tor-commits March 2018

tor-commits@lists.torproject.org

15 participants
1430 discussions

[tor/release-0.3.3] Merge remote-tracking branch 'dgoulet/bug24767_033_03' into maint-0.3.3
by nickm＠torproject.org 28 Mar '18

28 Mar '18

commit e0bbef48bf414574a9e1342accfd2fe0d2a5343e Merge: 3cc382b93 ab16f1e2a Author: Nick Mathewson <nickm(a)torproject.org> Date: Wed Mar 28 14:21:04 2018 -0400 Merge remote-tracking branch 'dgoulet/bug24767_033_03' into maint-0.3.3 changes/bug24767 | 5 + src/or/connection_or.c | 225 +++++++++++++++++++++++++++++++++++++++++++++ src/or/connection_or.h | 5 + src/or/nodelist.c | 4 +- src/or/nodelist.h | 2 +- src/test/test_connection.c | 97 +++++++++++++++++++ 6 files changed, 335 insertions(+), 3 deletions(-)

1 0

[tor/release-0.3.3] Merge branch 'maint-0.3.3' into release-0.3.3
by nickm＠torproject.org 28 Mar '18

28 Mar '18

commit 2504c2d4ae5b5f37be35b95c20dc0ee40dd3c3cd Merge: e31487b4e e0bbef48b Author: Nick Mathewson <nickm(a)torproject.org> Date: Wed Mar 28 14:23:06 2018 -0400 Merge branch 'maint-0.3.3' into release-0.3.3 changes/bug24767 | 5 + src/or/connection_or.c | 225 +++++++++++++++++++++++++++++++++++++++++++++ src/or/connection_or.h | 5 + src/or/nodelist.c | 4 +- src/or/nodelist.h | 2 +- src/test/test_connection.c | 97 +++++++++++++++++++ 6 files changed, 335 insertions(+), 3 deletions(-)

1 0

[tor/release-0.3.3] relay: Avoid connecting to down relays
by nickm＠torproject.org 28 Mar '18

28 Mar '18

commit f29d158330a8c6bcc91b71a888db741766135aaf Author: David Goulet <dgoulet(a)torproject.org> Date: Tue Feb 6 15:40:17 2018 -0500 relay: Avoid connecting to down relays If we failed to connect at the TCP level to a relay, note it down and refuse to connect again for another 60 seconds. Fixes #24767 Signed-off-by: David Goulet <dgoulet(a)torproject.org> --- changes/bug24767 | 5 ++ src/or/connection_or.c | 224 +++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 229 insertions(+) diff --git a/changes/bug24767 b/changes/bug24767 new file mode 100644 index 000000000..56fbe51a9 --- /dev/null +++ b/changes/bug24767 @@ -0,0 +1,5 @@ + o Major bugfixes (relay, connection): + - Refuse to connect again to a relay from which we failed previously with + a connection refused, timeout or error (at the TCP level). The relay + won't be retried for 60 seconds after the failure occured. Fixes bug + 24767; bugfix on 0.0.6. diff --git a/src/or/connection_or.c b/src/or/connection_or.c index 272a086a3..976889ad0 100644 --- a/src/or/connection_or.c +++ b/src/or/connection_or.c @@ -1122,6 +1122,216 @@ connection_or_group_set_badness_(smartlist_t *group, int force) } SMARTLIST_FOREACH_END(or_conn); } +/* Lifetime of a connection failure. After that, we'll retry. This is in + * seconds. */ +#define OR_CONNECT_FAILURE_LIFETIME 60 +/* The interval to use with when to clean up the failure cache. */ +#define OR_CONNECT_FAILURE_CLEANUP_INTERVAL 60 + +/* When is the next time we have to cleanup the failure map. We keep this + * because we clean it opportunistically. */ +static time_t or_connect_failure_map_next_cleanup_ts = 0; + +/* OR connection failure entry data structure. It is kept in the connection + * failure map defined below and indexed by OR identity digest, address and + * port. + * + * We need to identify a connection failure with these three values because we + * want to avoid to wrongfully blacklist a relay if someone is trying to + * extend to a known identity digest but with the wrong IP/port. For instance, + * it can happen if a relay changed its port but the client still has an old + * descriptor with the old port. We want to stop connecting to that + * IP/port/identity all together, not only the relay identity. */ +typedef struct or_connect_failure_entry_t { + HT_ENTRY(or_connect_failure_entry_t) node; + /* Identity digest of the connection where it is connecting to. */ + uint8_t identity_digest[DIGEST_LEN]; + /* This is the connection address from the base connection_t. After the + * connection is checked for canonicity, the base address should represent + * what we know instead of where we are connecting to. This is what we need + * so we can correlate known relays within the consensus. */ + tor_addr_t addr; + uint16_t port; + /* Last time we were unable to connect. */ + time_t last_failed_connect_ts; +} or_connect_failure_entry_t; + +/* Map where we keep connection failure entries. They are indexed by addr, + * port and identity digest. */ +static HT_HEAD(or_connect_failure_ht, or_connect_failure_entry_t) + or_connect_failures_map = HT_INITIALIZER(); + +/* Helper: Hashtable equal function. Return 1 if equal else 0. */ +static int +or_connect_failure_ht_eq(const or_connect_failure_entry_t *a, + const or_connect_failure_entry_t *b) +{ + return fast_memeq(a->identity_digest, b->identity_digest, DIGEST_LEN) && + tor_addr_eq(&a->addr, &b->addr) && + a->port == b->port; +} + +/* Helper: Return the hash for the hashtable of the given entry. For this + * table, it is a combination of address, port and identity digest. */ +static unsigned int +or_connect_failure_ht_hash(const or_connect_failure_entry_t *entry) +{ + size_t offset = 0, addr_size; + const void *addr_ptr; + /* Largest size is IPv6 and IPv4 is smaller so it is fine. */ + uint8_t data[16 + sizeof(uint16_t) + DIGEST_LEN]; + + /* Get the right address bytes depending on the family. */ + switch (tor_addr_family(&entry->addr)) { + case AF_INET: + addr_size = 4; + addr_ptr = &entry->addr.addr.in_addr.s_addr; + break; + case AF_INET6: + addr_size = 16; + addr_ptr = &entry->addr.addr.in6_addr.s6_addr; + break; + default: + tor_assert_nonfatal_unreached(); + return 0; + } + + memcpy(data, addr_ptr, addr_size); + offset += addr_size; + memcpy(data + offset, entry->identity_digest, DIGEST_LEN); + offset += DIGEST_LEN; + set_uint16(data + offset, entry->port); + offset += sizeof(uint16_t); + + return (unsigned int) siphash24g(data, offset); +} + +HT_PROTOTYPE(or_connect_failure_ht, or_connect_failure_entry_t, node, + or_connect_failure_ht_hash, or_connect_failure_ht_eq) + +HT_GENERATE2(or_connect_failure_ht, or_connect_failure_entry_t, node, + or_connect_failure_ht_hash, or_connect_failure_ht_eq, + 0.6, tor_reallocarray_, tor_free_) + +/* Initialize a given connect failure entry with the given identity_digest, + * addr and port. All field are optional except ocf. */ +static void +or_connect_failure_init(const char *identity_digest, const tor_addr_t *addr, + uint16_t port, or_connect_failure_entry_t *ocf) +{ + tor_assert(ocf); + if (identity_digest) { + memcpy(ocf->identity_digest, identity_digest, + sizeof(ocf->identity_digest)); + } + if (addr) { + tor_addr_copy(&ocf->addr, addr); + } + ocf->port = port; +} + +/* Return a newly allocated connection failure entry. It is initialized with + * the given or_conn data. This can't fail. */ +static or_connect_failure_entry_t * +or_connect_failure_new(const or_connection_t *or_conn) +{ + or_connect_failure_entry_t *ocf = tor_malloc_zero(sizeof(*ocf)); + or_connect_failure_init(or_conn->identity_digest, &or_conn->real_addr, + TO_CONN(or_conn)->port, ocf); + return ocf; +} + +/* Return a connection failure entry matching the given or_conn. NULL is + * returned if not found. */ +static or_connect_failure_entry_t * +or_connect_failure_find(const or_connection_t *or_conn) +{ + or_connect_failure_entry_t lookup; + tor_assert(or_conn); + or_connect_failure_init(or_conn->identity_digest, &TO_CONN(or_conn)->addr, + TO_CONN(or_conn)->port, &lookup); + return HT_FIND(or_connect_failure_ht, &or_connect_failures_map, &lookup); +} + +/* Note down in the connection failure cache that a failure occurred on the + * given or_conn. */ +static void +note_or_connect_failed(const or_connection_t *or_conn) +{ + or_connect_failure_entry_t *ocf = NULL; + + tor_assert(or_conn); + + ocf = or_connect_failure_find(or_conn); + if (ocf == NULL) { + ocf = or_connect_failure_new(or_conn); + HT_INSERT(or_connect_failure_ht, &or_connect_failures_map, ocf); + } + ocf->last_failed_connect_ts = approx_time(); +} + +/* Cleanup the connection failure cache and remove all entries below the + * given cutoff. */ +static void +or_connect_failure_map_cleanup(time_t cutoff) +{ + or_connect_failure_entry_t **ptr, **next, *entry; + + for (ptr = HT_START(or_connect_failure_ht, &or_connect_failures_map); + ptr != NULL; ptr = next) { + entry = *ptr; + if (entry->last_failed_connect_ts <= cutoff) { + next = HT_NEXT_RMV(or_connect_failure_ht, &or_connect_failures_map, ptr); + tor_free(entry); + } else { + next = HT_NEXT(or_connect_failure_ht, &or_connect_failures_map, ptr); + } + } +} + +/* Return true iff the given OR connection can connect to its destination that + * is the triplet identity_digest, address and port. + * + * The or_conn MUST have gone through connection_or_check_canonicity() so the + * base address is properly set to what we know or doesn't know. */ +static int +should_connect_to_relay(const or_connection_t *or_conn) +{ + time_t now, cutoff; + time_t connect_failed_since_ts = 0; + or_connect_failure_entry_t *ocf; + + tor_assert(or_conn); + + now = approx_time(); + cutoff = now - OR_CONNECT_FAILURE_LIFETIME; + + /* Opportunistically try to cleanup the failure cache. We do that at regular + * interval so it doesn't grow too big. */ + if (or_connect_failure_map_next_cleanup_ts <= now) { + or_connect_failure_map_cleanup(cutoff); + or_connect_failure_map_next_cleanup_ts = + now + OR_CONNECT_FAILURE_CLEANUP_INTERVAL; + } + + /* Look if we have failed previously to the same destination as this + * OR connection. */ + ocf = or_connect_failure_find(or_conn); + if (ocf) { + connect_failed_since_ts = ocf->last_failed_connect_ts; + } + /* If we do have an unable to connect timestamp and it is below cutoff, we + * can connect. Or we have never failed before so let it connect. */ + if (connect_failed_since_ts > cutoff) { + goto no_connect; + } + + /* Ok we can connect! */ + return 1; + no_connect: + return 0; +} + /** <b>conn</b> is in the 'connecting' state, and it failed to complete * a TCP connection. Send notifications appropriately. * @@ -1135,6 +1345,7 @@ connection_or_connect_failed(or_connection_t *conn, control_event_or_conn_status(conn, OR_CONN_EVENT_FAILED, reason); if (!authdir_mode_tests_reachability(get_options())) control_event_bootstrap_prob_or(msg, reason, conn); + note_or_connect_failed(conn); } /** <b>conn</b> got an error in connection_handle_read_impl() or @@ -1225,6 +1436,19 @@ connection_or_connect, (const tor_addr_t *_addr, uint16_t port, conn->chan = chan; chan->conn = conn; connection_or_init_conn_from_address(conn, &addr, port, id_digest, ed_id, 1); + + /* We have a proper OR connection setup, now check if we can connect to it + * that is we haven't had a failure earlier. This is to avoid to try to + * constantly connect to relays that we think are not reachable. */ + if (!should_connect_to_relay(conn)) { + log_info(LD_GENERAL, "Can't connect to identity %s at %s:%u because we " + "failed earlier. Refusing.", + hex_str(id_digest, DIGEST_LEN), fmt_addr(&TO_CONN(conn)->addr), + TO_CONN(conn)->port); + connection_free_(TO_CONN(conn)); + return NULL; + } + connection_or_change_state(conn, OR_CONN_STATE_CONNECTING); control_event_or_conn_status(conn, OR_CONN_EVENT_LAUNCHED, 0);

1 0

[translation/tor-launcher-properties] Update translations for tor-launcher-properties
by translation＠torproject.org 28 Mar '18

28 Mar '18

commit bf9f0fd54c102703e6c1c1499301ce27f36f17c0 Author: Translation commit bot <translation(a)torproject.org> Date: Wed Mar 28 16:16:24 2018 +0000 Update translations for tor-launcher-properties --- nb/torlauncher.properties | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/nb/torlauncher.properties b/nb/torlauncher.properties index 7e4a018e6..8e97a0dc1 100644 --- a/nb/torlauncher.properties +++ b/nb/torlauncher.properties @@ -32,10 +32,10 @@ torlauncher.error_bridge_bad_default_type=Det er ingen angitte broer som har til torlauncher.bridge_suffix.meek-amazon=(virker i Kina) torlauncher.bridge_suffix.meek-azure=(virker i Kina) -torlauncher.request_a_bridge=Request a Bridge… -torlauncher.request_a_new_bridge=Request a New Bridge… +torlauncher.request_a_bridge=Forespør en bro… +torlauncher.request_a_new_bridge=Forespør ny bro… torlauncher.contacting_bridgedb=Contacting BridgeDB. Please wait. -torlauncher.captcha_prompt=Solve the CAPTCHA to request a bridge. +torlauncher.captcha_prompt=Løs CAPTCHA-en for å forespørre en bro. torlauncher.bad_captcha_solution=The solution is not correct. Please try again. torlauncher.unable_to_get_bridge=Unable to obtain a bridge from BridgeDB.\n\n%S torlauncher.no_meek=This browser is not configured for meek, which is needed to obtain bridges.

1 0

[webwml/master] Update Tor Messenger activity status on volunteer page
by hiro＠torproject.org 28 Mar '18

28 Mar '18

commit 21f81f99eade9e633cc443a8f6d3b9459fb01745 Author: Sukhbir Singh <sukhbir(a)torproject.org> Date: Wed Mar 28 10:22:18 2018 -0400 Update Tor Messenger activity status on volunteer page Signed-off-by: hiro <hiro(a)torproject.org> --- getinvolved/en/volunteer.wml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/getinvolved/en/volunteer.wml b/getinvolved/en/volunteer.wml index 6d38a40f..22eeaa38 100644 --- a/getinvolved/en/volunteer.wml +++ b/getinvolved/en/volunteer.wml @@ -174,7 +174,7 @@ meetings around the world.</li> <td><a href="#project-tormessenger">Tor Messenger</a></td> <td>Bundle</td> <td>JavaScript, XUL, Scripting</td> - <td>Heavy</td> + <td>None</td> <td>arlolra, boklm, sukhe</td> </tr>

1 0

[torbirdy/master] Update the ChangeLog
by sukhbir＠torproject.org 28 Mar '18

28 Mar '18

commit 670d3488f82acce67df4859d23bc6c4a23ee2907 Author: Sukhbir Singh <sukhbir(a)torproject.org> Date: Wed Mar 28 11:33:51 2018 -0400 Update the ChangeLog --- ChangeLog | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/ChangeLog b/ChangeLog index d95c5d2..935e39d 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,6 +1,8 @@ 0.2.4, - * Update translations, adding all supported languages from Transifex + * Bug 22484: TB 52+ leaks installed dictionary + * Bug 22944: Disable fetching site icons * Added support for Thunderbird 58 and Thunderbird 59 + * Update translations, adding all supported languages from Transifex 0.2.3, 04 Aug 2017 * Bug 21880: Enable encrypted email headers for Enigmail (Memory Hole)

1 0

[torbirdy/master] Prevent leaking user's installed dictionary (Bug 22484)
by sukhbir＠torproject.org 28 Mar '18

28 Mar '18

commit 63fa6e508d5ed3218ca9f261a00347f0027b0913 Author: Sukhbir Singh <sukhbir(a)torproject.org> Date: Wed Mar 28 11:31:53 2018 -0400 Prevent leaking user's installed dictionary (Bug 22484) Also set the Content-Language to "en-US" to prevent leaking the user's default dictionary. Reported in https://bugs.torproject.org/22484 and discussed in Section 4 of RFC 3282. Thunderbird bug: https://bugzilla.mozilla.org/show_bug.cgi?id=1370217 --- components/torbirdy.js | 20 +++++++++++++++----- 1 file changed, 15 insertions(+), 5 deletions(-) diff --git a/components/torbirdy.js b/components/torbirdy.js index bfe2427..e8d2cd2 100644 --- a/components/torbirdy.js +++ b/components/torbirdy.js @@ -391,12 +391,14 @@ var TorBirdyOldPrefs = [ "network.proxy.http", ] -// sanitizeDateHeaders() -// Run this function to make sure that the Date header in a new message -// is rounded down to the nearest minute. -function sanitizeDateHeaders() { +// sanitizeHeaders() +// Sanitize the "Date" and "Content-Language" headers. +function sanitizeHeaders() { // Import the jsmime module that is used to generate mail headers. let { jsmime } = Cu.import("resource:///modules/jsmime.jsm"); + // Date + // Run this function to make sure that the Date header in a new message + // is rounded down to the nearest minute. // Inject our own structured encoder to the default header emitter, // to override the default Date encoder with a rounded-down version. jsmime.headeremitter.addStructuredEncoder("Date", function (date) { @@ -410,6 +412,14 @@ function sanitizeDateHeaders() { // We replace the "GMT" symbol with "+0000" because it is preferred. this.addText(roundedDate.toUTCString().replace(/GMT$/, "+0000"), false); }); + // Content-Language + // Also set the Content-Language to "en-US" to prevent leaking the user's + // default dictionary. Reported in https://bugs.torproject.org/22484 and + // discussed in Section 4 of RFC 3282. + // Thunderbird bug: https://bugzilla.mozilla.org/show_bug.cgi?id=1370217 + jsmime.headeremitter.addStructuredEncoder("Content-Language", function (locale) { + this.addText("en-US", false); + }); } function TorBirdy() { @@ -439,7 +449,7 @@ function TorBirdy() { this.setAccountPrefs(); this.setPrefs(); - sanitizeDateHeaders(); + sanitizeHeaders(); }

1 0

[metrics-web/master] Update to latest metrics-lib.
by karsten＠torproject.org 28 Mar '18

28 Mar '18

commit d85e4368a4d91846a1541aa1a72242bb001a5e3c Author: Karsten Loesing <karsten.loesing(a)gmx.net> Date: Wed Mar 28 16:43:16 2018 +0200 Update to latest metrics-lib. --- src/submods/metrics-lib | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/submods/metrics-lib b/src/submods/metrics-lib index 2e235ef..f826234 160000 --- a/src/submods/metrics-lib +++ b/src/submods/metrics-lib @@ -1 +1 @@ -Subproject commit 2e235ef8f38aac679817fa8e8b9171aadb3d5b9d +Subproject commit f826234e8064014487d871838e8074e83a142e42

1 0

[metrics-web/master] Make generate-specs target optional.
by karsten＠torproject.org 28 Mar '18

28 Mar '18

commit ac10cec5a86dbbfd821bb6572636fe4cb23cbdde Author: Karsten Loesing <karsten.loesing(a)gmx.net> Date: Wed Mar 28 16:41:19 2018 +0200 Make generate-specs target optional. Without this change, `ant war` would not succeed, because it finds the spec JSP files in two places with the same name, and that is not allowed. --- build.xml | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) diff --git a/build.xml b/build.xml index 0e05d12..3b4222e 100644 --- a/build.xml +++ b/build.xml @@ -151,7 +151,6 @@ </jar> <antcall target="generate-ml-javadoc" /> - <antcall target="generate-specs" /> <war destfile="${dist}/${warfile}" duplicate="fail" @@ -186,9 +185,6 @@ <zipfileset dir="${resources}/web/jsps" prefix="WEB-INF" includes="*.jsp"/> - <zipfileset dir="${specdir}" - prefix="WEB-INF" - includes="*.jsp"/> <zipfileset dir="${resources}" includes="*.properties" /> <fileset dir="${resources}" includes="jetty.xml" /> @@ -287,7 +283,7 @@ <exec executable="awk" dir="${specdir}" logError="true" - output="${specdir}/${specfile}.jsp" + output="${resources}/web/jsps/${specfile}.jsp" failonerror="true" > <arg value="-f" /> <arg value="${basedir}/${resources}/spec/convert.awk" />

1 0

[metrics-web/master] Update news.json to version 247 of doc/MetricsTimeline.
by karsten＠torproject.org 28 Mar '18

28 Mar '18

commit b5f3e402159a7f5ed9d955aa83324343123f0cb1 Author: Karsten Loesing <karsten.loesing(a)gmx.net> Date: Wed Mar 28 17:17:43 2018 +0200 Update news.json to version 247 of doc/MetricsTimeline. --- src/main/resources/web/json/news.json | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) diff --git a/src/main/resources/web/json/news.json b/src/main/resources/web/json/news.json index 2f72791..8081a2f 100644 --- a/src/main/resources/web/json/news.json +++ b/src/main/resources/web/json/news.json @@ -4515,6 +4515,24 @@ ] }, { + "start": "2018-03-26", + "protocols": [ + "snowflake" + ], + "shortDescription": "Release of Tor Browser 8.0a5", + "description": "Release of Tor Browser 8.0a5. Improves snowflake client performance.", + "links": [ + { + "label": "blog post", + "target": "https://blog.torproject.org/tor-browser-80a5-released" + }, + { + "label": "ticket", + "target": "https://bugs.torproject.org/21312" + } + ] + }, + { "start": "2016-02-24", "places": [ "tm"

1 0

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

tor-commits March 2018