devel@ntpsec.org said:
> There are no panaceas, but enabling this feature in Gitlab means that we can
> use it when we're ready for it.
I'm missing the big picture. I don't use merge requests. I just do a git
push.
What are the advantagesof a merge request if the submitter has write access?
> This was discussed on IRC and I wanted to make sure everyone was aware.
Was there anything interesting in the IRC discussion?
--
These are my opinions. I hate spam.
___
devel mailing list
devel@ntpsec.org
> I've fixed the pipelines. ...
My pipeline retries are failing. Have they remembered the old setup recipe?
--
These are my opinions. I hate spam.
___
devel mailing list
devel@ntpsec.org
http://lists.ntpsec.org/mailman/listinfo/devel
> ... now we need to explicitly install the netbase package...?
Should that be added to buildprep?
Should we be using buildprep to setup the build environment?
--
These are my opinions. I hate spam.
___
devel mailing list
devel@ntpsec.org
I just pushed a simple fix. GitLab tells me:
Subject: ntpsec | Pipeline #21417003 has failed for master | 7f1a350a
Anybody recognize this?
It works for me on Fedora 27. The tests that fail are clang-basic and
clang-refclocks and the same pair again and python-coverage.
The previous
Fedora 28 is out. It's using gcc 8 which has tightened some type checking.
Things like this will become much more common.
../../tests/ntpd/leapsec.c:434:19: warning: cast between incompatible
function types from âint (*)(FILE * restrict, const char * restrict,
...)â {aka âint
> The ANSI C function you look for is strncat, I'd think.
Actually, strncat checks for bogus source. We have constants for the source.
We want to check for overflowing the destination. Fortunately, we have
strlcat, even if the environment doesn't provide it.
Thanks for pointing me in the
This is from gcc:
gcc (GCC) 8.0.1 20180324 (Red Hat 8.0.1-0.20)
Which is shipping with Fedora 28 which is in beta. (I'm testing on a Pi 3.)
../../ntpd/ntp_loopfilter.c:366:13: warning: passing argument 1 to
restrict-qualified parameter aliases with argument 4 [-Wrestrict]
It's working now. The problem was seccomp with various help from operator
errors to confuse debugging.
--
These are my opinions. I hate spam.
___
devel mailing list
devel@ntpsec.org
http://lists.ntpsec.org/mailman/listinfo/devel
We've had a couple of bug reports involving musl. One recent. I think they
were on Alpine Linux, so I setup a box to see what would happen.
I've got things setup so it builds, but it doesn't run. Is anybody actually
running ntpsec on Alpine Linux?
Anybody recognize this?
Note that there
I made a fresh clone, fixed my .git/config, made the edit, commit-ed it, and
tried to push. It didn't work.
My "fixup" to .git/config ends up with:
[remote "origin"]
url = g...@gitlab.com:NTPsec/ntpsec.git
fetch = +refs/heads/*:refs/remotes/origin/*
push =
> You are not on master branch?
But git status said:
> On branch master
> What does this say:
> # git branch -la
* master
remotes/origin/HEAD -> origin/master
remotes/origin/master
remotes/origin/proto-refactor
remotes/origin/refgenfix
--
These are my opinions. I hate spam.
Anybody recognize this one"
[murray@hgm raw]$ git push
Everything up-to-date
[murray@hgm raw]$ git status
On branch master
Your branch is ahead of 'origin/master' by 1 commit.
(use "git push" to publish your local commits)
--
These are my opinions. I hate spam.
Gary said:
> Intesting, but now I gotta install a javascript interpreter. Yet another
> language and API to deal with. This will take a while...
Your web browser is probably already set up to handle javascript. I don't
know if you can use that to solve your problem, but it might be worth a
rlaa...@wiktel.com said:
> This should make notes push by default:
> git config --add remote.origin.push \
> '+refs/notes/commits:refs/notes/commits'
> Since this is the first set of notes, push them: git push
That said:
Counting objects: 3, done.
Delta compression using up to 4 threads.
rlaa...@wiktel.com said:
>> Are there any alternatives?
> See `git notes` to attach notes to objects (e.g.
> commits) after the fact without changing the object itself.
> Using the example from the man page, try something like this:
> git notes add -m 'The 674 should be 474.' c8c888f8
Thanks.
e...@thyrsus.com said:
> If it weren't in the public repo this would be easy - I'm enclosing a script
> that does this.
> Since it is, I'd have to edit it here, then unprotect the public master
> branch, then force-push, then reprotect it. This will take the public repo
> temporarily out of
The 674 should be 474.
commit c8c888f8f2ef295c0ea5f854127069b3812e4c09
Author: Hal Murray <mur...@shuksan.example.com>
Date: Fri Mar 23 01:11:17 2018 -0700
Add logging of year(s) on large clock steps. Issue #674
--
These are my opinions. I hat
Gary said:
> The raspberry pi has no RTC. When it starts cold, the time may well be in
> 1969. Somehow, not sure how, that becomes 2134. Then gpsd uses that as the
> GPS epoch, and things go downhill from there...
So gpsd does something stupid and you expect ntpd to figure out how to
> I want my clock not stuck in 2134. -g alone does not fix that.
Back to the beginning. Where does 2134 come from?
> 'tos minsane 3' fixes it, unless I'm offline, which is pretty common for
> RasPi.
If you are getting bogus time from a local source and you want to work
offline, you are
>> I'm not sure the case you are describing.
> Hangs, as in unresponsive, frozen. ntpmon fails to update its output.
Is that a problem in ntpd or ntpmon? Try ntpq -p from another window and/or
tail rawstats to see if it is getting any answers.
> Seems to me any daemon that crashes itself on
Gary said:
> I guess I did not wait long enough. ntpd, and ntpmon, hang for about three
> minutes, then wake up fine. Hang is at: 2134-03-01T17:57:35
> How do we explain that?
I'm not sure the case you are describing.
ntpd waits a while to be sure before stepping the clock. In general,
Gary said:
>> It's a flag, not a variable. -g sets it.
> Yeah, and most daemons let you put flags in the config file.
I'd be happy with a policy that said anything you can do on the command line
you can also do in the config file, but it doesn't seem like a big deal.
[tinker doc for panic and
Gary said:
> Also no way to put panicgate in the ntp.conf. At least the man page does
> not say how.
It's a flag, not a variable. -g sets it.
There are several tinker variables in this area. Details in
docs/includes/misc-options.txt
+tinker+ [+allan+ _allan_ | +dispersion+ _dispersion_ |
> It just marks ALL chimers and refclocks as invalid. Then hangs.
Is it really hanging? This usually indicates success:
03-22T17:39:49 ntpd[10392]: PROTO: 204.17.205.30 901a 8a sys_peer
--
These are my opinions. I hate spam.
___
devel mailing
Gary said:
>> Was there a successful jump in the log file?
> You tell me, below is all I have for one test cycle. I have many similar
> test cycles logs.
This looks like a successful big jump to me.
12-31T16:00:24 ntpd[1931]: PROTO: 0.0.0.0 c41c 0c clock_step
+902266367.623732 s
Eric said:
> I've never tried do describe that kind of testing because it's not easy to
> tell people without prior experience running the clients what a success/
> failure indication looks like. Of course alarm bells would go off on a
> crash, but the most definite thing I could say otherwise
> I'm thinking there were two jumps in the first little bit. Maybe 60
> seconds?
There should be something in the log file.
> Do you have a GPS clock that is prefer? Did you set the time to 2134,
> restart gpsd so it is in the wrong epoch? Then start ntpd? very repeatable
> for me. Even
e...@thyrsus.com said:
> Alas, the client tools are difficult ebnough to test-jig that I have to
> hand-test under Python 3 before releases. I have a routine for this and I
> assume Ian does as well.
How much testing do you do by hand? Is that written down anyplace? Should
it be on the
> Here is the log:
> 12-21T17:28:40 ntpd[6085]: PROTO: 0.0.0.0 0017 07 panic_stop +579076096 s;
> set clock manually within 1000 s.
> 12-21T17:28:40 ntpd[6085]: CLOCK: Panic: offset too big: 579076096.001
You only get one big jump with -g. Was there another jump earlier?
I looked at the code,
>From packaging/packaging.txt
> The shebang lines in our Python scripts point to
> "python". Part of our standard tests check that
> you can change that to "python3" without breaking
> anything.
What sort of testing is that? How do I run it on my setup?
--
These are my opinions. I hate
>> About time the NTP told the user what time is being served?
>Yes.
> It should got into an extension block.
It seems like a wild goose chase to me. We should focus on distributing UTC.
> It should got into an extension block.
> Two, one for time base, one for PPS source. ...
The PPS idea is
Gary said:
> Currently I am playing with a Javad GPS that lets you set your PPS to GPS,
> SBAS, GLONASS, GALILEO, BeiDou, QZSS, IRNSS, UTC(USNO), UTC(SU), UTC(SBAS),
> UTC(GALILEO), UTC(BeiDou), UTC(QZZSS) or UTC(IRNSS)
What is the theoretical spread on all those different flavors of time? How
I didn't see an announcement. In case you didn't notice...
commit 0b2beb1ffbebed5bd6d9ef615aac905df7e2f220 (tag: NTPsec_1_1_0)
Author: Mark Atwood
Date: Thu Mar 15 04:11:41 2018 +
version 1.1.0
Signed-off-by: Mark Atwood
--
Thanks for testing this area.
Do you have a HOWTO type checklist that I can use to try to reproduce this
problem?
Were there any errors/warnings, possibly a long time back early in startup?
The hardpps option uses code that is not normally included in the kernel.
What kernel/distro are you
I'm putting together a new system.
ntpq dies with:
ntpq: can't find Python NTP library -- check PYTHONPATH.
No module named util
Note is says "util", not "ntp.util". I've got ntpsec.pth and it works well
enough to fix the initial can't find error.
The build directory contains util.py and
INSTALL says:
You can browse a summary of differences from legacy NTP here:
https://docs.ntpsec.org/latest/ntpsec.html
How does that compare to NEWS? Does the web version get updated with each
release? ...
--
These are my opinions. I hate spam.
> Where do you want to document it? Is "./waf configure --help" sufficient?
I don't have a good suggestion. It's the sort of detail that belongs in an
overview section, but nobody ever reads those.
Do we have an overview section?
> (packaging/packaging.txt has several typos related to
>> How am I supposed to figure out that the format is
>> --build-epoch=$date
>> rather than
>> --build-epoch $date
> Both forms work.
Thanks.
> Do we suggest that only 1 form works somewhere?
./waf configure --help shows the = version
I don't know of any place that says anything about
> After a bit more, incoclusive, chat with the user, it seems that he somehow
> started root with a PATH that did not include /usr/sbin where scons lives.
Why would scons go in /usr/sbin?
Is that a typo or a strange setup?
--
These are my opinions. I hate spam.
How am I supposed to figure out that the format is
--build-epoch=$date
rather than
--build-epoch $date
None of the examples show the equal sign.
--
These are my opinions. I hate spam.
___
devel mailing list
devel@ntpsec.org
devel@ntpsec.org said:
> What kind of special labeling does ntpsnmpd require? Is putting
> "experimental" in the documentation sufficient? Does it need to give a
> warning on launch?
I'd put "experimental" in NEWS.
If you put something in the documentation it should probably say a bit about
devel@ntpsec.org said:
> Why wouldn't we require a certain openssl version as there are a number of
> security vulnerabilities in (older) openssl?
Do you have a pointer to a list of the insecure versions with a summary of
the bug so we can see if we use that feature?
--
These are my
Is something interesting going on here, or is this just leftover cruft?
Why would ntp_adjtime turn into SIGSYS? Would that be mentioned in a man
page? Do we care about running on those systems?
/*
* Use sigsetjmp() to save state and then call ntp_adjtime(); if
* it
> Do you have the truncate fix in?
Apologies for not sending a specific announcement.
Yes.
commit b01f1d658b11c4e8c24b307a7a79e8307364fbc2
Author: Hal Murray <mur...@shuksan.example.com>
Date: Fri Mar 2 00:38:49 2018 -0800
Truncate digests longer than 20 bytes.
--
The
Eric said:
>> I could imagine that we have tweaked mode6 enough to be interesting.
> There's really only one possible point of breakage - driver IDs for
> reclocks. I think we're safe there.
It wouldn't surprise me if we had added something interesting. I'm pretty
sure I have added things.
I think we should sort it out for the release.
--
These are my opinions. I hate spam.
___
devel mailing list
devel@ntpsec.org
http://lists.ntpsec.org/mailman/listinfo/devel
fallenpega...@gmail.com said:
> If Hal isn't happy, I'm not happy. I'll hold the release until this gets
> unsnarled. ..m
It will take a day or two to fix the truncate case. Maybe tonight.
It will take a week or so to add CMAC support. Waiting for that seems like a
good idea. It will
[truncate long digests]
> Bletch. No, we don't.
Except that others are already doing it, so I guess we should do it too.
I'll add a warning to the code that reads in keys.
--
These are my opinions. I hate spam.
___
devel mailing list
> I see no real blockers. We've got a bunch of little nits and documentation
> issues. I might try to push a fix for #446.
>From n...@ietf.org
> Please note that latest versions of ntp truncate long digests in MACs to 160
> bits, so the authentication should work with any hash function
devel@ntpsec.org said:
> I see no real blockers. We've got a bunch of little nits and documentation
> issues. I might try to push a fix for #446.
There is no problem unless you setup your keys file to use an algorithm with
a big digest.
The short term clean fix is to reject algorithms with
> Are we comfortable with the 1.0.1 release on March 3rd?
I'm not. My attempts at fixing #461 aren't working.
I think it should be simple. I think I understand what the problem is, but I
don't understand why my attempts at fixing it aren't working.
The root of the problem is this (from the
> Yeah, it pushes the web pages.
Should I just ignore it?
Or tell somebody? If so, who? What would they do?
Any idea why I'm getting pairs of messages?
--
These are my opinions. I hate spam.
___
devel mailing list
devel@ntpsec.org
Again, it's the second of a pair of messages. The first one said it
succeeded.
Subject: ntpsec | Pipeline #18114131 has failed for master | 8c34d988
...
Pipeline #18114131 ( https://gitlab.com/NTPsec/ntpsec/pipelines/18114131 )
triggered by Hal Murray ( https://gitlab.com/hal.murray )
had 1
#18075609 has failed for master | eef92d62
Pipeline #18075609 ( https://gitlab.com/NTPsec/ntpsec/pipelines/18075609 )
triggered by Hal Murray ( https://gitlab.com/hal.murray )
had 1 failed build.
Job #54445477
Stage: deploy
Name: pages:deploy
--
These are my opinions. I hate spam
If we are serious about security, it seems like a good tool to add to the
collection.
Is there a HOWTO set it up? I didn't find one for NTP. Are all jaildirs
sufficiently similar that a global HOWTO is all one needs to set one up for
NTP? I found one for FreeBSD, but not much for Linux.
e...@thyrsus.com said:
> Gary and Mark asked me for something like such an Apache-like feature
> because it makes life easier for configuration-assistant software and distro
> packages.
Is that true in our case? It seems like it might be, but our config file is
generally simple enough that a
> You need to be running an SNMP daemon and an NTP daemon.
I've got plenty of ntp servers to experiment with.
>> Is there a HOWTO that tells me how to set things up?
> I'll get to work on that.
There may be two targets for that document. One is SNMP wizards who don't
know much about ntpd.
>> Do we have any documentation describing a use case?
> Not exactly, but look at the snippets under etc/ntp.d/.
All those examples don't use the auto-grab feature. Most of the file names
don't end with .conf, and the one that does is a sample ntp.conf that could
be better placed up a level.
Do we have any documentation describing a use case?
--
These are my opinions. I hate spam.
___
devel mailing list
devel@ntpsec.org
http://lists.ntpsec.org/mailman/listinfo/devel
devel@ntpsec.org said:
> The only real blocker that I can see at this time is the need for broad
> testing. [reiteration of me requesting testers / reviewers goes here.]
Is there a HOWTO that tells me how to set things up?
Actually, I need something before that. Why is it interesting? What
Maybe build for ARM on an Intel box.
Is there a diff for compiled files? (with a flag to skip time stamps or
whatever would change if I just rebuilt with the same environment)
--
These are my opinions. I hate spam.
___
devel mailing list
> Well, on Gentoo, its mix and match as you wish.
> You've got clang and gcc for C compilers
> Mix with glibc, musl, uclibc and uclibc-ng for your C library.
> On amd64 you'll find gcc and glibc almost all the time.
> On arm you'll find the same, but musl gaining market share.
Thanks.
Should we
If different, the test-space of seccomp gets (much?) bigger.
--
These are my opinions. I hate spam.
___
devel mailing list
devel@ntpsec.org
http://lists.ntpsec.org/mailman/listinfo/devel
rlaa...@wiktel.com said:
> If you're going to move to time-based, you might consider quarterly
> releases?
I'd be happy with quarterly releases.
The next question is how seriously do we take the release date? I think
there are two approaches. The first is to try hard to release as scheduled.
> I've been looking at the code around mode 6 generation and discovered that
> in some areas it's still globals all the way down. Translating these
> globals will make future refactoring/translating easier.
I'm missing the big idea.
The current case is that we have a lot of global variables.
hmur...@megapathdsl.net said:
> You can get some interesting data if you kick up the memory for the MRU
> list.
> A US/NA server needs 325K to hold everything for a bit over a day.
Argh. I forgot a key piece of info.
That's on a box signed up for 100 megabits of IPv4 and IPv6.
--
These
> Two of my servers are in the NTP pool...
You can get some interesting data if you kick up the memory for the MRU list.
A US/NA server needs 325K to hold everything for a bit over a day.
Something like:
# 88200 = 86400 + 30*60
mru initmem 10 maxmem 50 maxage 88200 minage 3600
Thanks for the input.
> I'm a big fan of "always stable master" and time based releases.
I'd be happy with that. What sort of interval did you have in mind for "time
based"?
Our master is generally pretty stable, but we don't have a solid test setup.
We can tell if it builds, but that
devel@ntpsec.org said:
> So, I'm declaring an intention for the 1.0.1 release the weekend after next,
> about March 3rd.
Could you please say a bit more about how you picked that date?
I would expect either:
as soon as we finish feature X, or
as soon as we stop fixing minor things (like
There are two projects I've had my eye on for a while.
The first is to remove the input buffer queue. That's leftover from before
kernels supported time stamps on received network packets. (ntpd used to
grab the packets from an IO signal handler)
The other is to remove the table lookup in
devel@ntpsec.org said:
> Your first 'graph is something I didn't know. I think it removes the
> pressure to keep this feature. Go ahead and take it out, Hal.
OK. I'll take it out of the parser in time for the release. Cleaning up the
internals should wait until after the release.
--
> The big deal is whether we have closure on the Python installation mess.
The only loose end that I know about is PYTHONDIR vs PYTHONARCHDIR.
We now understand why what we have been expecting doesn't work.
We are trying to import ntp.ntpc. That's a two step process. First it looks
up ntp,
Is anybody using/testing it?
We don't support receiving broadcast.
It used to support a ttl option. That got broken/dropped somewhere along the
way. Should I restore that? Or maybe document that it is missing? ...
Context is that I'm cleaning up the mode/ttl mess. The mode for refclocks
> Poking around, I see that some of my servers are not responding.
Things are happy again.
I think I've figured it out, but I don't have a smoking gun, a clean fix, or
a reproducible test case.
I think the problem was that I have servers setup to depend on each other.
Several have good
because it gets better data over the net.
(Yes, the Ethernet is also USB, but it's faster USB.)
remote refid st t when poll reach delay offset
jitter
==
=
+192.168.1.3 192.168.1.33 2 u
Thanks for passing that on.
That code is long gone.
We only have one extra thread for doing DNS lookups. If you configure with
--disable-dns-lookup, it could build without threads.
We don't currently do anything about the stack size. Is that interesting? I
don't know of any usage in
What do people recommend for a GUI package to use with python?
I want to plot a graph of something and update it in real time by scrolling
all the old data to make room for the new samples as they arrive.
Google suggests that matpythonlib may be the right starting point, but I
haven't pulled
If I run waf build after a clean build, It does this:
...
'build' finished successfully (0.451s)
...
--- building host ---
Waf: Entering directory `/home/murray/ntpsec/play/hgm/host'
Waf: Leaving directory `/home/murray/ntpsec/play/hgm/host'
--- building main ---
Waf: Entering directory
Poking around, I see that some of my servers are not responding. (I test
various combinations and/or collect data with noselect.)
Has anybody noticed anything similar? Are you running the latest bits? ...
I haven't seen any problems recently so I assume it's due to a recent change.
All my
>> Config in /etc/ntpsntp.conf
> /etc/ntpsntpd.conf server, not client.
Interesting.. Thanks.
ntpd is both client and server. We don't use the "d" on config files.
Is there a client for ntpsnmpd? I'd expect the client side to be part of a
snmp package. Do we need samples for that? How
devel@ntpsec.org said:
> You know our users do not read man pages! Can you provide a script, or at
> least a detailed procedure?
Sure. If you look back in the message that started this thread there are
snippets of code.
The initial message was asking if there was any interest. (Or
devel@ntpsec.org said:
> There are two sets of configuration data: the usual server and logging
> settings, and the flags which control which notifications are active.
I'd put stuff in parallel with ntp/ntpd
Config in /etc/ntpsntp.conf
State to be saved over boot in /var/ntpsntp/
Cleaned over
> Doesn't ntpd need to be started as root to set that?
> But how does ntpd set its caps before it starts?
man 8 setcap
You set them on your ntpd when you mark it setuid as part of the install
process.
The capabilities on the file get OR-ed in to whatever they inherit from the
starting user.
>> Yes, please. I see no reason why ntpd should start up as root these
>> days.
> It needs to be able to read /dev/pps*, SHM(0) and SHM(1)
You don't need root for /dev/whatever if you set the owner to ntp:ntp before
starting ntpd.
Linux has split the root-does-everything permissions to
Should be fixed now.
Is anybody other than me testing the authentication stuff?
Poke me off list if you want to set something up.
--
These are my opinions. I hate spam.
___
devel mailing list
devel@ntpsec.org
Rats/sorry. I broke it last night.
Usually I test things better than this.
Should be fixed soon.
--
These are my opinions. I hate spam.
___
devel mailing list
devel@ntpsec.org
http://lists.ntpsec.org/mailman/listinfo/devel
symetricactive(1),
symetricpassive(2),
client(3),
server(4),
broadcastserver(5),
broadcastclient(6)
Those are close to the "mode" field in the packet.
client and server are mode 3+4.
symetricactive and symetricpassive are "peer, modes 1+2.
We don't send symetricactive mode any
devel@ntpsec.org said:
> From the possible values of ntpEntStatPktMode it would appear that the
> "modes" this table is talking about are not the normal NTP communication
> modes like mode6.
What are the possibilies?
--
These are my opinions. I hate spam.
I've been running on Linux with ntpd starting as non-root with reduced
capabilities. Do we want to merge this in?
It's not a big deal, but one more small step in the right direction. The
biggest disadvantage I can see is the increased complexity in the startup
scripts.
It will take a lot
rlaa...@wiktel.com said:
>> What's the right fix?
> As an individual, choose one of these:
...
Thanks.
It's just the same old mess that I didn't recognize because of the /usr/pkg/
rather than the familiar /usr/local/ slightly complicated by a version that
got installed someplace on the
Python on NetBSD doesn't search /usr/local/lib/
It looks in /usr/pkg/...
-bash-4.4$ python
Python 2.7.14 (default, Oct 15 2017, 00:57:13)
[GCC 4.8.4] on netbsd7
Type "help", "copyright", "credits" or "license" for more information.
>>> import sys
>>> print sys.path
['',
> I've just tried that again with both the uBlox-6 and uBlox-8 and while I can
> attach the PPS line discipline to the interface, no PPS ever gets generated
> on that device. If they can actually do that, I'd be interested in how to
> enable that capability.
You can build your own with PPS by
> PPS line discipline via USB virtual serial works and produces the expected
> ~1ms offsets due to the USB poll interval. It should be possible to remove
> that shift by doing a loopback measurement via RTS/CTS eventually. Jitter
> is roughly on par with the local stratum-1 over network.
The
> But maybe I don't understand what you're after.
I'm interested in learning whatever we can about the timing in the kernel PPS
area.
There is another approach that might be interesting. Use a loopback on some
modem control signals or gpio pins. Then a test program can grab the time,
devel@ntpsec.org said:
> As far as I understand the sources, setting up an echo allows a PPS client
> to register a callback function with the PPS device to be called when a PPS
> event happens. The default echo function simply outputs "echo: assert" or
> "echo: clear". I guess one could indeed
dropkic...@gmail.com said:
> I have 3 ntp servers on a subnet. I would like to aggregate stat files for
> each server to another server for the purpose of centralized analysis each
> ntp server would have its own directory for stat files. I'm currently using
> a python script that scp files
devel@ntpsec.org said:
> Well, maybe if I can think about a use of that measurement I might do it,
> but _creating_ a PPS from the rasPi and then measuring it externally with a
> TIC of sufficient resolution would be much more useful than teasing out this
> or that internal delay that is going to
>> I'd expect another step if the timing difference between
>> pulses is such that it changes from 1 interrupt to 2. The two
>> interrupt case will add the time to return from an interrupt
>> and take the second one. Maybe a one-shot with a
>> knob so you can adjust the delay and plot the
> For starters, the two PPS pulses should be close enough together to trigger
> a back-to-back queued interrupt, so the second will have to wait for the
> first handler to complete.
You might learn something by connecting the same PPS signal to two pins and
comparing the time stamps. The
1201 - 1300 of 2329 matches
Mail list logo