Journal fixing

See merge request !626

wire: don't use misaligned pointers

See merge request !625

It's the same as 9badca98.

wire_ctx: refactor to reduce duplication and UB

See merge request !622

Journal lmdb joint

See merge request !616

Adapt to the new afl to use __AFL_LOOP

See merge request !624

It was using misaligned pointers which causes undefined behavior,
at least theoretically.

API change: data in wire_ctx_read() is now passed as void* instead of
uint8_t*, and it zeroes the data on error.

...including configuration changes of max-journal-size, max-journal-usage, max-journal-depth ...

This reverts commit 7f14873e.

The reason is it's wrong and brings more problems than benefit.

doc: updated year in pre-built man templates

See merge request !621

Matthijs Mekking authored 8 years ago and Daniel Salzman committed 8 years ago

Notes:
 - To do: if the zone cut is deduced from a RRSIG record, we could also detect if
   there was a wildcard match and set the RRL_WILDCARD flag.
 - Perhaps only the name needs to be retrieved from the SOA record in case of
   a NXDOMAIN, since RRL will use the query name anyway for the other classifications.

Jan Včelák authored 8 years ago and Daniel Salzman committed 8 years ago

The RRL rate is measured and averaged over a period of time. This
measurement window was applied only for existing buckets. This change
enables this behavior for new buckets as well.

This should have no operational impact. But it will make our testing
a bit easier.