[tor-bugs] #16794 [Core Tor/Tor]: All cryptography unit test coverage should be over 95%; all should have test vectors

[Tor Bug Tracker & Wiki]

#16794: All cryptography unit test coverage should be over 95%; all should have
test vectors
-------------------------------------------------+-------------------------
 Reporter:  nickm                                |          Owner:  nickm
     Type:  enhancement                          |         Status:
 Priority:  Medium                               |  needs_review
Component:  Core Tor/Tor                         |      Milestone:  Tor:
 Severity:  Normal                               |  0.2.9.x-final
 Keywords:  testing, 028-triage, tor-tests-      |        Version:
  coverage, tor-tests-unit, TorCoreTeam201605,   |     Resolution:
  review-group-1                                 |  Actual Points:
Parent ID:  #16791                               |         Points:  medium
 Reviewer:  isis                                 |        Sponsor:
                                                 |  SponsorS-can
-------------------------------------------------+-------------------------

Comment (by nickm):

 Replying to [comment:25 isis]:
 > When we say "coverage should be over 95%", do we mean line coverage,
 function coverage, or branch coverage? I currently see line coverage for
 src/common/crypto.c at 93.3% (but function coverage is over 96%).

 I've been looking at overall line coverage of src/common/crypto*.c .
 Different compilers will instrument a bit differently, of course.

 > Review:
 >
 >  * `d1f2af57` LGTM.
 >  * `405b6375` LGTM.
 >  * `44a32481` I checked that the vectors match those in RFC7748. LGTM.
 >  * `54697fa4` I checked that the vectors match those in NIST SP800-38a
 §F.5 for AES-128 CTR. LGTM.
 >  * `5845c228` Again, I manually checked that the vectors match those in
 draft-irtf-cfrg-eddsa-05. LGTM.
 >   I'm not super familiar with IRTF drafts and their statuses
 (specifically what might change in them), but that draft is currently in
 "Waiting for Document Shepherd" state. How much should we worry that,
 through the shepherding process, the vectors might change? Should we be
 keeping a close eye on this?

 We should probably keep an eye on it, but I would be surprised if they
 decided to break backward compatibility with all the existing Ed25519
 implementations.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/16794#comment:26>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online