Pull Request Test Coverage Report for Build 12268281475

Details

• 3 of 5 (60.0%) changed or added relevant lines in 1 file are covered.
• 28 unchanged lines in 8 files lost coverage.
• Overall coverage increased (+3.2%) to 64.756%

Totals
Change from base Build 12257697323:	3.2%
Covered Lines:	125906
Relevant Lines:	163530

💛 - Coveralls

I am not sure these changes are correct. getSOAUncached may end up invoking getSOA and set db to a valid pointer.

On the other hand, regardless of whether this call succeeds or not, a final nullptr check for s_db.db should be done to exit before the rest of the logic runs.

In other words:

  if (s_db.db == nullptr) {
    ... existing code unchanged ...
  }

  if (s_db.db == nullptr)
    return;

  ... rest of the logic here

That's not necessary, as UeberBackend::getSOAUncached will only return true if db gets set to a valid pointer (which is handled by DNSBackend::getSOA). This entire PR is just wrong and fails to understand how this code works; this section needs no changes.

Then maybe adding a simple assert(s_db.db != nullptr) line would hint the static analyzer that the pointer is valid at this point.

After the forced push, this pull is reduced to a bad behavior change. Resulting in a huge performance hit and potentially broken NSEC responses.