[tor-bugs] #22006 [Core Tor/Tor]: prop224: Validate ed25519 pubkeys to remove torsion component

[Tor Bug Tracker & Wiki]

#22006: prop224: Validate ed25519 pubkeys to remove torsion component
------------------------------------+------------------------------------
 Reporter:  asn                     |          Owner:  asn
     Type:  defect                  |         Status:  needs_review
 Priority:  Medium                  |      Milestone:  Tor: 0.3.1.x-final
Component:  Core Tor/Tor            |        Version:
 Severity:  Normal                  |     Resolution:
 Keywords:  tor-hs prop224 ed25519  |  Actual Points:
Parent ID:  #21888                  |         Points:
 Reviewer:                          |        Sponsor:  SponsorR-can
------------------------------------+------------------------------------
Changes (by asn):

 * status:  needs_revision => needs_review


Comment:

 OK pushed a fixup commit that adds the `ALIGN` calls that Yawning
 suggested in the donna code. It seems to be a common idiom in the donna
 code so it should be the right thing. I tried to reproduce the SSE2 crash
 but I couldn't repro it. Yawning said that he didn't see the crash
 himself, but he thought that it would crash because of the lack of `ALIGN`
 calls.

 So now, the last issue is whether we want our validation (which includes a
 scalar mult) running 7k times everytime we receive a consensus, and also
 everytime we receive a descriptor etc. It's currently the case, since
 [https://lists.torproject.org/pipermail/tor-dev/2017-April/012230.html Ian
 suggested we do so] and he said the validation is not particularly
 expensive. I tested it on my laptop and I didn't notice any speed
 difference from validating 7k ed25519 keys when receiving the consensus,
 however perhaps this might be more battery-drain in mobile devices, etc.
 Nick any opinions here?

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/22006#comment:8>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online