[ntp:questions] Leap second data question

Stephen Yu syu at symmetricom.com
Thu Apr 11 17:16:53 UTC 2013

The 4.2.7 code uses the majority rule for leap indicator when processing network peers/servers/ref clock's LI bits (with the leap_vote counter). However, the same logic is not used when processing the leap data from peers/servers via auto-key protocol. In the "case CRYPTO_LEAP | CRYPTP_RESP" (ntp_crypto.c), the "leap_tai" and "leap_sec" variables are updated without needing a majority. My guess the majority check in this case is probably defined by the values of  "leap_tai" and "leap_sec" together (ep->pkt[-0] and ep->pkt[1]). Of course, the question is whether there is any need to have a major rule check here.


More information about the questions mailing list