March 2013 - tor-commits - lists.torproject.org

[tor/maint-0.2.4] Bug 8477: Don't warn if fromerly GENERAL circuits still have streams.
by nickm＠torproject.org 29 Mar '13

29 Mar '13

commit f6a2f088fdd3b3ed3ccc355c98dde8da37b70a09 Author: Mike Perry <mikeperry-git(a)fscked.org> Date: Tue Mar 26 11:43:40 2013 -0700 Bug 8477: Don't warn if fromerly GENERAL circuits still have streams. This can happen in various cases of network failure. --- src/or/connection_edge.c | 4 +++- 1 files changed, 3 insertions(+), 1 deletions(-) diff --git a/src/or/connection_edge.c b/src/or/connection_edge.c index 9c39c25..5075c47 100644 --- a/src/or/connection_edge.c +++ b/src/or/connection_edge.c @@ -651,7 +651,9 @@ connection_ap_expire_beginning(void) } continue; } - if (circ->purpose != CIRCUIT_PURPOSE_C_GENERAL) { + if (circ->purpose != CIRCUIT_PURPOSE_C_GENERAL && + circ->purpose != CIRCUIT_PURPOSE_C_MEASURE_TIMEOUT && + circ->purpose != CIRCUIT_PURPOSE_PATH_BIAS_TESTING) { log_warn(LD_BUG, "circuit->purpose == CIRCUIT_PURPOSE_C_GENERAL failed. " "The purpose on the circuit was %s; it was in state %s, " "path_state %s.",

1 0

[tor/maint-0.2.4] Bug #8196: Demote a path bias notice that can be caused by controllers.
by nickm＠torproject.org 29 Mar '13

29 Mar '13

commit 9117b142186941f99eae8e07626f12f0b60acd44 Author: Mike Perry <mikeperry-git(a)fscked.org> Date: Tue Mar 26 15:57:38 2013 -0700 Bug #8196: Demote a path bias notice that can be caused by controllers. We didn't see this in normal usage anyway. --- src/or/circuitbuild.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/src/or/circuitbuild.c b/src/or/circuitbuild.c index 3ab534b..7db2b70 100644 --- a/src/or/circuitbuild.c +++ b/src/or/circuitbuild.c @@ -1378,7 +1378,7 @@ pathbias_should_count(origin_circuit_t *circ) if (circ->build_state->desired_path_len != 1 || !circ->build_state->onehop_tunnel) { if ((rate_msg = rate_limit_log(&count_limit, approx_time()))) { - log_notice(LD_BUG, + log_info(LD_BUG, "One-hop circuit has length %d. Path state is %s. " "Circuit is a %s currently %s.%s", circ->build_state->desired_path_len,

1 0

[tor/maint-0.2.4] Bug 8419: Apply the badexit fix from #2203 to validatio too
by nickm＠torproject.org 29 Mar '13

29 Mar '13

commit 651e49713c38b8ec2bde285001e46b61c500d10c Author: Mike Perry <mikeperry-git(a)fscked.org> Date: Tue Mar 26 14:15:58 2013 -0700 Bug 8419: Apply the badexit fix from #2203 to validatio too This was causing dirauths to emit flag weight validation warns if there was a sufficiently large amount of badexit bandwidth to make a difference in flag weight results. --- src/or/dirvote.c | 4 ++-- src/or/dirvote.h | 3 +++ src/or/routerparse.c | 13 ++++++++++--- src/or/routerparse.h | 2 +- 4 files changed, 16 insertions(+), 6 deletions(-) diff --git a/src/or/dirvote.c b/src/or/dirvote.c index bcfe2b0..7043cef 100644 --- a/src/or/dirvote.c +++ b/src/or/dirvote.c @@ -1913,7 +1913,7 @@ networkstatus_compute_consensus(smartlist_t *votes, } /* Fix bug 2203: Do not count BadExit nodes as Exits for bw weights */ - if (consensus_method >= 11) { + if (consensus_method >= MIN_METHOD_TO_CUT_BADEXIT_WEIGHT) { is_exit = is_exit && !is_bad_exit; } @@ -2210,7 +2210,7 @@ networkstatus_compute_consensus(smartlist_t *votes, } // Verify balancing parameters if (consensus_method >= MIN_METHOD_FOR_BW_WEIGHTS && added_weights) { - networkstatus_verify_bw_weights(c); + networkstatus_verify_bw_weights(c, consensus_method); } networkstatus_vote_free(c); } diff --git a/src/or/dirvote.h b/src/or/dirvote.h index fbb61b6..8d036d6 100644 --- a/src/or/dirvote.h +++ b/src/or/dirvote.h @@ -34,6 +34,9 @@ /** Lowest consensus method that generates microdescriptors */ #define MIN_METHOD_FOR_MICRODESC 8 +/** Lowest consensus method that doesn't count bad exits as exits for weight */ +#define MIN_METHOD_TO_CUT_BADEXIT_WEIGHT 11 + /** Lowest consensus method that ensures a majority of authorities voted * for a param. */ #define MIN_METHOD_FOR_MAJORITY_PARAMS 12 diff --git a/src/or/routerparse.c b/src/or/routerparse.c index a2b4f9e..f8edb84 100644 --- a/src/or/routerparse.c +++ b/src/or/routerparse.c @@ -2257,7 +2257,7 @@ networkstatus_v2_parse_from_string(const char *s) /** Verify the bandwidth weights of a network status document */ int -networkstatus_verify_bw_weights(networkstatus_t *ns) +networkstatus_verify_bw_weights(networkstatus_t *ns, int consensus_method) { int64_t weight_scale; int64_t G=0, M=0, E=0, D=0, T=0; @@ -2343,14 +2343,21 @@ networkstatus_verify_bw_weights(networkstatus_t *ns) // Then, gather G, M, E, D, T to determine case SMARTLIST_FOREACH_BEGIN(ns->routerstatus_list, routerstatus_t *, rs) { + int is_exit = 0; + if (consensus_method >= MIN_METHOD_TO_CUT_BADEXIT_WEIGHT) { + /* Bug #2203: Don't count bad exits as exits for balancing */ + is_exit = rs->is_exit && !rs->is_bad_exit; + } else { + is_exit = rs->is_exit; + } if (rs->has_bandwidth) { T += rs->bandwidth; - if (rs->is_exit && rs->is_possible_guard) { + if (is_exit && rs->is_possible_guard) { D += rs->bandwidth; Gtotal += Wgd*rs->bandwidth; Mtotal += Wmd*rs->bandwidth; Etotal += Wed*rs->bandwidth; - } else if (rs->is_exit) { + } else if (is_exit) { E += rs->bandwidth; Mtotal += Wme*rs->bandwidth; Etotal += Wee*rs->bandwidth; diff --git a/src/or/routerparse.h b/src/or/routerparse.h index 859a691..7bf7d79 100644 --- a/src/or/routerparse.h +++ b/src/or/routerparse.h @@ -54,7 +54,7 @@ void dump_distinct_digest_count(int severity); int compare_routerstatus_entries(const void **_a, const void **_b); int compare_vote_routerstatus_entries(const void **_a, const void **_b); networkstatus_v2_t *networkstatus_v2_parse_from_string(const char *s); -int networkstatus_verify_bw_weights(networkstatus_t *ns); +int networkstatus_verify_bw_weights(networkstatus_t *ns, int); networkstatus_t *networkstatus_parse_vote_from_string(const char *s, const char **eos_out, networkstatus_type_t ns_type);

1 0

[tor/maint-0.2.4] Changes file for 8230, 8477 and 8419.
by nickm＠torproject.org 29 Mar '13

29 Mar '13

commit 80d8fb23e39e16764447f608428356bd5c49e4c6 Author: Mike Perry <mikeperry-git(a)fscked.org> Date: Tue Mar 26 14:22:23 2013 -0700 Changes file for 8230, 8477 and 8419. --- changes/log-noise | 11 +++++++++++ 1 files changed, 11 insertions(+), 0 deletions(-) diff --git a/changes/log-noise b/changes/log-noise new file mode 100644 index 0000000..bbbf0d2 --- /dev/null +++ b/changes/log-noise @@ -0,0 +1,11 @@ + o Minor bugfixes (log message reduction) + - Fix a path state issue that triggered a notice during relay startup. + Fixes bug #8320; bugfix on 0.2.4.10-alpha. + - Reduce occurrences of warns about circuit purpose in + connection_ap_expire_building(). Fixes bug #8477; bugfix on + 0.2.4.11-alpha. + - Fix a directory authority warn caused when we have a large amount + of badexit bandwidth. Fixes bug #8419; bugfix on 0.2.2.10-alpha. + - Reduce a path bias length check notice log to info. The notice + is triggered when creating controller circuits. Fixes bug #8196; + bugfix on 0.2.4.8-alpha.

1 0

[tor/maint-0.2.4] Merge branch '0.2.4-lognoise-fixes-rebased' into maint-0.2.4
by nickm＠torproject.org 29 Mar '13

29 Mar '13

commit 8e05001dc60f6dc1e803f19e33cc2ccad12efc82 Merge: 33f86fa 9117b14 Author: Nick Mathewson <nickm(a)torproject.org> Date: Fri Mar 29 12:24:45 2013 -0400 Merge branch '0.2.4-lognoise-fixes-rebased' into maint-0.2.4 changes/log-noise | 11 +++++++++++ src/or/circuitbuild.c | 10 +++++----- src/or/connection_edge.c | 4 +++- src/or/dirvote.c | 4 ++-- src/or/dirvote.h | 3 +++ src/or/routerparse.c | 13 ++++++++++--- src/or/routerparse.h | 2 +- 7 files changed, 35 insertions(+), 12 deletions(-)

1 0

[tor/master] Add changes file for bug 7799's log message changes.
by nickm＠torproject.org 29 Mar '13

29 Mar '13

commit cc641ff5b05b2f8bc831ff28fa5d4226933d18a0 Author: Mike Perry <mikeperry-git(a)fscked.org> Date: Wed Mar 27 15:33:58 2013 -0700 Add changes file for bug 7799's log message changes. Note this does not close bug 7799. --- changes/bug7799 | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-) diff --git a/changes/bug7799 b/changes/bug7799 new file mode 100644 index 0000000..1bb5597 --- /dev/null +++ b/changes/bug7799 @@ -0,0 +1,3 @@ + o Minor changes (log clarification) + - Add more detail to a log message about relaxed timeouts. Hopefully + this additional detail will allow us to diagnose the cause of bug 7799.

1 0

[tor/master] Add detail to log messages related to bug 7799.
by nickm＠torproject.org 29 Mar '13

29 Mar '13

commit 66586da9bc023bf60b2d18e02004a4e278b73359 Author: Mike Perry <mikeperry-git(a)fscked.org> Date: Wed Mar 27 15:34:54 2013 -0700 Add detail to log messages related to bug 7799. Note this does not solve bug 7799, it is only to help us diagnose it. --- src/or/circuituse.c | 21 +++++++++++++++------ 1 files changed, 15 insertions(+), 6 deletions(-) diff --git a/src/or/circuituse.c b/src/or/circuituse.c index 14576f1..bbf219a 100644 --- a/src/or/circuituse.c +++ b/src/or/circuituse.c @@ -536,8 +536,13 @@ circuit_expire_building(void) int first_hop_succeeded = TO_ORIGIN_CIRCUIT(victim)->cpath->state == CPATH_STATE_OPEN; log_info(LD_CIRC, - "No circuits are opened. Relaxing timeout for " - "a circuit with channel state %s. %d guards are live.", + "No circuits are opened. Relaxing timeout for circuit %d " + "(a %s %d-hop circuit in state %s with channel state %s). " + "%d guards are live.", + TO_ORIGIN_CIRCUIT(victim)->global_identifier, + circuit_purpose_to_string(victim->purpose), + TO_ORIGIN_CIRCUIT(victim)->build_state->desired_path_len, + circuit_state_to_string(victim->state), channel_state_to_string(victim->n_chan->state), num_live_entry_guards(0)); @@ -552,10 +557,14 @@ circuit_expire_building(void) } else { static ratelim_t relax_timeout_limit = RATELIM_INIT(3600); log_fn_ratelim(&relax_timeout_limit, LOG_NOTICE, LD_CIRC, - "No circuits are opened. Relaxed timeout for " - "a circuit with channel state %s to %ldms. " - "However, it appears the circuit has timed out anyway. " - "%d guards are live.", + "No circuits are opened. Relaxed timeout for circuit %d " + "(a %s %d-hop circuit in state %s with channel state %s) to " + "%ldms. However, it appears the circuit has timed out " + "anyway. %d guards are live.", + TO_ORIGIN_CIRCUIT(victim)->global_identifier, + circuit_purpose_to_string(victim->purpose), + TO_ORIGIN_CIRCUIT(victim)->build_state->desired_path_len, + circuit_state_to_string(victim->state), channel_state_to_string(victim->n_chan->state), (long)circ_times.close_ms, num_live_entry_guards(0)); }

1 0

[tor/master] Don't relax the timeout for already opened 1-hop circuits.
by nickm＠torproject.org 29 Mar '13

29 Mar '13

commit d39e6736fe1608028223604f49d5d091ef23bb27 Author: Mike Perry <mikeperry-git(a)fscked.org> Date: Wed Mar 27 17:43:27 2013 -0700 Don't relax the timeout for already opened 1-hop circuits. --- src/or/circuituse.c | 7 ++++++- 1 files changed, 6 insertions(+), 1 deletions(-) diff --git a/src/or/circuituse.c b/src/or/circuituse.c index bbf219a..8fb70f5 100644 --- a/src/or/circuituse.c +++ b/src/or/circuituse.c @@ -529,7 +529,12 @@ circuit_expire_building(void) if (timercmp(&victim->timestamp_began, &cutoff, >)) continue; /* it's still young, leave it alone */ - if (!any_opened_circs) { + /* We need to double-check the opened state here because + * we don't want to consider opened 1-hop dircon circuits for + * deciding when to relax the timeout, but we *do* want to relax + * those circuits too if nothing else is opened *and* they still + * aren't either. */ + if (!any_opened_circs && victim->state != CIRCUIT_STATE_OPEN) { /* It's still young enough that we wouldn't close it, right? */ if (timercmp(&victim->timestamp_began, &close_cutoff, >)) { if (!TO_ORIGIN_CIRCUIT(victim)->relaxed_timeout) {

1 0

[tor/master] Merge branch 'bug7799_rebased' into maint-0.2.4
by nickm＠torproject.org 29 Mar '13

29 Mar '13

commit 33f86faa5a20c3c2b7c6be386186bb01b5883739 Merge: 96c0a42 d39e673 Author: Nick Mathewson <nickm(a)torproject.org> Date: Fri Mar 29 12:17:35 2013 -0400 Merge branch 'bug7799_rebased' into maint-0.2.4 changes/bug7799 | 7 +++++++ src/or/circuituse.c | 28 +++++++++++++++++++++------- 2 files changed, 28 insertions(+), 7 deletions(-)

1 0

[tor/master] Update the changes file for bug7799.
by nickm＠torproject.org 29 Mar '13

29 Mar '13

commit 1d49ba84a82c6dfb8873cf32fdecdf5af0b65c90 Author: Mike Perry <mikeperry-git(a)fscked.org> Date: Wed Mar 27 17:43:05 2013 -0700 Update the changes file for bug7799. Still not sure it's actually fixed yet... --- changes/bug7799 | 4 ++++ 1 files changed, 4 insertions(+), 0 deletions(-) diff --git a/changes/bug7799 b/changes/bug7799 index 1bb5597..ed45701 100644 --- a/changes/bug7799 +++ b/changes/bug7799 @@ -1,3 +1,7 @@ o Minor changes (log clarification) - Add more detail to a log message about relaxed timeouts. Hopefully this additional detail will allow us to diagnose the cause of bug 7799. + o Minor bugfixes + - Don't attempt to relax the timeout of already opened 1-hop circuits. + They might never timeout. This should eliminate some/all cases of + the relaxed timeout log message.

1 0