[tor-bugs] #12204 [Tor]: Revisit flags passed to entry_is_live() around entrynodes.c

[Tor Bug Tracker & Wiki]

#12204: Revisit flags passed to entry_is_live() around entrynodes.c
-----------------------+---------------------
 Reporter:  asn        |          Owner:
     Type:  defect     |         Status:  new
 Priority:  normal     |      Milestone:
Component:  Tor        |        Version:
 Keywords:  tor-guard  |  Actual Points:
Parent ID:             |         Points:
-----------------------+---------------------
 It seems that we are almost always passing `need_capacity=True` to
 `entry_is_live()`.

 The only case where we don't, is in `choose_random_entry_impl()` where we
 actually pick an entry guard. In that case, we pass whatever
 `cpath_build_state_t.need_capacity` is set to (which makes sense).

 However, in other calls, like the one at
 `entry_guard_register_connect_status()`, we always call `entry_is_live()`
 with `need_capacity` set to true. Is this needed?

 In `entry_guard_register_connect_status()`, if we just connected for the
 first time to a new guard, we re-activate all the previous live guards
 since it might be a sign that our network was down and it just came back
 up. However, since `entry_is_live()` is used with `need_capacity` we only
 reactivate the '''Fast''' guards.

 However, as I understand it, there is no requirement that guards need to
 be '''Fast'''. So it might be the case, that if our primary guard is not
 fast, we don't reactivate it.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/12204>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online