> From: Cliff Spradlin [mailto:csprad...@waymo.com] > Sent: Wednesday, August 1, 2018 9:55 AM > > The main goal of this series it to have one L3 VLAN interface and multiple > > L2 > ports in the VLAN (we have tested this on the switch with 64 ports). > > OK, thanks for the more detailed overview. It seems like it would be useful > for ptp4l to support this large-network use case. >
True, mainly running on a switch/router acting as boundary clock with many ports and VLAN domains. There is a patch missing to handle link up/down event on the VLAN netdev and support for IPv6 that we will share. Will be happy to get feedback about the current patch-set. > > We thought about using Linux namespaces also, but PTP4L does not > support them. > > Our solution was much easier to implement. > > I see, maybe the biggest problem with namespaces for ptp4l is its main poll > loop(), which could only wait for events on a single namespace. > > Would it be too complicated to run a separate ptp4l per-interface? > Each instance of ptp4l would run in its own network namespace (such as via > "ip netns"). We were looking for a solution not to handcraft the Eth/VLAN/IP headers and as part of that we were thinking that we can have a sub netdev in different namespace with the same IP address as the VLAN but this again broke as we didn't have a solution for the ARPs Regardless, we are also working to modify the event loop to work with namespaces (and / or lite VRF) to allow the boundary clock run in different L4 domains (and not only VLAN). What do you think about replacing the event loop with epoll? ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ Linuxptp-devel mailing list Linuxptp-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxptp-devel