Previous with full dname copy and chopping was really inefficient.
Yields approx. 20% lower relative cost, 6% throughput increase.

Change-Id: Icaa49572f75769739ae61f48d2ca958e38db6104

Change-Id: I3cfa7f07a71b1d8b9e77f93ef16b683c15f9fe24

Change-Id: I571cc82f393d526e43ba17ade19f57ce067b5c52

Coccinelle patch used to perform this cleanup is included.

Change-Id: Ia367764245c845839fd22034b2b5f58b93bd54e8

Change-Id: Ib942dde21c1a386cbde98ab422608abc2f1b35cc

Change-Id: If7b7a325ed20432d8fc3da73e931e0fa44a37add

Change-Id: I6dd8a10278383e8c97e399562fec3562cbb848d8

Change-Id: I4eedcaf1e9fef0e43494314e16c68b0bbb2d3107

Change-Id: Id352f59b8ef9011e0379455bc485072d9b67b029

Change-Id: I2683becadd439cc55de64e6eea96a9e2a47667b0

Change-Id: I80c06b6d17a532d3185c73b17da1177646f3b64c

Change-Id: I777d3986ac9197ec438b0b7c00b2098ba60d9437

This allows creating of multiple signatures without reloading the key
every time when a new signature is needed.

refs #2413

Change-Id: I1392a6a697e9a599eaceee21dbe4410ae01c6f46

Change-Id: I4e314b414f0ce94ff691f00a461653350b10370e

Handler destructor was called only if handler was brought down
while the server was running. Also it didn't free the context struct.

Change-Id: I8e71f4abc2bb314528416eeba063b48f2fb79773

Previously it was set to maximum size and msg_namelen was never
reset, which could cause problems with mixed IPv6/IPv4.

Change-Id: I2b4c58c33324fbe702a5b2aa74db2b530119c4ab

refs #2137

Change-Id: Ibccdf2203e79fd2b9f24e75d54bfedc1eed605c0

refs #2137

Change-Id: I983291e35d76531cd82c7feb7467cd8fbe0cbbf7

refs #2137

Change-Id: I3cf4a2d379846f06656cbbc15a2e793c1f1fbc53

Change-Id: I415c5e449f5f2f046ce8970a72341ba13543bb05

Change-Id: I7d4597e875a42805c307b2c1ed75ed309bf3b37a

Fixed missing RCU unlock in early exit.
Expire event synchronized RCU before marking event as finished,
this could result in deadlocking threads holding RCU read lock while
waiting for event cancellation.
Transfer requests usually hold zone reference, so no RCU locking is
required.

Change-Id: I05caf7feb789c78d253b25812666c639f76ffb96

- Better warning messages.
- Small fixes.

Change-Id: Ief9b250c9203b716a07223b34063c4d840521bee

Change-Id: Iad8fce7118770ae49643cc57b510487a087c5242

Change-Id: Ib2ab2c50babf7623b936a48837c1d6096c9af861

- New parser does not report empty file as an error (and nor should it, I guess), so an extra check had to be added.

Change-Id: I80ca06cbd496ae9069be7fbed3e096d64b01f758

- It makes no sense to move compression offset for RDLENGTH bytes for each type, the value was totally wrong.
- It would work fine for single RRs, but if there were more than one, and if a comprimable DNAME appeared for the first time in the packet, later references to this DNAME would be broken.

Change-Id: Ic02351625d5a4e120b492946759607583502967f

refs #2137

Change-Id: Iecacf01d0f88c5774d02eddf1db53199f4357608

Change-Id: I429b3833c5db9e78220beb1ea72980b557ba06c4

Change-Id: Idb34e5dc231760fa1b0eb8e28250c2f390b43f71

Fallback for x86/amd64 architectures using full barrier.

Change-Id: I80e3beaa721653f760f2d0a6d26a80dd422ab618

Change-Id: Ieb580a23436b227fb889197cb3f4857f03ccf7ae

- Simplified and decoupled I/O from processing code.
- Single reader with multiple workers.
- Reader is bound to specific CPU for lowest possible latency.
- Reverted to select() instead of fdset wrapper as epoll/kqueue could
  be quite slower for a single or handful of sockets.
- Reader I/O calls are non-blocking.

General idea is:
- Reader takes free request from a rx_queue
- Receives data and pushes it to the tx_queue
- Worker processes the packets, sends answers and returns the request
  to the rx_queue

Semi lock-free queue is used for communication.
This is measured to be slightly better than singlethreading, as socket
contention could be better predicted - the reader cannot check sockets
for new events when tx_queue is full.

Change-Id: Ifc40889b355404a9e78e10d77c26cb3fc4c58386

Author: Rusty Russel <rusty@rustcorp.com.au>
Licensed under MIT-BSD.

Extended with consumer sleep/wakeup for our use,
where we want the consumers to sleep instead of
racing for new packets.

Change-Id: I37b57befcd98e1da6a485aaf48433dc06c68fa5e

Change-Id: Ib080da4896ab797a6d950545cc6d426fd54fde51

Change-Id: Iceb9613b3bbd4e56e66a52ba8b712f572eb3d81f

NOTE: Full verification will be added in the next release.

refs #2137

Change-Id: I092a8caf1640b5221815051e5cda1d1ec02e7098

refs #2137

Change-Id: I12da030a6f612bba6b9aef3a495448571f876c3d

The PID file needed to be created before with knotc to indicate
starting instance. This is better than knotd creating it, but
it could result in either dropping privileges before starting knotd
or creating PID file in an inaccessible directory.
As of now, knotd drops privileges and creates empty PID file before
loading zones.

Change-Id: Ifb1507b2a922217d4d4608accb1f71dd7be5c0ee

refs #2137

Change-Id: Iaca6cb6f9d4e99a0a9b399d0206cc9773ae6c285