The MAC latches a hardware receive timestamp later than the frame
reaches the wire, because of the PHY and the clock domain crossing.
This fixed delay is the ingressLatency port parameter of IEEE 1588 and
shows up as a constant phase error between the local and the master
clock.
Subtract the configured latency from every hardware receive timestamp
in ptp_getrxtime(), the single place where they enter the daemon, so
Sync, Delay_Resp and the peer delay messages are all corrected.
- Add CONFIG_NETUTILS_PTPD_INGRESS_LATENCY_NS (default 0, which applies
no compensation).
- Add the -I option to override it at run time.
- Add ingress_latency_ns to struct ptpd_config_s.
Software timestamps are not affected.
Assisted-by: Claude:claude-sonnet-5
Signed-off-by: Daniel P. Carvalho <danieloak@gmail.com>
A single drift-rate sample computed between two consecutive sync
updates was clamped against CLOCK_ADJTIME_SLEWLIMIT_PPM - the
hardware's slew-rate safety limit, not a bound on how large a real
crystal-oscillator drift measurement can plausibly be. An abnormally
short or long measurement interval (e.g. right after a clock
source outage/reconnect, or a burst of closely spaced sync packets
following packet loss) could therefore produce a wildly implausible
sample that still passed the check and corrupted the long-term
drift_ppb average.
Add CONFIG_NETUTILS_PTPD_MAX_DRIFT_PPB (default 500000, well above
any real crystal's few-hundred-ppm drift) as a dedicated plausibility
bound, intentionally much tighter than CLOCK_ADJTIME_SLEWLIMIT_PPM.
A sample outside this bound is discarded and the previous averaged
drift_ppb is kept unchanged instead of being corrupted.
Assisted-by: Claude:claude-sonnet-5
Signed-off-by: Daniel P. Carvalho <danieloak@gmail.com>
Replace the sigqueue + shared-memory IPC mechanism in ptpd_status()
with a file-based approach:
- Daemon side: on SIGUSR1, write a binary ptpd_status_s struct to
a temp file and atomically rename it to the status path.
- Client side: send kill(SIGUSR1), poll for the file to appear,
then read the struct back.
This removes the CONFIG_BUILD_FLAT restriction (the old code returned
-ENOTSUP for Protected and Kernel builds) and avoids passing pointers
across address spaces via sigqueue. The status file path is
configurable via NETUTILS_PTPD_STATUSFILE (default /tmp/ptpd.status).
The atomic temp + rename pattern ensures readers never see a partial
write.
Signed-off-by: wenquan1 <wenquan1@xiaomi.com>
fix compile warning when only enable PTPD_CLIENT or PTPD_SERVER
ptpd.c:493:38: error: 'CONFIG_NETUTILS_PTPD_PRIORITY1' undeclared (first
use in this function); did you mean 'CONFIG_NETUTILS_PTPD_CLIENT'?
ptpd.c:494:39: error: 'CONFIG_NETUTILS_PTPD_CLASS' undeclared (first use
in this function); did you mean 'CONFIG_NETUTILS_PTPD_CLIENT'?
ptpd.c:495:39: error: 'CONFIG_NETUTILS_PTPD_ACCURACY' undeclared (first
use in this function); did you mean 'CONFIG_NETUTILS_PTPD_DEBUG'?
ptpd.c:498:38: error: 'CONFIG_NETUTILS_PTPD_PRIORITY2' undeclared (first
use in this function); did you mean 'CONFIG_NETUTILS_PTPD_CLIENT'?
ptpd.c:502:36: error: 'CONFIG_NETUTILS_PTPD_CLOCKSOURCE' undeclared
(first use in this function); did you mean
'CONFIG_NETUTILS_PTPD_STACKSIZE'?
Signed-off-by: dongjiuzhu1 <dongjiuzhu1@xiaomi.com>
Fixed bug where negative offsets were not properly handled because
clock_timespec_subtract clamps values to zero.
Implement support for SO_TIMESTAMP to get accurate packet
reception timestamp.
Implemented delay requests for measuring packet transfer delay.
Implemented clock drift estimation to bring the clocks closer to
sync and to filter out measurement jitter.
If multicast PTP packets are not being received, rejoin the multicast group.
This automatically recovers from situations such as rebooting a network switch.
What works:
- Basic server & client operation
- Transmission and reception of announce, sync and follow-up
Still missing:
- SO_TIMINGS for getting more precise packet timestamps
- Implementation of delay_req and delay_resp packets
- Status and stop interfaces for the daemon
