[tor-commits] [tor/master] shared_random: Improve fallback for client no-live-consensus case.

[asn at torproject.org]

commit a5bc08579ff67a06441478016025eb4f3fd879b0
Author: Nick Mathewson <nickm at torproject.org>
Date:   Wed Feb 26 11:33:03 2020 -0500

    shared_random: Improve fallback for client no-live-consensus case.
    
    In this case, when we're looking for the voting interval, we should
    try looking at the _latest_ consensus if we have one.  When we're
    looking for the start of the current voting period, we can use our
    existing fallback logic without complaint, since the voting interval
    code will already have given us a reasonable voting interval, and we
    want to have a round starting time based on the current time.
---
 src/feature/hs_common/shared_random_client.c | 22 +++++++++++++++++++---
 1 file changed, 19 insertions(+), 3 deletions(-)

diff --git a/src/feature/hs_common/shared_random_client.c b/src/feature/hs_common/shared_random_client.c
index ece6e101a..3f46321be 100644
--- a/src/feature/hs_common/shared_random_client.c
+++ b/src/feature/hs_common/shared_random_client.c
@@ -58,10 +58,20 @@ get_voting_interval(void)
   networkstatus_t *consensus = networkstatus_get_live_consensus(time(NULL));
 
   if (consensus) {
+    /* Ideally we have a live consensus and we can just use that. */
     interval = (int)(consensus->fresh_until - consensus->valid_after);
   } else if (authdir_mode(get_options()) || ASSUME_AUTHORITY_SCHEDULING) {
+    /* If we don't have a live consensus and we're an authority,
+     * we should believe our own view of what the schedule ought to be. */
     interval = dirauth_sched_get_configured_interval();
+  } else if ((consensus = networkstatus_get_latest_consensus())) {
+    /* If we're a client, then maybe a latest consensus is good enough?
+     * It's better than falling back to the non-consensus case. */
+    interval = (int)(consensus->fresh_until - consensus->valid_after);
   } else {
+    /* We should never be reaching this point, since a client should never
+     * call this code unless they have some kind of a consensus. All we can
+     * do is hope that this network is using the default voting interval. */
     tor_assert_nonfatal_unreached_once();
     interval = DEFAULT_NETWORK_VOTING_INTERVAL;
   }
@@ -235,15 +245,21 @@ sr_state_get_start_time_of_current_protocol_run(void)
   /* This function is not used for voting purposes, so if we have a live
      consensus, use its valid-after as the beginning of the current round.
      If we have no consensus but we're an authority, use our own
-     schedule. Otherwise, we have a bug somewhere, so we fall back to the
-     default voting interval. */
+     schedule.  Otherwise, try using our view of the voting interval
+     to figure out when the current round _should_ be starting.
+  */
   networkstatus_t *ns = networkstatus_get_live_consensus(approx_time());
   if (ns) {
     beginning_of_curr_round = ns->valid_after;
   } else if (authdir_mode(get_options()) || ASSUME_AUTHORITY_SCHEDULING) {
     beginning_of_curr_round = dirauth_sched_get_cur_valid_after_time();
   } else {
-    tor_assert_nonfatal_unreached_once();
+    /* voting_interval comes from get_voting_interval(), so if we're in
+     * this case as a client, we already tried to get the voting interval
+     * from the latest_consensus and gave a bug warning if we couldn't.
+     *
+     * We wouldn't want to look at the latest consensus's valid_after time,
+     * since that would be out of date. */
     beginning_of_curr_round = voting_sched_get_start_of_interval_after(
                                              approx_time() - voting_interval,
                                              voting_interval,