IXFR SOA response implemented.

Fixed packet maxsize calculation.
Support for unlimited (only to 64k) packets for TCP.
Fixed optional authority section.

HAT-trie zone database wasn't effective for small number of zones,
because hashing presented a constant penalty. New zone database
is represented by an array of zones, sorted by number of labels and
then lexicographically. Zones are then grouped by the label count
into stack, so obviously mismatching names are pruned when searching.
Each labelcount group is then search either linearly or using binary
search based on the array length.

Fixes some still-reachable memory leaks. The function should be
called upon termination by all tools dealing with crypto (currently
knotd, knsec3hash, and knsupdate).

refs #4

refs #4

Slight cleanup and API changes included.

refs #4

Also removes the restriction that NSEC3 only algorithms can
be used only with NSEC3.

refs #4

refs #4

libknot/dnssec now contains interface for:
- key loading
- low level DNSSEC signing
- NSEC3 hash computation
- SIG(0)
- generating NSEC/NSEC3 records for zones
- generating RRSIG for zones

refs #2413

Change-Id: I0b0a572eb5cc11787e332404226b0b079ed85f88

refs #2413

Change-Id: I071d0a4560ded8051352ee6d30513a8c10b51a02

Also simplified, cleaned and fixed.

refs #109

The reasons are discussed in #71.
Namely:
- transfers are self-paced, TCP clients are divided among threads
- due to the nature of DNS, most of the connections should be active
- if not, it is probably an attack, long idling connections are disconnected
- even then the performance degradation is not so severe

fixes #71

refs #54

- Works, but there are leaks
- Missing doxygens
- The estimate is not accurate - it reports about a half of the actual consumption on a 32bit system

Refs #37

Marek Vavrusa authored 11 years ago and Daniel Salzman committed 11 years ago

Two reasons:
- There's some rare race condition in the queue implementation
- Doesn't scale and current performance is almost the same (-~2%)
- Created branch with the queue still retained for future inspection

Two reasons:
- There's some rare race condition in the queue implementation
- Doesn't scale and current performance is almost the same (-~2%)
- Created branch with the queue still retained for future inspection