- tor-commits - lists.torproject.org

[torbutton/master] Bug #3960: Don't disable zoom.siteSpecific on TBB
by mikeperry＠torproject.org 08 Sep '11

08 Sep '11

commit 20cb5232a2b76ab1edd7991127fb0a4df440bb28 Author: Mike Perry <mikeperry-git(a)fscked.org> Date: Wed Sep 7 21:58:09 2011 -0700 Bug #3960: Don't disable zoom.siteSpecific on TBB Bug 3229's fix no longer requires us to disable this pref --- src/chrome/content/torbutton.js | 4 +++- 1 files changed, 3 insertions(+), 1 deletions(-) diff --git a/src/chrome/content/torbutton.js b/src/chrome/content/torbutton.js index b6f7f3b..3998770 100644 --- a/src/chrome/content/torbutton.js +++ b/src/chrome/content/torbutton.js @@ -1699,8 +1699,10 @@ function torbutton_update_status(mode, force_update) { torbutton_setBoolPref("browser.cache.offline.enable", "offline_cache", !mode, mode, changed); - torbutton_setBoolPref("browser.zoom.siteSpecific", "zoom_specific", + if (!m_tb_tbb) { + torbutton_setBoolPref("browser.zoom.siteSpecific", "zoom_specific", !mode, mode, changed); + } // Disable safebrowsing in Tor for FF2. It fetches some info in // cleartext with no HMAC (Firefox Bug 360387)

1 0

[torspec/master] Oops; I never added the refillintervals proposal
by nickm＠torproject.org 08 Sep '11

08 Sep '11

commit 87ad7d40a58c1cb902fcc87aa53123ef9eb93f30 Author: Nick Mathewson <nickm(a)torproject.org> Date: Wed Sep 7 22:07:49 2011 -0400 Oops; I never added the refillintervals proposal --- proposals/000-index.txt | 2 + proposals/183-refillintervals.txt | 89 +++++++++++++++++++++++++++++++++++++ 2 files changed, 91 insertions(+), 0 deletions(-) diff --git a/proposals/000-index.txt b/proposals/000-index.txt index 5e6cd41..c4106e8 100644 --- a/proposals/000-index.txt +++ b/proposals/000-index.txt @@ -103,6 +103,7 @@ Proposals by number: 180 Pluggable transports for circumvention [OPEN] 181 Optimistic Data for Tor: Client Side [CLOSED] 182 Credit Bucket [DRAFT] +183 Refill Intervals [CLOSED] Proposals by status: @@ -186,6 +187,7 @@ Proposals by status: 171 Separate streams across circuits by connection metadata [in 0.2.3.3-alpha] 174 Optimistic Data for Tor: Server Side [in 0.2.3.1-alpha] 181 Optimistic Data for Tor: Client Side [in 0.2.3.3-alpha] + 183 Refill Intervals [in 0.2.3.4-alpha] SUPERSEDED: 112 Bring Back Pathlen Coin Weight 113 Simplifying directory authority administration diff --git a/proposals/183-refillintervals.txt b/proposals/183-refillintervals.txt new file mode 100644 index 0000000..c823d40 --- /dev/null +++ b/proposals/183-refillintervals.txt @@ -0,0 +1,89 @@ +Filename: 183-refillintervals.txt +Title: Refill Intervals +Author: Florian Tschorsch and Björn Scheuermann +Created: 03-Dec-2010 +Status: Open + +Overview: + + In order to avoid additional queuing and bursty traffic, the refill + interval of the token bucket algorithm should be shortened. Thus we + propose a configurable parameter that sets the refill interval + accordingly. + +Motivation and Background: + + In Tor there exist multiple token buckets on different logical levels. They + all work independently. They are used to limit the up- and downstream of an + onion router. All token buckets are refilled every second with a constant + amount of tokens that depends on the configured bandwidth limits. The very + coarse-grained refill interval of one second has detrimental effects. + + First, consider an onion router with multiple TLS connections over which + cells arrive. If there is high activity (i.e., many incoming cells in + total), then the coarse refill interval will cause unfairness. Assume (just + for simplicity) that C doesn't share its TLS connection with any other + circuit. Moreover, assume that C hasn't transmitted any data for some time + (e.g., due a typical bursty HTTP traffic pattern). Consequently, there are + no cells from this circuit in the incoming socket buffers. When the buckets + are refilled, the incoming token bucket will immediately spend all its + tokens on other incoming connections. Now assume that cells from C arrive + soon after. For fairness' sake, these cells should be serviced timely -- + circuit C hasn't received any bandwidth for a significant time before. + However, it will take a very long time (one refill interval) before the + current implementation will fetch these cells from the incoming TLS + connection, because the token bucket will remain empty for a long time. Just + because the cells happened to arrive at the "wrong" point in time, they must + wait. Such situations may occur even though the configured admissible + incoming data rate is not exceeded by incoming cells: the long refill + intervals often lead to an operational state where all the cells that were + admissible during a given one-second period are queued until the end of this + second, before the onion router even just starts processing them. This + results in unnecessary, long queuing delays in the incoming socket buffers. + These delays are not visible in the Tor circuit queue delay statistics [1]. + + Finally, the coarse-grained refill intervals result in a very bursty outgoing + traffic pattern at the onion routers (one large chunk of data once per + second, instead of smooth transmission progress). This is undesirable, since + such a traffic pattern can interfere with TCP's control mechanisms and can + be the source of suboptimal TCP performance on the TLS links between onion + routers. + +Specific Changes: + + The token buckets should be refilled more often, with a correspondingly + smaller amount of tokens. For instance, the buckets might be refilled every + 10 milliseconds with one-hundredth of the amount of data admissible per + second. This will help to overcome the problem of unfairness when reading + from the incoming socket buffers. At the same time it smoothes the traffic + leaving the onion routers. We are aware that this latter change has + apparently been discussed before [2]; we are not sure why this change has + not been implemented yet. + + In particular we need to change the current implementation in Tor which + triggers refilling always after exactly one second. Instead the refill event + should fire more frequently. The smaller time intervals between each refill + action need to be taken into account for the number of tokens that are added + to the bucket. + + With libevent 2.x and bufferevents enabled, smaller refill intervals are + already considered but hard coded. This should be changed to a configurable + parameter, too. + +Conclusion: + + This proposal can be implemented with moderate effort and requires changes + only at the points where the token bucket operations are currently + performed. + + This change will also be a good starting point for further enhancements + to improve queuing times in Tor. I.e. it will pave the ground for other means + that tackle this problem. + + Feedback is highly appreciated. + +References: + + [1] Karsten Loesing. Analysis of Circuit Queues in Tor. August 25, 2009. + [2] https://trac.torproject.org/projects/tor/wiki/sponsors/SponsorD/June2011 +

1 0

[tor/master] Fix whitespace issues in patches merged today so far
by nickm＠torproject.org 08 Sep '11

08 Sep '11

commit d3ff167e098fa44e4ec30c7e6f50127a8c99dae7 Author: Nick Mathewson <nickm(a)torproject.org> Date: Wed Sep 7 20:26:58 2011 -0400 Fix whitespace issues in patches merged today so far --- src/or/connection.c | 3 ++- src/or/connection_edge.c | 16 +++++++++------- src/or/connection_edge.h | 1 - src/or/or.h | 4 ++-- src/or/relay.c | 25 +++++++++++++------------ 5 files changed, 26 insertions(+), 23 deletions(-) diff --git a/src/or/connection.c b/src/or/connection.c index deb6d2c..790383b 100644 --- a/src/or/connection.c +++ b/src/or/connection.c @@ -970,7 +970,8 @@ connection_create_listener(const struct sockaddr *listensockaddr, log_warn(LD_NET,"Unable to chown() %s socket: user %s not found.", address, options->User); } else if (chown(address, pw->pw_uid, pw->pw_gid) < 0) { - log_warn(LD_NET,"Unable to chown() %s socket: %s.", address, strerror(errno)); + log_warn(LD_NET,"Unable to chown() %s socket: %s.", + address, strerror(errno)); goto err; } } diff --git a/src/or/connection_edge.c b/src/or/connection_edge.c index 0d2c10f..f59f44c 100644 --- a/src/or/connection_edge.c +++ b/src/or/connection_edge.c @@ -140,7 +140,7 @@ connection_edge_process_inbuf(edge_connection_t *conn, int package_partial) switch (conn->_base.state) { case AP_CONN_STATE_SOCKS_WAIT: - if (connection_ap_handshake_process_socks(EDGE_TO_ENTRY_CONN(conn)) < 0) { + if (connection_ap_handshake_process_socks(EDGE_TO_ENTRY_CONN(conn)) <0) { /* already marked */ return -1; } @@ -619,7 +619,8 @@ connection_ap_expire_beginning(void) if (connection_ap_detach_retriable(entry_conn, TO_ORIGIN_CIRCUIT(circ), END_STREAM_REASON_TIMEOUT)<0) { if (!base_conn->marked_for_close) - connection_mark_unattached_ap(entry_conn, END_STREAM_REASON_CANT_ATTACH); + connection_mark_unattached_ap(entry_conn, + END_STREAM_REASON_CANT_ATTACH); } } SMARTLIST_FOREACH_END(base_conn); } @@ -743,7 +744,8 @@ circuit_discard_optional_exit_enclaves(extend_info_t *info) * Returns -1 on err, 1 on success, 0 on not-yet-sure. */ int -connection_ap_detach_retriable(entry_connection_t *conn, origin_circuit_t *circ, +connection_ap_detach_retriable(entry_connection_t *conn, + origin_circuit_t *circ, int reason) { control_event_stream_status(conn, STREAM_EVENT_FAILED_RETRIABLE, reason); @@ -939,8 +941,8 @@ addressmap_clear_excluded_trackexithosts(const or_options_t *options) if (len < 6) continue; /* malformed. */ dot = target + len - 6; /* dot now points to just before .exit */ - while(dot > target && *dot != '.') - dot--; + while (dot > target && *dot != '.') + dot--; if (*dot == '.') dot++; nodename = tor_strndup(dot, len-5-(dot-target));; node = node_get_by_nickname(nodename, 0); @@ -2455,8 +2457,8 @@ connection_ap_handshake_send_begin(entry_connection_t *ap_conn) connection_ap_supports_optimistic_data(ap_conn)) { log_info(LD_APP, "Sending up to %ld + %ld bytes of queued-up data", (long)connection_get_inbuf_len(base_conn), - ap_conn->sending_optimistic_data ? - (long)generic_buffer_len(ap_conn->sending_optimistic_data) : 0); + ap_conn->sending_optimistic_data ? + (long)generic_buffer_len(ap_conn->sending_optimistic_data) : 0); if (connection_edge_package_raw_inbuf(edge_conn, 1, NULL) < 0) { connection_mark_for_close(base_conn); } diff --git a/src/or/connection_edge.h b/src/or/connection_edge.h index 75a54e8..830667e 100644 --- a/src/or/connection_edge.h +++ b/src/or/connection_edge.h @@ -105,7 +105,6 @@ hostname_type_t parse_extended_hostname(char *address, int allowdotexit); int get_pf_socket(void); #endif - int connection_edge_compatible_with_circuit(const entry_connection_t *conn, const origin_circuit_t *circ); int connection_edge_update_circuit_isolation(const entry_connection_t *conn, diff --git a/src/or/or.h b/src/or/or.h index ff94a3e..bf404df 100644 --- a/src/or/or.h +++ b/src/or/or.h @@ -1389,8 +1389,8 @@ static edge_connection_t *TO_EDGE_CONN(connection_t *); /** Convert a connection_t* to an entry_connection_t*; assert if the cast is * invalid. */ static entry_connection_t *TO_ENTRY_CONN(connection_t *); -/** Convert a edge_connection_t* to an entry_connection_t*; assert if the cast is - * invalid. */ +/** Convert a edge_connection_t* to an entry_connection_t*; assert if the cast + * is invalid. */ static entry_connection_t *EDGE_TO_ENTRY_CONN(edge_connection_t *); /** Convert a connection_t* to an control_connection_t*; assert if the cast is * invalid. */ diff --git a/src/or/relay.c b/src/or/relay.c index 5bbcf0a..318c6c7 100644 --- a/src/or/relay.c +++ b/src/or/relay.c @@ -715,8 +715,8 @@ connection_ap_process_end_not_open( (void) layer_hint; /* unused */ if (rh->length > 0 && edge_reason_is_retriable(reason) && - !connection_edge_is_rendezvous_stream(edge_conn) /* avoid retry if rend */ - ) { + /* avoid retry if rend */ + !connection_edge_is_rendezvous_stream(edge_conn)) { const char *chosen_exit_digest = circ->build_state->chosen_exit->identity_digest; log_info(LD_APP,"Address '%s' refused due to '%s'. Considering retrying.", @@ -999,7 +999,8 @@ connection_edge_process_relay_cell_not_open( connection_ap_handshake_socks_resolved(entry_conn, RESOLVED_TYPE_ERROR_TRANSIENT, 0, NULL, 0, TIME_MAX); - connection_mark_unattached_ap(entry_conn,END_STREAM_REASON_TORPROTOCOL); + connection_mark_unattached_ap(entry_conn, + END_STREAM_REASON_TORPROTOCOL); return 0; } } @@ -1356,7 +1357,7 @@ int connection_edge_package_raw_inbuf(edge_connection_t *conn, int package_partial, int *max_cells) { - size_t amount_to_process, length; + size_t bytes_to_process, length; char payload[CELL_PAYLOAD_SIZE]; circuit_t *circ; const unsigned domain = conn->_base.type == CONN_TYPE_AP ? LD_APP : LD_EXIT; @@ -1403,26 +1404,26 @@ connection_edge_package_raw_inbuf(edge_connection_t *conn, int package_partial, entry_conn->sending_optimistic_data != NULL; if (PREDICT_UNLIKELY(sending_from_optimistic)) { - amount_to_process = generic_buffer_len(entry_conn->sending_optimistic_data); - if (PREDICT_UNLIKELY(!amount_to_process)) { + bytes_to_process = generic_buffer_len(entry_conn->sending_optimistic_data); + if (PREDICT_UNLIKELY(!bytes_to_process)) { log_warn(LD_BUG, "sending_optimistic_data was non-NULL but empty"); - amount_to_process = connection_get_inbuf_len(TO_CONN(conn)); + bytes_to_process = connection_get_inbuf_len(TO_CONN(conn)); sending_from_optimistic = 0; } } else { - amount_to_process = connection_get_inbuf_len(TO_CONN(conn)); + bytes_to_process = connection_get_inbuf_len(TO_CONN(conn)); } - if (!amount_to_process) + if (!bytes_to_process) return 0; - if (!package_partial && amount_to_process < RELAY_PAYLOAD_SIZE) + if (!package_partial && bytes_to_process < RELAY_PAYLOAD_SIZE) return 0; - if (amount_to_process > RELAY_PAYLOAD_SIZE) { + if (bytes_to_process > RELAY_PAYLOAD_SIZE) { length = RELAY_PAYLOAD_SIZE; } else { - length = amount_to_process; + length = bytes_to_process; } stats_n_data_bytes_packaged += length; stats_n_data_cells_packaged += 1;

1 0

[tor/master] chown() sockets when User option is set
by nickm＠torproject.org 08 Sep '11

08 Sep '11

commit 58a0afe30f390988b142b0497d2e78e66c3caf45 Author: Jérémy Bobbio <lunar(a)debian.org> Date: Wed Sep 7 18:51:24 2011 +0200 chown() sockets when User option is set Fixes bug 3421 --- changes/bug3421 | 6 ++++++ src/or/connection.c | 24 ++++++++++++++++++++++-- 2 files changed, 28 insertions(+), 2 deletions(-) diff --git a/changes/bug3421 b/changes/bug3421 new file mode 100644 index 0000000..8a40729 --- /dev/null +++ b/changes/bug3421 @@ -0,0 +1,6 @@ + o Minor features: + - When starting as root and then changing our UID via the User + control option, if we are running with ControlSocket, make sure + that the ControlSocket is owned by the same account that Tor will + run under. Implements ticket 3421; fix by Jérémy Bobbio. + diff --git a/src/or/connection.c b/src/or/connection.c index 97989c0..deb6d2c 100644 --- a/src/or/connection.c +++ b/src/or/connection.c @@ -40,6 +40,10 @@ #include <event2/event.h> #endif +#ifdef HAVE_PWD_H +#include <pwd.h> +#endif + static connection_t *connection_create_listener( const struct sockaddr *listensockaddr, socklen_t listensocklen, int type, @@ -859,6 +863,10 @@ connection_create_listener(const struct sockaddr *listensockaddr, listener_connection_t *lis_conn; connection_t *conn; tor_socket_t s; /* the socket we're going to make */ + or_options_t const *options = get_options(); +#if defined(HAVE_PWD_H) && defined(HAVE_SYS_UN_H) + struct passwd *pw = NULL; +#endif uint16_t usePort = 0, gotPort = 0; int start_reading = 0; static int global_next_session_group = SESSION_GROUP_FIRST_AUTO; @@ -931,7 +939,7 @@ connection_create_listener(const struct sockaddr *listensockaddr, * and listeners at the same time */ tor_assert(type == CONN_TYPE_CONTROL_LISTENER); - if (check_location_for_unix_socket(get_options(), address) < 0) + if (check_location_for_unix_socket(options, address) < 0) goto err; log_notice(LD_NET, "Opening %s on %s", @@ -955,7 +963,19 @@ connection_create_listener(const struct sockaddr *listensockaddr, tor_socket_strerror(tor_socket_errno(s))); goto err; } - if (get_options()->ControlSocketsGroupWritable) { +#ifdef HAVE_PWD_H + if (options->User) { + pw = getpwnam(options->User); + if (pw == NULL) { + log_warn(LD_NET,"Unable to chown() %s socket: user %s not found.", + address, options->User); + } else if (chown(address, pw->pw_uid, pw->pw_gid) < 0) { + log_warn(LD_NET,"Unable to chown() %s socket: %s.", address, strerror(errno)); + goto err; + } + } +#endif + if (options->ControlSocketsGroupWritable) { /* We need to use chmod; fchmod doesn't work on sockets on all * platforms. */ if (chmod(address, 0660) < 0) {

1 0

[tor/master] Clean up HTTP request header generation a little
by nickm＠torproject.org 08 Sep '11

08 Sep '11

commit ed463404e94d294ac474e6dd6921fdda54415daf Author: Nick Mathewson <nickm(a)torproject.org> Date: Fri Jun 24 16:38:44 2011 -0400 Clean up HTTP request header generation a little Use a list of headers rather than trying to printf every header that might exist. --- src/or/directory.c | 45 +++++++++++++++++++++------------------------ 1 files changed, 21 insertions(+), 24 deletions(-) diff --git a/src/or/directory.c b/src/or/directory.c index cd0b9b4..6055ae3 100644 --- a/src/or/directory.c +++ b/src/or/directory.c @@ -1117,11 +1117,11 @@ directory_send_command(dir_connection_t *conn, time_t if_modified_since) { char proxystring[256]; - char proxyauthstring[256]; char hoststring[128]; - char imsstring[RFC1123_TIME_LEN+32]; + smartlist_t *headers = smartlist_create(); char *url; char request[8192]; + char *header; const char *httpcommand = NULL; size_t len; @@ -1141,12 +1141,11 @@ directory_send_command(dir_connection_t *conn, } /* Format if-modified-since */ - if (!if_modified_since) { - imsstring[0] = '\0'; - } else { + if (if_modified_since) { char b[RFC1123_TIME_LEN+1]; format_rfc1123_time(b, if_modified_since); - tor_snprintf(imsstring, sizeof(imsstring), "\r\nIf-Modified-Since: %s", b); + tor_asprintf(&header, "If-Modified-Since: %s\r\n", b); + smartlist_add(headers, header); } /* come up with some proxy lines, if we're using one. */ @@ -1161,16 +1160,14 @@ directory_send_command(dir_connection_t *conn, log_warn(LD_BUG, "Encoding http authenticator failed"); } if (base64_authenticator) { - tor_snprintf(proxyauthstring, sizeof(proxyauthstring), - "\r\nProxy-Authorization: Basic %s", + tor_asprintf(&header, + "Proxy-Authorization: Basic %s\r\n", base64_authenticator); tor_free(base64_authenticator); - } else { - proxyauthstring[0] = 0; + smartlist_add(headers, header); } } else { proxystring[0] = 0; - proxyauthstring[0] = 0; } switch (purpose) { @@ -1287,26 +1284,26 @@ directory_send_command(dir_connection_t *conn, connection_write_to_buf(url, strlen(url), TO_CONN(conn)); tor_free(url); - if (!strcmp(httpcommand, "GET") && !payload) { - tor_snprintf(request, sizeof(request), - " HTTP/1.0\r\nHost: %s%s%s\r\n\r\n", - hoststring, - imsstring, - proxyauthstring); - } else { - tor_snprintf(request, sizeof(request), - " HTTP/1.0\r\nContent-Length: %lu\r\nHost: %s%s%s\r\n\r\n", - payload ? (unsigned long)payload_len : 0, - hoststring, - imsstring, - proxyauthstring); + if (!strcmp(httpcommand, "POST") || payload) { + tor_asprintf(&header, "Content-Length: %lu\r\n", + payload ? (unsigned long)payload_len : 0); + smartlist_add(headers, header); } + + header = smartlist_join_strings(headers, "", 0, NULL); + tor_snprintf(request, sizeof(request), " HTTP/1.0\r\nHost: %s\r\n%s\r\n", + hoststring, header); + tor_free(header); + connection_write_to_buf(request, strlen(request), TO_CONN(conn)); if (payload) { /* then send the payload afterwards too */ connection_write_to_buf(payload, payload_len, TO_CONN(conn)); } + + SMARTLIST_FOREACH(headers, char *, h, tor_free(h)); + smartlist_free(headers); } /** Parse an HTTP request string <b>headers</b> of the form

1 0

[tor/master] Report reason for generating descriptor in an HTTP header
by nickm＠torproject.org 08 Sep '11

08 Sep '11

commit 3e3480d70406d9deab79c478868ba5acb44255e7 Author: Nick Mathewson <nickm(a)torproject.org> Date: Fri Jun 24 16:43:08 2011 -0400 Report reason for generating descriptor in an HTTP header Suggested by arma; based on 3327. --- changes/bug3327 | 6 ++++++ src/or/directory.c | 8 +++++++- src/or/router.c | 17 +++++++++++++++++ src/or/router.h | 1 + 4 files changed, 31 insertions(+), 1 deletions(-) diff --git a/changes/bug3327 b/changes/bug3327 index 32a8b9c..454eb31 100644 --- a/changes/bug3327 +++ b/changes/bug3327 @@ -6,3 +6,9 @@ to re-publish itself because it has done so too recently, even though the authorities decided not to list its recent-enough descriptor. Fix for bug 3327. + + o Minor features: + - Relays now include a reason for regenerating their descriptors + an HTTP header when uploading to the authorities. This will + make it easier to debug descriptor-upload issues in the future. + diff --git a/src/or/directory.c b/src/or/directory.c index 6055ae3..f56a63c 100644 --- a/src/or/directory.c +++ b/src/or/directory.c @@ -1228,12 +1228,18 @@ directory_send_command(dir_connection_t *conn, httpcommand = "GET"; tor_asprintf(&url, "/tor/micro/%s", resource); break; - case DIR_PURPOSE_UPLOAD_DIR: + case DIR_PURPOSE_UPLOAD_DIR: { + const char *why = router_get_descriptor_gen_reason(); tor_assert(!resource); tor_assert(payload); httpcommand = "POST"; url = tor_strdup("/tor/"); + if (why) { + tor_asprintf(&header, "X-Desc-Gen-Reason: %s\r\n", why); + smartlist_add(headers, header); + } break; + } case DIR_PURPOSE_UPLOAD_VOTE: tor_assert(!resource); tor_assert(payload); diff --git a/src/or/router.c b/src/or/router.c index ccaa139..07a6599 100644 --- a/src/or/router.c +++ b/src/or/router.c @@ -1220,9 +1220,14 @@ router_get_advertised_dir_port(const or_options_t *options, uint16_t dirport) static routerinfo_t *desc_routerinfo = NULL; /** My extrainfo */ static extrainfo_t *desc_extrainfo = NULL; +/** Why did we most recently decide to regenerate our descriptor? Used to + * tell the authorities why we're sending it to them. */ +static const char *desc_gen_reason = NULL; /** Since when has our descriptor been "clean"? 0 if we need to regenerate it * now. */ static time_t desc_clean_since = 0; +/** Why did we mark the descriptor dirty? */ +static const char *desc_dirty_reason = NULL; /** Boolean: do we need to regenerate the above? */ static int desc_needs_upload = 0; @@ -1389,6 +1394,14 @@ router_get_my_extrainfo(void) return desc_extrainfo; } +/** Return a human-readable string describing what triggered us to generate + * our current descriptor, or NULL if we don't know. */ +const char * +router_get_descriptor_gen_reason(void) +{ + return desc_gen_reason; +} + /** A list of nicknames that we've warned about including in our family * declaration verbatim rather than as digests. */ static smartlist_t *warned_nonexistent_family = NULL; @@ -1606,6 +1619,8 @@ router_rebuild_descriptor(int force) desc_clean_since = time(NULL); desc_needs_upload = 1; + desc_gen_reason = desc_dirty_reason; + desc_dirty_reason = NULL; control_event_my_descriptor_changed(); return 0; } @@ -1662,6 +1677,8 @@ mark_my_descriptor_dirty(const char *reason) { desc_clean_since = 0; log_info(LD_OR, "Decided to publish new relay descriptor: %s", reason); + if (!desc_dirty_reason) + desc_dirty_reason = reason; } /** How frequently will we republish our descriptor because of large (factor diff --git a/src/or/router.h b/src/or/router.h index af202e6..f9d156c 100644 --- a/src/or/router.h +++ b/src/or/router.h @@ -73,6 +73,7 @@ int router_my_exit_policy_is_reject_star(void); const routerinfo_t *router_get_my_routerinfo(void); extrainfo_t *router_get_my_extrainfo(void); const char *router_get_my_descriptor(void); +const char *router_get_descriptor_gen_reason(void); int router_digest_is_me(const char *digest); int router_extrainfo_digest_is_me(const char *digest); int router_is_me(const routerinfo_t *router);

1 0

[tor/master] Merge branch 'bug3327_squashed'
by nickm＠torproject.org 08 Sep '11

08 Sep '11

commit 1ed1bf7f07350bce0eb4a73f29e1d8a2d7ffdbc8 Merge: 41eef66 3e3480d Author: Nick Mathewson <nickm(a)torproject.org> Date: Wed Sep 7 15:03:54 2011 -0400 Merge branch 'bug3327_squashed' changes/bug3327 | 14 ++++++++++++ src/or/directory.c | 53 +++++++++++++++++++++++--------------------- src/or/main.c | 6 +---- src/or/router.c | 61 +++++++++++++++++++++++++++++++++++++++++++++++++-- src/or/router.h | 3 +- 5 files changed, 103 insertions(+), 34 deletions(-)

1 0

[tor/master] fix a const warning
by nickm＠torproject.org 08 Sep '11

08 Sep '11

commit 5c53a0f867cd3628e37ccffa30b3a534a8f0c74a Author: Nick Mathewson <nickm(a)torproject.org> Date: Wed Sep 7 15:06:01 2011 -0400 fix a const warning --- src/or/router.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/src/or/router.c b/src/or/router.c index 07a6599..ba6648b 100644 --- a/src/or/router.c +++ b/src/or/router.c @@ -1639,7 +1639,7 @@ void mark_my_descriptor_dirty_if_too_old(time_t now) { networkstatus_t *ns; - routerstatus_t *rs; + const routerstatus_t *rs; const char *retry_fast_reason = NULL; /* Set if we should retry frequently */ const time_t slow_cutoff = now - FORCE_REGENERATE_DESCRIPTOR_INTERVAL; const time_t fast_cutoff = now - FAST_RETRY_DESCRIPTOR_INTERVAL;

1 0

[tor/master] Upload descriptors more often when recent desc is unlisted
by nickm＠torproject.org 08 Sep '11

08 Sep '11

commit 1f4b6944c06fce5befc3b8a21d252ba946632f63 Author: Nick Mathewson <nickm(a)torproject.org> Date: Wed Jun 22 12:27:27 2011 -0400 Upload descriptors more often when recent desc is unlisted Right now we only force a new descriptor upload every 18 hours. This can make servers become unlisted if they upload a descriptor at time T which the authorities reject as being "too similar" to one they uploaded before. Nothing will actually make the server upload a new descriptor later on, until another 18 hours have passed. This patch changes the upload behavior so that the 18 hour interval applies only when we're listed in a live consensus with a descriptor published within the last 18 hours. Otherwise--if we're not listed in the live consensus, or if we're listed with a publication time over 18 hours in the past--we upload a new descriptor every 90 minutes. This is an attempted bugfix for #3327. If we merge it, it should obsolete #535. --- changes/bug3327 | 8 ++++++++ src/or/main.c | 6 +----- src/or/router.c | 44 +++++++++++++++++++++++++++++++++++++++++--- src/or/router.h | 2 +- 4 files changed, 51 insertions(+), 9 deletions(-) diff --git a/changes/bug3327 b/changes/bug3327 new file mode 100644 index 0000000..32a8b9c --- /dev/null +++ b/changes/bug3327 @@ -0,0 +1,8 @@ + o Major features: + - Relays now try regenerating and uploading their descriptor more + frequently if they are not listed in the consensus, or if the + version of their descriptor listed in the consensus is too + old. This fix should prevent situations where a server declines + to re-publish itself because it has done so too recently, even + though the authorities decided not to list its recent-enough + descriptor. Fix for bug 3327. diff --git a/src/or/main.c b/src/or/main.c index 2cdbe71..af769c4 100644 --- a/src/or/main.c +++ b/src/or/main.c @@ -1326,11 +1326,7 @@ run_scheduled_events(time_t now) time_to_check_ipaddress = now + CHECK_IPADDRESS_INTERVAL; check_descriptor_ipaddress_changed(now); } -/** If our router descriptor ever goes this long without being regenerated - * because something changed, we force an immediate regenerate-and-upload. */ -#define FORCE_REGENERATE_DESCRIPTOR_INTERVAL (18*60*60) - mark_my_descriptor_dirty_if_older_than( - now - FORCE_REGENERATE_DESCRIPTOR_INTERVAL); + mark_my_descriptor_dirty_if_too_old(now); consider_publishable_server(0); /* also, check religiously for reachability, if it's within the first * 20 minutes of our uptime. */ diff --git a/src/or/router.c b/src/or/router.c index 531d3fb..ccaa139 100644 --- a/src/or/router.c +++ b/src/or/router.c @@ -1610,12 +1610,50 @@ router_rebuild_descriptor(int force) return 0; } -/** Mark descriptor out of date if it's older than <b>when</b> */ +/** If our router descriptor ever goes this long without being regenerated + * because something changed, we force an immediate regenerate-and-upload. */ +#define FORCE_REGENERATE_DESCRIPTOR_INTERVAL (18*60*60) + +/** If our router descriptor seems to be missing or unacceptable according + * to the authorities, regenerate and reupload it _this_ often. */ +#define FAST_RETRY_DESCRIPTOR_INTERVAL (90*60) + +/** Mark descriptor out of date if it's been "too long" since we last tried + * to upload one. */ void -mark_my_descriptor_dirty_if_older_than(time_t when) +mark_my_descriptor_dirty_if_too_old(time_t now) { - if (desc_clean_since < when) + networkstatus_t *ns; + routerstatus_t *rs; + const char *retry_fast_reason = NULL; /* Set if we should retry frequently */ + const time_t slow_cutoff = now - FORCE_REGENERATE_DESCRIPTOR_INTERVAL; + const time_t fast_cutoff = now - FAST_RETRY_DESCRIPTOR_INTERVAL; + + /* If it's already dirty, don't mark it. */ + if (! desc_clean_since) + return; + + /* If it's older than FORCE_REGENERATE_DESCRIPTOR_INTERVAL, it's always + * time to rebuild it. */ + if (desc_clean_since < slow_cutoff) { mark_my_descriptor_dirty("time for new descriptor"); + return; + } + /* Now we see whether we want to be retrying frequently or no. The + * rule here is that we'll retry frequently if we aren't listed in the + * live consensus we have, or if the publication time of the + * descriptor listed for us in the consensus is very old. */ + ns = networkstatus_get_live_consensus(now); + if (ns) { + rs = networkstatus_vote_find_entry(ns, server_identitykey_digest); + if (rs == NULL) + retry_fast_reason = "not listed in consensus"; + else if (rs->published_on < slow_cutoff) + retry_fast_reason = "version listed in consensus is quite old"; + } + + if (retry_fast_reason && desc_clean_since < fast_cutoff) + mark_my_descriptor_dirty(retry_fast_reason); } /** Call when the current descriptor is out of date. */ diff --git a/src/or/router.h b/src/or/router.h index f6d3c12..af202e6 100644 --- a/src/or/router.h +++ b/src/or/router.h @@ -62,7 +62,7 @@ void consider_publishable_server(int force); int should_refuse_unknown_exits(const or_options_t *options); void router_upload_dir_desc_to_dirservers(int force); -void mark_my_descriptor_dirty_if_older_than(time_t when); +void mark_my_descriptor_dirty_if_too_old(time_t now); void mark_my_descriptor_dirty(const char *reason); void check_descriptor_bandwidth_changed(time_t now); void check_descriptor_ipaddress_changed(time_t now);

1 0

[tor/master] Merge remote-tracking branch 'public/bug2649_squashed' into maint-0.2.2
by nickm＠torproject.org 08 Sep '11

08 Sep '11

commit dfa6cde4d4de57476ea8381199dfbac1afc245d5 Merge: 5f9cdee f45261c Author: Nick Mathewson <nickm(a)torproject.org> Date: Wed Sep 7 14:43:06 2011 -0400 Merge remote-tracking branch 'public/bug2649_squashed' into maint-0.2.2 changes/bug2649a | 5 +++++ changes/bug2649b | 5 +++++ doc/tor.1.txt | 7 ++++++- src/or/config.c | 3 ++- src/or/dirserv.c | 16 +++++++++++----- src/or/dirvote.h | 2 +- src/or/networkstatus.c | 2 +- src/or/or.h | 3 +++ 8 files changed, 34 insertions(+), 9 deletions(-)

1 0