:
        ecrt_domain_queue(domain1);
clock_gettime(CLOCK_TO_USE, &time);
ecrt_master_application_time(master, TIMESPEC2NS(time));
ecrt_master_sync_reference_clock(master);
ecrt_master_sync_slave_clocks(master);
        ecrt_master_send(master);
    The sync functions are called after ecrt_domain_queue() as suggested in other mail threads. 

    I have also tried the igb driver instead of generic, but the problem still. 

Can anyone help explain what is happening and what should I do to get rid of these errors please?

Thanks in advance,

Yours,
Jancon


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.etherlab.org/pipermail/etherlab-users/attachments/20181117/49729d49/attachment-0002.htm>