In PTP Redundant T-BC setup, when primary ptp4l instance is down,
os clock is expected to sync to secondary ptp instance. In contrary,
os clock is not converging and it is observed swinging offset.
This is caused by recent change where we increased PMC poll timeout
from 0 to 1000 ms, as 0 ms was too low and it was causing timeout
frequently on pmc polling while querying utc_offset, DDS and PDS.
Here we are changing the timeout back to closer to 0 ms. That value
has been determined 20 ms, by experimentation.
TEST PLAN:
- Deploy PTP Redundant T-BC with two PTP4l instances
- with both PTP4l instances started, there are no
"timeout reading pmc" in user.log
- stop primary ptp instance: systemctl stop ptp4l@ptp1
check HA switchover with:
echo -n "status" | socat -t 1 - UNIX-CONNECT:/var/run/phc-ha
check os_clock locked to secondary ptp instance
Story: 2011370
Task: 52157
Change-Id: Ib2872fafd496caed8009f0737822569ae915d762
Signed-off-by: Tara Nath Subedi <tara.subedi@windriver.com>
886a9a476b