Re: how to emulate wrong data in shm to cheat ntpd

[Gary E. Miller]

Yo артур!

On Tue, 02 Jun 2020 23:01:54 +0300
артур хайруллин <ya_dinamovec@mail.ru> wrote:

> I am interested in opportunity to cheat ntpd by putting wrong data to
> shm like gpsd does.

Data is data.  gpsd passes on what it gets.  Garbage in, garbage out.

> The problem is that gps receivers i use have a strange habit to lie
> about the time periodically.

I would like to get a binary capture of an even like that.  just
use "gpsd -R > tmp.raw" to grab the raw data.

> Gpsd daemon communicates with ntpd
> daemon through SHM and provides data to adjust system clock. Wrong
> data taken from gps receivers cheat ntpd and my system for some time
> move forward or back on time chart for some seconds or couple of
> years.

This should not happen.  What ntpd are you using?  ntpd usually
wants at least 3 sources to agree on a time before using it.  Are
all three of your sources failing the same way?  Also, ntpd does
not make large changes after startup, so something is not working
as designed.


> I want to add to ntp.conf another ntp servers(internet) with low
> stratum to block that situation.

Stratum does not really matter.  Jitter matters.  You should always
have more than 3.

> But before i have to research the
> opportunity to cheat ntpd with gpsfake utility(i have long logs of
> gpspipe).

Hopefully raw logs, please send them.

> I made some fixes to python scripts to make them correct
> and now gpsfake emulates old data. I see it in cgps utility. 

Fixes for what?  gpsfake is used for the regression tests for
data going back 20 years.


> But gpsfake(as i saw yesterday) doesn't push this data to SHM.

Intentional.  Otherwise regression tests, and other tests, mess up the
clients.  If there is a way to send bad data to the clients then
many people would try it, and that would be bad.

> I want
> to know (before some code changes - if it will be necessary) if there
> is some mean to do that in gpsd project?

No way, currently, to get gpsfake to send data to SHM.  It might
be easier to just write a small SHM simulator.

RGDS
GARY
---------------------------------------------------------------------------
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97703
        gem@rellim.com  Tel:+1 541 382 8588

            Veritas liberabit vos. -- Quid est veritas?
    "If you can't measure it, you can't improve it." - Lord Kelvin

pgpqndRZDEV41.pgp
Description: OpenPGP digital signature