[ntp:hackers] NTP Development Snapshot 4.2.5p163 Released

David Mills mills at udel.edu
Wed Apr 22 20:36:58 UTC 2009


Brian,

 From my point of view 164 is a good candidate. The mitigation rules, 
which were formerly an awesome weedpatch where every driver had a 
different failover scenario I hope is resolved. The PPSAPI code is in a 
common place accessable to any drive with practically zilch overburden 
and now works correctly with Solaris. Even Dave Hart's fixup of the 
Solaris_timepps.h has been tested. The documentation has been updated as 
well.

The pps_sample() routine, which is in fact in the release version of 
refclock_atom.c, is about 20 lines of straightforward code that has now 
and then nothing to do with the atom driver. The only driver that needs 
it is refclock_parse.c and that for unknown purpose. The best thing 
would be for Frank to toss it in the driver; it would be rude for 
somebody else to do that and maybe even dangerous.

Even if I put the code back in the atom driver, some poor soul might 
thin out the driver herd needing only the parse driver, but would be 
mightily surprised when the build failed to link.

Dave

Harlan Stenn wrote:

>Adding Dave Mills as a Cc.
>
>Brian Utterback wrote:
>  
>
>>I see that p164 has now been released. I am in the final approach to 
>>integrating NTP into OpenSolaris. I have been working with p161 and 
>>have been reluctant to update past that due to the pps_sample issues. 
>>Obviously I can't integrate something that cannot compile.
>>    
>>
>
>Agreed.
>
>  
>
>>So, question one, has this been resolved in p164? I know Dave Hart 
>>provided fixes to get the compile to work again, but do all of the 
>>refclocks work?
>>    
>>
>
>The problem is still there.  We are hoping that with Dave's latest
>PPS/PPSAPI reorganization it will be easy for Frank or somebody to get
>the parse refclocks working again.
>
>  
>
>>Question two. I know that Dave did some work on the PPS stuff. Does 
>>anybody have p164 running with PPS on a Solaris system, preferably 10 
>>or OpenSolaris. Perhaps something in the flock?
>>    
>>
>
>I don't know.
>
>  
>
>>I would really like to get the Orphan improvements before locking in, 
>>but I need to know if the sanity checking has already happened, at 
>>least on Solaris. I don't have a lot of leeway here.
>>    
>>
>
>I'm not sure and I am still working toward starting the 4.2.6 RC cycle
>within the next week.  It will happen as soon as I get the OK from Dave
>on including the disabled-by-default Samba patches.  There are a couple
>more bugfixes that are needed, but I am hoping they can be resolved
>quickly.
>
>H
>  
>



More information about the hackers mailing list