October 22Oct 22 I got Orange PI RV2 many months earlier to experiment with RISCV. I installed initially Orange PI Ubuntu and recently Armbian, I was able to cross compile RV2 image using RPI5 which is great. To my big surprise the RV2 board ethernet show it is supporting hardware timestamping by looking at ethtool -T root@orangepirv2:/sys/class/net/eth0/queues# ethtool -T eth0 Time stamping parameters for eth0: Capabilities: hardware-transmit software-transmit hardware-receive software-receive software-system-clock hardware-raw-clock Hardware timestamp provider index: 0 Hardware timestamp provider qualifier: Precise (IEEE 1588 quality) Hardware Transmit Timestamp Modes: off on Hardware Receive Filter Modes: none all ptpv1-l4-event ptpv1-l4-sync ptpv1-l4-delay-req ptpv2-l4-event ptpv2-l4-sync ptpv2-l4-delay-req ptpv2-event ptpv2-sync ptpv2-delay-req However when trying to run ptp4l in either master or slave with "ptp4l -i eth0 -2 -m --tx_timestamp_timeout 100 -s" or "ptp4l -i eth0 -2 -m --tx_timestamp_timeout 100" ptp4l complained it cannot receive timestamp when in slave mode when SYNC is received, or in master when DELAY_REQ is received. ptp4l in master mode can send SYNC and receive timestamp for subsequent FOLLOWUP packet, so for tx RV2 ethernet can return the tx timestamp, but for receive it does not work. I check the datasheet of K1 ethernet and PTP section registers, nothing seem wrong. I just suspect maybe the filter for PTP packet to be timestamped maybe not working. root@orangepirv2:/sys/class/net/eth0/queues# /root/Downloads/linuxptp-code/ptp4l -i eth0 -2 -m --tx_timestamp_timeout 100 ptp4l[120447.510]: selected /dev/ptp0 as PTP clock ptp4l[120447.512]: port 1 (eth0): INITIALIZING to LISTENING on INIT_COMPLETE ptp4l[120447.513]: port 0 (/var/run/ptp/ptp4l): INITIALIZING to LISTENING on INIT_COMPLETE ptp4l[120447.513]: port 0 (/var/run/ptp/ptp4lro): INITIALIZING to LISTENING on INIT_COMPLETE ptp4l[120453.808]: port 1 (eth0): LISTENING to MASTER on ANNOUNCE_RECEIPT_TIMEOUT_EXPIRES ptp4l[120453.808]: selected local clock c0742b.fffe.fcd1e0 as best master ptp4l[120453.808]: port 1 (eth0): assuming the grand master role ptp4l[120492.717]: port 1 (eth0): received DELAY_REQ without timestamp ptp4l[120493.027]: port 1 (eth0): received DELAY_REQ without timestamp ptp4l[120494.187]: port 1 (eth0): received DELAY_REQ without timestamp ptp4l[120494.332]: port 1 (eth0): received DELAY_REQ without timestamp ptp4l[120495.579]: port 1 (eth0): received DELAY_REQ without timestamp ptp4l[120496.234]: port 1 (eth0): received DELAY_REQ without timestamp ptp4l[120496.659]: port 1 (eth0): received DELAY_REQ without timestamp ptp4l[120498.583]: port 1 (eth0): received DELAY_REQ without timestamp root@orangepirv2:/sys/class/net/eth0/queues# /root/Downloads/linuxptp-code/ptp4l -i eth0 -2 -m --tx_timestamp_timeout 100 -s ptp4l[120550.932]: selected /dev/ptp0 as PTP clock ptp4l[120550.934]: port 1 (eth0): INITIALIZING to LISTENING on INIT_COMPLETE ptp4l[120550.935]: port 0 (/var/run/ptp/ptp4l): INITIALIZING to LISTENING on INIT_COMPLETE ptp4l[120550.935]: port 0 (/var/run/ptp/ptp4lro): INITIALIZING to LISTENING on INIT_COMPLETE ptp4l[120550.996]: port 1 (eth0): received SYNC without timestamp ptp4l[120551.996]: port 1 (eth0): received SYNC without timestamp ptp4l[120551.997]: port 1 (eth0): new foreign master 2ccf67.fffe.1f909a-1 ptp4l[120552.996]: port 1 (eth0): received SYNC without timestamp ptp4l[120553.996]: port 1 (eth0): received SYNC without timestamp ptp4l[120554.996]: port 1 (eth0): received SYNC without timestamp ptp4l[120555.996]: port 1 (eth0): received SYNC without timestamp ptp4l[120555.997]: selected best master clock 2ccf67.fffe.1f909a ptp4l[120555.997]: port 1 (eth0): LISTENING to UNCALIBRATED on RS_SLAVE ptp4l[120556.996]: port 1 (eth0): received SYNC without timestamp ptp4l[120557.996]: port 1 (eth0): received SYNC without timestamp the driver is driver/net/ethernet/spacemit for Armbian, kernel 6.18 from Orangepi they use older kernel 6.6 and driver/net/ethernet/k1 the two drivers are identical with minor changes, the section for ptp handling looks identical. I hope to find kernel dev here who has tried ptp4l with the Spacemit ethernet with PTP support for help. I work a lot on PTP so anytime I see a board supporting hardware timestamping, I just want to try. So far only boards I know with hw timestamping is RPI5 and BeagleBoneBlack (even if only 100M). Have a good day! Chi-Thanh
Monday at 11:45 PM4 days Could you try the same L2 client test with software timestamps: ptp4l -i eth0 -2 -m -s -S? Run it instead of the hardware-timestamping instance on the test setup. If that works on the same peer and link, it narrows the problem to the hardware RX timestamp path. The ethtool -T list shows supported modes, so for a driver report it would also help to capture the RX filter the driver actually selected when ptp4l configured it. The missing SYNC and DELAY_REQ timestamps are on receive, so increasing tx_timestamp_timeout won’t address those messages.
Tuesday at 10:15 PM4 days Author I already tested without hardware timestamping on RV2, that case works, kind of expected since everything is software.I already narrow down the issue to be generating HW timestamp when receiving SYNC or DELAY_REQ.The tx_timestamp_timeout is needed when you are a master clock and you generate SYNC/FOLLOWUP, if there is no tx_timestamp_timeout like I putted, you would fail to receive the timestamp when you transmit the SYNC so problem to get hw timestamp is in both direction, I already found the fix for TX direction, now it is RX.I manually program the k1 mac register to enable the rx filter in the hardware but no luck, I hope Spacemit people can chime in if they had tested successfully that PTP feature.
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.