Always disable the clock sanity check when the ntpshm servo is used,
because it doesn't know when or how is the clock adjusted. Disable also
the kernel leap option as the process controlling the clock is
responsible for setting the kernel flag.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Introduce a new function to inform the servo about upcoming leap second.
This is used when the kernel leap is disabled to allow the servo to
handle better the sudden 1 second step in the offset when the leap
second is inserted or deleted.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
After moving the leap second check away from update_pmc(), the time
stamp is used only to control the pmc update interval. Switch to
CLOCK_MONOTONIC to keep the interval stable when CLOCK_REALTIME is
stepped.
When synchronizing the system clock and the PTP UTC offset is valid and
traceable, set the TAI offset of the clock to have correct CLOCK_TAI
(which is implemented in the kernel as CLOCK_REALTIME + TAI offset).
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
This simplifies passing of pending leap seconds to the clocks and it
will also allow to apply leap second to other clocks than system clock
if needed in future.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
This is a simple servo that provides the NTP SHM reference clock. It
doesn't make any clock adjustments and it always returns with the
unlocked state. It writes all samples to the SHM segment and another
process (e.g. chronyd or ntpd) is needed to read the samples and
actually synchronize the clock. The SHM segment number is set to the PTP
domain number to allow multiple SHM reference clocks running at the same
time.
This is mainly useful when other time sources are available on the
system (e.g. NTP, hardware reference clocks, or other PTP domains)
and a fallback to/from PTP is needed.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
When grandmaster does not use PTP timescale but the previous one did, we
need to reset sync offset and leap flag.
Signed-off-by: Jiri Benc <jbenc@redhat.com>
By default, do not synchronize CLOCK_REALTIME. To do it, -r option is
needed. That will only consider CLOCK_REALTIME as the destination. To
consider it also as a possible time source, use -rr.
Signed-off-by: Jiri Benc <jbenc@redhat.com>
Recognize errors returned in MANAGEMENT_ERROR_STATUS TLV and return
a distinct value from run_pmc in case such error is received.
Signed-off-by: Jiri Benc <jbenc@redhat.com>
Add support for subscribing to events (run_pmc_subscribe) and receiving and
handling of received events (run_pmc_events).
Add initial support for port status changes.
Signed-off-by: Jiri Benc <jbenc@redhat.com>
Do not call clock_open to open a clock device but let clock_add do that and
return the newly created struct. Also, store the device (interface) name in
struct clock.
Signed-off-by: Jiri Benc <jbenc@redhat.com>
This just moves code around to have related functions together and forward
declaration at the beginning of the file. No code changes.
Signed-off-by: Jiri Benc <jbenc@redhat.com>
For now, only CLOCK_REALTIME can be UTC. This may stay this way forever but
now we have a clean separation between codepaths where CLOCK_REALTIME is
required and codepaths any UTC clock should take.
The main motiviation behind this change is removal of sync_offset_direction.
It has to be computed on the fly based on the source and destination when we
have multiple clocks supported and automatic following of ptp4l state
changes implemented.
Signed-off-by: Jiri Benc <jbenc@redhat.com>
Split members that apply to all synchronized clocks and members that apply
to an individual clock. Keep all clocks in a list, with a pointer to the
source clock. This will allow to support multiple clocks synchronization.
Signed-off-by: Jiri Benc <jbenc@redhat.com>
Split the generic (global) part of update_sync_offset and the part that
affects individual clocks. This is in preparation for phc2sys handling
synchronization of more clocks.
Signed-off-by: Jiri Benc <jbenc@redhat.com>
Make run_pmc usable for any kind of management message. Create wrappers for
waiting for ptp4l and for getting UTC offset.
Signed-off-by: Jiri Benc <jbenc@redhat.com>
Most PHC drivers implement stepping (ADJ_SETOFFSET) by reading the
clock, adjusting the value by the offset and writing it back. This is
not perfectly accurate and if the operation is slow (e.g. due to PCIe
latencies), the error can be in microseconds.
Increase the default first step threshold from 100 nanoseconds to 20
microseconds to step only when the initial offset is larger than
the error in the step.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
This servo uses linear regression to estimate current time and
frequency error. The number of points used in the regression is
variable (from 4 to 64 in powers of 2) and is selected by a long-term
statistic of the prediction error.
Future improvements could include tracking of sudden frequency changes
(e.g. due to temperature variations), better stability of the error
statistic when a large offset is corrected, options to set the speed of
the adaptation, minimum and maximum number of points, or an option to
prefer frequency accuracy over time accuracy.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
These settings will be useful for all implemented servos, so move them
to the common servo code to avoid duplication. The configuration options
are renamed, but the they can be still set by their old names.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Check the sanity of the synchronized clock by comparing its uncorrected
frequency with the system monotonic clock. When the measured frequency
offset is larger than the value of the sanity_freq_limit option (20% by
default), a warning message will be printed and the servo will be reset.
Setting the option to zero disables the check.
This is useful to detect when the clock is broken or adjusted by another
program.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Following command produces unexpected update rate.
# phc2sys -s eth0 -q -m -O0 -R1.0842021724855044e-19
This is caused by wrong validation of phc_interval related to
significant digits of double precision and phc_interval_tp.tv_sec
overflow.
To avoid these unwanted trouble, this patch limits lower bound of
phc_rate to 1e-9.
There is no profound meaning in the lower bound value. I just think
it's enough to actual use and it doesn't cause phc_interval_tp.tv_sec
overflow in not only 64bit environment but also 32bit environment.
Thereby, messy validation of phc_interval is no longer needed.
Signed-off-by: Ken ICHIKAWA <ichikawa.ken@jp.fujitsu.com>
This patch makes the original behavior of sending the
TLV values for GET actions with a length of zero.
Signed-off-by: Richard Cochran <richardcochran@gmail.com>
Current pi servo steps clock without any condition on start.
This patch adds a new servo option "configured_pi_f_offset". The option is similar
to configured_pi_offset but only affects in the first clock update. Therefore,
if this option is set as 0.0, we can prevent clock step on start.
The new servo option can be specified from phc2sys by using -F option.
This feature is usefull when we need to restart phc2sys without system
clock jump. Restarting phc2sys is needed to change its configuration.
changes since v2:
- manual page fix.
- also apply max_offset along with max_f_offset in servo step1.
- add a variable to check if first update is done.
changes since v1:(http://sourceforge.net/mailarchive/message.php?msg_id=31039874)
- remake as a new servo option.
Signed-off-by: Ken ICHIKAWA <ichikawa.ken@jp.fujitsu.com>
The default zero offset can lead to misalignment between system clocks or wrong
time to be broadcast to the domain. Therefore we require setting offset upon
invocation.
Signed-off-by: Libor Pechacek <lpechacek@suse.cz>
This patch updates phc2sys usage reporting to give a slightly better indication
of why the program was unable to run.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
This patch modifies phc2sys to enable the use of interface names in clock_open
rather than having to do that by hand. This enables cleaner use of the -s and -c
options as they can accept interface names. This also enables the user to set
the slave clock by network interface as well.
-v2-
* fix clock_open as it used device instead of phc_device in the final call to phc_open
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
[RC: use CLOCK_MONOTONIC as suggested on the list.]
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Signed-off-by: Richard Cochran <richardcochran@gmail.com>
PPS output from a PHC has to be enabled by PTP_ENABLE_PPS ioctl. Call
the ioctl when both PHC device and PPS device are specified and PPS is
supported by the PHC.
Signed-off-by: Jiri Benc <jbenc@redhat.com>
Reset the STA_UNSYNC flag and maxerror value with every frequency update
to let the kernel synchronize the RTC to the system clock (11 minute mode).
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
This patch adds help messages for -l, -m and -q options.
Also it swaps -h for -v because ptp4l's help is this order.
Signed-off-by: Ken ICHIKAWA <ichikawa.ken@jp.fujitsu.com>
Add kernel_leap option for ptp4l and -x option for phc2sys to disable
setting of the STA_INS/STA_DEL bit to slowly correct the one-second
offset by servo.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Update the currentUtcOffset and leap61/59 values at one minute
interval. When a leap second is detected, set the STA_INS/STA_DEL
bit for the system clock.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Modify the pmc to allow non-blocking operation. Run it on each clock
update to have the sync offset updated from currentUtcOffset with every
other call.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Add new options to ptp4l and phc2sys to print summary statistics of the
clock instead of the individual samples.
[ RC - Fix () function prototype with (void).
- Use unsigned for sample counter.
- Fix over-zealous line breaks. ]
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Signed-off-by: Richard Cochran <richardcochran@gmail.com>
If the delay is known, print it together with the offset and frequency.
Remove the time stamp from the output to fit it into 80 chars.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Change the label of the frequency offset in the clock messages printed
by ptp4l and phc2sys from "adj" to "freq" to indicate it's a frequency
value.
Also, modify clock_no_adjust() to print the frequency offset instead of
the ratio and use PRId64 instead of lld to print int64_t values.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Use pr_* functions to print messages and add -m, -q, -l options to allow
configuration of the printing level and where should be the messages sent.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
There really is no such state, but there probably should have been one.
In any case, we do have one just to make the code simpler, but this should
not appear in the management responses. This patch fixes the issue by
covering over our tracks before sending a response.
Signed-off-by: Richard Cochran <richardcochran@gmail.com>
Add a new option to wait for ptp4l to be in a synchronized state.
Periodically check PORT_DATA_SET and wait until there is a port in
SLAVE, MASTER or GRAND_MASTER state. Also, set the default
synchronization offset according to the currentUtcOffset value from
TIME_PROPERTIES_DATA_SET and the direction of the clock synchronization.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
The PPS time stamps are always made by the system clock, don't allow
running the PPS loop with other clocks.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
In the PPS loop, instead of setting the system clock from the PHC only
once on start, read PHC with each PPS sample and use the time stamp to
get the whole number of seconds in the offset. This will prevent phc2sys
from losing track of the system clock.
Also, check if the PPS is synchronized to the PHC.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
With non-PPS loops let the servo make the inital correction. Move the
code to the PPS loop and change it to use the sample filtering to reduce
the error in the initial correction.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Instead of always starting at zero frequency offset, read the currently
stored value on start and pass it to the servo. As the read may silently
fail and return zero, set the clock frequency back to the read value to
make sure it's always equal to the actual frequency of the clock.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
We use the PTP_SYS_OFFSET ioctl method in preference to doing
clock_gettime from user space, the assumption being that the kernel space
measurement will always be more accurate.
Signed-off-by: Richard Cochran <richardcochran@gmail.com>
this patch changes sk_interface_phc to sk_get_ts_info, by allowing the function
to store all the data returned by Ethtool's get_ts_info IOCTL in a struct. A new
struct "sk_ts_info" contains the same data as well as a field for specifying the
structure as valid (in order to support old kernels without the IOCTL). The
valid field should be set only when the IOCTL successfully populates the fields.
A follow-on patch will add new functionality possible because of these
changes. This patch only updates the programs which use the call to perform the
minimum they already do, using the new interface.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
This patch adds support for sk_interface_phc to lookup the /dev/ptpX device
using ethtool's get_ts_info ioctl
-v2-
* set ethdev to null initially
* rebased on top of phc2sys changes
-v3-
* fixed typo regarding slash
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
The main loop has become a bit messy, mixing two completely disjunct code
paths. This patch separates the two use cases for the sake of clarity.
Signed-off-by: Richard Cochran <richardcochran@gmail.com>
As some PHC hardware/drivers don't provide PPS, it may be useful to keep the
system clock synchronized to PHC via clock_gettime(). While the precision
of the clock is only in microsecond range, the error seems to be quite stable.
The -d parameter now can be omitted if -s is provided.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
As some PHC hardware/drivers don't provide PPS, it may be useful to keep the
system clock synchronized to PHC via clock_gettime(). While the precision
of the clock is only in microsecond range, the error seems to be quite stable.
The -d parameter now can be omitted if -s is provided.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Miroslav Lichvar
Jiri Benc
Ken ICHIKAWA
Richard Cochran
Libor Pechacek
Jacob Keller