[ntp:commitlogs] incoming pull in whimsy.udel.edu:/backroom/ntp-stable

Harlan Stenn stenn at whimsy.udel.edu
Fri Jan 9 23:30:13 PST 2004


VAR       CLIENT              SERVER              
===       ======              ======              
USER      stenn               stenn               
HOST      whimsy.udel.edu     pogo.udel.edu       
ROOT      /backroom/ntp-stable/pogo/users/stenn/ntp-stable
LEVEL     1                   1                   
TIME_T    1042609353          1042609353          
UTC       20030115054233      20030115054233      
VERSION   bk-3.0.1            bk-3.0.1            

stenn at whimsy.udel.edu fired the post-incoming--010commitlogs trigger in /backroom/ntp-stable
    Received the following changesets
    
    ChangeSet
      1.1186 04/01/10 02:27:07 stenn at pogo.udel.edu +2 -0
      Fixes from Reg Clemens:
      
      Cleanup.
      [Bug 233]
      [Bug 246]
      [Bug 260]
    
      ntpd/refclock_oncore.c
        1.47 04/01/10 02:27:01 stenn at pogo.udel.edu +104 -137
        Fixes from Reg Clemens:
        
        some (minor) changes:
           (a) some changes to messages printed.
           (b) can now find config file in /etc/ntp/ as well as /etc/
        
        Bug 233: New LP64 Patches
        patch imported from original author
        
        Bug 246: 3 issues with the ONCORE driver
        (1) You can reset the ONCORE driver back to defaults by *reading*
            the variables from across the net.
        
        This bug was introduced when I (foolishly) 'fixed' the code so it
        would understand the FLAG variables that corresponded to its own
        input variables from the ntp.ONCORE.x file.  Since I do NOT understand
        all the implications of ntp_control, I will remove this 'fix'.
        The variables can now only be set in ntp.ONCORE.  They can NOT be
        changed dynamically.  The documentation will be changed back.  Perhaps
        at some time in the future I will dig through the ntp code to
        understand ntp_control.
        
        (2) Fudge acted upon AFTER ntp.ONCORE.
        
        Fixed by (1) the FUDGE flag/time variables are ignored, use the
        ntp.ONCORE file.
        
        (3) In the driver, ONCORE_start and a few other procedures do not
            produce any output in the clockstats file.
        
        The originator of the bug is running FreeBSD 5.1.
        I have looked at the output from MY copy of FreeBSD 5.1, and several
        Linux versions (7.3 -> 9), and Fedora and do NOT see any missing
        messages.  The messages start with an "ONCORE DRIVER -- CONFIGURING"
        message written when the driver is first entered, and contain all the
        messages I would expect after that.
        
        I have no idea why he is not seeing the messages.
        
        No action.
        
        Bug 260, Patches for item (1) in the above.
        
        The patch purports to fix the problem in Bug 246.
        
        As noted I don't understand the details of what is going on behind the
        curtain, so I would rather remove the offending code, than try to fix it.
        
        There is also an increase buffer size so that longer messages can be
        sent to the clockstats file.  Ill leave that for the user to do on his
        own if he wants to change the messages, but with slightly more effort
        he could do the same using SHMEM.
        
        He found an error in determining the ONCORE M12/M12+T. (thanks)
        Rather than his #ifdef, I have corrected the code.
        
    
      html/drivers/driver30.html
        1.16 04/01/10 02:27:01 stenn at pogo.udel.edu +8 -8
        Fixes from Reg Clemens:
        
        I have changed the documentation page for the ONCORE so it no longer
        references Flag2/Flag3.
    



More information about the commitlogs mailing list