[tor-bugs] #18318 [Tor]: Make sure keys and IP:Ports are unique in a consensus
Tor Bug Tracker & Wiki
blackhole at torproject.org
Sat Mar 19 23:56:32 UTC 2016
#18318: Make sure keys and IP:Ports are unique in a consensus
-------------------------------------------------+-------------------------
Reporter: teor | Owner: nickm
Type: defect | Status:
Priority: Very High | needs_review
Component: Tor | Milestone: Tor:
Severity: Blocker | 0.2.8.x-final
Keywords: TorCoreTeam201602, must-fix- | Version:
before-028-rc, 027-backport | Resolution:
Parent ID: #17668 | Actual Points:
Reviewer: | Points:
| Sponsor: None
-------------------------------------------------+-------------------------
Comment (by teor):
Can we do the tie-breaker the same we we did in #17688?
Or is that not an issue for IP addresses and ports?
(I haven't had a chance to review this code yet, but I'd like to.)
From comments 38/44 in #17688, labelled T3 in that issue:
> > This check preserves the existing digestmap entry if the published
times are equal. But different authorities could choose different
descriptors-with-identical-published-times, depending on the order they
arrived at the authority.
> > So let's find a consistent way of selecting a descriptor to omit if
their published times are identical - how about comparing
signed_descriptor_digest? (And if the digests are identical, it really
doesn't matter which descriptor we omit, because they have the same
content.)
>
> Good idea; Calling it `T3`.
>
> > We should do the same tie-breaking in the IP:Port checks too.
> > {{{
> > + if ((ri2 = digest256map_get(by_ed_key, pk))) {
> > + /* Duplicate; must omit one. Omit the earlier. */
> > + if (ri2->cache_info.published_on < ri->cache_info.published_on)
{
> > + digest256map_set(by_ed_key, pk, ri);
> > + ri2->omit_from_vote = 1;
> > + } else {
> > + ri->omit_from_vote = 1;
> > + }
> > }}}
--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/18318#comment:18>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online
More information about the tor-bugs
mailing list