Provide additional answers in NAPTR queries

[miodvallat]

Short description

This adds additional answers with replacement names found in NAPTR queries when known, and NAPTR records have the S or A flags. This is similar to what the recursor already does, and is based upon its logic.

Takes over #12966 (with @Habbie's approval)

Checklist

I have:

• read the CONTRIBUTING.md document
• compiled this code
• tested this code
• included documentation (including possible behaviour changes)
• documented the code
• added or modified regression test(s)
• added or modified unit test(s)
• ticked at least 4 boxes above

[coveralls]

Pull Request Test Coverage Report for Build 13116764149

Details

• 27 of 27 (100.0%) changed or added relevant lines in 1 file are covered.
• 33 unchanged lines in 12 files lost coverage.
• Overall coverage increased (+0.01%) to 64.711%

---

Files with Coverage Reduction	New Missed Lines	%
modules/gpgsqlbackend/gpgsqlbackend.cc	1	88.62%
pdns/validate.cc	1	68.24%
pdns/dnsdistdist/dnsdist.hh	1	84.18%
pdns/misc.cc	2	62.93%
modules/lmdbbackend/lmdbbackend.cc	2	72.71%
pdns/rcpgenerator.cc	2	90.1%
pdns/tcpreceiver.cc	2	66.82%
pdns/opensslsigners.cc	3	61.34%
pdns/shuffle.cc	4	53.93%
pdns/recursordist/pdns_recursor.cc	4	72.59%

Totals
Change from base Build 13115942971:	0.01%
Covered Lines:	127906
Relevant Lines:	166554

---

💛 - Coveralls

[Habbie]

if I recall correctly, the "S" case should look for SRV, and then follow those through to get to A/AAAA

[miodvallat]

if I recall correctly, the "S" case should look for SRV, and then follow those through to get to A/AAAA

That makes perfect sense.

How about this new, uglier, logic?

[Habbie]

How about this new, uglier, logic?

(besides review comments), not bad! If the process was recursive, we could just push <name,SRV> there, but it's not.

[Habbie]

one open question for somebody to answer. If the answer to the cache question is good, this is approved.

[Habbie]

somebody (you? me?) needs to check if this is a cache hit if the previous query for the name was QType::ANY

[miodvallat]

Adding a quick accessor to the d_cached überfield shows that, apart from the first time this code path is hit, the lookup is found in cache. That's probably not precise enough.

[miodvallat]

If a perform an ANY query to the target of SRV (_double._tcp.dc.test.com) before the NAPTR query of ensm, all SRV lookups are found in cache. Does this answer your question?