47 // We keep track of the number of unicast DNS servers and log a message when we exceed 64.
158     // Don't increment when we resurrect a DNS server that is not marked with DNSServer_FlagDelete.
159     // We have already accounted for it when it was added for the first time. This case happens when
201             // when we receive the first response to a DNSSEC question, we set
210     // We always update the ID (not just when we allocate a new instance) because we could
217 // PenalizeDNSServer is called when the number of queries to the unicast
218 // DNS server exceeds MAX_UCAST_UNANSWERED_QUERIES or when we receive an
289             // times when we receive SERVFAIL from delayed packets in the network e.g., DNS server
306             // the interval. We do this only the first time when we try all the DNS servers. Once we reached the end of
308             // use the normal backoff which is done in uDNS_CheckCurrentQuestion when we send the packet out.
324             // so that when the cache expires, we can restart the query.  We defer this up until we generate
327             // Be careful not to touch the ThisQInterval here. For a normal question, when we answer the question
478 // Keep track of when to request/refresh the external address using NAT-PMP or UPnP/IGD,
479 // and do so when necessary
539     // be called when we need to send the request(s)
670     mDNSu32 when = NonZeroTime(m->timenow + waitTicks);
676         n->retryPortMap  = when;
688     m->retryGetAddr = when;
734         // We should change n->NewAddress only when n is one of:
739         // 2) an address-only operation that did not succeed via PCP, because when such an
1004     //!!!KRS when we implement multiple llqs per message, we'll need to memmove anything past the question section
1030 // so this is fine, except when the TCP connection ends up going over a VPN tunnel instead.
1043 // May be called with llq NULL, when retransmitting a lost Challenge Response
1064         llqBuf.err      = LLQErr_NoError;   // Don't need to tell server UDP notification port when sending over UDP
1477                 // We get here when we fail to establish a new TCP/TLS connection that would have been used for a new LLQ request or an LLQ renewal.
1478                 // Note that ThisQInterval is also zero when sendChallengeResponse resends the LLQ request on an extant TCP/TLS connection.
1687             llqData.err   = LLQErr_NoError; // Don't need to tell server UDP notification port when sending over UDP
1823 // <rdar://problem/5687667> BTMM: Don't fallback to unencrypted operations when SRV lookup fails
1827             zd->ZonePrivate = mDNSfalse;    // Causes ZoneDataSRV() to yield a different SRV name when building the query
1871         // The caller needs to free the memory when done with zone data
2126     // SRVChanged is set when when the target of the SRV record changes (See UpdateOneSRVRecord).
2196 // when the target is resolvable. Similarly, when host name changes, we enter regState_NoTarget
2241     // when it entered NoTarget state, it did not go through complete initialization
2301         // In these states, the SRV has either not yet been registered (it will get up-to-date information when it is)
2479                 mDNSPlatformMemFree(hi);    // free hi when both v4 and v6 AuthRecs deallocated
2487         // don't unlink or free - we can retry when we get a new address/router
2693     // setting router to zero indicates that nat mappings must be reestablished when router is reset
2981         // We never call this function when there is no zone information . Log a message if it ever happens.
3371             // again when we return to CheckGroupRecordUpdates.
3499     // Later when need to send an update, we will get the zone data again. Thus we avoid
3503     // in some cases. For example, when we get out regState_NoTarget state e.g., move out
3505     // _autotunnel6 and host records are registered/deregistered when NAT state changes.
3511         // We always expect the question to be stopped when we get a valid response from the server.
3536     // no target, it will be deregistered. Hence, the updating to the right state should not matter when going
4108     // so we may not be able to reference it (most importantly it's AuthInfo) when we actually send the message
4130             // ThisQInterval is not adjusted when we return from here which means that we will get called back
4504     // It is also possible to reenter the NoTarget state when we move to a network with a NAT that has
4512         // This function may be called during sleep when there are no sleep proxy servers
4523     //   in the resource record are same. We don't want to free in that case. It will be freed when "rdata"
4564     // find the TCP below there. This case can happen only when tcp is trying to actively retransmit
4604     // data when it encounters this record.
4885             // where we want to return timely feedback to the user when no network is available.
4915                     // MaxQuestionInterval when we answer the question e.g., LongLived, we need to keep backing off
4919                     // Pick a new DNS server now. Otherwise, when the cache is 80% of its expiry, we will try
5140                     //    matter because when the answer comes back, RecordRegistrationGotZoneData resets the interval
5231             // one will be registered when we come back.
5416     // and when we get a response, we add a LocalOnly entry b._dns-sd._udp.local whose RDATA
5814 // Construction of Default Browse domain list (i.e. when clients pass NULL) is as follows:

Indexes created Tue Jul 24 14:28:13 CEST 2018