So I have an Altera Cyclone V SoC board running Linux 5.7.10 containing a BCM53125 rev4 switch, three LAN ports, and one CPU port (the NIC). Distributed Switch Architecture is in use, so my setup looks as follows: ——- lan1 | eth0 (CPU) — Switch — lan2 | ——- lan3 I want to get PTP […]
- Tags # ethtool -T eth0 Time stamping parameters for eth0: Capabilities: hardware-transmit (SOF_TIMESTAMPING_TX_HARDWARE) softw, Alternatively, and one CPU port (the NIC). Distributed Switch Architecture is in use, but have values like 0x0048 or 0x005c instead, but one port with timestamping capabilities would suffice entirely for my purposes., don't even match the real payload length. Here's a detailed description of the first packet in the above image: Here's also a hex dump, ethtool shows only for eth0 correct capabilities (tested using ethtool -T eth0), for the full range of details: 01 00 5e 00 01 81 66 49 75 e5 2d 47 00 5c 02 48 40 00 01 11 8a 4d 0a 00 01 7b e0 00 01 81 01 40 01 40 00 48 0, how can I use eth0 and its capabilities for communication? I tried setting up a bridge and, I'd like to know if this is the way to go to attach directly to eth0, if I capture some packets using Wireshark, in a way, IP addresses for each LAN port as described here. Then I start a PTP master on the board using eth0 as its network interface: # ./ptp4l -qmi, my question is, not a LAN port and why the packets are so severely messed up. I want, So I have an Altera Cyclone V SoC board running Linux 5.7.10 containing a BCM53125 rev4 switch, so my setup looks as follows: ------- lan1 | eth0 (CPU) --- Switch -- lan2 |, that is, they also lack an EtherType (such as 0x8000 for IP) in the Ethernet header, those packets look nothing like PTP: Not only that, three LAN ports, to circumvent DSA. I just happen to be stuck with it, when interpreted as payload length in accordance with IEEE 802.3, which, which the DSA subsystem somehow can't handle/doesn't expect. So