December 2012 - tor-commits - lists.torproject.org

[tor/master] Update changes file.
by nickm＠torproject.org 26 Dec '12

26 Dec '12

commit 08da247042a106fbf8a437b7cd25460b0f2ee045 Author: Mike Perry <mikeperry-git(a)fscked.org> Date: Sun Dec 9 23:58:01 2012 -0800 Update changes file. --- changes/bug7157 | 11 +++++++++++ 1 files changed, 11 insertions(+), 0 deletions(-) diff --git a/changes/bug7157 b/changes/bug7157 index fad3977..4f0e3b3 100644 --- a/changes/bug7157 +++ b/changes/bug7157 @@ -13,3 +13,14 @@ in combination with PathBiasExtremeRate. - Increase the default values for PathBiasScaleThreshold and PathBiasCircThreshold from 200 and 20 to 300 and 150, respectively. + - Add in circuit usage accounting to path bias. If we try to use a + built circuit but fail for any reason, it counts as path bias. + Certain classes of circuits where the adversary gets to pick your + destination node are exempt from this accounting. Usage accounting + can be specifically disabled via consensus parameter or torrc. + - Convert all internal path bias state to double-precision floating + point, to avoid roundoff error and other issues. + - Only record path bias information for circuits that have completed + *two* hops. Assuming end-to-end tagging is the attack vector, this + makes us more resilient to ambient circuit failure without any + detection capability loss.

1 0

[tor/master] Update manpage.
by nickm＠torproject.org 26 Dec '12

26 Dec '12

commit 43a00877cf01888229d5d34085a2b75748848af1 Author: Mike Perry <mikeperry-git(a)fscked.org> Date: Mon Dec 10 00:13:55 2012 -0800 Update manpage. --- doc/tor.1.txt | 12 +++++++++--- 1 files changed, 9 insertions(+), 3 deletions(-) diff --git a/doc/tor.1.txt b/doc/tor.1.txt index 734bc64..7ceefb6 100644 --- a/doc/tor.1.txt +++ b/doc/tor.1.txt @@ -1183,12 +1183,18 @@ The following options are useful only for clients (that is, if **PathBiasMultFactor** __NUM__ + -**PathBiasScaleFactor** __NUM__:: +**PathBiasScaleFactor** __NUM__ + + +**PathBiasUseCloseCounts** __NUM__:: These options override the default behavior of Tor's (**currently experimental**) path bias detection algorithm. To try to find broken or misbehaving guard nodes, Tor looks for nodes where more than a certain - fraction of circuits through that node fail after the first hop. The - PathBiasCircThreshold option controls how many circuits we need to build + fraction of circuits through that guard fail to get built. If + PathBiasUseCloseCounts is set to 1 (the default), usage-based accounting is + performed, and circuits that fail to carry streams are also counted as + failures. + + + + The PathBiasCircThreshold option controls how many circuits we need to build through a guard before we make these checks. The PathBiasNoticeRate, PathBiasWarnRate and PathBiasExtremeRate options control what fraction of circuits must succeed through a guard so we won't write log messages.

1 0

[tor/master] Add a missing comment.
by nickm＠torproject.org 26 Dec '12

26 Dec '12

commit c1bc6a112498ba53243cfc8ec259a9e20124d86a Author: Mike Perry <mikeperry-git(a)fscked.org> Date: Mon Dec 10 00:36:10 2012 -0800 Add a missing comment. --- src/or/circuitbuild.c | 8 +++++++- 1 files changed, 7 insertions(+), 1 deletions(-) diff --git a/src/or/circuitbuild.c b/src/or/circuitbuild.c index d39817c..cc43797 100644 --- a/src/or/circuitbuild.c +++ b/src/or/circuitbuild.c @@ -1608,7 +1608,13 @@ pathbias_count_timeout(origin_circuit_t *circ) } } -// XXX: DOCDOC +/** + * Return the number of circuits counted as successfully closed for + * this guard. + * + * Also add in the currently open circuits to give them the benefit + * of the doubt. + */ double pathbias_get_closed_count(entry_guard_t *guard) {

1 0

[tor/master] More log message and space fixups.
by nickm＠torproject.org 26 Dec '12

26 Dec '12

commit d409c8a90d876c2f45a1c4ea14ddae44fa7c8f18 Author: Mike Perry <mikeperry-git(a)fscked.org> Date: Mon Dec 10 00:28:07 2012 -0800 More log message and space fixups. --- src/or/circuitbuild.c | 22 ++++++++++++++-------- src/or/relay.c | 2 +- 2 files changed, 15 insertions(+), 9 deletions(-) diff --git a/src/or/circuitbuild.c b/src/or/circuitbuild.c index f903bbf..d39817c 100644 --- a/src/or/circuitbuild.c +++ b/src/or/circuitbuild.c @@ -1436,18 +1436,22 @@ pathbias_check_close(origin_circuit_t *ocirc, int reason) * streams could be bias */ log_info(LD_CIRC, "Circuit %d closed without successful use for reason %d. " - "Circuit purpose %d currently %s.", + "Circuit purpose %d currently %d,%s. Len %d.", ocirc->global_identifier, - reason, circ->purpose, circuit_state_to_string(circ->state)); + reason, circ->purpose, ocirc->has_opened, + circuit_state_to_string(circ->state), + ocirc->build_state->desired_path_len); pathbias_count_unusable(ocirc); } else { if (reason & END_CIRC_REASON_FLAG_REMOTE) { /* Unused remote circ close reasons all could be bias */ log_info(LD_CIRC, "Circuit %d remote-closed without successful use for reason %d. " - "Circuit purpose %d currently %s.", + "Circuit purpose %d currently %d,%s. Len %d.", ocirc->global_identifier, - reason, circ->purpose, circuit_state_to_string(circ->state)); + reason, circ->purpose, ocirc->has_opened, + circuit_state_to_string(circ->state), + ocirc->build_state->desired_path_len); pathbias_count_collapse(ocirc); } else if ((reason & ~END_CIRC_REASON_FLAG_REMOTE) == END_CIRC_REASON_CHANNEL_CLOSED && @@ -1455,14 +1459,16 @@ pathbias_check_close(origin_circuit_t *ocirc, int reason) circ->n_chan->reason_for_closing != CHANNEL_CLOSE_REQUESTED) { /* If we didn't close the channel ourselves, it could be bias */ - /* FIXME: Only count bias if the network is live? + /* XXX: Only count bias if the network is live? * What about clock jumps/suspends? */ log_info(LD_CIRC, "Circuit %d's channel closed without successful use for reason " - "%d, channel reason %d. Circuit purpose %d currently %s.", - ocirc->global_identifier, + "%d, channel reason %d. Circuit purpose %d currently %d,%s. Len " + "%d.", ocirc->global_identifier, reason, circ->n_chan->reason_for_closing, - circ->purpose, circuit_state_to_string(circ->state)); + circ->purpose, ocirc->has_opened, + circuit_state_to_string(circ->state), + ocirc->build_state->desired_path_len); pathbias_count_collapse(ocirc); } else { pathbias_count_successful_close(ocirc); diff --git a/src/or/relay.c b/src/or/relay.c index 6ed4b93..fd8f857 100644 --- a/src/or/relay.c +++ b/src/or/relay.c @@ -697,7 +697,7 @@ connection_ap_process_end_not_open( /* Path bias: If we get a valid reason code from the exit, * it wasn't due to tagging */ // XXX: This relies on recognized+digest being strong enough not - // to be spoofable.. Is that a valid assumption? + // to be spoofable.. Is that a valid assumption? // Or more accurately: is it better than nothing? Can the attack // be done offline? circ->path_state = PATH_STATE_USE_SUCCEEDED;

1 0

[tor/master] Fix whitespace
by nickm＠torproject.org 26 Dec '12

26 Dec '12

commit 8324824d8f7cd9a68b2c2d6d50aa7ad28b9ba761 Author: Nick Mathewson <nickm(a)torproject.org> Date: Tue Dec 25 23:34:16 2012 -0500 Fix whitespace --- src/or/circuitbuild.c | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/src/or/circuitbuild.c b/src/or/circuitbuild.c index 804782d..55f7461 100644 --- a/src/or/circuitbuild.c +++ b/src/or/circuitbuild.c @@ -1078,7 +1078,7 @@ pathbias_get_scale_threshold(const or_options_t *options) /** * The scale factor is the denominator for our scaling - * of circuit counts for our path bias window. + * of circuit counts for our path bias window. * * Note that our use of doubles for the path bias state * file means that powers of 2 work best here. @@ -1605,7 +1605,7 @@ pathbias_count_timeout(origin_circuit_t *circ) /** * Return the number of circuits counted as successfully closed for * this guard. - * + * * Also add in the currently open circuits to give them the benefit * of the doubt. */

1 0

[tor/master] Tags on relay cells can result in certain reason codes.
by nickm＠torproject.org 26 Dec '12

26 Dec '12

commit ccaeef22e168af34e9b6a63d65ce17e58dd702e2 Author: Mike Perry <mikeperry-git(a)fscked.org> Date: Tue Dec 11 17:49:12 2012 -0800 Tags on relay cells can result in certain reason codes. Close the circuit (it's probably junk anyways), and make sure we don't probe it/count it as a success. --- src/or/circuitbuild.c | 2 ++ src/or/or.h | 8 +++++++- src/or/relay.c | 24 +++++++++++++++++------- 3 files changed, 26 insertions(+), 8 deletions(-) diff --git a/src/or/circuitbuild.c b/src/or/circuitbuild.c index 3e2568c..f93b04f 100644 --- a/src/or/circuitbuild.c +++ b/src/or/circuitbuild.c @@ -1140,6 +1140,8 @@ pathbias_state_to_string(path_state_t state) return "build succeeded"; case PATH_STATE_USE_SUCCEEDED: return "use succeeded"; + case PATH_STATE_USE_FAILED: + return "use failed"; } return "unknown"; diff --git a/src/or/or.h b/src/or/or.h index aaf817d..ccc20b9 100644 --- a/src/or/or.h +++ b/src/or/or.h @@ -2779,6 +2779,12 @@ typedef enum { * just tag at a later point. */ PATH_STATE_USE_SUCCEEDED = 3, + + /** + * This is a special state to indicate that we got a corrupted + * relay cell on a circuit and we don't intend to probe it. + */ + PATH_STATE_USE_FAILED = 4, } path_state_t; /** An origin_circuit_t holds data necessary to build and use a circuit. @@ -2816,7 +2822,7 @@ typedef struct origin_circuit_t { /** Kludge to help us prevent the warn in bug #6475 and eventually * debug why we are not seeing first hops in some cases. */ - path_state_t path_state : 2; + path_state_t path_state : 3; /** Set iff this is a hidden-service circuit which has timed out * according to our current circuit-build timeout, but which has diff --git a/src/or/relay.c b/src/or/relay.c index fd8f857..b4b7700 100644 --- a/src/or/relay.c +++ b/src/or/relay.c @@ -694,13 +694,23 @@ connection_ap_process_end_not_open( (void) layer_hint; /* unused */ if (rh->length > 0) { - /* Path bias: If we get a valid reason code from the exit, - * it wasn't due to tagging */ - // XXX: This relies on recognized+digest being strong enough not - // to be spoofable.. Is that a valid assumption? - // Or more accurately: is it better than nothing? Can the attack - // be done offline? - circ->path_state = PATH_STATE_USE_SUCCEEDED; + if (reason == END_STREAM_REASON_TORPROTOCOL || + reason == END_STREAM_REASON_INTERNAL || + reason == END_STREAM_REASON_DESTROY) { + /* All three of these reasons could mean a failed tag + * hit the exit and it shat itself. Do not probe. + * Fail the circuit. */ + circ->path_state = PATH_STATE_USE_FAILED; + return -END_CIRC_REASON_TORPROTOCOL; + } else { + /* Path bias: If we get a valid reason code from the exit, + * it wasn't due to tagging */ + // XXX: This relies on recognized+digest being strong enough not + // to be spoofable.. Is that a valid assumption? + // Or more accurately: is it better than nothing? Can the attack + // be done offline? + circ->path_state = PATH_STATE_USE_SUCCEEDED; + } } if (rh->length > 0 && edge_reason_is_retriable(reason) &&

1 0

[tor/master] Changes from Nick's code review 'part 1'
by nickm＠torproject.org 26 Dec '12

26 Dec '12

commit b0fc18c37e0d30d9941109382df4b2b44f0f0ff8 Author: Mike Perry <mikeperry-git(a)fscked.org> Date: Tue Dec 18 12:39:03 2012 -0800 Changes from Nick's code review 'part 1' I think this is actually his third code review of this branch so far. --- src/or/circuitbuild.c | 90 +++++++++++++++++++++++++--------------------- src/or/config.c | 4 +- src/or/connection_edge.c | 2 +- src/or/entrynodes.c | 2 +- src/or/relay.c | 12 +++--- 5 files changed, 59 insertions(+), 51 deletions(-) diff --git a/src/or/circuitbuild.c b/src/or/circuitbuild.c index f93b04f..58020f2 100644 --- a/src/or/circuitbuild.c +++ b/src/or/circuitbuild.c @@ -1015,7 +1015,7 @@ pathbias_get_notice_rate(const or_options_t *options) /* XXXX024 I'd like to have this be static again, but entrynodes.c needs it. */ /** The circuit success rate below which we issue a warn */ -double +static double pathbias_get_warn_rate(const or_options_t *options) { #define DFLT_PATH_BIAS_WARN_PCT 50 @@ -1055,7 +1055,7 @@ pathbias_get_dropguards(const or_options_t *options) return options->PathBiasDropGuards; else return networkstatus_get_param(NULL, "pb_dropguards", - DFLT_PATH_BIAS_DROP_GUARDS, 0, 100)/100.0; + DFLT_PATH_BIAS_DROP_GUARDS, 0, 1); } /** @@ -1078,9 +1078,10 @@ pathbias_get_scale_threshold(const or_options_t *options) /** * The scale factor is the denominator for our scaling - * of circuit counts for our path bias window. Note that - * we must be careful of the values we use here, as the - * code only scales in the event of no integer truncation. + * of circuit counts for our path bias window. + * + * Note that our use of doubles for the path bias state + * file means that powers of 2 work best here. */ static int pathbias_get_scale_factor(const or_options_t *options) @@ -1301,7 +1302,7 @@ pathbias_count_circ_attempt(origin_circuit_t *circ) } else { if ((rate_msg = rate_limit_log(&circ_attempt_notice_limit, approx_time()))) { - log_info(LD_BUG, + log_info(LD_CIRC, "Unopened circuit has no known guard. " "Circuit is a %s currently %s.%s", circuit_purpose_to_string(circ->base_.purpose), @@ -1378,7 +1379,7 @@ pathbias_count_build_success(origin_circuit_t *circ) } else if (circ->base_.purpose != CIRCUIT_PURPOSE_C_MEASURE_TIMEOUT) { if ((rate_msg = rate_limit_log(&success_notice_limit, approx_time()))) { - log_info(LD_BUG, + log_info(LD_CIRC, "Completed circuit has no known guard. " "Circuit is a %s currently %s.%s", circuit_purpose_to_string(circ->base_.purpose), @@ -1490,7 +1491,7 @@ pathbias_count_successful_close(origin_circuit_t *circ) /* In rare cases, CIRCUIT_PURPOSE_TESTING can get converted to * CIRCUIT_PURPOSE_C_MEASURE_TIMEOUT and have no guards here. * No need to log that case. */ - log_info(LD_BUG, + log_info(LD_CIRC, "Successfully closed circuit has no known guard. " "Circuit is a %s currently %s", circuit_purpose_to_string(circ->base_.purpose), @@ -1526,7 +1527,7 @@ pathbias_count_collapse(origin_circuit_t *circ) /* In rare cases, CIRCUIT_PURPOSE_TESTING can get converted to * CIRCUIT_PURPOSE_C_MEASURE_TIMEOUT and have no guards here. * No need to log that case. */ - log_info(LD_BUG, + log_info(LD_CIRC, "Destroyed circuit has no known guard. " "Circuit is a %s currently %s", circuit_purpose_to_string(circ->base_.purpose), @@ -1554,7 +1555,7 @@ pathbias_count_unusable(origin_circuit_t *circ) /* In rare cases, CIRCUIT_PURPOSE_TESTING can get converted to * CIRCUIT_PURPOSE_C_MEASURE_TIMEOUT and have no guards here. * No need to log that case. */ - log_info(LD_BUG, + log_info(LD_CIRC, "Stream-failing circuit has no known guard. " "Circuit is a %s currently %s", circuit_purpose_to_string(circ->base_.purpose), @@ -1620,10 +1621,9 @@ pathbias_get_closed_count(entry_guard_t *guard) continue; if (ocirc->path_state >= PATH_STATE_BUILD_SUCCEEDED && - (memcmp(guard->identity, + fast_memeq(guard->identity, ocirc->cpath->extend_info->identity_digest, - DIGEST_LEN) - == 0)) { + DIGEST_LEN)) { open_circuits++; } } @@ -1670,15 +1670,17 @@ entry_guard_inc_circ_attempt_count(entry_guard_t *guard) "Your Guard %s=%s is failing an extremely large amount of " "circuits. To avoid potential route manipluation attacks, " "Tor has disabled use of this guard. " - "Success counts are %d/%d. %d circuits completed, %d " - "were unusable, %d collapsed, and %d timed out. For " + "Success counts are %ld/%ld. %ld circuits completed, %ld " + "were unusable, %ld collapsed, and %ld timed out. For " "reference, your timeout cutoff is %ld seconds.", guard->nickname, hex_str(guard->identity, DIGEST_LEN), - (int)pathbias_get_closed_count(guard), - (int)guard->circ_attempts, (int)guard->circ_successes, - (int)guard->unusable_circuits, - (int)guard->collapsed_circuits, (int)guard->timeouts, - (long)circ_times.close_ms/1000); + tor_lround(pathbias_get_closed_count(guard)), + tor_lround(guard->circ_attempts), + tor_lround(guard->circ_successes), + tor_lround(guard->unusable_circuits), + tor_lround(guard->collapsed_circuits), + tor_lround(guard->timeouts), + tor_lround(circ_times.close_ms/1000)); guard->path_bias_disabled = 1; guard->bad_since = approx_time(); return -1; @@ -1689,15 +1691,17 @@ entry_guard_inc_circ_attempt_count(entry_guard_t *guard) "Your Guard %s=%s is failing an extremely large amount of " "circuits. This could indicate a route manipulation attack, " "extreme network overload, or a bug. " - "Success counts are %d/%d. %d circuits completed, %d " - "were unusable, %d collapsed, and %d timed out. For " + "Success counts are %ld/%ld. %ld circuits completed, %ld " + "were unusable, %ld collapsed, and %ld timed out. For " "reference, your timeout cutoff is %ld seconds.", guard->nickname, hex_str(guard->identity, DIGEST_LEN), - (int)pathbias_get_closed_count(guard), - (int)guard->circ_attempts, (int)guard->circ_successes, - (int)guard->unusable_circuits, - (int)guard->collapsed_circuits, (int)guard->timeouts, - (long)circ_times.close_ms/1000); + tor_lround(pathbias_get_closed_count(guard)), + tor_lround(guard->circ_attempts), + tor_lround(guard->circ_successes), + tor_lround(guard->unusable_circuits), + tor_lround(guard->collapsed_circuits), + tor_lround(guard->timeouts), + tor_lround(circ_times.close_ms/1000)); } } else if (pathbias_get_success_count(guard)/((double)guard->circ_attempts) < pathbias_get_warn_rate(options)) { @@ -1708,15 +1712,17 @@ entry_guard_inc_circ_attempt_count(entry_guard_t *guard) "circuits. Most likely this means the Tor network is " "overloaded, but it could also mean an attack against " "you or the potentially the guard itself. " - "Success counts are %d/%d. %d circuits completed, %d " - "were unusable, %d collapsed, and %d timed out. For " + "Success counts are %ld/%ld. %ld circuits completed, %ld " + "were unusable, %ld collapsed, and %ld timed out. For " "reference, your timeout cutoff is %ld seconds.", guard->nickname, hex_str(guard->identity, DIGEST_LEN), - (int)pathbias_get_closed_count(guard), - (int)guard->circ_attempts, (int)guard->circ_successes, - (int)guard->unusable_circuits, - (int)guard->collapsed_circuits, (int)guard->timeouts, - (long)circ_times.close_ms/1000); + tor_lround(pathbias_get_closed_count(guard)), + tor_lround(guard->circ_attempts), + tor_lround(guard->circ_successes), + tor_lround(guard->unusable_circuits), + tor_lround(guard->collapsed_circuits), + tor_lround(guard->timeouts), + tor_lround(circ_times.close_ms/1000)); } } else if (pathbias_get_success_count(guard)/((double)guard->circ_attempts) < pathbias_get_notice_rate(options)) { @@ -1725,15 +1731,17 @@ entry_guard_inc_circ_attempt_count(entry_guard_t *guard) log_notice(LD_CIRC, "Your Guard %s=%s is failing more circuits than usual. " "Most likely this means the Tor network is overloaded. " - "Success counts are %d/%d. %d circuits completed, %d " - "were unusable, %d collapsed, and %d timed out. For " + "Success counts are %ld/%ld. %ld circuits completed, %ld " + "were unusable, %ld collapsed, and %ld timed out. For " "reference, your timeout cutoff is %ld seconds.", guard->nickname, hex_str(guard->identity, DIGEST_LEN), - (int)pathbias_get_closed_count(guard), - (int)guard->circ_attempts, (int)guard->circ_successes, - (int)guard->unusable_circuits, - (int)guard->collapsed_circuits, (int)guard->timeouts, - (long)circ_times.close_ms/1000); + tor_lround(pathbias_get_closed_count(guard)), + tor_lround(guard->circ_attempts), + tor_lround(guard->circ_successes), + tor_lround(guard->unusable_circuits), + tor_lround(guard->collapsed_circuits), + tor_lround(guard->timeouts), + tor_lround(circ_times.close_ms/1000)); } } } diff --git a/src/or/config.c b/src/or/config.c index 79725cb..1d81c54 100644 --- a/src/or/config.c +++ b/src/or/config.c @@ -322,8 +322,8 @@ static config_var_t option_vars_[] = { V(PathBiasScaleThreshold, INT, "-1"), V(PathBiasScaleFactor, INT, "-1"), V(PathBiasMultFactor, INT, "-1"), - V(PathBiasDropGuards, BOOL, "0"), - V(PathBiasUseCloseCounts, BOOL, "1"), + V(PathBiasDropGuards, AUTOBOOL, "0"), + V(PathBiasUseCloseCounts, AUTOBOOL, "1"), OBSOLETE("PathlenCoinWeight"), V(PerConnBWBurst, MEMUNIT, "0"), diff --git a/src/or/connection_edge.c b/src/or/connection_edge.c index 570ffe4..fa91c3a 100644 --- a/src/or/connection_edge.c +++ b/src/or/connection_edge.c @@ -2189,7 +2189,7 @@ connection_ap_handshake_socks_reply(entry_connection_t *conn, char *reply, // DNS remaps can trigger this. So can failed hidden service // lookups. log_info(LD_BUG, - "No origin circuit for successful SOCKS stream %ld. Reason: " + "No origin circuit for successful SOCKS stream %lu. Reason: " "%d", ENTRY_TO_CONN(conn)->global_identifier, endreason); } else { TO_ORIGIN_CIRCUIT(conn->edge_.on_circuit)->path_state diff --git a/src/or/entrynodes.c b/src/or/entrynodes.c index 066dbec..fc7f6ed 100644 --- a/src/or/entrynodes.c +++ b/src/or/entrynodes.c @@ -1194,7 +1194,7 @@ entry_guards_update_state(or_state_t *state) d, e->chosen_by_version, t); next = &(line->next); } - if (e->circ_attempts) { + if (e->circ_attempts > 0) { *next = line = tor_malloc_zero(sizeof(config_line_t)); line->key = tor_strdup("EntryGuardPathBias"); /* In the long run: circuit_success ~= successful_circuit_close + diff --git a/src/or/relay.c b/src/or/relay.c index b4b7700..8b3b27f 100644 --- a/src/or/relay.c +++ b/src/or/relay.c @@ -698,17 +698,17 @@ connection_ap_process_end_not_open( reason == END_STREAM_REASON_INTERNAL || reason == END_STREAM_REASON_DESTROY) { /* All three of these reasons could mean a failed tag - * hit the exit and it shat itself. Do not probe. + * hit the exit and it complained. Do not probe. * Fail the circuit. */ circ->path_state = PATH_STATE_USE_FAILED; return -END_CIRC_REASON_TORPROTOCOL; } else { /* Path bias: If we get a valid reason code from the exit, - * it wasn't due to tagging */ - // XXX: This relies on recognized+digest being strong enough not - // to be spoofable.. Is that a valid assumption? - // Or more accurately: is it better than nothing? Can the attack - // be done offline? + * it wasn't due to tagging. + * + * We rely on recognized+digest being strong enough to make + * tags unlikely to allow us to get tagged, yet 'recognized' + * reason codes here. */ circ->path_state = PATH_STATE_USE_SUCCEEDED; } }

1 0

[tor/master] Woops, this log message triggers with the 2-hop bias commit.
by nickm＠torproject.org 26 Dec '12

26 Dec '12

commit af9011f82430a5fac0a6db368f1afb1aa4bbc9f6 Author: Mike Perry <mikeperry-git(a)fscked.org> Date: Tue Dec 11 17:19:39 2012 -0800 Woops, this log message triggers with the 2-hop bias commit. --- src/or/circuitbuild.c | 16 ---------------- 1 files changed, 0 insertions(+), 16 deletions(-) diff --git a/src/or/circuitbuild.c b/src/or/circuitbuild.c index cc43797..3e2568c 100644 --- a/src/or/circuitbuild.c +++ b/src/or/circuitbuild.c @@ -1309,22 +1309,6 @@ pathbias_count_circ_attempt(origin_circuit_t *circ) } } } - } else { - /* Help track down the real cause of bug #6475: */ - if (circ->path_state == PATH_STATE_NEW_CIRC) { - if ((rate_msg = rate_limit_log(&circ_attempt_notice_limit, - approx_time()))) { - log_info(LD_BUG, - "A %s circuit is in cpath state %d (opened: %d). " - "Circuit is a %s currently %s.%s", - pathbias_state_to_string(circ->path_state), - circ->cpath->state, circ->has_opened, - circuit_purpose_to_string(circ->base_.purpose), - circuit_state_to_string(circ->base_.state), - rate_msg); - tor_free(rate_msg); - } - } } return 0;

1 0

[tor/master] Merge remote-tracking branch 'mikeperry/209-path-bias-changes'
by nickm＠torproject.org 26 Dec '12

26 Dec '12

commit 885e8d35c7af65b3e1e598d6f2e951ba84e65928 Merge: 265aab2 406d59a Author: Nick Mathewson <nickm(a)torproject.org> Date: Tue Dec 25 23:30:28 2012 -0500 Merge remote-tracking branch 'mikeperry/209-path-bias-changes' changes/bug7157 | 26 ++ doc/tor.1.txt | 41 ++- src/or/circuitbuild.c | 720 +++++++++++++++++++++++++++++++++++++--------- src/or/circuitbuild.h | 5 +- src/or/circuitlist.c | 12 +- src/or/circuituse.c | 33 +++ src/or/config.c | 7 +- src/or/connection_edge.c | 25 ++ src/or/entrynodes.c | 61 +++- src/or/entrynodes.h | 22 ++- src/or/or.h | 32 ++- src/or/relay.c | 20 ++ src/or/rendclient.c | 11 + src/or/rendservice.c | 8 + 14 files changed, 838 insertions(+), 185 deletions(-) diff --cc src/or/entrynodes.h index ae5d230,2686a4f..2a9a99c --- a/src/or/entrynodes.h +++ b/src/or/entrynodes.h @@@ -31,11 -31,14 +31,15 @@@ typedef struct entry_guard_t * router, 1 if we have. */ unsigned int can_retry : 1; /**< Should we retry connecting to this entry, * in spite of having it marked as unreachable?*/ - unsigned int path_bias_notice : 1; /**< Did we alert the user about path bias + unsigned int path_bias_noticed : 1; /**< Did we alert the user about path + * bias for this node already? */ + unsigned int path_bias_warned : 1; /**< Did we alert the user about path bias * for this node already? */ + unsigned int path_bias_extreme : 1; /**< Did we alert the user about path + * bias for this node already? */ unsigned int path_bias_disabled : 1; /**< Have we disabled this node because * of path bias issues? */ + unsigned int is_dir_cache : 1; /**< Is this node a directory cache? */ time_t bad_since; /**< 0 if this guard is currently usable, or the time at * which it was observed to become (according to the * directory or the user configuration) unusable. */

1 0

[tor/master] Fix compilation warning: must not format u64 as long.
by nickm＠torproject.org 26 Dec '12

26 Dec '12

commit 01a09e8f8601ce61be500c92ea200c29e4389c7a Author: Nick Mathewson <nickm(a)torproject.org> Date: Tue Dec 25 23:34:38 2012 -0500 Fix compilation warning: must not format u64 as long. --- src/or/connection_edge.c | 6 ++++-- 1 files changed, 4 insertions(+), 2 deletions(-) diff --git a/src/or/connection_edge.c b/src/or/connection_edge.c index a7c201f..defe944 100644 --- a/src/or/connection_edge.c +++ b/src/or/connection_edge.c @@ -2199,8 +2199,10 @@ connection_ap_handshake_socks_reply(entry_connection_t *conn, char *reply, // DNS remaps can trigger this. So can failed hidden service // lookups. log_info(LD_BUG, - "No origin circuit for successful SOCKS stream %lu. Reason: " - "%d", ENTRY_TO_CONN(conn)->global_identifier, endreason); + "No origin circuit for successful SOCKS stream "U64_FORMAT + ". Reason: %d", + U64_PRINTF_ARG(ENTRY_TO_CONN(conn)->global_identifier), + endreason); } else { TO_ORIGIN_CIRCUIT(conn->edge_.on_circuit)->path_state = PATH_STATE_USE_SUCCEEDED;

1 0