I managed to assemble a test fixture similar to the embedded system
I am evaluating deploying linuxptp on, as shown on the diagram below.
System clocks of the three main components are working in synchrony
(apparently) and a change on the system clock of the GM is propagated
after a few seconds to the other two members of the gang.
Next step is checking how well synchronized they are and how well sync
is kept in the long run.
My question is about the configuration of the SOM component. I thought
of configuring the interfaces as a BC, but could not make it work. In
one of the runs that succeeded (i.e., produced no errors), ptp4l running
on the AM complained that Master clock quality received is greater
than configured, ignoring master!". I searched the mailing list about
boundary clock but I could not find information on how to use BC; I am
under the impression that BC depends on hardware support (a centralized
clock the net interfaces use) for it to work.
So I ended up with the setup below, using one instance of ptp4l per
interface and four instances of phc2sys to synchronize system
clocks and interface.
My question is if I am overcomplicating it, IOW, if there is not a
simpler way to set the SOM, including using BC correctly, if
it is ever possible with a X86 based architecture.
FWIW I tried using ts2phc to synchronize the eth_am interface with
eth_head, but I could not make it work, so I ended up using
the second instance of phc2sys instead.
Sys clock (GM)
|
v
phc2sys phc2sys -> Sys clock SOM Sys clock AM
| ^ | ^
| | phc2sys |
| | | |
v | v |
enp3s0 eth_head eth_am eth0 (SW ts)
GM -----------------------> SOM <---------------------> AM
enp3s0 eth_head eth_am eth0
ptp4l(M) ptp4l(S) ptp4l(M) ptp4l (S/SW)
_______________________________________________
Linuxptp-users mailing list
Linuxptp-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linuxptp-users
