Changes in version 0.2.4.1-alpha - 2012-09-05 Tor 0.2.4.1-alpha lets bridges publish their pluggable transports to bridgedb; lets relays use IPv6 addresses and directory authorities advertise them; and switches to a cleaner build interface. This is the first alpha release in a new series, so expect there to be bugs. Users who would rather test out a more stable branch should stay with 0.2.3.x for now. o Major features (bridges): - Bridges now report the pluggable transports they support to the bridge authority, so it can pass the supported transports on to bridgedb and/or eventually do reachability testing. Implements ticket 3589. o Major features (IPv6): - Bridge authorities now accept IPv6 bridge addresses and include them in network status documents. Implements ticket 5534. - Clients who set "ClientUseIPv6 1" may connect to entry nodes over IPv6. Set "ClientPreferIPv6ORPort 1" to make this even more likely to happen. Implements ticket 5535. - All kind of relays, not just bridges, can now advertise an IPv6 OR port. Implements ticket 6362. - Directory authorities vote on IPv6 OR ports using the new consensus method 14. Implements ticket 6363. o Major features (build): - Switch to a nonrecursive Makefile structure. Now instead of each Makefile.am invoking other Makefile.am's, there is a master Makefile.am that includes the others. This change makes our build process slightly more maintainable, and improves parallelism for building with make -j. Original patch by Stewart Smith; various fixes by Jim Meyering. - Where available, we now use automake's "silent" make rules by default, so that warnings are easier to spot. You can get the old behavior with "make V=1". Patch by Stewart Smith for ticket 6522. o Minor features (code security and spec conformance): - Clear keys and key-derived material left on the stack in rendservice.c and rendclient.c. Check return value of crypto_pk_write_private_key_to_string() in end_service_load_keys(). These fixes should make us more forward-secure against cold-boot attacks and the like. Fixes bug 2385. - Reject EXTEND cells sent to nonexistent streams. According to the spec, an EXTEND cell sent to _any_ nonzero stream ID is invalid, but we were only checking for stream IDs that were currently in use. Found while hunting for more instances of bug 6271. Bugfix on 0.0.2pre8, which introduced incremental circuit construction. o Minor features (streamlining); - No longer include the "opt" prefix when generating routerinfos or v2 directories: it has been needless since Tor 0.1.2. Closes ticket 5124. - Remove some now-needless code that tried to aggressively flush OR connections as data was added to them. Since 0.2.0.1-alpha, our cell queue logic has saved us from the failure mode that this code was supposed to prevent. Removing this code will limit the number of baroque control flow paths through Tor's network logic. Reported pseudonymously on IRC. Fixes bug 6468; bugfix on 0.2.0.1-alpha. o Minor features (controller): - Add a "GETINFO signal/names" control port command. Implements ticket 3842. - Provide default values for all options via "GETINFO config/defaults". Implements ticket 4971. o Minor features (IPv6): - New config option "AuthDirHasIPv6Connectivity 1" that directory authorities should set if they have IPv6 connectivity and want to do reachability tests for IPv6 relays. Implements feature 5974. - A relay with an IPv6 OR port now sends that address in NETINFO cells (in addition to its other address). Implements ticket 6364. o Minor features (log messages): - Omit the first heartbeat log message, because it never has anything useful to say, and it clutters up the bootstrapping messages. Resolves ticket 6758. - Don't log about reloading the microdescriptor cache at startup. Our bootstrap warnings are supposed to tell the user when there's a problem, and our bootstrap notices say when there isn't. Resolves ticket 6759; bugfix on 0.2.2.6-alpha. - Don't log "I learned some more directory information" when we're reading cached directory information. Reserve it for when new directory information arrives in response to a fetch. Resolves ticket 6760. - Prevent rounding error in path bias counts when scaling them down, and use the correct scale factor default. Also demote some path bias related log messages down a level and make others less scary sounding. Fixes bug 6647. Bugfix against 0.2.3.17-beta. - We no longer warn so much when generating manpages from their asciidoc source. o Code simplifications and refactoring: - Enhance our internal sscanf replacement so that we can eliminate the last remaining uses of the system sscanf. (Though those uses of sscanf were safe, sscanf itself is generally error prone, so we want to eliminate when we can.) Fixes ticket 4195 and Coverity CID 448. - Move ipv6_preferred from routerinfo_t to node_t. Addresses bug 4620. - Move last_reachable and testing_since from routerinfo_t to node_t. Implements ticket 5529. - Add replaycache_t structure, functions and unit tests, then refactor rend_service_introduce() to be more clear to read, improve, debug, and test. Resolves bug 6177. - Finally remove support for malloc_good_size and malloc_usable_size. We had hoped that these functions would let us eke a little more memory out of our malloc implementation. Unfortunately, the only implementations that provided these functions are also ones that are already efficient about not overallocation: they never got us more than 7 or so bytes per allocation. Removing them saves us a little code complexity and a nontrivial amount of build complexity. o New requirements: - Tor maintainers now require Automake version 1.9 or later to build Tor from the Git repository. (Automake is not required when building from a source distribution.) Changes in version 0.2.3.21-rc - 2012-09-05 Tor 0.2.3.21-rc is the fourth release candidate for the Tor 0.2.3.x series. It fixes a trio of potential security bugs, fixes a bug where we were leaving some of the fast relays out of the microdescriptor consensus, resumes interpreting "ORPort 0" and "DirPort 0" correctly, and cleans up other smaller issues. o Major bugfixes (security): - Tear down the circuit if we get an unexpected SENDME cell. Clients could use this trick to make their circuits receive cells faster than our flow control would have allowed, or to gum up the network, or possibly to do targeted memory denial-of-service attacks on entry nodes. Fixes bug 6252. Bugfix on the 54th commit on Tor -- from July 2002, before the release of Tor 0.0.0. We had committed this patch previously, but we had to revert it because of bug 6271. Now that 6271 is fixed, this patch appears to work. - Reject any attempt to extend to an internal address. Without this fix, a router could be used to probe addresses on an internal network to see whether they were accepting connections. Fixes bug 6710; bugfix on 0.0.8pre1. - Do not crash when comparing an address with port value 0 to an address policy. This bug could have been used to cause a remote assertion failure by or against directory authorities, or to allow some applications to crash clients. Fixes bug 6690; bugfix on 0.2.1.10-alpha. o Major bugfixes: - Remove the upper bound on microdescriptor length. We were hitting the limit for routers with complex exit policies or family declarations, causing clients to not use them. Fixes the first piece of bug 6404; fix on 0.2.2.6-alpha. - Detect "ORPort 0" as meaning, uniformly, that we're not running as a relay. Previously, some of our code would treat the presence of any ORPort line as meaning that we should act like a relay, even though our new listener code would correctly not open any ORPorts for ORPort 0. Similar bugs in other Port options are also fixed. Fixes the first half of bug 6507; bugfix on 0.2.3.3-alpha. o Minor bugfixes: - Avoid a pair of double-free and use-after-mark bugs that can occur with certain timings in canceled and re-received DNS requests. Fixes bug 6472; bugfix on 0.0.7rc1. - Fix build and 64-bit compile warnings from --enable-openbsd-malloc. Fixes bug 6379. Bugfix on 0.2.0.20-rc. - Allow one-hop directory fetching circuits the full "circuit build timeout" period, rather than just half of it, before failing them and marking the relay down. This fix should help reduce cases where clients declare relays (or worse, bridges) unreachable because the TLS handshake takes a few seconds to complete. Fixes bug 6743; bugfix on 0.2.2.2-alpha, where we changed the timeout from a static 30 seconds. - Authorities no longer include any router in their microdescriptor consensuses for which they couldn't generate or agree on a microdescriptor. Fixes the second piece of bug 6404; fix on 0.2.2.6-alpha. - Detect and reject attempts to specify both "FooPort" and "FooPort 0" in the same configuration domain. (It's still okay to have a FooPort in your configuration file, and use "FooPort 0" on the command line to disable it.) Fixes the second half of bug 6507; bugfix on 0.2.3.3-alpha. - Make wildcarded addresses (that is, ones beginning with "*.") work when provided via the controller's MapAddress command. Previously, they were accepted, but we never actually noticed that they were wildcards. Fixes bug 6244; bugfix on 0.2.3.9-alpha. - Avoid crashing on a malformed state file where EntryGuardPathBias precedes EntryGuard. Fix for bug 6774; bugfix on 0.2.3.17-beta. - Add a (probably redundant) memory clear between iterations of the router status voting loop, to prevent future coding errors where data might leak between iterations of the loop. Resolves ticket 6514. o Minor bugfixes (log messages): - Downgrade "set buildtimeout to low value" messages to "info" severity; they were never an actual problem, there was never anything reasonable to do about them, and they tended to spam logs from time to time. Fixes bug 6251; bugfix on 0.2.2.2-alpha. - Downgrade path-bias warning messages to "info". We'll try to get them working better in 0.2.4. Add internal circuit construction state to protect against the noisy warn message "Unexpectedly high circuit_successes". Also add some additional rate-limited notice messages to help determine the root cause of the warn. Fixes bug 6475. Bugfix against 0.2.3.17-beta. - Move log message when unable to find a microdesc in a routerstatus entry to parse time. Previously we'd spam this warning every time we tried to figure out which microdescriptors to download. Fixes the third piece of bug 6404; fix on 0.2.3.18-rc. o Minor features: - Consider new, removed or changed IPv6 OR ports a non-cosmetic change when the authority is deciding whether to accept a newly uploaded descriptor. Implements ticket 6423. - Add missing documentation for consensus and microdesc files. Resolves ticket 6732. Changes in version 0.2.2.38 - 2012-08-12 Tor 0.2.2.38 fixes a remotely triggerable crash bug, and fixes a timing attack that could in theory leak path information. o Security fixes: - Avoid an uninitialized memory read when reading a vote or consensus document that has an unrecognized flavor name. This read could lead to a remote crash bug. Fixes bug 6530; bugfix on 0.2.2.6-alpha. - Try to leak less information about what relays a client is choosing to a side-channel attacker. Previously, a Tor client would stop iterating through the list of available relays as soon as it had chosen one, thus finishing a little earlier when it picked a router earlier in the list. If an attacker can recover this timing information (nontrivial but not proven to be impossible), they could learn some coarse-grained information about which relays a client was picking (middle nodes in particular are likelier to be affected than exits). The timing attack might be mitigated by other factors (see bug 6537 for some discussion), but it's best not to take chances. Fixes bug 6537; bugfix on 0.0.8rc1. Changes in version 0.2.3.20-rc - 2012-08-05 Tor 0.2.3.20-rc is the third release candidate for the Tor 0.2.3.x series. It fixes a pair of code security bugs and a potential anonymity issue, updates our RPM spec files, and cleans up other smaller issues. o Security fixes: - Avoid read-from-freed-memory and double-free bugs that could occur when a DNS request fails while launching it. Fixes bug 6480; bugfix on 0.2.0.1-alpha. - Avoid an uninitialized memory read when reading a vote or consensus document that has an unrecognized flavor name. This read could lead to a remote crash bug. Fixes bug 6530; bugfix on 0.2.2.6-alpha. - Try to leak less information about what relays a client is choosing to a side-channel attacker. Previously, a Tor client would stop iterating through the list of available relays as soon as it had chosen one, thus finishing a little earlier when it picked a router earlier in the list. If an attacker can recover this timing information (nontrivial but not proven to be impossible), they could learn some coarse-grained information about which relays a client was picking (middle nodes in particular are likelier to be affected than exits). The timing attack might be mitigated by other factors (see bug 6537 for some discussion), but it's best not to take chances. Fixes bug 6537; bugfix on 0.0.8rc1. o Minor features: - Try to make the warning when giving an obsolete SOCKSListenAddress a little more useful. - Terminate active server managed proxies if Tor stops being a relay. Addresses parts of bug 6274; bugfix on 0.2.3.6-alpha. - Provide a better error message about possible OSX Asciidoc failure reasons. Fixes bug 6436. - Warn when Tor is configured to use accounting in a way that can link a hidden service to some other hidden service or public address. Resolves ticket 6490. o Minor bugfixes: - Check return value of fputs() when writing authority certificate file. Fixes Coverity issue 709056; bugfix on 0.2.0.1-alpha. - Ignore ServerTransportPlugin lines when Tor is not configured as a relay. Fixes bug 6274; bugfix on 0.2.3.6-alpha. - When disabling guards for having too high a proportion of failed circuits, make sure to look at each guard. Fixes bug 6397; bugfix on 0.2.3.17-beta. o Packaging (RPM): - Update our default RPM spec files to work with mock and rpmbuild on RHEL/Fedora. They have an updated set of dependencies and conflicts, a fix for an ancient typo when creating the "_tor" user, and better instructions. Thanks to Ondrej Mikle for the patch series. Fixes bug 6043. o Testing: - Make it possible to set the TestingTorNetwork configuration option using AlternateDirAuthority and AlternateBridgeAuthority as an alternative to setting DirServer. Addresses ticket 6377. o Documentation: - Clarify the documentation for the Alternate*Authority options. Fixes bug 6387. - Fix some typos in the manpages. Patch from A. Costa. Fixes bug 6500. o Code simplification and refactoring: - Do not use SMARTLIST_FOREACH for any loop whose body exceeds 10 lines. Also, don't nest them. Doing so in the past has led to hard-to-debug code. The new style is to use the SMARTLIST_FOREACH_{BEGIN,END} pair. Addresses issue 6400. Changes in version 0.2.3.19-rc - 2012-07-06 Tor 0.2.3.19-rc is the second release candidate for the Tor 0.2.3.x series. It fixes the compile on Windows, reverts to a GeoIP database that isn't as broken, and fixes a flow control bug that has been around since the beginning of Tor. o Major bugfixes: - Fix a bug handling SENDME cells on nonexistent streams that could result in bizarre window values. Report and patch contributed pseudonymously. Fixes part of bug 6271. This bug was introduced before the first Tor release, in svn commit r152. - Revert to the May 1 2012 Maxmind GeoLite Country database. In the June 2012 database, Maxmind marked many Tor relays as country "A1", which will cause risky behavior for clients that set EntryNodes or ExitNodes. Addresses bug 6334; bugfix on 0.2.3.17-beta. - Instead of ENOBUFS on Windows, say WSAENOBUFS. Fixes compilation on Windows. Fixes bug 6296; bugfix on 0.2.3.18-rc. o Minor bugfixes: - Fix wrong TCP port range in parse_port_range(). Fixes bug 6218; bugfix on 0.2.1.10-alpha. Changes in version 0.2.3.18-rc - 2012-06-28 Tor 0.2.3.18-rc is the first release candidate for the Tor 0.2.3.x series. It fixes a few smaller bugs, but generally appears stable. Please test it and let us know whether it is! o Major bugfixes: - Allow wildcarded mapaddress targets to be specified on the controlport. Partially fixes bug 6244; bugfix on 0.2.3.9-alpha. - Make our linker option detection code more robust against linkers such as on FreeBSD 8, where a bad combination of options completes successfully but makes an unrunnable binary. Fixes bug 6173; bugfix on 0.2.3.17-beta. o Minor bugfixes (on 0.2.2.x and earlier): - Avoid a false positive in the util/threads unit test by increasing the maximum timeout time. Fixes bug 6227; bugfix on 0.2.0.4-alpha. - Replace "Sending publish request" log messages with "Launching upload", so that they no longer confusingly imply that we're sending something to a directory we might not even be connected to yet. Fixes bug 3311; bugfix on 0.2.0.10-alpha. - Make sure to set *socket_error in all error cases in connection_connect(), so it can't produce a warning about errno being zero from errno_to_orconn_end_reason(). Bugfix on 0.2.1.1-alpha; resolves ticket 6028. - Downgrade "Got a certificate, but we already have it" log messages from warning to info, except when we're a dirauth. Fixes bug 5238; bugfix on 0.2.1.7-alpha. - When checking for requested signatures on the latest consensus before serving it to a client, make sure to check the right consensus flavor. Bugfix on 0.2.2.6-alpha. - Downgrade "eventdns rejected address" message to LOG_PROTOCOL_WARN. Fixes bug 5932; bugfix on 0.2.2.7-alpha. o Minor bugfixes (on 0.2.3.x): - Make format_helper_exit_status() avoid unnecessary space padding and stop confusing log_from_pipe(). Fixes ticket 5557; bugfix on 0.2.3.1-alpha. - Downgrade a message about cleaning the microdescriptor cache to "info" from "notice". Fixes bug 6238; bugfix on 0.2.3.1-alpha. - Log a BUG message at severity INFO if we have a networkstatus with a missing entry for some microdescriptor. Continues on a patch to 0.2.3.2-alpha. - Improve the log message when a managed proxy fails to launch. Fixes bug 5099; bugfix on 0.2.3.6-alpha. - Don't do DNS lookups when parsing corrupted managed proxy protocol messages. Fixes bug 6226; bugfix on 0.2.3.6-alpha. - When formatting wildcarded address mappings for the controller, be sure to include "*." as appropriate. Partially fixes bug 6244; bugfix on 0.2.3.9-alpha. - Avoid a warning caused by using strcspn() from glibc with clang 3.0. Bugfix on 0.2.3.13-alpha. - Stop logging messages about running with circuit timeout learning enabled at severity LD_BUG. Fixes bug 6169; bugfix on 0.2.3.17-beta. - Disable a spurious warning about reading on a marked and flushing connection. We shouldn't be doing that, but apparently we sometimes do. Fixes bug 6203; bugfix on 0.2.3.17-beta. - Fix a bug that stopped AllowDotExit from working on addresses that had an entry in the DNS cache. Fixes bug 6211; bugfix on 0.2.3.17-beta. o Code simplification, refactoring, unit tests: - Move tor_gettimeofday_cached() into compat_libevent.c, and use Libevent's notion of cached time when possible. - Remove duplicate code for invoking getrlimit() from control.c. - Add a unit test for the environment_variable_names_equal function. o Documentation: - Document the --defaults-torrc option, and the new (in 0.2.3) semantics for overriding, extending, and clearing lists of options. Closes bug 4748. Changes in version 0.2.3.17-beta - 2012-06-15 Tor 0.2.3.17-beta enables compiler and linker hardening by default, gets our TLS handshake back on track for being able to blend in with Firefox, fixes a big bug in 0.2.3.16-alpha that broke Tor's interaction with Vidalia, and otherwise continues to get us closer to a release candidate. o Major features: - Enable gcc and ld hardening by default. Resolves ticket 5210. - Update TLS cipher list to match Firefox 8 and later. Resolves ticket 4744. - Implement the client side of proposal 198: remove support for clients falsely claiming to support standard ciphersuites that they can actually provide. As of modern OpenSSL versions, it's not necessary to fake any standard ciphersuite, and doing so prevents us from using better ciphersuites in the future, since servers can't know whether an advertised ciphersuite is really supported or not. Some hosts -- notably, ones with very old versions of OpenSSL or where OpenSSL has been built with ECC disabled -- will stand out because of this change; TBB users should not be affected. o Major bugfixes: - Change the default value for DynamicDHGroups (introduced in 0.2.3.9-alpha) to 0. This feature can make Tor relays less identifiable by their use of the mod_ssl DH group, but at the cost of some usability (#4721) and bridge tracing (#6087) regressions. Resolves ticket 5598. - Send a CRLF at the end of each STATUS_* control protocol event. This bug tickled a bug in Vidalia which would make it freeze. Fixes bug 6094; bugfix on 0.2.3.16-alpha. o Minor bugfixes: - Disable writing on marked-for-close connections when they are blocked on bandwidth, to prevent busy-looping in Libevent. Fixes bug 5263; bugfix on 0.0.2pre13, where we first added a special case for flushing marked connections. - Detect SSL handshake even when the initial attempt to write the server hello fails. Fixes bug 4592; bugfix on 0.2.0.13-alpha. - Change the AllowDotExit rules so they should actually work. We now enforce AllowDotExit only immediately after receiving an address via SOCKS or DNSPort: other sources are free to provide .exit addresses after the resolution occurs. Fixes bug 3940; bugfix on 0.2.2.1-alpha. - Fix a (harmless) integer overflow in cell statistics reported by some fast relays. Fixes bug 5849; bugfix on 0.2.2.1-alpha. - Make sure circuitbuild.c checks LearnCircuitBuildTimeout in all the right places and never depends on the consensus parameters or computes adaptive timeouts when it is disabled. Fixes bug 5049; bugfix on 0.2.2.14-alpha. - When building Tor on Windows with -DUNICODE (not default), ensure that error messages, filenames, and DNS server names are always NUL-terminated when we convert them to a single-byte encoding. Fixes bug 5909; bugfix on 0.2.2.16-alpha. - Make Tor build correctly again with -DUNICODE -D_UNICODE defined. Fixes bug 6097; bugfix on 0.2.2.16-alpha. - Fix an edge case where TestingTorNetwork is set but the authorities and relays all have an uptime of zero, where the private Tor network could briefly lack support for hidden services. Fixes bug 3886; bugfix on 0.2.2.18-alpha. - Correct the manpage's descriptions for the default values of DirReqStatistics and ExtraInfoStatistics. Fixes bug 2865; bugfix on 0.2.3.1-alpha. - Fix the documentation for the --hush and --quiet command line options, which changed their behavior back in 0.2.3.3-alpha. - Fix compilation warning with clang 3.1. Fixes bug 6141; bugfix on 0.2.3.11-alpha. o Minor features: - Rate-limit the "Weighted bandwidth is 0.000000" message, and add more information to it, so that we can track it down in case it returns again. Mitigates bug 5235. - Check CircuitBuildTimeout and LearnCircuitBuildTimeout in options_validate(); warn if LearnCircuitBuildTimeout is disabled and CircuitBuildTimeout is set unreasonably low. Resolves ticket 5452. - Warn the user when HTTPProxy, but no other proxy type, is configured. This can cause surprising behavior: it doesn't send all of Tor's traffic over the HTTPProxy -- it sends unencrypted directory traffic only. Resolves ticket 4663. - Issue a notice if a guard completes less than 40% of your circuits. Threshold is configurable by torrc option PathBiasNoticeRate and consensus parameter pb_noticepct. There is additional, off-by- default code to disable guards which fail too many circuits. Addresses ticket 5458. - Update to the June 6 2012 Maxmind GeoLite Country database. o Code simplifications and refactoring: - Remove validate_pluggable_transports_config(): its warning message is now handled by connection_or_connect(). Changes in version 0.2.2.37 - 2012-06-06 Tor 0.2.2.37 introduces a workaround for a critical renegotiation bug in OpenSSL 1.0.1 (where 20% of the Tor network can't talk to itself currently). o Major bugfixes: - Work around a bug in OpenSSL that broke renegotiation with TLS 1.1 and TLS 1.2. Without this workaround, all attempts to speak the v2 Tor connection protocol when both sides were using OpenSSL 1.0.1 would fail. Resolves ticket 6033. - When waiting for a client to renegotiate, don't allow it to add any bytes to the input buffer. This fixes a potential DoS issue. Fixes bugs 5934 and 6007; bugfix on 0.2.0.20-rc. - Fix an edge case where if we fetch or publish a hidden service descriptor, we might build a 4-hop circuit and then use that circuit for exiting afterwards -- even if the new last hop doesn't obey our ExitNodes config option. Fixes bug 5283; bugfix on 0.2.0.10-alpha. o Minor bugfixes: - Fix a build warning with Clang 3.1 related to our use of vasprintf. Fixes bug 5969. Bugfix on 0.2.2.11-alpha. o Minor features: - Tell GCC and Clang to check for any errors in format strings passed to the tor_v*(print|scan)f functions. Changes in version 0.2.3.16-alpha - 2012-06-05 Tor 0.2.3.16-alpha introduces a workaround for a critical renegotiation bug in OpenSSL 1.0.1 (where 20% of the Tor network can't talk to itself currently). It also fixes a variety of smaller bugs and other cleanups that get us closer to a release candidate. o Major bugfixes (general): - Work around a bug in OpenSSL that broke renegotiation with TLS 1.1 and TLS 1.2. Without this workaround, all attempts to speak the v2 Tor connection protocol when both sides were using OpenSSL 1.0.1 would fail. Resolves ticket 6033. - When waiting for a client to renegotiate, don't allow it to add any bytes to the input buffer. This fixes a potential DoS issue. Fixes bugs 5934 and 6007; bugfix on 0.2.0.20-rc. - Pass correct OR address to managed proxies (like obfsproxy), even when ORListenAddress is used. Fixes bug 4865; bugfix on 0.2.3.9-alpha. - The advertised platform of a router now includes only its operating system's name (e.g., "Linux", "Darwin", "Windows 7"), and not its service pack level (for Windows) or its CPU architecture (for Unix). We also no longer include the "git-XYZ" tag in the version. Resolves part of bug 2988. o Major bugfixes (clients): - If we are unable to find any exit that supports our predicted ports, stop calling them predicted, so that we don't loop and build hopeless circuits indefinitely. Fixes bug 3296; bugfix on 0.0.9pre6, which introduced predicted ports. - Fix an edge case where if we fetch or publish a hidden service descriptor, we might build a 4-hop circuit and then use that circuit for exiting afterwards -- even if the new last hop doesn't obey our ExitNodes config option. Fixes bug 5283; bugfix on 0.2.0.10-alpha. - Check at each new consensus whether our entry guards were picked long enough ago that we should rotate them. Previously, we only did this check at startup, which could lead to us holding a guard indefinitely. Fixes bug 5380; bugfix on 0.2.1.14-rc. - When fetching a bridge descriptor from a bridge authority, always do so anonymously, whether we have been able to open circuits or not. Partial fix for bug 1938; bugfix on 0.2.0.7-alpha. This behavior makes it *safer* to use UpdateBridgesFromAuthority, but we'll need to wait for bug 6010 before it's actually usable. o Major bugfixes (directory authorities): - When computing weight parameters, behave more robustly in the presence of a bad bwweightscale value. Previously, the authorities would crash if they agreed on a sufficiently broken weight_scale value: now, they use a reasonable default and carry on. Partial fix for 5786; bugfix on 0.2.2.17-alpha. - Check more thoroughly to prevent a rogue authority from double-voting on any consensus directory parameter. Previously, authorities would crash in this case if the total number of votes for any parameter exceeded the number of active voters, but would let it pass otherwise. Partial fix for bug 5786; bugfix on 0.2.2.2-alpha. o Minor features: - Rate-limit log messages when asked to connect anonymously to a private address. When these hit, they tended to hit fast and often. Also, don't bother trying to connect to addresses that we are sure will resolve to 127.0.0.1: getting 127.0.0.1 in a directory reply makes us think we have been lied to, even when the address the client tried to connect to was "localhost." Resolves ticket 2822. - Allow packagers to insert an extra string in server descriptor platform lines by setting the preprocessor variable TOR_BUILD_TAG. Resolves the rest of ticket 2988. - Raise the threshold of server descriptors needed (75%) and exit server descriptors needed (50%) before we will declare ourselves bootstrapped. This will make clients start building circuits a little later, but makes the initially constructed circuits less skewed and less in conflict with further directory fetches. Fixes ticket 3196. - Close any connection that sends unrecognized junk before the handshake. Solves an issue noted in bug 4369. - Improve log messages about managed transports. Resolves ticket 5070. - Tag a bridge's descriptor as "never to be sent unencrypted". This shouldn't matter, since bridges don't open non-anonymous connections to the bridge authority and don't allow unencrypted directory connections from clients, but we might as well make sure. Closes bug 5139. - Expose our view of whether we have gone dormant to the controller, via a new "GETINFO dormant" value. Torbutton and other controllers can use this to avoid doing periodic requests through Tor while it's dormant (bug 4718). Fixes bug 5954. - Tell GCC and Clang to check for any errors in format strings passed to the tor_v*(print|scan)f functions. - Update to the May 1 2012 Maxmind GeoLite Country database. o Minor bugfixes (already included in 0.2.2.36): - Reject out-of-range times like 23:59:61 in parse_rfc1123_time(). Fixes bug 5346; bugfix on 0.0.8pre3. - Correct parsing of certain date types in parse_http_time(). Without this patch, If-Modified-Since would behave incorrectly. Fixes bug 5346; bugfix on 0.2.0.2-alpha. Patch from Esteban Manchado Velázques. - Make our number-parsing functions always treat too-large values as an error, even when those values exceed the width of the underlying type. Previously, if the caller provided these functions with minima or maxima set to the extreme values of the underlying integer type, these functions would return those values on overflow rather than treating overflow as an error. Fixes part of bug 5786; bugfix on 0.0.9. - If we hit the error case where routerlist_insert() replaces an existing (old) server descriptor, make sure to remove that server descriptor from the old_routers list. Fix related to bug 1776. Bugfix on 0.2.2.18-alpha. - Clarify the behavior of MaxCircuitDirtiness with hidden service circuits. Fixes issue 5259. o Minor bugfixes (coding cleanup, on 0.2.2.x and earlier): - Prevent a null-pointer dereference when receiving a data cell for a nonexistent stream when the circuit in question has an empty deliver window. We don't believe this is triggerable, since we don't currently allow deliver windows to become empty, but the logic is tricky enough that it's better to make the code robust. Fixes bug 5541; bugfix on 0.0.2pre14. - Fix a memory leak when trying to launch a DNS request when the network is disabled or the nameservers are unconfigurable. Fixes bug 5916; bugfix on Tor 0.1.2.1-alpha (for the unconfigurable nameserver case) and on 0.2.3.9-alpha (for the DisableNetwork case). - Don't hold a Windows file handle open for every file mapping; the file mapping handle is sufficient. Fixes bug 5951; bugfix on 0.1.2.1-alpha. - Avoid O(n^2) performance characteristics when parsing a large extrainfo cache. Fixes bug 5828; bugfix on 0.2.0.1-alpha. - Format more doubles with %f, not %lf. Patch from grarpamp to make Tor build correctly on older BSDs again. Fixes bug 3894; bugfix on Tor 0.2.0.8-alpha. - Make our replacement implementation of strtok_r() compatible with the standard behavior of strtok_r(). Patch by nils. Fixes bug 5091; bugfix on 0.2.2.1-alpha. - Fix a NULL-pointer dereference on a badly formed SETCIRCUITPURPOSE command. Found by mikeyc. Fixes bug 5796; bugfix on 0.2.2.9-alpha. - Fix a build warning with Clang 3.1 related to our use of vasprintf. Fixes bug 5969. Bugfix on 0.2.2.11-alpha. - Defensively refactor rend_mid_rendezvous() so that protocol violations and length checks happen in the beginning. Fixes bug 5645. - Set _WIN32_WINNT to 0x0501 consistently throughout the code, so that IPv6 stuff will compile on MSVC, and compilation issues will be easier to track down. Fixes bug 5861. o Minor bugfixes (correctness, on 0.2.2.x and earlier): - Exit nodes now correctly report EADDRINUSE and EADDRNOTAVAIL as resource exhaustion, so that clients can adjust their load to try other exits. Fixes bug 4710; bugfix on 0.1.0.1-rc, which started using END_STREAM_REASON_RESOURCELIMIT. - Don't check for whether the address we're using for outbound connections has changed until after the outbound connection has completed. On Windows, getsockname() doesn't succeed until the connection is finished. Fixes bug 5374; bugfix on 0.1.1.14-alpha. - If the configuration tries to set MyFamily on a bridge, refuse to do so, and warn about the security implications. Fixes bug 4657; bugfix on 0.2.0.3-alpha. - If the client fails to set a reasonable set of ciphersuites during its v2 handshake renegotiation, allow the renegotiation to continue nevertheless (i.e. send all the required certificates). Fixes bug 4591; bugfix on 0.2.0.20-rc. - When we receive a SIGHUP and the controller __ReloadTorrcOnSIGHUP option is set to 0 (which Vidalia version 0.2.16 now does when a SAVECONF attempt fails), perform other actions that SIGHUP usually causes (like reopening the logs). Fixes bug 5095; bugfix on 0.2.1.9-alpha. - If we fail to write a microdescriptor to the disk cache, do not continue replacing the old microdescriptor file. Fixes bug 2954; bugfix on 0.2.2.6-alpha. - Exit nodes don't need to fetch certificates for authorities that they don't recognize; only directory authorities, bridges, and caches need to do that. Fixes part of bug 2297; bugfix on 0.2.2.11-alpha. - Correctly handle checking the permissions on the parent directory of a control socket in the root directory. Bug found by Esteban Manchado Velázquez. Fixes bug 5089; bugfix on Tor 0.2.2.26-beta. - When told to add a bridge with the same digest as a preexisting bridge but a different addr:port, change the addr:port as requested. Previously we would not notice the change. Fixes half of bug 5603; fix on 0.2.2.26-beta. - End AUTHCHALLENGE error messages (in the control protocol) with a CRLF. Fixes bug 5760; bugfix on 0.2.2.36 and 0.2.3.13-alpha. o Minor bugfixes (on 0.2.3.x): - Turn an assertion (that the number of handshakes received as a server is not < 1) into a warning. Fixes bug 4873; bugfix on 0.2.3.1-alpha. - Format IPv4 addresses correctly in ADDRMAP events. (Previously, we had reversed them when the answer was cached.) Fixes bug 5723; bugfix on 0.2.3.1-alpha. - Work correctly on Linux systems with accept4 support advertised in their headers, but without accept4 support in the kernel. Fix by murb. Fixes bug 5762; bugfix on 0.2.3.1-alpha. - When told to add a bridge with the same addr:port as a preexisting bridge but a different transport, change the transport as requested. Previously we would not notice the change. Fixes half of bug 5603; fix on 0.2.3.2-alpha. - Avoid a "double-reply" warning when replying to a SOCKS request with a parse error. Patch from Fabian Keil. Fixes bug 4108; bugfix on 0.2.3.4-alpha. - Fix a bug where a bridge authority crashes if it has seen no directory requests when it's time to write statistics to disk. Fixes bug 5891; bugfix on 0.2.3.6-alpha. Also fixes bug 5508 in a better way. - Don't try to open non-control listeners when DisableNetwork is set. Previously, we'd open all listeners, then immediately close them. Fixes bug 5604; bugfix on 0.2.3.9-alpha. - Don't abort the managed proxy protocol if the managed proxy sends us an unrecognized line; ignore it instead. Fixes bug 5910; bugfix on 0.2.3.9-alpha. - Fix a compile warning in crypto.c when compiling with clang 3.1. Fixes bug 5969, bugfix on 0.2.3.9-alpha. - Fix a compilation issue on GNU Hurd, which doesn't have PATH_MAX. Fixes bug 5355; bugfix on 0.2.3.11-alpha. - Remove bogus definition of "_WIN32" from src/win32/orconfig.h, to unbreak the MSVC build. Fixes bug 5858; bugfix on 0.2.3.12-alpha. - Resolve numerous small warnings and build issues with MSVC. Resolves bug 5859. o Documentation fixes: - Improve the manual's documentation for the NT Service command-line options. Addresses ticket 3964. - Clarify SessionGroup documentation slightly; resolves ticket 5437. - Document the changes to the ORPort and DirPort options, and the fact that {OR/Dir}ListenAddress is now unnecessary (and therefore deprecated). Resolves ticket 5597. o Removed files: - Remove the torrc.bridge file: we don't use it for anything, and it had become badly desynchronized from torrc.sample. Resolves bug 5622. Changes in version 0.2.2.36 - 2012-05-24 Tor 0.2.2.36 updates the addresses for two of the eight directory authorities, fixes some potential anonymity and security issues, and fixes several crash bugs. Tor 0.2.1.x has reached its end-of-life. Those Tor versions have many known flaws, and nobody should be using them. You should upgrade. If you're using a Linux or BSD and its packages are obsolete, stop using those packages and upgrade anyway. o Directory authority changes: - Change IP address for maatuska (v3 directory authority). - Change IP address for ides (v3 directory authority), and rename it to turtles. o Security fixes: - When building or running with any version of OpenSSL earlier than 0.9.8s or 1.0.0f, disable SSLv3 support. These OpenSSL versions have a bug (CVE-2011-4576) in which their block cipher padding includes uninitialized data, potentially leaking sensitive information to any peer with whom they make a SSLv3 connection. Tor does not use SSL v3 by default, but a hostile client or server could force an SSLv3 connection in order to gain information that they shouldn't have been able to get. The best solution here is to upgrade to OpenSSL 0.9.8s or 1.0.0f (or later). But when building or running with a non-upgraded OpenSSL, we disable SSLv3 entirely to make sure that the bug can't happen. - Never use a bridge or a controller-supplied node as an exit, even if its exit policy allows it. Found by wanoskarnet. Fixes bug 5342. Bugfix on 0.1.1.15-rc (for controller-purpose descriptors) and 0.2.0.3-alpha (for bridge-purpose descriptors). - Only build circuits if we have a sufficient threshold of the total descriptors that are marked in the consensus with the "Exit" flag. This mitigates an attack proposed by wanoskarnet, in which all of a client's bridges collude to restrict the exit nodes that the client knows about. Fixes bug 5343. - Provide controllers with a safer way to implement the cookie authentication mechanism. With the old method, if another locally running program could convince a controller that it was the Tor process, then that program could trick the controller into telling it the contents of an arbitrary 32-byte file. The new "SAFECOOKIE" authentication method uses a challenge-response approach to prevent this attack. Fixes bug 5185; implements proposal 193. o Major bugfixes: - Avoid logging uninitialized data when unable to decode a hidden service descriptor cookie. Fixes bug 5647; bugfix on 0.2.1.5-alpha. - Avoid a client-side assertion failure when receiving an INTRODUCE2 cell on a general purpose circuit. Fixes bug 5644; bugfix on 0.2.1.6-alpha. - Fix builds when the path to sed, openssl, or sha1sum contains spaces, which is pretty common on Windows. Fixes bug 5065; bugfix on 0.2.2.1-alpha. - Correct our replacements for the timeradd() and timersub() functions on platforms that lack them (for example, Windows). The timersub() function is used when expiring circuits, while timeradd() is currently unused. Bug report and patch by Vektor. Fixes bug 4778; bugfix on 0.2.2.24-alpha. - Fix the SOCKET_OK test that we use to tell when socket creation fails so that it works on Win64. Fixes part of bug 4533; bugfix on 0.2.2.29-beta. Bug found by wanoskarnet. o Minor bugfixes: - Reject out-of-range times like 23:59:61 in parse_rfc1123_time(). Fixes bug 5346; bugfix on 0.0.8pre3. - Make our number-parsing functions always treat too-large values as an error, even when those values exceed the width of the underlying type. Previously, if the caller provided these functions with minima or maxima set to the extreme values of the underlying integer type, these functions would return those values on overflow rather than treating overflow as an error. Fixes part of bug 5786; bugfix on 0.0.9. - Older Linux kernels erroneously respond to strange nmap behavior by having accept() return successfully with a zero-length socket. When this happens, just close the connection. Previously, we would try harder to learn the remote address: but there was no such remote address to learn, and our method for trying to learn it was incorrect. Fixes bugs 1240, 4745, and 4747. Bugfix on 0.1.0.3-rc. Reported and diagnosed by "r1eo". - Correct parsing of certain date types in parse_http_time(). Without this patch, If-Modified-Since would behave incorrectly. Fixes bug 5346; bugfix on 0.2.0.2-alpha. Patch from Esteban Manchado Velázques. - Change the BridgePassword feature (part of the "bridge community" design, which is not yet implemented) to use a time-independent comparison. The old behavior might have allowed an adversary to use timing to guess the BridgePassword value. Fixes bug 5543; bugfix on 0.2.0.14-alpha. - Detect and reject certain misformed escape sequences in configuration values. Previously, these values would cause us to crash if received in a torrc file or over an authenticated control port. Bug found by Esteban Manchado Velázquez, and independently by Robert Connolly from Matta Consulting who further noted that it allows a post-authentication heap overflow. Patch by Alexander Schrijver. Fixes bugs 5090 and 5402 (CVE 2012-1668); bugfix on 0.2.0.16-alpha. - Fix a compile warning when using the --enable-openbsd-malloc configure option. Fixes bug 5340; bugfix on 0.2.0.20-rc. - During configure, detect when we're building with clang version 3.0 or lower and disable the -Wnormalized=id and -Woverride-init CFLAGS. clang doesn't support them yet. - When sending an HTTP/1.1 proxy request, include a Host header. Fixes bug 5593; bugfix on 0.2.2.1-alpha. - Fix a NULL-pointer dereference on a badly formed SETCIRCUITPURPOSE command. Found by mikeyc. Fixes bug 5796; bugfix on 0.2.2.9-alpha. - If we hit the error case where routerlist_insert() replaces an existing (old) server descriptor, make sure to remove that server descriptor from the old_routers list. Fix related to bug 1776. Bugfix on 0.2.2.18-alpha. o Minor bugfixes (documentation and log messages): - Fix a typo in a log message in rend_service_rendezvous_has_opened(). Fixes bug 4856; bugfix on Tor 0.0.6. - Update "ClientOnly" man page entry to explain that there isn't really any point to messing with it. Resolves ticket 5005. - Document the GiveGuardFlagTo_CVE_2011_2768_VulnerableRelays directory authority option (introduced in Tor 0.2.2.34). - Downgrade the "We're missing a certificate" message from notice to info: people kept mistaking it for a real problem, whereas it is seldom the problem even when we are failing to bootstrap. Fixes bug 5067; bugfix on 0.2.0.10-alpha. - Correctly spell "connect" in a log message on failure to create a controlsocket. Fixes bug 4803; bugfix on 0.2.2.26-beta. - Clarify the behavior of MaxCircuitDirtiness with hidden service circuits. Fixes issue 5259. o Minor features: - Directory authorities now reject versions of Tor older than 0.2.1.30, and Tor versions between 0.2.2.1-alpha and 0.2.2.20-alpha inclusive. These versions accounted for only a small fraction of the Tor network, and have numerous known security issues. Resolves issue 4788. - Update to the May 1 2012 Maxmind GeoLite Country database. - Feature removal: - When sending or relaying a RELAY_EARLY cell, we used to convert it to a RELAY cell if the connection was using the v1 link protocol. This was a workaround for older versions of Tor, which didn't handle RELAY_EARLY cells properly. Now that all supported versions can handle RELAY_EARLY cells, and now that we're enforcing the "no RELAY_EXTEND commands except in RELAY_EARLY cells" rule, remove this workaround. Addresses bug 4786. Changes in version 0.2.3.15-alpha - 2012-04-30 Tor 0.2.3.15-alpha fixes a variety of smaller bugs, including making the development branch build on Windows again. o Minor bugfixes (on 0.2.2.x and earlier): - Make sure that there are no unhandled pending TLS errors before reading from a TLS stream. We had checks in 0.1.0.3-rc, but lost them in 0.1.0.5-rc when we refactored read_to_buf_tls(). Bugfix on 0.1.0.5-rc; fixes bug 4528. - Fix an assert that directory authorities could trigger on sighup during some configuration state transitions. We now don't treat it as a fatal error when the new descriptor we just generated in init_keys() isn't accepted. Fixes bug 4438; bugfix on 0.2.1.9-alpha. - After we pick a directory mirror, we would refuse to use it if it's in our ExcludeExitNodes list, resulting in mysterious failures to bootstrap for people who just wanted to avoid exiting from certain locations. Fixes bug 5623; bugfix on 0.2.2.25-alpha. - When building with --enable-static-tor on OpenBSD, do not erroneously attempt to link -lrt. Fixes bug 5103. o Minor bugfixes (on 0.2.3.x): - When Tor is built with kernel headers from a recent (last few years) Linux kernel, do not fail to run on older (pre-2.6.28 Linux kernels). Fixes bug 5112; bugfix on 0.2.3.1-alpha. - Fix cross-compilation issues with mingw. Bugfixes on 0.2.3.6-alpha and 0.2.3.12-alpha. - Fix compilation with miniupnpc version 1.6; patch from Anthony G. Basile. Fixes bug 5434; bugfix on 0.2.3.12-alpha. - Fix compilation with MSVC, which had defined MS_WINDOWS. Bugfix on 0.2.3.13-alpha; found and fixed by Gisle Vanem. - Fix compilation on platforms without unistd.h, or where environ is defined in stdlib.h. Fixes bug 5704; bugfix on 0.2.3.13-alpha. o Minor features: - Directory authorities are now a little more lenient at accepting older router descriptors, or newer router descriptors that don't make big changes. This should help ameliorate past and future issues where routers think they have uploaded valid descriptors, but the authorities don't think so. Fix for ticket 2479. - Make the code that clients use to detect an address change be IPv6-aware, so that it won't fill clients' logs with error messages when trying to get the IPv4 address of an IPv6 connection. Implements ticket 5537. o Removed features: - Remove the GiveGuardFlagTo_CVE_2011_2768_VulnerableRelays option; authorities needed to use it for a while to keep the network working as people upgraded to 0.2.1.31, 0.2.2.34, or 0.2.3.6-alpha, but that was six months ago. As of now, it should no longer be needed or used. Changes in version 0.2.3.14-alpha - 2012-04-23 Tor 0.2.3.14-alpha fixes yet more bugs to get us closer to a release candidate. It also dramatically speeds up AES: fast relays should consider switching to the newer OpenSSL library. o Directory authority changes: - Change IP address for ides (v3 directory authority), and rename it to turtles. o Major bugfixes: - Avoid logging uninitialized data when unable to decode a hidden service descriptor cookie. Fixes bug 5647; bugfix on 0.2.1.5-alpha. - Avoid a client-side assertion failure when receiving an INTRODUCE2 cell on a general purpose circuit. Fixes bug 5644; bugfix on 0.2.1.6-alpha. - If authorities are unable to get a v2 consensus document from other directory authorities, they no longer fall back to fetching them from regular directory caches. Fixes bug 5635; bugfix on 0.2.2.26-beta, where routers stopped downloading v2 consensus documents entirely. - When we start a Tor client with a normal consensus already cached, be willing to download a microdescriptor consensus. Fixes bug 4011; fix on 0.2.3.1-alpha. o Major features (performance): - When built to use OpenSSL 1.0.1, and built for an x86 or x86_64 instruction set, take advantage of OpenSSL's AESNI, bitsliced, or vectorized AES implementations as appropriate. These can be much, much faster than other AES implementations. o Minor bugfixes (0.2.2.x and earlier): - Don't launch more than 10 service-side introduction-point circuits for a hidden service in five minutes. Previously, we would consider launching more introduction-point circuits if at least one second had passed without any introduction-point circuits failing. Fixes bug 4607; bugfix on 0.0.7pre1. - Change the BridgePassword feature (part of the "bridge community" design, which is not yet implemented) to use a time-independent comparison. The old behavior might have allowed an adversary to use timing to guess the BridgePassword value. Fixes bug 5543; bugfix on 0.2.0.14-alpha. - Enforce correct return behavior of tor_vsscanf() when the '%%' pattern is used. Fixes bug 5558. Bugfix on 0.2.1.13. - When sending an HTTP/1.1 proxy request, include a Host header. Fixes bug 5593; bugfix on 0.2.2.1-alpha. - Don't log that we have "decided to publish new relay descriptor" unless we are actually publishing a descriptor. Fixes bug 3942; bugfix on 0.2.2.28-beta. o Minor bugfixes (0.2.3.x): - Fix a bug where a bridge authority crashes (on a failed assert) if it has seen no directory requests when it's time to write statistics to disk. Fixes bug 5508. Bugfix on 0.2.3.6-alpha. - Fix bug stomping on ORPort option NoListen and ignoring option NoAdvertise. Fixes bug 5151; bugfix on 0.2.3.9-alpha. - In the testsuite, provide a large enough buffer in the tor_sscanf unit test. Otherwise we'd overrun that buffer and crash during the unit tests. Found by weasel. Fixes bug 5449; bugfix on 0.2.3.12-alpha. - Make sure we create the keys directory if it doesn't exist and we're about to store the dynamic Diffie-Hellman parameters. Fixes bug 5572; bugfix on 0.2.3.13-alpha. - Fix a small memory leak when trying to decode incorrect base16 authenticator during SAFECOOKIE authentication. Found by Coverity Scan. Fixes CID 507. Bugfix on 0.2.3.13-alpha. o Minor features: - Add more information to a log statement that might help track down bug 4091. If you're seeing "Bug: tor_addr_is_internal() called with a non-IP address" messages (or any Bug messages, for that matter!), please let us know about it. - Relays now understand an IPv6 address when they get one from a directory server. Resolves ticket 4875. - Resolve IPv6 addresses in bridge and entry statistics to country code "??" which means we at least count them. Resolves ticket 5053; improves on 0.2.3.9-alpha. - Update to the April 3 2012 Maxmind GeoLite Country database. - Begin a doc/state-contents.txt file to explain the contents of the Tor state file. Fixes bug 2987. o Default torrc changes: - Stop listing "socksport 9050" in torrc.sample. We open a socks port on 9050 by default anyway, so this should not change anything in practice. - Stop mentioning the deprecated *ListenAddress options in torrc.sample. Fixes bug 5438. - Document unit of bandwidth related options in sample torrc. Fixes bug 5621. o Removed features: - The "torify" script no longer supports the "tsocks" socksifier tool, since tsocks doesn't support DNS and UDP right for Tor. Everyone should be using torsocks instead. Fixes bugs 3530 and 5180. Based on a patch by "ugh". o Code refactoring: - Change the symmetric cipher interface so that creating and initializing a stream cipher are no longer separate functions. - Remove all internal support for unpadded RSA. We never used it, and it would be a bad idea to start. Changes in version 0.2.3.13-alpha - 2012-03-26 Tor 0.2.3.13-alpha fixes a variety of stability and correctness bugs in managed pluggable transports, as well as providing other cleanups that get us closer to a release candidate. o Directory authority changes: - Change IP address for maatuska (v3 directory authority). o Security fixes: - Provide controllers with a safer way to implement the cookie authentication mechanism. With the old method, if another locally running program could convince a controller that it was the Tor process, then that program could trick the controller into telling it the contents of an arbitrary 32-byte file. The new "SAFECOOKIE" authentication method uses a challenge-response approach to prevent this attack. Fixes bug 5185, implements proposal 193. - Never use a bridge or a controller-supplied node as an exit, even if its exit policy allows it. Found by wanoskarnet. Fixes bug 5342. Bugfix on 0.1.1.15-rc (for controller-purpose descriptors) and 0.2.0.3-alpha (for bridge-purpose descriptors). - Only build circuits if we have a sufficient threshold of the total descriptors that are marked in the consensus with the "Exit" flag. This mitigates an attack proposed by wanoskarnet, in which all of a client's bridges collude to restrict the exit nodes that the client knows about. Fixes bug 5343. o Major bugfixes (on Tor 0.2.3.x): - Avoid an assert when managed proxies like obfsproxy are configured, and we receive HUP signals or setconf attempts too rapidly. This situation happens most commonly when Vidalia tries to attach to Tor or tries to configure the Tor it's attached to. Fixes bug 5084; bugfix on 0.2.3.6-alpha. - Fix a relay-side pluggable transports bug where managed proxies were unreachable from the Internet, because Tor asked them to bind on localhost. Fixes bug 4725; bugfix on 0.2.3.9-alpha. - Stop discarding command-line arguments when TestingTorNetwork is set. Discovered by Kevin Bauer. Fixes bug 5373; bugfix on 0.2.3.9-alpha, where task 4552 added support for two layers of torrc files. - Resume allowing the unit tests to run in gdb. This was accidentally made impossible when the DisableDebuggerAttachment option was introduced. Fixes bug 5448; bugfix on 0.2.3.9-alpha. - Resume building with nat-pmp support. Fixes bug 4955; bugfix on 0.2.3.11-alpha. Reported by Anthony G. Basile. o Minor bugfixes (on 0.2.2.x and earlier): - Ensure we don't cannibalize circuits that are longer than three hops already, so we don't end up making circuits with 5 or more hops. Patch contributed by wanoskarnet. Fixes bug 5231; bugfix on 0.1.0.1-rc which introduced cannibalization. - Detect and reject certain misformed escape sequences in configuration values. Previously, these values would cause us to crash if received in a torrc file or over an authenticated control port. Bug found by Esteban Manchado Velázquez, and independently by Robert Connolly from Matta Consulting who further noted that it allows a post-authentication heap overflow. Patch by Alexander Schrijver. Fixes bugs 5090 and 5402 (CVE 2012-1668); bugfix on 0.2.0.16-alpha. - Fix a compile warning when using the --enable-openbsd-malloc configure option. Fixes bug 5340; bugfix on 0.2.0.20-rc. - Directory caches no longer refuse to clean out descriptors because of missing v2 networkstatus documents, unless they're configured to retrieve v2 networkstatus documents. Fixes bug 4838; bugfix on 0.2.2.26-beta. Patch by Daniel Bryg. - Update to the latest version of the tinytest unit testing framework. This includes a couple of bugfixes that can be relevant for running forked unit tests on Windows, and removes all reserved identifiers. o Minor bugfixes (on 0.2.3.x): - On a failed pipe() call, don't leak file descriptors. Fixes bug 4296; bugfix on 0.2.3.1-alpha. - Spec conformance: on a v3 handshake, do not send a NETINFO cell until after we have received a CERTS cell. Fixes bug 4361; bugfix on 0.2.3.6-alpha. Patch by "frosty". - When binding to an IPv6 address, set the IPV6_V6ONLY socket option, so that the IP stack doesn't decide to use it for IPv4 too. Fixes bug 4760; bugfix on 0.2.3.9-alpha. - Ensure that variables set in Tor's environment cannot override environment variables that Tor passes to a managed pluggable-transport proxy. Previously, Tor would pass every variable in its environment to managed proxies along with the new ones, in such a way that on many operating systems, the inherited environment variables would override those which Tor tried to explicitly set. Bugfix on 0.2.3.12-alpha for most Unixoid systems; bugfix on 0.2.3.9-alpha for Windows. o Minor features: - A wide variety of new unit tests by Esteban Manchado Velázquez. - Shorten links in the tor-exit-notice file. Patch by Christian Kujau. - Update to the March 6 2012 Maxmind GeoLite Country database. Changes in version 0.2.3.12-alpha - 2012-02-13 Tor 0.2.3.12-alpha lets fast exit relays scale better, allows clients to use bridges that run Tor 0.2.2.x, and resolves several big bugs when Tor is configured to use a pluggable transport like obfsproxy. o Major bugfixes: - Fix builds when the path to sed, openssl, or sha1sum contains spaces, which is pretty common on Windows. Fixes bug 5065; bugfix on 0.2.2.1-alpha. - Set the SO_REUSEADDR socket option before we call bind() on outgoing connections. This change should allow busy exit relays to stop running out of available sockets as quickly. Fixes bug 4950; bugfix on 0.2.2.26-beta. - Allow 0.2.3.x clients to use 0.2.2.x bridges. Previously the client would ask the bridge for microdescriptors, which are only supported in 0.2.3.x, and then fail to bootstrap when it didn't get the answers it wanted. Fixes bug 4013; bugfix on 0.2.3.2-alpha. - Properly set up obfsproxy's environment when in managed mode. The Tor Browser Bundle needs LD_LIBRARY_PATH to be passed to obfsproxy, and when you run your Tor as a daemon, there's no HOME. Fixes bugs 5076 and 5082; bugfix on 0.2.3.6-alpha. o Minor features: - Use the dead_strip option when building Tor on OS X. This reduces binary size by almost 19% when linking openssl and libevent statically, which we do for Tor Browser Bundle. - Fix broken URLs in the sample torrc file, and tell readers about the OutboundBindAddress, ExitPolicyRejectPrivate, and PublishServerDescriptor options. Addresses bug 4652. - Update to the February 7 2012 Maxmind GeoLite Country database. o Minor bugfixes: - Downgrade the "We're missing a certificate" message from notice to info: people kept mistaking it for a real problem, whereas it is seldom the problem even when we are failing to bootstrap. Fixes bug 5067; bugfix on 0.2.0.10-alpha. - Don't put "TOR_PT_EXTENDED_SERVER_PORT=127.0.0.1:4200" in a managed pluggable transport server proxy's environment. Previously, we would put it there, even though Tor doesn't implement an 'extended server port' yet, and even though Tor almost certainly isn't listening at that address. For now, we set it to an empty string to avoid crashing older obfsproxies. Bugfix on 0.2.3.6-alpha. - Log the heartbeat message every HeartbeatPeriod seconds, not every HeartbeatPeriod + 1 seconds. Fixes bug 4942; bugfix on 0.2.3.1-alpha. Bug reported by Scott Bennett. - Calculate absolute paths correctly on Windows. Fixes bug 4973; bugfix on 0.2.3.11-alpha. - Update "ClientOnly" man page entry to explain that there isn't really any point to messing with it. Resolves ticket 5005. - Use the correct CVE number for CVE-2011-4576 in our comments and log messages. Found by "fermenthor". Resolves bug 5066; bugfix on 0.2.3.11-alpha. o Code simplifications and refactoring: - Use the _WIN32 macro throughout our code to detect Windows. (Previously we had used the obsolete 'WIN32' and the idiosyncratic 'MS_WINDOWS'.) Changes in version 0.2.3.11-alpha - 2012-01-22 Tor 0.2.3.11-alpha marks feature-freeze for the 0.2.3 tree. It deploys the last step of the plan to limit maximum circuit length, includes a wide variety of hidden service performance and correctness fixes, works around an OpenSSL security flaw if your distro is too stubborn to upgrade, and fixes a bunch of smaller issues. o Major features: - Now that Tor 0.2.0.x is completely deprecated, enable the final part of "Proposal 110: Avoiding infinite length circuits" by refusing all circuit-extend requests that do not use a relay_early cell. This change helps Tor resist a class of denial-of-service attacks by limiting the maximum circuit length. - Adjust the number of introduction points that a hidden service will try to maintain based on how long its introduction points remain in use and how many introductions they handle. Fixes part of bug 3825. - Try to use system facilities for enumerating local interface addresses, before falling back to our old approach (which was binding a UDP socket, and calling getsockname() on it). That approach was scaring OS X users whose draconian firewall software warned about binding to UDP sockets, regardless of whether packets were sent. Now we try to use getifaddrs(), SIOCGIFCONF, or GetAdaptersAddresses(), depending on what the system supports. Resolves ticket 1827. o Major security workaround: - When building or running with any version of OpenSSL earlier than 0.9.8s or 1.0.0f, disable SSLv3 support. These OpenSSL versions have a bug (CVE-2011-4576) in which their block cipher padding includes uninitialized data, potentially leaking sensitive information to any peer with whom they make a SSLv3 connection. Tor does not use SSL v3 by default, but a hostile client or server could force an SSLv3 connection in order to gain information that they shouldn't have been able to get. The best solution here is to upgrade to OpenSSL 0.9.8s or 1.0.0f (or later). But when building or running with a non-upgraded OpenSSL, we disable SSLv3 entirely to make sure that the bug can't happen. o Major bugfixes: - Fix the SOCKET_OK test that we use to tell when socket creation fails so that it works on Win64. Fixes part of bug 4533; bugfix on 0.2.2.29-beta. Bug found by wanoskarnet. - Correct our replacements for the timeradd() and timersub() functions on platforms that lack them (for example, Windows). The timersub() function is used when expiring circuits, while timeradd() is currently unused. Bug report and patch by Vektor. Fixes bug 4778; bugfix on 0.2.2.24-alpha and 0.2.3.1-alpha. - Do not use OpenSSL 1.0.0's counter mode: it has a critical bug that was fixed in OpenSSL 1.0.0a. We test for the counter mode bug at runtime, not compile time, because some distributions hack their OpenSSL to mis-report its version. Fixes bug 4779; bugfix on 0.2.3.9-alpha. Found by Pascal. o Minor features (controller): - Use absolute path names when reporting the torrc filename in the control protocol, so a controller can more easily find the torrc file. Resolves bug 1101. - Extend the control protocol to report flags that control a circuit's path selection in CIRC events and in replies to 'GETINFO circuit-status'. Implements part of ticket 2411. - Extend the control protocol to report the hidden service address and current state of a hidden-service-related circuit in CIRC events and in replies to 'GETINFO circuit-status'. Implements part of ticket 2411. - When reporting the path to the cookie file to the controller, give an absolute path. Resolves ticket 4881. - Allow controllers to request an event notification whenever a circuit is cannibalized or its purpose is changed. Implements part of ticket 3457. - Include the creation time of a circuit in CIRC and CIRC2 control-port events and the list produced by the 'GETINFO circuit-status' control-port command. o Minor features (directory authorities): - Directory authorities now reject versions of Tor older than 0.2.1.30, and Tor versions between 0.2.2.1-alpha and 0.2.2.20-alpha inclusive. These versions accounted for only a small fraction of the Tor network, and have numerous known security issues. Resolves issue 4788. - Authority operators can now vote for all relays in a given set of countries to be BadDir/BadExit/Invalid/Rejected. - Provide two consensus parameters (FastFlagMinThreshold and FastFlagMaxThreshold) to control the range of allowable bandwidths for the Fast directory flag. These allow authorities to run experiments on appropriate requirements for being a "Fast" node. The AuthDirFastGuarantee config value still applies. - Document the GiveGuardFlagTo_CVE_2011_2768_VulnerableRelays directory authority option (introduced in Tor 0.2.2.34). o Minor features (other): - Don't disable the DirPort when we cannot exceed our AccountingMax limit during this interval because the effective bandwidthrate is low enough. This is useful in a situation where AccountMax is only used as an additional safeguard or to provide statistics. - Prepend an informative header to generated dynamic_dh_params files. - If EntryNodes are given, but UseEntryGuards is set to 0, warn that EntryNodes will have no effect. Resolves issue 2571. - Log more useful messages when we fail to disable debugger attachment. - Log which authority we're missing votes from when we go to fetch them from the other auths. - Log (at debug level) whenever a circuit's purpose is changed. - Add missing documentation for the MaxClientCircuitsPending, UseMicrodescriptors, UserspaceIOCPBuffers, and _UseFilteringSSLBufferevents options, all introduced during the 0.2.3.x series. - Update to the January 3 2012 Maxmind GeoLite Country database. o Minor bugfixes (hidden services): - Don't close hidden service client circuits which have almost finished connecting to their destination when they reach the normal circuit-build timeout. Previously, we would close introduction circuits which are waiting for an acknowledgement from the introduction point, and rendezvous circuits which have been specified in an INTRODUCE1 cell sent to a hidden service, after the normal CBT. Now, we mark them as 'timed out', and launch another rendezvous attempt in parallel. This behavior change can be disabled using the new CloseHSClientCircuitsImmediatelyOnTimeout option. Fixes part of bug 1297; bugfix on 0.2.2.2-alpha. - Don't close hidden-service-side rendezvous circuits when they reach the normal circuit-build timeout. This behaviour change can be disabled using the new CloseHSServiceRendCircuitsImmediatelyOnTimeout option. Fixes the remaining part of bug 1297; bugfix on 0.2.2.2-alpha. - Make sure we never mark the wrong rendezvous circuit as having had its introduction cell acknowleged by the introduction-point relay. Previously, when we received an INTRODUCE_ACK cell on a client-side hidden-service introduction circuit, we might have marked a rendezvous circuit other than the one we specified in the INTRODUCE1 cell as INTRO_ACKED, which would have produced a warning message and interfered with the hidden service connection-establishment process. Fixes bug 4759; bugfix on 0.2.3.3-alpha, when we added the stream-isolation feature which might cause Tor to open multiple rendezvous circuits for the same hidden service. - Don't trigger an assertion failure when we mark a new client-side hidden-service introduction circuit for close during the process of creating it. Fixes bug 4796; bugfix on 0.2.3.6-alpha. Reported by murb. o Minor bugfixes (log messages): - Correctly spell "connect" in a log message on failure to create a controlsocket. Fixes bug 4803; bugfix on 0.2.2.26-beta and 0.2.3.2-alpha. - Fix a typo in a log message in rend_service_rendezvous_has_opened(). Fixes bug 4856; bugfix on Tor 0.0.6. - Fix the log message describing how we work around discovering that our version is the ill-fated OpenSSL 0.9.8l. Fixes bug 4837; bugfix on 0.2.2.9-alpha. - When logging about a disallowed .exit name, do not also call it an "invalid onion address". Fixes bug 3325; bugfix on 0.2.2.9-alpha. o Minor bugfixes (build fixes): - During configure, detect when we're building with clang version 3.0 or lower and disable the -Wnormalized=id and -Woverride-init CFLAGS. clang doesn't support them yet. - During configure, search for library containing cos function as libm lives in libcore on some platforms (BeOS/Haiku). Linking against libm was hard-coded before. Fixes the first part of bug 4727; bugfix on 0.2.2.2-alpha. Patch and analysis by Martin Hebnes Pedersen. - Detect attempts to build Tor on (as yet hypothetical) versions of Windows where sizeof(intptr_t) != sizeof(SOCKET). Partial fix for bug 4533. Bugfix on 0.2.2.28-beta. - Preprocessor directives should not be put inside the arguments of a macro. This would break compilation with GCC releases prior to version 3.3. We would never recommend such an old GCC version, but it is apparently required for binary compatibility on some platforms (namely, certain builds of Haiku). Fixes the other part of bug 4727; bugfix on 0.2.3.3-alpha. Patch and analysis by Martin Hebnes Pedersen. o Minor bugfixes (other): - Older Linux kernels erroneously respond to strange nmap behavior by having accept() return successfully with a zero-length socket. When this happens, just close the connection. Previously, we would try harder to learn the remote address: but there was no such remote address to learn, and our method for trying to learn it was incorrect. Fixes bugs 1240, 4745, and 4747. Bugfix on 0.1.0.3-rc. Reported and diagnosed by "r1eo". - Fix null-pointer access that could occur if TLS allocation failed. Fixes bug 4531; bugfix on 0.2.0.20-rc. Found by "troll_un". This was erroneously listed as fixed in 0.2.3.9-alpha, but the fix had accidentally been reverted. - Fix our implementation of crypto_random_hostname() so it can't overflow on ridiculously large inputs. (No Tor version has ever provided this kind of bad inputs, but let's be correct in depth.) Fixes bug 4413; bugfix on 0.2.2.9-alpha. Fix by Stephen Palmateer. - Find more places in the code that should have been testing for invalid sockets using the SOCKET_OK macro. Required for a fix for bug 4533. Bugfix on 0.2.2.28-beta. - Fix an assertion failure when, while running with bufferevents, a connection finishes connecting after it is marked for close, but before it is closed. Fixes bug 4697; bugfix on 0.2.3.1-alpha. - test_util_spawn_background_ok() hardcoded the expected value for ENOENT to 2. This isn't portable as error numbers are platform specific, and particularly the hurd has ENOENT at 0x40000002. Construct expected string at runtime, using the correct value for ENOENT. Fixes bug 4733; bugfix on 0.2.3.1-alpha. - Reject attempts to disable DisableDebuggerAttachment while Tor is running. Fixes bug 4650; bugfix on 0.2.3.9-alpha. - Use an appropriate-width type for sockets in tor-fw-helper on win64. Fixes bug 1983 at last. Bugfix on 0.2.3.9-alpha. o Feature removal: - When sending or relaying a RELAY_EARLY cell, we used to convert it to a RELAY cell if the connection was using the v1 link protocol. This was a workaround for older versions of Tor, which didn't handle RELAY_EARLY cells properly. Now that all supported versions can handle RELAY_EARLY cells, and now that we're enforcing the "no RELAY_EXTEND commands except in RELAY_EARLY cells" rule, remove this workaround. Addresses bug 4786. o Code simplifications and refactoring: - Use OpenSSL's built-in SSL_state_string_long() instead of our own homebrewed ssl_state_to_string() replacement. Patch from Emile Snyder. Fixes bug 4653. - Use macros to indicate OpenSSL versions, so we don't need to worry about accidental hexadecimal bit shifts. - Remove some workaround code for OpenSSL 0.9.6 (which is no longer supported). - Convert more instances of tor_snprintf+tor_strdup into tor_asprintf. - Use the smartlist_add_asprintf() alias more consistently. - Use a TOR_INVALID_SOCKET macro when initializing a socket to an invalid value, rather than just -1. - Rename a handful of old identifiers, mostly related to crypto structures and crypto functions. By convention, our "create an object" functions are called "type_new()", our "free an object" functions are called "type_free()", and our types indicate that they are types only with a final "_t". But a handful of older types and functions broke these rules, with function names like "type_create" or "subsystem_op_type", or with type names like type_env_t. Changes in version 0.2.3.10-alpha - 2011-12-16 Tor 0.2.3.10-alpha fixes a critical heap-overflow security issue in Tor's buffers code. Absolutely everybody should upgrade. The bug relied on an incorrect calculation when making data continuous in one of our IO buffers, if the first chunk of the buffer was misaligned by just the wrong amount. The miscalculation would allow an attacker to overflow a piece of heap-allocated memory. To mount this attack, the attacker would need to either open a SOCKS connection to Tor's SocksPort (usually restricted to localhost), or target a Tor instance configured to make its connections through a SOCKS proxy (which Tor does not do by default). Good security practice requires that all heap-overflow bugs should be presumed to be exploitable until proven otherwise, so we are treating this as a potential code execution attack. Please upgrade immediately! This bug does not affect bufferevents-based builds of Tor. Special thanks to "Vektor" for reporting this issue to us! This release also contains a few minor bugfixes for issues discovered in 0.2.3.9-alpha. o Major bugfixes: - Fix a heap overflow bug that could occur when trying to pull data into the first chunk of a buffer, when that chunk had already had some data drained from it. Fixes CVE-2011-2778; bugfix on 0.2.0.16-alpha. Reported by "Vektor". o Minor bugfixes: - If we can't attach streams to a rendezvous circuit when we finish connecting to a hidden service, clear the rendezvous circuit's stream-isolation state and try to attach streams again. Previously, we cleared rendezvous circuits' isolation state either too early (if they were freshly built) or not at all (if they had been built earlier and were cannibalized). Bugfix on 0.2.3.3-alpha; fixes bug 4655. - Fix compilation of the libnatpmp helper on non-Windows. Bugfix on 0.2.3.9-alpha; fixes bug 4691. Reported by Anthony G. Basile. - Fix an assertion failure when a relay with accounting enabled starts up while dormant. Fixes bug 4702; bugfix on 0.2.3.9-alpha. o Minor features: - Update to the December 6 2011 Maxmind GeoLite Country database. Changes in version 0.2.2.35 - 2011-12-16 Tor 0.2.2.35 fixes a critical heap-overflow security issue in Tor's buffers code. Absolutely everybody should upgrade. The bug relied on an incorrect calculation when making data continuous in one of our IO buffers, if the first chunk of the buffer was misaligned by just the wrong amount. The miscalculation would allow an attacker to overflow a piece of heap-allocated memory. To mount this attack, the attacker would need to either open a SOCKS connection to Tor's SocksPort (usually restricted to localhost), or target a Tor instance configured to make its connections through a SOCKS proxy (which Tor does not do by default). Good security practice requires that all heap-overflow bugs should be presumed to be exploitable until proven otherwise, so we are treating this as a potential code execution attack. Please upgrade immediately! This bug does not affect bufferevents-based builds of Tor. Special thanks to "Vektor" for reporting this issue to us! Tor 0.2.2.35 also fixes several bugs in previous versions, including crash bugs for unusual configurations, and a long-term bug that would prevent Tor from starting on Windows machines with draconian AV software. With this release, we remind everyone that 0.2.0.x has reached its formal end-of-life. Those Tor versions have many known flaws, and nobody should be using them. You should upgrade -- ideally to the 0.2.2.x series. If you're using a Linux or BSD and its packages are obsolete, stop using those packages and upgrade anyway. The Tor 0.2.1.x series is also approaching its end-of-life: it will no longer receive support after some time in early 2012. o Major bugfixes: - Fix a heap overflow bug that could occur when trying to pull data into the first chunk of a buffer, when that chunk had already had some data drained from it. Fixes CVE-2011-2778; bugfix on 0.2.0.16-alpha. Reported by "Vektor". - Initialize Libevent with the EVENT_BASE_FLAG_NOLOCK flag enabled, so that it doesn't attempt to allocate a socketpair. This could cause some problems on Windows systems with overzealous firewalls. Fix for bug 4457; workaround for Libevent versions 2.0.1-alpha through 2.0.15-stable. - If we mark an OR connection for close based on a cell we process, don't process any further cells on it. We already avoid further reads on marked-for-close connections, but now we also discard the cells we'd already read. Fixes bug 4299; bugfix on 0.2.0.10-alpha, which was the first version where we might mark a connection for close based on processing a cell on it. - Correctly sanity-check that we don't underflow on a memory allocation (and then assert) for hidden service introduction point decryption. Bug discovered by Dan Rosenberg. Fixes bug 4410; bugfix on 0.2.1.5-alpha. - Fix a memory leak when we check whether a hidden service descriptor has any usable introduction points left. Fixes bug 4424. Bugfix on 0.2.2.25-alpha. - Don't crash when we're running as a relay and don't have a GeoIP file. Bugfix on 0.2.2.34; fixes bug 4340. This backports a fix we've had in the 0.2.3.x branch already. - When running as a client, do not print a misleading (and plain wrong) log message that we're collecting "directory request" statistics: clients don't collect statistics. Also don't create a useless (because empty) stats file in the stats/ directory. Fixes bug 4353; bugfix on 0.2.2.34. o Minor bugfixes: - Detect failure to initialize Libevent. This fix provides better detection for future instances of bug 4457. - Avoid frequent calls to the fairly expensive cull_wedged_cpuworkers function. This was eating up hideously large amounts of time on some busy servers. Fixes bug 4518; bugfix on 0.0.9.8. - Resolve an integer overflow bug in smartlist_ensure_capacity(). Fixes bug 4230; bugfix on Tor 0.1.0.1-rc. Based on a patch by Mansour Moufid. - Don't warn about unused log_mutex in log.c when building with --disable-threads using a recent GCC. Fixes bug 4437; bugfix on 0.1.0.6-rc which introduced --disable-threads. - When configuring, starting, or stopping an NT service, stop immediately after the service configuration attempt has succeeded or failed. Fixes bug 3963; bugfix on 0.2.0.7-alpha. - When sending a NETINFO cell, include the original address received for the other side, not its canonical address. Found by "troll_un"; fixes bug 4349; bugfix on 0.2.0.10-alpha. - Fix a typo in a hibernation-related log message. Fixes bug 4331; bugfix on 0.2.2.23-alpha; found by "tmpname0901". - Fix a memory leak in launch_direct_bridge_descriptor_fetch() that occurred when a client tried to fetch a descriptor for a bridge in ExcludeNodes. Fixes bug 4383; bugfix on 0.2.2.25-alpha. - Backport fixes for a pair of compilation warnings on Windows. Fixes bug 4521; bugfix on 0.2.2.28-beta and on 0.2.2.29-beta. - If we had ever tried to call tor_addr_to_str on an address of unknown type, we would have done a strdup on an uninitialized buffer. Now we won't. Fixes bug 4529; bugfix on 0.2.1.3-alpha. Reported by "troll_un". - Correctly detect and handle transient lookup failures from tor_addr_lookup. Fixes bug 4530; bugfix on 0.2.1.5-alpha. Reported by "troll_un". - Fix null-pointer access that could occur if TLS allocation failed. Fixes bug 4531; bugfix on 0.2.0.20-rc. Found by "troll_un". - Use tor_socket_t type for listener argument to accept(). Fixes bug 4535; bugfix on 0.2.2.28-beta. Found by "troll_un". o Minor features: - Add two new config options for directory authorities: AuthDirFastGuarantee sets a bandwidth threshold for guaranteeing the Fast flag, and AuthDirGuardBWGuarantee sets a bandwidth threshold that is always sufficient to satisfy the bandwidth requirement for the Guard flag. Now it will be easier for researchers to simulate Tor networks with different values. Resolves ticket 4484. - When Tor ignores a hidden service specified in its configuration, include the hidden service's directory in the warning message. Previously, we would only tell the user that some hidden service was ignored. Bugfix on 0.0.6; fixes bug 4426. - Update to the December 6 2011 Maxmind GeoLite Country database. o Packaging changes: - Make it easier to automate expert package builds on Windows, by removing an absolute path from makensis.exe command. Changes in version 0.2.3.9-alpha - 2011-12-08 Tor 0.2.3.9-alpha introduces initial IPv6 support for bridges, adds a "DisableNetwork" security feature that bundles can use to avoid touching the network until bridges are configured, moves forward on the pluggable transport design, fixes a flaw in the hidden service design that unnecessarily prevented clients with wrong clocks from reaching hidden services, and fixes a wide variety of other issues. o Major features: - Clients can now connect to private bridges over IPv6. Bridges still need at least one IPv4 address in order to connect to other relays. Note that we don't yet handle the case where the user has two bridge lines for the same bridge (one IPv4, one IPv6). Implements parts of proposal 186. - New "DisableNetwork" config option to prevent Tor from launching any connections or accepting any connections except on a control port. Bundles and controllers can set this option before letting Tor talk to the rest of the network, for example to prevent any connections to a non-bridge address. Packages like Orbot can also use this option to instruct Tor to save power when the network is off. - Clients and bridges can now be configured to use a separate "transport" proxy. This approach makes the censorship arms race easier by allowing bridges to use protocol obfuscation plugins. It implements the "managed proxy" part of proposal 180 (ticket 3472). - When using OpenSSL 1.0.0 or later, use OpenSSL's counter mode implementation. It makes AES_CTR about 7% faster than our old one (which was about 10% faster than the one OpenSSL used to provide). Resolves ticket 4526. - Add a "tor2web mode" for clients that want to connect to hidden services non-anonymously (and possibly more quickly). As a safety measure to try to keep users from turning this on without knowing what they are doing, tor2web mode must be explicitly enabled at compile time, and a copy of Tor compiled to run in tor2web mode cannot be used as a normal Tor client. Implements feature 2553. - Add experimental support for running on Windows with IOCP and no kernel-space socket buffers. This feature is controlled by a new "UserspaceIOCPBuffers" config option (off by default), which has no effect unless Tor has been built with support for bufferevents, is running on Windows, and has enabled IOCP. This may, in the long run, help solve or mitigate bug 98. - Use a more secure consensus parameter voting algorithm. Now at least three directory authorities or a majority of them must vote on a given parameter before it will be included in the consensus. Implements proposal 178. o Major bugfixes: - Hidden services now ignore the timestamps on INTRODUCE2 cells. They used to check that the timestamp was within 30 minutes of their system clock, so they could cap the size of their replay-detection cache, but that approach unnecessarily refused service to clients with wrong clocks. Bugfix on 0.2.1.6-alpha, when the v3 intro-point protocol (the first one which sent a timestamp field in the INTRODUCE2 cell) was introduced; fixes bug 3460. - Only use the EVP interface when AES acceleration is enabled, to avoid a 5-7% performance regression. Resolves issue 4525; bugfix on 0.2.3.8-alpha. o Privacy/anonymity features (bridge detection): - Make bridge SSL certificates a bit more stealthy by using random serial numbers, in the same fashion as OpenSSL when generating self-signed certificates. Implements ticket 4584. - Introduce a new config option "DynamicDHGroups", enabled by default, which provides each bridge with a unique prime DH modulus to be used during SSL handshakes. This option attempts to help against censors who might use the Apache DH modulus as a static identifier for bridges. Addresses ticket 4548. o Minor features (new/different config options): - New configuration option "DisableDebuggerAttachment" (on by default) to prevent basic debugging attachment attempts by other processes. Supports Mac OS X and Gnu/Linux. Resolves ticket 3313. - Allow MapAddress directives to specify matches against super-domains, as in "MapAddress *.torproject.org *.torproject.org.torserver.exit". Implements issue 933. - Slightly change behavior of "list" options (that is, config options that can appear more than once) when they appear both in torrc and on the command line. Previously, the command-line options would be appended to the ones from torrc. Now, the command-line options override the torrc options entirely. This new behavior allows the user to override list options (like exit policies and ports to listen on) from the command line, rather than simply appending to the list. - You can get the old (appending) command-line behavior for "list" options by prefixing the option name with a "+". - You can remove all the values for a "list" option from the command line without adding any new ones by prefixing the option name with a "/". - Add experimental support for a "defaults" torrc file to be parsed before the regular torrc. Torrc options override the defaults file's options in the same way that the command line overrides the torrc. The SAVECONF controller command saves only those options which differ between the current configuration and the defaults file. HUP reloads both files. (Note: This is an experimental feature; its behavior will probably be refined in future 0.2.3.x-alpha versions to better meet packagers' needs.) Implements task 4552. o Minor features: - Try to make the introductory warning message that Tor prints on startup more useful for actually finding help and information. Resolves ticket 2474. - Running "make version" now displays the version of Tor that we're about to build. Idea from katmagic; resolves issue 4400. - Expire old or over-used hidden service introduction points. Required by fix for bug 3460. - Move the replay-detection cache for the RSA-encrypted parts of INTRODUCE2 cells to the introduction point data structures. Previously, we would use one replay-detection cache per hidden service. Required by fix for bug 3460. - Reduce the lifetime of elements of hidden services' Diffie-Hellman public key replay-detection cache from 60 minutes to 5 minutes. This replay-detection cache is now used only to detect multiple INTRODUCE2 cells specifying the same rendezvous point, so we can avoid launching multiple simultaneous attempts to connect to it. o Minor bugfixes (on Tor 0.2.2.x and earlier): - Resolve an integer overflow bug in smartlist_ensure_capacity(). Fixes bug 4230; bugfix on Tor 0.1.0.1-rc. Based on a patch by Mansour Moufid. - Fix a minor formatting issue in one of tor-gencert's error messages. Fixes bug 4574. - Prevent a false positive from the check-spaces script, by disabling the "whitespace between function name and (" check for functions named 'op()'. - Fix a log message suggesting that people contact a non-existent email address. Fixes bug 3448. - Fix null-pointer access that could occur if TLS allocation failed. Fixes bug 4531; bugfix on 0.2.0.20-rc. Found by "troll_un". - Report a real bootstrap problem to the controller on router identity mismatch. Previously we just said "foo", which probably made a lot of sense at the time. Fixes bug 4169; bugfix on 0.2.1.1-alpha. - If we had ever tried to call tor_addr_to_str() on an address of unknown type, we would have done a strdup() on an uninitialized buffer. Now we won't. Fixes bug 4529; bugfix on 0.2.1.3-alpha. Reported by "troll_un". - Correctly detect and handle transient lookup failures from tor_addr_lookup(). Fixes bug 4530; bugfix on 0.2.1.5-alpha. Reported by "troll_un". - Use tor_socket_t type for listener argument to accept(). Fixes bug 4535; bugfix on 0.2.2.28-beta. Found by "troll_un". - Initialize conn->addr to a valid state in spawn_cpuworker(). Fixes bug 4532; found by "troll_un". o Minor bugfixes (on Tor 0.2.3.x): - Fix a compile warning in tor_inet_pton(). Bugfix on 0.2.3.8-alpha; fixes bug 4554. - Don't send two ESTABLISH_RENDEZVOUS cells when opening a new circuit for use as a hidden service client's rendezvous point. Fixes bugs 4641 and 4171; bugfix on 0.2.3.3-alpha. Diagnosed with help from wanoskarnet. - Restore behavior of overriding SocksPort, ORPort, and similar options from the command line. Bugfix on 0.2.3.3-alpha. o Build fixes: - Properly handle the case where the build-tree is not the same as the source tree when generating src/common/common_sha1.i, src/or/micro-revision.i, and src/or/or_sha1.i. Fixes bug 3953; bugfix on 0.2.0.1-alpha. o Code simplifications, cleanups, and refactorings: - Remove the pure attribute from all functions that used it previously. In many cases we assigned it incorrectly, because the functions might assert or call impure functions, and we don't have evidence that keeping the pure attribute is worthwhile. Implements changes suggested in ticket 4421. - Remove some dead code spotted by coverity. Fixes cid 432. Bugfix on 0.2.3.1-alpha, closes bug 4637. Changes in version 0.2.3.8-alpha - 2011-11-22 Tor 0.2.3.8-alpha fixes some crash and assert bugs, including a socketpair-related bug that has been bothering Windows users. It adds support to serve microdescriptors to controllers, so Vidalia's network map can resume listing relays (once Vidalia implements its side), and adds better support for hardware AES acceleration. Finally, it starts the process of adjusting the bandwidth cutoff for getting the "Fast" flag from 20KB to (currently) 32KB -- preliminary results show that tiny relays harm performance more than they help network capacity. o Major bugfixes: - Initialize Libevent with the EVENT_BASE_FLAG_NOLOCK flag enabled, so that it doesn't attempt to allocate a socketpair. This could cause some problems on Windows systems with overzealous firewalls. Fix for bug 4457; workaround for Libevent versions 2.0.1-alpha through 2.0.15-stable. - Correctly sanity-check that we don't underflow on a memory allocation (and then assert) for hidden service introduction point decryption. Bug discovered by Dan Rosenberg. Fixes bug 4410; bugfix on 0.2.1.5-alpha. - Remove the artificially low cutoff of 20KB to guarantee the Fast flag. In the past few years the average relay speed has picked up, and while the "top 7/8 of the network get the Fast flag" and "all relays with 20KB or more of capacity get the Fast flag" rules used to have the same result, now the top 7/8 of the network has a capacity more like 32KB. Bugfix on 0.2.1.14-rc. Fixes bug 4489. - Fix a rare assertion failure when checking whether a v0 hidden service descriptor has any usable introduction points left, and we don't have enough information to build a circuit to the first intro point named in the descriptor. The HS client code in 0.2.3.x no longer uses v0 HS descriptors, but this assertion can trigger on (and crash) v0 HS authorities. Fixes bug 4411. Bugfix on 0.2.3.1-alpha; diagnosed by frosty_un. - Make bridge authorities not crash when they are asked for their own descriptor. Bugfix on 0.2.3.7-alpha, reported by Lucky Green. - When running as a client, do not print a misleading (and plain wrong) log message that we're collecting "directory request" statistics: clients don't collect statistics. Also don't create a useless (because empty) stats file in the stats/ directory. Fixes bug 4353; bugfix on 0.2.2.34 and 0.2.3.7-alpha. o Major features: - Allow Tor controllers like Vidalia to obtain the microdescriptor for a relay by identity digest or nickname. Previously, microdescriptors were only available by their own digests, so a controller would have to ask for and parse the whole microdescriptor consensus in order to look up a single relay's microdesc. Fixes bug 3832; bugfix on 0.2.3.1-alpha. - Use OpenSSL's EVP interface for AES encryption, so that all AES operations can use hardware acceleration (if present). Resolves ticket 4442. o Minor bugfixes (on 0.2.2.x and earlier): - Detect failure to initialize Libevent. This fix provides better detection for future instances of bug 4457. - Avoid frequent calls to the fairly expensive cull_wedged_cpuworkers function. This was eating up hideously large amounts of time on some busy servers. Fixes bug 4518; bugfix on 0.0.9.8. - Don't warn about unused log_mutex in log.c when building with --disable-threads using a recent GCC. Fixes bug 4437; bugfix on 0.1.0.6-rc which introduced --disable-threads. - Allow manual 'authenticate' commands to the controller interface from netcat (nc) as well as telnet. We were rejecting them because they didn't come with the expected whitespace at the end of the command. Bugfix on 0.1.1.1-alpha; fixes bug 2893. - Fix some (not actually triggerable) buffer size checks in usage of tor_inet_ntop. Fixes bug 4434; bugfix on Tor 0.2.0.1-alpha. Patch by Anders Sundman. - Fix parsing of some corner-cases with tor_inet_pton(). Fixes bug 4515; bugfix on 0.2.0.1-alpha; fix by Anders Sundman. - When configuring, starting, or stopping an NT service, stop immediately after the service configuration attempt has succeeded or failed. Fixes bug 3963; bugfix on 0.2.0.7-alpha. - When sending a NETINFO cell, include the original address received for the other side, not its canonical address. Found by "troll_un"; fixes bug 4349; bugfix on 0.2.0.10-alpha. - Rename the bench_{aes,dmap} functions to test_*, so that tinytest can pick them up when the tests aren't disabled. Bugfix on 0.2.2.4-alpha which introduced tinytest. - Fix a memory leak when we check whether a hidden service descriptor has any usable introduction points left. Fixes bug 4424. Bugfix on 0.2.2.25-alpha. - Fix a memory leak in launch_direct_bridge_descriptor_fetch() that occurred when a client tried to fetch a descriptor for a bridge in ExcludeNodes. Fixes bug 4383; bugfix on 0.2.2.25-alpha. o Minor bugfixes (on 0.2.3.x): - Make util unit tests build correctly with MSVC. Bugfix on 0.2.3.3-alpha. Patch by Gisle Vanem. - Successfully detect AUTH_CHALLENGE cells with no recognized authentication type listed. Fixes bug 4367; bugfix on 0.2.3.6-alpha. Found by frosty_un. - If a relay receives an AUTH_CHALLENGE cell it can't answer, it should still send a NETINFO cell to allow the connection to become open. Fixes bug 4368; fix on 0.2.3.6-alpha; bug found by "frosty". - Log less loudly when we get an invalid authentication certificate from a source other than a directory authority: it's not unusual to see invalid certs because of clock skew. Fixes bug 4370; bugfix on 0.2.3.6-alpha. - Tolerate servers with more clock skew in their authentication certificates than previously. Fixes bug 4371; bugfix on 0.2.3.6-alpha. - Fix a couple of compile warnings on Windows. Fixes bug 4469; bugfix on 0.2.3.4-alpha and 0.2.3.6-alpha. o Minor features: - Add two new config options for directory authorities: AuthDirFastGuarantee sets a bandwidth threshold for guaranteeing the Fast flag, and AuthDirGuardBWGuarantee sets a bandwidth threshold that is always sufficient to satisfy the bandwidth requirement for the Guard flag. Now it will be easier for researchers to simulate Tor networks with different values. Resolves ticket 4484. - When Tor ignores a hidden service specified in its configuration, include the hidden service's directory in the warning message. Previously, we would only tell the user that some hidden service was ignored. Bugfix on 0.0.6; fixes bug 4426. - When we fail to initialize Libevent, retry with IOCP disabled so we don't need to turn on multi-threading support in Libevent, which in turn requires a working socketpair(). This is a workaround for bug 4457, which affects Libevent versions from 2.0.1-alpha through 2.0.15-stable. - Detect when we try to build on a platform that doesn't define AF_UNSPEC to 0. We don't work there, so refuse to compile. - Update to the November 1 2011 Maxmind GeoLite Country database. o Packaging changes: - Make it easier to automate expert package builds on Windows, by removing an absolute path from makensis.exe command. o Code simplifications and refactoring: - Remove some redundant #include directives throughout the code. Patch from Andrea Gelmini. - Unconditionally use OpenSSL's AES implementation instead of our old built-in one. OpenSSL's AES has been better for a while, and relatively few servers should still be on any version of OpenSSL that doesn't have good optimized assembly AES. - Use the name "CERTS" consistently to refer to the new cell type; we were calling it CERT in some places and CERTS in others. o Testing: - Numerous new unit tests for functions in util.c and address.c by Anders Sundman. - The long-disabled benchmark tests are now split into their own ./src/test/bench binary. - The benchmark tests can now use more accurate timers than gettimeofday() when such timers are available. Changes in version 0.2.3.7-alpha - 2011-10-30 Tor 0.2.3.7-alpha fixes a crash bug in 0.2.3.6-alpha introduced by the new v3 handshake. It also resolves yet another bridge address enumeration issue. o Major bugfixes: - If we mark an OR connection for close based on a cell we process, don't process any further cells on it. We already avoid further reads on marked-for-close connections, but now we also discard the cells we'd already read. Fixes bug 4299; bugfix on 0.2.0.10-alpha, which was the first version where we might mark a connection for close based on processing a cell on it. - Fix a double-free bug that would occur when we received an invalid certificate in a CERT cell in the new v3 handshake. Fixes bug 4343; bugfix on 0.2.3.6-alpha. - Bridges no longer include their address in NETINFO cells on outgoing OR connections, to allow them to blend in better with clients. Removes another avenue for enumerating bridges. Reported by "troll_un". Fixes bug 4348; bugfix on 0.2.0.10-alpha, when NETINFO cells were introduced. o Trivial fixes: - Fixed a typo in a hibernation-related log message. Fixes bug 4331; bugfix on 0.2.2.23-alpha; found by "tmpname0901". Changes in version 0.2.3.6-alpha - 2011-10-26 Tor 0.2.3.6-alpha includes the fix from 0.2.2.34 for a critical anonymity vulnerability where an attacker can deanonymize Tor users. Everybody should upgrade. This release also features support for a new v3 connection handshake protocol, and fixes to make hidden service connections more robust. o Major features: - Implement a new handshake protocol (v3) for authenticating Tors to each other over TLS. It should be more resistant to fingerprinting than previous protocols, and should require less TLS hacking for future Tor implementations. Implements proposal 176. - Allow variable-length padding cells to disguise the length of Tor's TLS records. Implements part of proposal 184. o Privacy/anonymity fixes (clients): - Clients and bridges no longer send TLS certificate chains on outgoing OR connections. Previously, each client or bridge would use the same cert chain for all outgoing OR connections until its IP address changes, which allowed any relay that the client or bridge contacted to determine which entry guards it is using. Fixes CVE-2011-2768. Bugfix on 0.0.9pre5; found by "frosty_un". - If a relay receives a CREATE_FAST cell on a TLS connection, it no longer considers that connection as suitable for satisfying a circuit EXTEND request. Now relays can protect clients from the CVE-2011-2768 issue even if the clients haven't upgraded yet. - Directory authorities no longer assign the Guard flag to relays that haven't upgraded to the above "refuse EXTEND requests to client connections" fix. Now directory authorities can protect clients from the CVE-2011-2768 issue even if neither the clients nor the relays have upgraded yet. There's a new "GiveGuardFlagTo_CVE_2011_2768_VulnerableRelays" config option to let us transition smoothly, else tomorrow there would be no guard relays. o Major bugfixes (hidden services): - Improve hidden service robustness: when an attempt to connect to a hidden service ends, be willing to refetch its hidden service descriptors from each of the HSDir relays responsible for them immediately. Previously, we would not consider refetching the service's descriptors from each HSDir for 15 minutes after the last fetch, which was inconvenient if the hidden service was not running during the first attempt. Bugfix on 0.2.0.18-alpha; fixes bug 3335. - When one of a hidden service's introduction points appears to be unreachable, stop trying it. Previously, we would keep trying to build circuits to the introduction point until we lost the descriptor, usually because the user gave up and restarted Tor. Partly fixes bug 3825. - Don't launch a useless circuit after failing to use one of a hidden service's introduction points. Previously, we would launch a new introduction circuit, but not set the hidden service which that circuit was intended to connect to, so it would never actually be used. A different piece of code would then create a new introduction circuit correctly. Bug reported by katmagic and found by Sebastian Hahn. Bugfix on 0.2.1.13-alpha; fixes bug 4212. o Major bugfixes (other): - Bridges now refuse CREATE or CREATE_FAST cells on OR connections that they initiated. Relays could distinguish incoming bridge connections from client connections, creating another avenue for enumerating bridges. Fixes CVE-2011-2769. Bugfix on 0.2.0.3-alpha. Found by "frosty_un". - Don't update the AccountingSoftLimitHitAt state file entry whenever tor gets started. This prevents a wrong average bandwidth estimate, which would cause relays to always start a new accounting interval at the earliest possible moment. Fixes bug 2003; bugfix on 0.2.2.7-alpha. Reported by BryonEldridge, who also helped immensely in tracking this bug down. - Fix a crash bug when changing node restrictions while a DNS lookup is in-progress. Fixes bug 4259; bugfix on 0.2.2.25-alpha. Bugfix by "Tey'". o Minor bugfixes (on 0.2.2.x and earlier): - When a hidden service turns an extra service-side introduction circuit into a general-purpose circuit, free the rend_data and intro_key fields first, so we won't leak memory if the circuit is cannibalized for use as another service-side introduction circuit. Bugfix on 0.2.1.7-alpha; fixes bug 4251. - Rephrase the log message emitted if the TestSocks check is successful. Patch from Fabian Keil; fixes bug 4094. - Bridges now skip DNS self-tests, to act a little more stealthily. Fixes bug 4201; bugfix on 0.2.0.3-alpha, which first introduced bridges. Patch by "warms0x". - Remove a confusing dollar sign from the example fingerprint in the man page, and also make the example fingerprint a valid one. Fixes bug 4309; bugfix on 0.2.1.3-alpha. - Fix internal bug-checking logic that was supposed to catch failures in digest generation so that it will fail more robustly if we ask for a nonexistent algorithm. Found by Coverity Scan. Bugfix on 0.2.2.1-alpha; fixes Coverity CID 479. - Report any failure in init_keys() calls launched because our IP address has changed. Spotted by Coverity Scan. Bugfix on 0.1.1.4-alpha; fixes CID 484. o Minor bugfixes (on 0.2.3.x): - Fix a bug in configure.in that kept it from building a configure script with autoconf versions earlier than 2.61. Fixes bug 2430; bugfix on 0.2.3.1-alpha. - Don't warn users that they are exposing a client port to the Internet if they have specified an RFC1918 address. Previously, we would warn if the user had specified any non-loopback address. Bugfix on 0.2.3.3-alpha. Fixes bug 4018; reported by Tas. - Fix memory leaks in the failing cases of the new SocksPort and ControlPort code. Found by Coverity Scan. Bugfix on 0.2.3.3-alpha; fixes coverity CIDs 485, 486, and 487. o Minor features: - When a hidden service's introduction point times out, consider trying it again during the next attempt to connect to the HS. Previously, we would not try it again unless a newly fetched descriptor contained it. Required by fixes for bugs 1297 and 3825. - The next version of Windows will be called Windows 8, and it has a major version of 6, minor version of 2. Correctly identify that version instead of calling it "Very recent version". Resolves ticket 4153; reported by funkstar. - The Bridge Authority now writes statistics on how many bridge descriptors it gave out in total, and how many unique descriptors it gave out. It also lists how often the most and least commonly fetched descriptors were given out, as well as the median and 25th/75th percentile. Implements tickets 4200 and 4294. - Update to the October 4 2011 Maxmind GeoLite Country database. o Code simplifications and refactoring: - Remove some old code to remember statistics about which descriptors we've served as a directory mirror. The feature wasn't used and is outdated now that microdescriptors are around. - Rename Tor functions that turn strings into addresses, so that "parse" indicates that no hostname resolution occurs, and "lookup" indicates that hostname resolution may occur. This should help prevent mistakes in the future. Fixes bug 3512. Changes in version 0.2.2.34 - 2011-10-26 Tor 0.2.2.34 fixes a critical anonymity vulnerability where an attacker can deanonymize Tor users. Everybody should upgrade. The attack relies on four components: 1) Clients reuse their TLS cert when talking to different relays, so relays can recognize a user by the identity key in her cert. 2) An attacker who knows the client's identity key can probe each guard relay to see if that identity key is connected to that guard relay right now. 3) A variety of active attacks in the literature (starting from "Low-Cost Traffic Analysis of Tor" by Murdoch and Danezis in 2005) allow a malicious website to discover the guard relays that a Tor user visiting the website is using. 4) Clients typically pick three guards at random, so the set of guards for a given user could well be a unique fingerprint for her. This release fixes components #1 and #2, which is enough to block the attack; the other two remain as open research problems. Special thanks to "frosty_un" for reporting the issue to us! Clients should upgrade so they are no longer recognizable by the TLS certs they present. Relays should upgrade so they no longer allow a remote attacker to probe them to test whether unpatched clients are currently connected to them. This release also fixes several vulnerabilities that allow an attacker to enumerate bridge relays. Some bridge enumeration attacks still remain; see for example proposal 188. o Privacy/anonymity fixes (clients): - Clients and bridges no longer send TLS certificate chains on outgoing OR connections. Previously, each client or bridge would use the same cert chain for all outgoing OR connections until its IP address changes, which allowed any relay that the client or bridge contacted to determine which entry guards it is using. Fixes CVE-2011-2768. Bugfix on 0.0.9pre5; found by "frosty_un". - If a relay receives a CREATE_FAST cell on a TLS connection, it no longer considers that connection as suitable for satisfying a circuit EXTEND request. Now relays can protect clients from the CVE-2011-2768 issue even if the clients haven't upgraded yet. - Directory authorities no longer assign the Guard flag to relays that haven't upgraded to the above "refuse EXTEND requests to client connections" fix. Now directory authorities can protect clients from the CVE-2011-2768 issue even if neither the clients nor the relays have upgraded yet. There's a new "GiveGuardFlagTo_CVE_2011_2768_VulnerableRelays" config option to let us transition smoothly, else tomorrow there would be no guard relays. o Privacy/anonymity fixes (bridge enumeration): - Bridge relays now do their directory fetches inside Tor TLS connections, like all the other clients do, rather than connecting directly to the DirPort like public relays do. Removes another avenue for enumerating bridges. Fixes bug 4115; bugfix on 0.2.0.35. - Bridges relays now build circuits for themselves in a more similar way to how clients build them. Removes another avenue for enumerating bridges. Fixes bug 4124; bugfix on 0.2.0.3-alpha, when bridges were introduced. - Bridges now refuse CREATE or CREATE_FAST cells on OR connections that they initiated. Relays could distinguish incoming bridge connections from client connections, creating another avenue for enumerating bridges. Fixes CVE-2011-2769. Bugfix on 0.2.0.3-alpha. Found by "frosty_un". o Major bugfixes: - Fix a crash bug when changing node restrictions while a DNS lookup is in-progress. Fixes bug 4259; bugfix on 0.2.2.25-alpha. Bugfix by "Tey'". - Don't launch a useless circuit after failing to use one of a hidden service's introduction points. Previously, we would launch a new introduction circuit, but not set the hidden service which that circuit was intended to connect to, so it would never actually be used. A different piece of code would then create a new introduction circuit correctly. Bug reported by katmagic and found by Sebastian Hahn. Bugfix on 0.2.1.13-alpha; fixes bug 4212. o Minor bugfixes: - Change an integer overflow check in the OpenBSD_Malloc code so that GCC is less likely to eliminate it as impossible. Patch from Mansour Moufid. Fixes bug 4059. - When a hidden service turns an extra service-side introduction circuit into a general-purpose circuit, free the rend_data and intro_key fields first, so we won't leak memory if the circuit is cannibalized for use as another service-side introduction circuit. Bugfix on 0.2.1.7-alpha; fixes bug 4251. - Bridges now skip DNS self-tests, to act a little more stealthily. Fixes bug 4201; bugfix on 0.2.0.3-alpha, which first introduced bridges. Patch by "warms0x". - Fix internal bug-checking logic that was supposed to catch failures in digest generation so that it will fail more robustly if we ask for a nonexistent algorithm. Found by Coverity Scan. Bugfix on 0.2.2.1-alpha; fixes Coverity CID 479. - Report any failure in init_keys() calls launched because our IP address has changed. Spotted by Coverity Scan. Bugfix on 0.1.1.4-alpha; fixes CID 484. o Minor bugfixes (log messages and documentation): - Remove a confusing dollar sign from the example fingerprint in the man page, and also make the example fingerprint a valid one. Fixes bug 4309; bugfix on 0.2.1.3-alpha. - The next version of Windows will be called Windows 8, and it has a major version of 6, minor version of 2. Correctly identify that version instead of calling it "Very recent version". Resolves ticket 4153; reported by funkstar. - Downgrade log messages about circuit timeout calibration from "notice" to "info": they don't require or suggest any human intervention. Patch from Tom Lowenthal. Fixes bug 4063; bugfix on 0.2.2.14-alpha. o Minor features: - Turn on directory request statistics by default and include them in extra-info descriptors. Don't break if we have no GeoIP database. Backported from 0.2.3.1-alpha; implements ticket 3951. - Update to the October 4 2011 Maxmind GeoLite Country database. Changes in version 0.2.1.31 - 2011-10-26 Tor 0.2.1.31 backports important security and privacy fixes for oldstable. This release is intended only for package maintainers and others who cannot use the 0.2.2 stable series. All others should be using Tor 0.2.2.x or newer. o Security fixes (also included in 0.2.2.x): - Replace all potentially sensitive memory comparison operations with versions whose runtime does not depend on the data being compared. This will help resist a class of attacks where an adversary can use variations in timing information to learn sensitive data. Fix for one case of bug 3122. (Safe memcmp implementation by Robert Ransom based partially on code by DJB.) - Fix an assert in parsing router descriptors containing IPv6 addresses. This one took down the directory authorities when somebody tried some experimental code. Bugfix on 0.2.1.3-alpha. o Privacy/anonymity fixes (also included in 0.2.2.x): - Clients and bridges no longer send TLS certificate chains on outgoing OR connections. Previously, each client or bridge would use the same cert chain for all outgoing OR connections until its IP address changes, which allowed any relay that the client or bridge contacted to determine which entry guards it is using. Fixes CVE-2011-2768. Bugfix on 0.0.9pre5; found by "frosty_un". - If a relay receives a CREATE_FAST cell on a TLS connection, it no longer considers that connection as suitable for satisfying a circuit EXTEND request. Now relays can protect clients from the CVE-2011-2768 issue even if the clients haven't upgraded yet. - Bridges now refuse CREATE or CREATE_FAST cells on OR connections that they initiated. Relays could distinguish incoming bridge connections from client connections, creating another avenue for enumerating bridges. Fixes CVE-2011-2769. Bugfix on 0.2.0.3-alpha. Found by "frosty_un". - When receiving a hidden service descriptor, check that it is for the hidden service we wanted. Previously, Tor would store any hidden service descriptors that a directory gave it, whether it wanted them or not. This wouldn't have let an attacker impersonate a hidden service, but it did let directories pre-seed a client with descriptors that it didn't want. Bugfix on 0.0.6. - Avoid linkability based on cached hidden service descriptors: forget all hidden service descriptors cached as a client when processing a SIGNAL NEWNYM command. Fixes bug 3000; bugfix on 0.0.6. - Make the bridge directory authority refuse to answer directory requests for "all" descriptors. It used to include bridge descriptors in its answer, which was a major information leak. Found by "piebeer". Bugfix on 0.2.0.3-alpha. - Don't attach new streams to old rendezvous circuits after SIGNAL NEWNYM. Previously, we would keep using an existing rendezvous circuit if it remained open (i.e. if it were kept open by a long-lived stream, or if a new stream were attached to it before Tor could notice that it was old and no longer in use). Bugfix on 0.1.1.15-rc; fixes bug 3375. o Minor bugfixes (also included in 0.2.2.x): - When we restart our relay, we might get a successful connection from the outside before we've started our reachability tests, triggering a warning: "ORPort found reachable, but I have no routerinfo yet. Failing to inform controller of success." This bug was harmless unless Tor is running under a controller like Vidalia, in which case the controller would never get a REACHABILITY_SUCCEEDED status event. Bugfix on 0.1.2.6-alpha; fixes bug 1172. - Build correctly on OSX with zlib 1.2.4 and higher with all warnings enabled. Fixes bug 1526. - Remove undocumented option "-F" from tor-resolve: it hasn't done anything since 0.2.1.16-rc. - Avoid signed/unsigned comparisons by making SIZE_T_CEILING unsigned. None of the cases where we did this before were wrong, but by making this change we avoid warnings. Fixes bug 2475; bugfix on 0.2.1.28. - Fix a rare crash bug that could occur when a client was configured with a large number of bridges. Fixes bug 2629; bugfix on 0.2.1.2-alpha. Bugfix by trac user "shitlei". - Correct the warning displayed when a rendezvous descriptor exceeds the maximum size. Fixes bug 2750; bugfix on 0.2.1.5-alpha. Found by John Brooks. - Fix an uncommon assertion failure when running with DNSPort under heavy load. Fixes bug 2933; bugfix on 0.2.0.1-alpha. - When warning about missing zlib development packages during compile, give the correct package names. Bugfix on 0.2.0.1-alpha. - Require that introduction point keys and onion keys have public exponent 65537. Bugfix on 0.2.0.10-alpha. - Do not crash when our configuration file becomes unreadable, for example due to a permissions change, between when we start up and when a controller calls SAVECONF. Fixes bug 3135; bugfix on 0.0.9pre6. - Fix warnings from GCC 4.6's "-Wunused-but-set-variable" option. Fixes bug 3208. - Always NUL-terminate the sun_path field of a sockaddr_un before passing it to the kernel. (Not a security issue: kernels are smart enough to reject bad sockaddr_uns.) Found by Coverity; CID #428. Bugfix on Tor 0.2.0.3-alpha. - Don't stack-allocate the list of supplementary GIDs when we're about to log them. Stack-allocating NGROUPS_MAX gid_t elements could take up to 256K, which is way too much stack. Found by Coverity; CID #450. Bugfix on 0.2.1.7-alpha. o Minor bugfixes (only in 0.2.1.x): - Resume using micro-version numbers in 0.2.1.x: our Debian packages rely on them. Bugfix on 0.2.1.30. - Use git revisions instead of svn revisions when generating our micro-version numbers. Bugfix on 0.2.1.15-rc; fixes bug 2402. o Minor features (also included in 0.2.2.x): - Adjust the expiration time on our SSL session certificates to better match SSL certs seen in the wild. Resolves ticket 4014. - Allow nameservers with IPv6 address. Resolves bug 2574. - Update to the October 4 2011 Maxmind GeoLite Country database. Changes in version 0.2.3.5-alpha - 2011-09-28 Tor 0.2.3.5-alpha fixes two bugs that make it possible to enumerate bridge relays; fixes an assertion error that many users started hitting today; and adds the ability to refill token buckets more often than once per second, allowing significant performance improvements. o Security fixes: - Bridge relays now do their directory fetches inside Tor TLS connections, like all the other clients do, rather than connecting directly to the DirPort like public relays do. Removes another avenue for enumerating bridges. Fixes bug 4115; bugfix on 0.2.0.35. - Bridges relays now build circuits for themselves in a more similar way to how clients build them. Removes another avenue for enumerating bridges. Fixes bug 4124; bugfix on 0.2.0.3-alpha, when bridges were introduced. o Major bugfixes: - Fix an "Assertion md->held_by_node == 1 failed" error that could occur when the same microdescriptor was referenced by two node_t objects at once. Fix for bug 4118; bugfix on Tor 0.2.3.1-alpha. o Major features (networking): - Add a new TokenBucketRefillInterval option to refill token buckets more frequently than once per second. This should improve network performance, alleviate queueing problems, and make traffic less bursty. Implements proposal 183; closes ticket 3630. Design by Florian Tschorsch and Björn Scheuermann; implementation by Florian Tschorsch. o Minor bugfixes: - Change an integer overflow check in the OpenBSD_Malloc code so that GCC is less likely to eliminate it as impossible. Patch from Mansour Moufid. Fixes bug 4059. o Minor bugfixes (usability): - Downgrade log messages about circuit timeout calibration from "notice" to "info": they don't require or suggest any human intervention. Patch from Tom Lowenthal. Fixes bug 4063; bugfix on 0.2.2.14-alpha. o Minor features (diagnostics): - When the system call to create a listener socket fails, log the error message explaining why. This may help diagnose bug 4027. Changes in version 0.2.3.4-alpha - 2011-09-13 Tor 0.2.3.4-alpha includes the fixes from 0.2.2.33, including a slight tweak to Tor's TLS handshake that makes relays and bridges that run this new version reachable from Iran again. It also fixes a few new bugs in 0.2.3.x, and teaches relays to recognize when they're not listed in the network consensus and republish. o Major bugfixes (also part of 0.2.2.33): - Avoid an assertion failure when reloading a configuration with TrackExitHosts changes. Found and fixed by 'laruldan'. Fixes bug 3923; bugfix on 0.2.2.25-alpha. o Minor features (security, also part of 0.2.2.33): - Check for replays of the public-key encrypted portion of an INTRODUCE1 cell, in addition to the current check for replays of the g^x value. This prevents a possible class of active attacks by an attacker who controls both an introduction point and a rendezvous point, and who uses the malleability of AES-CTR to alter the encrypted g^x portion of the INTRODUCE1 cell. We think that these attacks is infeasible (requiring the attacker to send on the order of zettabytes of altered cells in a short interval), but we'd rather block them off in case there are any classes of this attack that we missed. Reported by Willem Pinckaers. o Minor features (also part of 0.2.2.33): - Adjust the expiration time on our SSL session certificates to better match SSL certs seen in the wild. Resolves ticket 4014. - Change the default required uptime for a relay to be accepted as a HSDir (hidden service directory) from 24 hours to 25 hours. Improves on 0.2.0.10-alpha; resolves ticket 2649. - Add a VoteOnHidServDirectoriesV2 config option to allow directory authorities to abstain from voting on assignment of the HSDir consensus flag. Related to bug 2649. - Update to the September 6 2011 Maxmind GeoLite Country database. o Minor bugfixes (also part of 0.2.2.33): - Demote the 'replay detected' log message emitted when a hidden service receives the same Diffie-Hellman public key in two different INTRODUCE2 cells to info level. A normal Tor client can cause that log message during its normal operation. Bugfix on 0.2.1.6-alpha; fixes part of bug 2442. - Demote the 'INTRODUCE2 cell is too {old,new}' log message to info level. There is nothing that a hidden service's operator can do to fix its clients' clocks. Bugfix on 0.2.1.6-alpha; fixes part of bug 2442. - Clarify a log message specifying the characters permitted in HiddenServiceAuthorizeClient client names. Previously, the log message said that "[A-Za-z0-9+-_]" were permitted; that could have given the impression that every ASCII character between "+" and "_" was permitted. Now we say "[A-Za-z0-9+_-]". Bugfix on 0.2.1.5-alpha. o Build fixes (also part of 0.2.2.33): - Clean up some code issues that prevented Tor from building on older BSDs. Fixes bug 3894; reported by "grarpamp". - Search for a platform-specific version of "ar" when cross-compiling. Should fix builds on iOS. Resolves bug 3909, found by Marco Bonetti. o Major bugfixes: - Fix a bug where the SocksPort option (for example) would get ignored and replaced by the default if a SocksListenAddress option was set. Bugfix on 0.2.3.3-alpha; fixes bug 3936. Fix by Fabian Keil. 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. o Minor features: - Relays now include a reason for regenerating their descriptors in an HTTP header when uploading to the authorities. This will make it easier to debug descriptor-upload issues in the future. - When starting as root and then changing our UID via the User control option, and we have a ControlSocket configured, make sure that the ControlSocket is owned by the same account that Tor will run under. Implements ticket 3421; fix by Jérémy Bobbio. o Minor bugfixes: - Abort if tor_vasprintf fails in connection_printf_to_buf (a utility function used in the control-port code). This shouldn't ever happen unless Tor is completely out of memory, but if it did happen and Tor somehow recovered from it, Tor could have sent a log message to a control port in the middle of a reply to a controller command. Fixes part of bug 3428; bugfix on 0.1.2.3-alpha. - Make 'FetchUselessDescriptors' cause all descriptor types and all consensus types (including microdescriptors) to get fetched. Fixes bug 3851; bugfix on 0.2.3.1-alpha. o Code refactoring: - Make a new "entry connection" struct as an internal subtype of "edge connection", to simplify the code and make exit connections smaller. Changes in version 0.2.2.33 - 2011-09-13 Tor 0.2.2.33 fixes several bugs, and includes a slight tweak to Tor's TLS handshake that makes relays and bridges that run this new version reachable from Iran again. o Major bugfixes: - Avoid an assertion failure when reloading a configuration with TrackExitHosts changes. Found and fixed by 'laruldan'. Fixes bug 3923; bugfix on 0.2.2.25-alpha. o Minor features (security): - Check for replays of the public-key encrypted portion of an INTRODUCE1 cell, in addition to the current check for replays of the g^x value. This prevents a possible class of active attacks by an attacker who controls both an introduction point and a rendezvous point, and who uses the malleability of AES-CTR to alter the encrypted g^x portion of the INTRODUCE1 cell. We think that these attacks are infeasible (requiring the attacker to send on the order of zettabytes of altered cells in a short interval), but we'd rather block them off in case there are any classes of this attack that we missed. Reported by Willem Pinckaers. o Minor features: - Adjust the expiration time on our SSL session certificates to better match SSL certs seen in the wild. Resolves ticket 4014. - Change the default required uptime for a relay to be accepted as a HSDir (hidden service directory) from 24 hours to 25 hours. Improves on 0.2.0.10-alpha; resolves ticket 2649. - Add a VoteOnHidServDirectoriesV2 config option to allow directory authorities to abstain from voting on assignment of the HSDir consensus flag. Related to bug 2649. - Update to the September 6 2011 Maxmind GeoLite Country database. o Minor bugfixes (documentation and log messages): - Correct the man page to explain that HashedControlPassword and CookieAuthentication can both be set, in which case either method is sufficient to authenticate to Tor. Bugfix on 0.2.0.7-alpha, when we decided to allow these config options to both be set. Issue raised by bug 3898. - Demote the 'replay detected' log message emitted when a hidden service receives the same Diffie-Hellman public key in two different INTRODUCE2 cells to info level. A normal Tor client can cause that log message during its normal operation. Bugfix on 0.2.1.6-alpha; fixes part of bug 2442. - Demote the 'INTRODUCE2 cell is too {old,new}' log message to info level. There is nothing that a hidden service's operator can do to fix its clients' clocks. Bugfix on 0.2.1.6-alpha; fixes part of bug 2442. - Clarify a log message specifying the characters permitted in HiddenServiceAuthorizeClient client names. Previously, the log message said that "[A-Za-z0-9+-_]" were permitted; that could have given the impression that every ASCII character between "+" and "_" was permitted. Now we say "[A-Za-z0-9+_-]". Bugfix on 0.2.1.5-alpha. o Build fixes: - Provide a substitute implementation of lround() for MSVC, which apparently lacks it. Patch from Gisle Vanem. - Clean up some code issues that prevented Tor from building on older BSDs. Fixes bug 3894; reported by "grarpamp". - Search for a platform-specific version of "ar" when cross-compiling. Should fix builds on iOS. Resolves bug 3909, found by Marco Bonetti. Changes in version 0.2.3.3-alpha - 2011-09-01 Tor 0.2.3.3-alpha adds a new "stream isolation" feature to improve Tor's security, and provides client-side support for the microdescriptor and optimistic data features introduced earlier in the 0.2.3.x series. It also includes numerous critical bugfixes in the (optional) bufferevent-based networking backend. o Major features (stream isolation): - You can now configure Tor so that streams from different applications are isolated on different circuits, to prevent an attacker who sees your streams as they leave an exit node from linking your sessions to one another. To do this, choose some way to distinguish the applications: have them connect to different SocksPorts, or have one of them use SOCKS4 while the other uses SOCKS5, or have them pass different authentication strings to the SOCKS proxy. Then, use the new SocksPort syntax to configure the degree of isolation you need. This implements Proposal 171. - There's a new syntax for specifying multiple client ports (such as SOCKSPort, TransPort, DNSPort, NATDPort): you can now just declare multiple *Port entries with full addr:port syntax on each. The old *ListenAddress format is still supported, but you can't mix it with the new *Port syntax. o Major features (other): - Enable microdescriptor fetching by default for clients. This allows clients to download a much smaller amount of directory information. To disable it (and go back to the old-style consensus and descriptors), set "UseMicrodescriptors 0" in your torrc file. - Tor's firewall-helper feature, introduced in 0.2.3.1-alpha (see the "PortForwarding" config option), now supports Windows. - When using an exit relay running 0.2.3.x, clients can now "optimistically" send data before the exit relay reports that the stream has opened. This saves a round trip when starting connections where the client speaks first (such as web browsing). This behavior is controlled by a consensus parameter (currently disabled). To turn it on or off manually, use the "OptimisticData" torrc option. Implements proposal 181; code by Ian Goldberg. o Major bugfixes (bufferevents, fixes on 0.2.3.1-alpha): - When using IOCP on Windows, we need to enable Libevent windows threading support. - The IOCP backend now works even when the user has not specified the (internal, debugging-only) _UseFilteringSSLBufferevents option. Fixes part of bug 3752. - Correctly record the bytes we've read and written when using bufferevents, so that we can include them in our bandwidth history and advertised bandwidth. Fixes bug 3803. - Apply rate-limiting only at the bottom of a chain of filtering bufferevents. This prevents us from filling up internal read buffers and violating rate-limits when filtering bufferevents are enabled. Fixes part of bug 3804. - Add high-watermarks to the output buffers for filtered bufferevents. This prevents us from filling up internal write buffers and wasting CPU cycles when filtering bufferevents are enabled. Fixes part of bug 3804. - Correctly notice when data has been written from a bufferevent without flushing it completely. Fixes bug 3805. - Fix a bug where server-side tunneled bufferevent-based directory streams would get closed prematurely. Fixes bug 3814. - Fix a use-after-free error with per-connection rate-limiting buckets. Fixes bug 3888. o Major bugfixes (also part of 0.2.2.31-rc): - If we're configured to write our ControlPorts to disk, only write them after switching UID and creating the data directory. This way, we don't fail when starting up with a nonexistent DataDirectory and a ControlPortWriteToFile setting based on that directory. Fixes bug 3747; bugfix on Tor 0.2.2.26-beta. o Minor features: - Added a new CONF_CHANGED event so that controllers can be notified of any configuration changes made by other controllers, or by the user. Implements ticket 1692. - Use evbuffer_copyout() in inspect_evbuffer(). This fixes a memory leak when using bufferevents, and lets Libevent worry about how to best copy data out of a buffer. - Replace files in stats/ rather than appending to them. Now that we include statistics in extra-info descriptors, it makes no sense to keep old statistics forever. Implements ticket 2930. o Minor features (build compatibility): - Limited, experimental support for building with nmake and MSVC. - Provide a substitute implementation of lround() for MSVC, which apparently lacks it. Patch from Gisle Vanem. o Minor features (also part of 0.2.2.31-rc): - Update to the August 2 2011 Maxmind GeoLite Country database. o Minor bugfixes (on 0.2.3.x-alpha): - Fix a spurious warning when parsing SOCKS requests with bufferevents enabled. Fixes bug 3615; bugfix on 0.2.3.2-alpha. - Get rid of a harmless warning that could happen on relays running with bufferevents. The warning was caused by someone doing an http request to a relay's orport. Also don't warn for a few related non-errors. Fixes bug 3700; bugfix on 0.2.3.1-alpha. o Minor bugfixes (on 2.2.x and earlier): - Correct the man page to explain that HashedControlPassword and CookieAuthentication can both be set, in which case either method is sufficient to authenticate to Tor. Bugfix on 0.2.0.7-alpha, when we decided to allow these config options to both be set. Issue raised by bug 3898. - The "--quiet" and "--hush" options now apply not only to Tor's behavior before logs are configured, but also to Tor's behavior in the absense of configured logs. Fixes bug 3550; bugfix on 0.2.0.10-alpha. o Minor bugfixes (also part of 0.2.2.31-rc): - Write several files in text mode, on OSes that distinguish text mode from binary mode (namely, Windows). These files are: 'buffer-stats', 'dirreq-stats', and 'entry-stats' on relays that collect those statistics; 'client_keys' and 'hostname' for hidden services that use authentication; and (in the tor-gencert utility) newly generated identity and signing keys. Previously, we wouldn't specify text mode or binary mode, leading to an assertion failure. Fixes bug 3607. Bugfix on 0.2.1.1-alpha (when the DirRecordUsageByCountry option which would have triggered the assertion failure was added), although this assertion failure would have occurred in tor-gencert on Windows in 0.2.0.1-alpha. - Selectively disable deprecation warnings on OS X because Lion started deprecating the shipped copy of openssl. Fixes bug 3643. - Remove an extra pair of quotation marks around the error message in control-port STATUS_GENERAL BUG events. Bugfix on 0.1.2.6-alpha; fixes bug 3732. - When unable to format an address as a string, report its value as "???" rather than reusing the last formatted address. Bugfix on 0.2.1.5-alpha. o Code simplifications and refactoring: - Rewrite the listener-selection logic so that parsing which ports we want to listen on is now separate from binding to the ports we want. o Build changes: - Building Tor with bufferevent support now requires Libevent 2.0.13-stable or later. Previous versions of Libevent had bugs in SSL-related bufferevents and related issues that would make Tor work badly with bufferevents. Requiring 2.0.13-stable also allows Tor with bufferevents to take advantage of Libevent APIs introduced after 2.0.8-rc. Changes in version 0.2.2.32 - 2011-08-27 The Tor 0.2.2 release series is dedicated to the memory of Andreas Pfitzmann (1958-2010), a pioneer in anonymity and privacy research, a founder of the PETS community, a leader in our field, a mentor, and a friend. He left us with these words: "I had the possibility to contribute to this world that is not as it should be. I hope I could help in some areas to make the world a better place, and that I could also encourage other people to be engaged in improving the world. Please, stay engaged. This world needs you, your love, your initiative -- now I cannot be part of that anymore." Tor 0.2.2.32, the first stable release in the 0.2.2 branch, is finally ready. More than two years in the making, this release features improved client performance and hidden service reliability, better compatibility for Android, correct behavior for bridges that listen on more than one address, more extensible and flexible directory object handling, better reporting of network statistics, improved code security, and many many other features and bugfixes. Changes in version 0.2.2.31-rc - 2011-08-17 Tor 0.2.2.31-rc is the second and hopefully final release candidate for the Tor 0.2.2.x series. o Major bugfixes: - Remove an extra pair of quotation marks around the error message in control-port STATUS_GENERAL BUG events. Bugfix on 0.1.2.6-alpha; fixes bug 3732. - If we're configured to write our ControlPorts to disk, only write them after switching UID and creating the data directory. This way, we don't fail when starting up with a nonexistent DataDirectory and a ControlPortWriteToFile setting based on that directory. Fixes bug 3747; bugfix on Tor 0.2.2.26-beta. o Minor features: - Update to the August 2 2011 Maxmind GeoLite Country database. o Minor bugfixes: - Allow GETINFO fingerprint to return a fingerprint even when we have not yet built a router descriptor. Fixes bug 3577; bugfix on 0.2.0.1-alpha. - Write several files in text mode, on OSes that distinguish text mode from binary mode (namely, Windows). These files are: 'buffer-stats', 'dirreq-stats', and 'entry-stats' on relays that collect those statistics; 'client_keys' and 'hostname' for hidden services that use authentication; and (in the tor-gencert utility) newly generated identity and signing keys. Previously, we wouldn't specify text mode or binary mode, leading to an assertion failure. Fixes bug 3607. Bugfix on 0.2.1.1-alpha (when the DirRecordUsageByCountry option which would have triggered the assertion failure was added), although this assertion failure would have occurred in tor-gencert on Windows in 0.2.0.1-alpha. - Selectively disable deprecation warnings on OS X because Lion started deprecating the shipped copy of openssl. Fixes bug 3643. - When unable to format an address as a string, report its value as "???" rather than reusing the last formatted address. Bugfix on 0.2.1.5-alpha. Changes in version 0.2.3.2-alpha - 2011-07-18 Tor 0.2.3.2-alpha introduces two new experimental features: microdescriptors and pluggable transports. It also continues cleaning up a variety of recently introduced features. o Major features: - Clients can now use microdescriptors instead of regular descriptors to build circuits. Microdescriptors are authority-generated summaries of regular descriptors' contents, designed to change very rarely (see proposal 158 for details). This feature is designed to save bandwidth, especially for clients on slow internet connections. It's off by default for now, since nearly no caches support it, but it will be on-by-default for clients in a future version. You can use the UseMicrodescriptors option to turn it on. - Tor clients using bridges can now be configured to use a separate 'transport' proxy for each bridge. This approach helps to resist censorship by allowing bridges to use protocol obfuscation plugins. It implements part of proposal 180. Implements ticket 2841. - While we're trying to bootstrap, record how many TLS connections fail in each state, and report which states saw the most failures in response to any bootstrap failures. This feature may speed up diagnosis of censorship events. Implements ticket 3116. o Major bugfixes (on 0.2.3.1-alpha): - When configuring a large set of nodes in EntryNodes (as with 'EntryNodes {cc}' or 'EntryNodes 1.1.1.1/16'), choose only a random subset to be guards, and choose them in random order. Fixes bug 2798. - Tor could crash when remembering a consensus in a non-used consensus flavor without having a current consensus set. Fixes bug 3361. - Comparing an unknown address to a microdescriptor's shortened exit policy would always give a "rejected" result. Fixes bug 3599. - Using microdescriptors as a client no longer prevents Tor from uploading and downloading hidden service descriptors. Fixes bug 3601. o Minor features: - Allow nameservers with IPv6 address. Resolves bug 2574. - Accept attempts to include a password authenticator in the handshake, as supported by SOCKS5. This handles SOCKS clients that don't know how to omit a password when authenticating. Resolves bug 1666. - When configuring a large set of nodes in EntryNodes, and there are enough of them listed as Guard so that we don't need to consider the non-guard entries, prefer the ones listed with the Guard flag. - Check for and recover from inconsistency in the microdescriptor cache. This will make it harder for us to accidentally free a microdescriptor without removing it from the appropriate data structures. Fixes issue 3135; issue noted by "wanoskarnet". - Log SSL state transitions at log level DEBUG, log domain HANDSHAKE. This can be useful for debugging censorship events. Implements ticket 3264. - Add port 6523 (Gobby) to LongLivedPorts. Patch by intrigeri; implements ticket 3439. o Minor bugfixes (on 0.2.3.1-alpha): - Do not free all general-purpose regular descriptors just because microdescriptor use is enabled. Fixes bug 3113. - Correctly link libevent_openssl when --enable-static-libevent is passed to configure. Fixes bug 3118. - Bridges should not complain during their heartbeat log messages that they are unlisted in the consensus: that's more or less the point of being a bridge. Fixes bug 3183. - Report a SIGNAL event to controllers when acting on a delayed SIGNAL NEWNYM command. Previously, we would report a SIGNAL event to the controller if we acted on a SIGNAL NEWNYM command immediately, and otherwise not report a SIGNAL event for the command at all. Fixes bug 3349. - Fix a crash when handling the SIGNAL controller command or reporting ERR-level status events with bufferevents enabled. Found by Robert Ransom. Fixes bug 3367. - Always ship the tor-fw-helper manpage in our release tarballs. Fixes bug 3389. Reported by Stephen Walker. - Fix a class of double-mark-for-close bugs when bufferevents are enabled. Fixes bug 3403. - Update tor-fw-helper to support libnatpmp-20110618. Fixes bug 3434. - Add SIGNAL to the list returned by the 'GETINFO events/names' control-port command. Fixes part of bug 3465. - Prevent using negative indices during unit test runs when read_all() fails. Spotted by coverity. - Fix a rare memory leak when checking the nodelist without it being present. Found by coverity. - Only try to download a microdescriptor-flavored consensus from a directory cache that provides them. o Minor bugfixes (on 0.2.2.x and earlier): - Assert that hidden-service-related operations are not performed using single-hop circuits. Previously, Tor would assert that client-side streams are not attached to single-hop circuits, but not that other sensitive operations on the client and service side are not performed using single-hop circuits. Fixes bug 3332; bugfix on 0.0.6. - Don't publish a new relay descriptor when we reload our onion key, unless the onion key has actually changed. Fixes bug 3263 and resolves another cause of bug 1810. Bugfix on 0.1.1.11-alpha. - Allow GETINFO fingerprint to return a fingerprint even when we have not yet built a router descriptor. Fixes bug 3577; bugfix on 0.2.0.1-alpha. - Make 'tor --digests' list hashes of all Tor source files. Bugfix on 0.2.2.4-alpha; fixes bug 3427. o Code simplification and refactoring: - Use tor_sscanf() in place of scanf() in more places through the code. This makes us a little more locale-independent, and should help shut up code-analysis tools that can't tell a safe sscanf string from a dangerous one. - Use tt_assert(), not tor_assert(), for checking for test failures. This makes the unit tests more able to go on in the event that one of them fails. - Split connection_about_to_close() into separate functions for each connection type. o Build changes: - On Windows, we now define the _WIN32_WINNT macros only if they are not already defined. This lets the person building Tor decide, if they want, to require a later version of Windows. Changes in version 0.2.2.30-rc - 2011-07-07 Tor 0.2.2.30-rc is the first release candidate for the Tor 0.2.2.x series. It fixes a few smaller bugs, but generally appears stable. Please test it and let us know whether it is! o Minor bugfixes: - Send a SUCCEEDED stream event to the controller when a reverse resolve succeeded. Fixes bug 3536; bugfix on 0.0.8pre1. Issue discovered by katmagic. - Always NUL-terminate the sun_path field of a sockaddr_un before passing it to the kernel. (Not a security issue: kernels are smart enough to reject bad sockaddr_uns.) Found by Coverity; CID #428. Bugfix on Tor 0.2.0.3-alpha. - Don't stack-allocate the list of supplementary GIDs when we're about to log them. Stack-allocating NGROUPS_MAX gid_t elements could take up to 256K, which is way too much stack. Found by Coverity; CID #450. Bugfix on 0.2.1.7-alpha. - Add BUILDTIMEOUT_SET to the list returned by the 'GETINFO events/names' control-port command. Bugfix on 0.2.2.9-alpha; fixes part of bug 3465. - Fix a memory leak when receiving a descriptor for a hidden service we didn't ask for. Found by Coverity; CID #30. Bugfix on 0.2.2.26-beta. o Minor features: - Update to the July 1 2011 Maxmind GeoLite Country database. Changes in version 0.2.2.29-beta - 2011-06-20 Tor 0.2.2.29-beta reverts an accidental behavior change for users who have bridge lines in their torrc but don't want to use them; gets us closer to having the control socket feature working on Debian; and fixes a variety of smaller bugs. o Major bugfixes: - Revert the UseBridges option to its behavior before 0.2.2.28-beta. When we changed the default behavior to "use bridges if any are listed in the torrc", we surprised users who had bridges in their torrc files but who didn't actually want to use them. Partial resolution for bug 3354. o Privacy fixes: - Don't attach new streams to old rendezvous circuits after SIGNAL NEWNYM. Previously, we would keep using an existing rendezvous circuit if it remained open (i.e. if it were kept open by a long-lived stream, or if a new stream were attached to it before Tor could notice that it was old and no longer in use). Bugfix on 0.1.1.15-rc; fixes bug 3375. o Minor bugfixes: - Fix a bug when using ControlSocketsGroupWritable with User. The directory's group would be checked against the current group, not the configured group. Patch by Jérémy Bobbio. Fixes bug 3393; bugfix on 0.2.2.26-beta. - Make connection_printf_to_buf()'s behaviour sane. Its callers expect it to emit a CRLF iff the format string ends with CRLF; it actually emitted a CRLF iff (a) the format string ended with CRLF or (b) the resulting string was over 1023 characters long or (c) the format string did not end with CRLF *and* the resulting string was 1021 characters long or longer. Bugfix on 0.1.1.9-alpha; fixes part of bug 3407. - Make send_control_event_impl()'s behaviour sane. Its callers expect it to always emit a CRLF at the end of the string; it might have emitted extra control characters as well. Bugfix on 0.1.1.9-alpha; fixes another part of bug 3407. - Make crypto_rand_int() check the value of its input correctly. Previously, it accepted values up to UINT_MAX, but could return a negative number if given a value above INT_MAX+1. Found by George Kadianakis. Fixes bug 3306; bugfix on 0.2.2pre14. - Avoid a segfault when reading a malformed circuit build state with more than INT_MAX entries. Found by wanoskarnet. Bugfix on 0.2.2.4-alpha. - When asked about a DNS record type we don't support via a client DNSPort, reply with NOTIMPL rather than an empty reply. Patch by intrigeri. Fixes bug 3369; bugfix on 2.0.1-alpha. - Fix a rare memory leak during stats writing. Found by coverity. o Minor features: - Update to the June 1 2011 Maxmind GeoLite Country database. o Code simplifications and refactoring: - Remove some dead code as indicated by coverity. - Remove a few dead assignments during router parsing. Found by coverity. - Add some forgotten return value checks during unit tests. Found by coverity. - Don't use 1-bit wide signed bit fields. Found by coverity. Changes in version 0.2.2.28-beta - 2011-06-04 Tor 0.2.2.28-beta makes great progress towards a new stable release: we fixed a big bug in whether relays stay in the consensus consistently, we moved closer to handling bridges and hidden services correctly, and we started the process of better handling the dreaded "my Vidalia died, and now my Tor demands a password when I try to reconnect to it" usability issue. o Major bugfixes: - Don't decide to make a new descriptor when receiving a HUP signal. This bug has caused a lot of 0.2.2.x relays to disappear from the consensus periodically. Fixes the most common case of triggering bug 1810; bugfix on 0.2.2.7-alpha. - Actually allow nameservers with IPv6 addresses. Fixes bug 2574. - Don't try to build descriptors if "ORPort auto" is set and we don't know our actual ORPort yet. Fix for bug 3216; bugfix on 0.2.2.26-beta. - Resolve a crash that occurred when setting BridgeRelay to 1 with accounting enabled. Fixes bug 3228; bugfix on 0.2.2.18-alpha. - Apply circuit timeouts to opened hidden-service-related circuits based on the correct start time. Previously, we would apply the circuit build timeout based on time since the circuit's creation; it was supposed to be applied based on time since the circuit entered its current state. Bugfix on 0.0.6; fixes part of bug 1297. - Use the same circuit timeout for client-side introduction circuits as for other four-hop circuits, rather than the timeout for single-hop directory-fetch circuits; the shorter timeout may have been appropriate with the static circuit build timeout in 0.2.1.x and earlier, but caused many hidden service access attempts to fail with the adaptive CBT introduced in 0.2.2.2-alpha. Bugfix on 0.2.2.2-alpha; fixes another part of bug 1297. - In ticket 2511 we fixed a case where you could use an unconfigured bridge if you had configured it as a bridge the last time you ran Tor. Now fix another edge case: if you had configured it as a bridge but then switched to a different bridge via the controller, you would still be willing to use the old one. Bugfix on 0.2.0.1-alpha; fixes bug 3321. o Major features: - Add an __OwningControllerProcess configuration option and a TAKEOWNERSHIP control-port command. Now a Tor controller can ensure that when it exits, Tor will shut down. Implements feature 3049. - If "UseBridges 1" is set and no bridges are configured, Tor will now refuse to build any circuits until some bridges are set. If "UseBridges auto" is set, Tor will use bridges if they are configured and we are not running as a server, but otherwise will make circuits as usual. The new default is "auto". Patch by anonym, so the Tails LiveCD can stop automatically revealing you as a Tor user on startup. o Minor bugfixes: - Fix warnings from GCC 4.6's "-Wunused-but-set-variable" option. - Remove a trailing asterisk from "exit-policy/default" in the output of the control port command "GETINFO info/names". Bugfix on 0.1.2.5-alpha. - Use a wide type to hold sockets when built for 64-bit Windows builds. Fixes bug 3270. - Warn when the user configures two HiddenServiceDir lines that point to the same directory. Bugfix on 0.0.6 (the version introducing HiddenServiceDir); fixes bug 3289. - Remove dead code from rend_cache_lookup_v2_desc_as_dir. Fixes part of bug 2748; bugfix on 0.2.0.10-alpha. - Log malformed requests for rendezvous descriptors as protocol warnings, not warnings. Also, use a more informative log message in case someone sees it at log level warning without prior info-level messages. Fixes the other part of bug 2748; bugfix on 0.2.0.10-alpha. - Clear the table recording the time of the last request for each hidden service descriptor from each HS directory on SIGNAL NEWNYM. Previously, we would clear our HS descriptor cache on SIGNAL NEWNYM, but if we had previously retrieved a descriptor (or tried to) from every directory responsible for it, we would refuse to fetch it again for up to 15 minutes. Bugfix on 0.2.2.25-alpha; fixes bug 3309. - Fix a log message that said "bits" while displaying a value in bytes. Found by wanoskarnet. Fixes bug 3318; bugfix on 0.2.0.1-alpha. - When checking for 1024-bit keys, check for 1024 bits, not 128 bytes. This allows Tor to correctly discard keys of length 1017 through 1023. Bugfix on 0.0.9pre5. o Minor features: - Relays now log the reason for publishing a new relay descriptor, so we have a better chance of hunting down instances of bug 1810. Resolves ticket 3252. - Revise most log messages that refer to nodes by nickname to instead use the "$key=nickname at address" format. This should be more useful, especially since nicknames are less and less likely to be unique. Resolves ticket 3045. - Log (at info level) when purging pieces of hidden-service-client state because of SIGNAL NEWNYM. o Removed options: - Remove undocumented option "-F" from tor-resolve: it hasn't done anything since 0.2.1.16-rc. Changes in version 0.2.2.27-beta - 2011-05-18 Tor 0.2.2.27-beta fixes a bridge-related stability bug in the previous release, and also adds a few more general bugfixes. o Major bugfixes: - Fix a crash bug when changing bridges in a running Tor process. Fixes bug 3213; bugfix on 0.2.2.26-beta. - When the controller configures a new bridge, don't wait 10 to 60 seconds before trying to fetch its descriptor. Bugfix on 0.2.0.3-alpha; fixes bug 3198 (suggested by 2355). o Minor bugfixes: - Require that onion keys have exponent 65537 in microdescriptors too. Fixes more of bug 3207; bugfix on 0.2.2.26-beta. - Tor used to limit HttpProxyAuthenticator values to 48 characters. Changed the limit to 512 characters by removing base64 newlines. Fixes bug 2752. Fix by Michael Yakubovich. - When a client starts or stops using bridges, never use a circuit that was built before the configuration change. This behavior could put at risk a user who uses bridges to ensure that her traffic only goes to the chosen addresses. Bugfix on 0.2.0.3-alpha; fixes bug 3200. Changes in version 0.2.2.26-beta - 2011-05-17 Tor 0.2.2.26-beta fixes a variety of potential privacy problems. It also introduces a new "socksport auto" approach that should make it easier to run multiple Tors on the same system, and does a lot of cleanup to get us closer to a release candidate. o Security/privacy fixes: - Replace all potentially sensitive memory comparison operations with versions whose runtime does not depend on the data being compared. This will help resist a class of attacks where an adversary can use variations in timing information to learn sensitive data. Fix for one case of bug 3122. (Safe memcmp implementation by Robert Ransom based partially on code by DJB.) - When receiving a hidden service descriptor, check that it is for the hidden service we wanted. Previously, Tor would store any hidden service descriptors that a directory gave it, whether it wanted them or not. This wouldn't have let an attacker impersonate a hidden service, but it did let directories pre-seed a client with descriptors that it didn't want. Bugfix on 0.0.6. - On SIGHUP, do not clear out all TrackHostExits mappings, client DNS cache entries, and virtual address mappings: that's what NEWNYM is for. Fixes bug 1345; bugfix on 0.1.0.1-rc. o Major features: - The options SocksPort, ControlPort, and so on now all accept a value "auto" that opens a socket on an OS-selected port. A new ControlPortWriteToFile option tells Tor to write its actual control port or ports to a chosen file. If the option ControlPortFileGroupReadable is set, the file is created as group-readable. Now users can run two Tor clients on the same system without needing to manually mess with parameters. Resolves part of ticket 3076. - Set SO_REUSEADDR on all sockets, not just listeners. This should help busy exit nodes avoid running out of useable ports just because all the ports have been used in the near past. Resolves issue 2850. o Minor features: - New "GETINFO net/listeners/(type)" controller command to return a list of addresses and ports that are bound for listeners for a given connection type. This is useful when the user has configured "SocksPort auto" and the controller needs to know which port got chosen. Resolves another part of ticket 3076. - Add a new ControlSocketsGroupWritable configuration option: when it is turned on, ControlSockets are group-writeable by the default group of the current user. Patch by Jérémy Bobbio; implements ticket 2972. - Tor now refuses to create a ControlSocket in a directory that is world-readable (or group-readable if ControlSocketsGroupWritable is 0). This is necessary because some operating systems do not enforce permissions on an AF_UNIX sockets. Permissions on the directory holding the socket, however, seems to work everywhere. - Rate-limit a warning about failures to download v2 networkstatus documents. Resolves part of bug 1352. - Backport code from 0.2.3.x that allows directory authorities to clean their microdescriptor caches. Needed to resolve bug 2230. - When an HTTPS proxy reports "403 Forbidden", we now explain what it means rather than calling it an unexpected status code. Closes bug 2503. Patch from Michael Yakubovich. - Update to the May 1 2011 Maxmind GeoLite Country database. o Minor bugfixes: - Authorities now clean their microdesc cache periodically and when reading from disk initially, not only when adding new descriptors. This prevents a bug where we could lose microdescriptors. Bugfix on 0.2.2.6-alpha. Fixes bug 2230. - Do not crash when our configuration file becomes unreadable, for example due to a permissions change, between when we start up and when a controller calls SAVECONF. Fixes bug 3135; bugfix on 0.0.9pre6. - Avoid a bug that would keep us from replacing a microdescriptor cache on Windows. (We would try to replace the file while still holding it open. That's fine on Unix, but Windows doesn't let us do that.) Bugfix on 0.2.2.6-alpha; bug found by wanoskarnet. - Add missing explanations for the authority-related torrc options RephistTrackTime, BridgePassword, and V3AuthUseLegacyKey in the man page. Resolves issue 2379. - As an authority, do not upload our own vote or signature set to ourself. It would tell us nothing new, and as of 0.2.2.24-alpha, it would get flagged as a duplicate. Resolves bug 3026. - Accept hidden service descriptors if we think we might be a hidden service directory, regardless of what our consensus says. This helps robustness, since clients and hidden services can sometimes have a more up-to-date view of the network consensus than we do, and if they think that the directory authorities list us a HSDir, we might actually be one. Related to bug 2732; bugfix on 0.2.0.10-alpha. - When a controller changes TrackHostExits, remove mappings for hosts that should no longer have their exits tracked. Bugfix on 0.1.0.1-rc. - When a controller changes VirtualAddrNetwork, remove any mappings for hosts that were automapped to the old network. Bugfix on 0.1.1.19-rc. - When a controller changes one of the AutomapHosts* options, remove any mappings for hosts that should no longer be automapped. Bugfix on 0.2.0.1-alpha. - Do not reset the bridge descriptor download status every time we re-parse our configuration or get a configuration change. Fixes bug 3019; bugfix on 0.2.0.3-alpha. o Minor bugfixes (code cleanup): - When loading the microdesc journal, remember its current size. In 0.2.2, this helps prevent the microdesc journal from growing without limit on authorities (who are the only ones to use it in 0.2.2). Fixes a part of bug 2230; bugfix on 0.2.2.6-alpha. Fix posted by "cypherpunks." - The microdesc journal is supposed to get rebuilt only if it is at least _half_ the length of the store, not _twice_ the length of the store. Bugfix on 0.2.2.6-alpha; fixes part of bug 2230. - Fix a potential null-pointer dereference while computing a consensus. Bugfix on tor-0.2.0.3-alpha, found with the help of clang's analyzer. - Avoid a possible null-pointer dereference when rebuilding the mdesc cache without actually having any descriptors to cache. Bugfix on 0.2.2.6-alpha. Issue discovered using clang's static analyzer. - If we fail to compute the identity digest of a v3 legacy keypair, warn, and don't use a buffer-full of junk instead. Bugfix on 0.2.1.1-alpha; fixes bug 3106. - Resolve an untriggerable issue in smartlist_string_num_isin(), where if the function had ever in the future been used to check for the presence of a too-large number, it would have given an incorrect result. (Fortunately, we only used it for 16-bit values.) Fixes bug 3175; bugfix on 0.1.0.1-rc. - Require that introduction point keys and onion handshake keys have a public exponent of 65537. Starts to fix bug 3207; bugfix on 0.2.0.10-alpha. o Removed features: - Caches no longer download and serve v2 networkstatus documents unless FetchV2Networkstatus flag is set: these documents haven't haven't been used by clients or relays since 0.2.0.x. Resolves bug 3022. Changes in version 0.2.3.1-alpha - 2011-05-05 Tor 0.2.3.1-alpha adds some new experimental features, including support for an improved network IO backend, IOCP networking on Windows, microdescriptor caching, "fast-start" support for streams, and automatic home router configuration. There are also numerous internal improvements to try to make the code easier for developers to work with. This is the first alpha release in a new series, so expect there to be bugs. Users who would rather test out a more stable branch should stay with 0.2.2.x for now. o Major features: - Tor can now optionally build with the "bufferevents" buffered IO backend provided by Libevent 2. To use this feature, make sure you have the latest possible version of Libevent, and pass the --enable-bufferevents flag to configure when building Tor from source. This feature will make our networking code more flexible, let us stack layers on each other, and let us use more efficient zero-copy transports where available. - As an experimental feature, Tor can use IOCP for networking on Windows. Once this code is tuned and optimized, it promises much better performance than the select-based backend we've used in the past. To try this feature, you must build Tor with Libevent 2, configure Tor with the "bufferevents" buffered IO backend, and add "DisableIOCP 0" to your torrc. There are known bugs here: only try this if you can help debug it as it breaks. - The EntryNodes option can now include country codes like {de} or IP addresses or network masks. Previously we had disallowed these options because we didn't have an efficient way to keep the list up to date. Fixes bug 1982, but see bug 2798 for an unresolved issue here. - Exit nodes now accept and queue data on not-yet-connected streams. Previously, the client wasn't allowed to send data until the stream was connected, which slowed down all connections. This change will enable clients to perform a "fast-start" on streams and send data without having to wait for a confirmation that the stream has opened. (Patch from Ian Goldberg; implements the server side of Proposal 174.) - Tor now has initial support for automatic port mapping on the many home routers that support NAT-PMP or UPnP. (Not yet supported on Windows). To build the support code, you'll need to have libnatpnp library and/or the libminiupnpc library, and you'll need to enable the feature specifically by passing "--enable-upnp" and/or "--enable-natpnp" to configure. To turn it on, use the new PortForwarding option. - Caches now download, cache, and serve multiple "flavors" of the consensus, including a flavor that describes microdescriptors. - Caches now download, cache, and serve microdescriptors -- small summaries of router descriptors that are authenticated by all of the directory authorities. Once enough caches are running this code, clients will be able to save significant amounts of directory bandwidth by downloading microdescriptors instead of router descriptors. o Minor features: - Make logging resolution configurable with a new LogTimeGranularity option, and change the default from 1 millisecond to 1 second. Implements enhancement 1668. - We log which torrc file we're using on startup. Implements ticket 2444. - Ordinarily, Tor does not count traffic from private addresses (like 127.0.0.1 or 10.0.0.1) when calculating rate limits or accounting. There is now a new option, CountPrivateBandwidth, to disable this behavior. Patch from Daniel Cagara. - New --enable-static-tor configure option for building Tor as statically as possible. Idea, general hackery and thoughts from Alexei Czeskis, John Gilmore, Jacob Appelbaum. Implements ticket 2702. - If you set the NumCPUs option to 0, Tor will now try to detect how many CPUs you have. This is the new default behavior. - Turn on directory request statistics by default and include them in extra-info descriptors. Don't break if we have no GeoIP database. - Relays that set "ConnDirectionStatistics 1" write statistics on the bidirectional use of connections to disk every 24 hours. - Add a GeoIP file digest to the extra-info descriptor. Implements enhancement 1883. - The NodeFamily option -- which let you declare that you want to consider nodes to be part of a family whether they list themselves that way or not -- now allows IP address ranges and country codes. - Add a new 'Heartbeat' log message type to periodically log a message describing Tor's status at level Notice. This feature is meant for operators who log at notice, and want to make sure that their Tor server is still working. Implementation by George Kadianakis. o Minor bugfixes (on 0.2.2.25-alpha): - When loading the microdesc journal, remember its current size. In 0.2.2, this helps prevent the microdesc journal from growing without limit on authorities (who are the only ones to use it in 0.2.2). Fixes a part of bug 2230; bugfix on 0.2.2.6-alpha. Fix posted by "cypherpunks." - The microdesc journal is supposed to get rebuilt only if it is at least _half_ the length of the store, not _twice_ the length of the store. Bugfix on 0.2.2.6-alpha; fixes part of bug 2230. - If as an authority we fail to compute the identity digest of a v3 legacy keypair, warn, and don't use a buffer-full of junk instead. Bugfix on 0.2.1.1-alpha; fixes bug 3106. - Authorities now clean their microdesc cache periodically and when reading from disk initially, not only when adding new descriptors. This prevents a bug where we could lose microdescriptors. Bugfix on 0.2.2.6-alpha. o Minor features (controller): - Add a new SIGNAL event to the controller interface so that controllers can be notified when Tor handles a signal. Resolves issue 1955. Patch by John Brooks. - Add a new GETINFO option to get total bytes read and written. Patch from pipe, revised by atagar. Resolves ticket 2345. - Implement some GETINFO controller fields to provide information about the Tor process's pid, euid, username, and resource limits. o Build changes: - Our build system requires automake 1.6 or later to create the Makefile.in files. Previously, you could have used 1.4. This only affects developers and people building Tor from git; people who build Tor from the source distribution without changing the Makefile.am files should be fine. - Our autogen.sh script uses autoreconf to launch autoconf, automake, and so on. This is more robust against some of the failure modes associated with running the autotools pieces on their own. o Minor packaging issues: - On OpenSUSE, create the /var/run/tor directory on startup if it is not already created. Patch from Andreas Stieger. Fixes bug 2573. o Code simplifications and refactoring: - A major revision to our internal node-selecting and listing logic. Tor already had at least two major ways to look at the question of "which Tor servers do we know about": a list of router descriptors, and a list of entries in the current consensus. With microdescriptors, we're adding a third. Having so many systems without an abstraction layer over them was hurting the codebase. Now, we have a new "node_t" abstraction that presents a consistent interface to a client's view of a Tor node, and holds (nearly) all of the mutable state formerly in routerinfo_t and routerstatus_t. - The helper programs tor-gencert, tor-resolve, and tor-checkkey no longer link against Libevent: they never used it, but our library structure used to force them to link it. o Removed features: - Remove some old code to work around even older versions of Tor that used forked processes to handle DNS requests. Such versions of Tor are no longer in use as servers. o Documentation fixes: - Correct a broken faq link in the INSTALL file. Fixes bug 2307. - Add missing documentation for the authority-related torrc options RephistTrackTime, BridgePassword, and V3AuthUseLegacyKey. Resolves issue 2379. Changes in version 0.2.2.25-alpha - 2011-04-29 Tor 0.2.2.25-alpha fixes many bugs: hidden service clients are more robust, routers no longer overreport their bandwidth, Win7 should crash a little less, and NEWNYM (as used by Vidalia's "new identity" button) now prevents hidden service-related activity from being linkable. It provides more information to Vidalia so you can see if your bridge is working. Also, 0.2.2.25-alpha revamps the Entry/Exit/ExcludeNodes and StrictNodes configuration options to make them more reliable, more understandable, and more regularly applied. If you use those options, please see the revised documentation for them in the manual page. o Major bugfixes: - Relays were publishing grossly inflated bandwidth values because they were writing their state files wrong--now they write the correct value. Also, resume reading bandwidth history from the state file correctly. Fixes bug 2704; bugfix on 0.2.2.23-alpha. - Improve hidden service robustness: When we find that we have extended a hidden service's introduction circuit to a relay not listed as an introduction point in the HS descriptor we currently have, retry with an introduction point from the current descriptor. Previously we would just give up. Fixes bugs 1024 and 1930; bugfix on 0.2.0.10-alpha. - Clients now stop trying to use an exit node associated with a given destination by TrackHostExits if they fail to reach that exit node. Fixes bug 2999. Bugfix on 0.2.0.20-rc. - Fix crash bug on platforms where gmtime and localtime can return NULL. Windows 7 users were running into this one. Fixes part of bug 2077. Bugfix on all versions of Tor. Found by boboper. o Security and stability fixes: - Don't double-free a parsable, but invalid, microdescriptor, even if it is followed in the blob we're parsing by an unparsable microdescriptor. Fixes an issue reported in a comment on bug 2954. Bugfix on 0.2.2.6-alpha; fix by "cypherpunks". - If the Nickname configuration option isn't given, Tor would pick a nickname based on the local hostname as the nickname for a relay. Because nicknames are not very important in today's Tor and the "Unnamed" nickname has been implemented, this is now problematic behavior: It leaks information about the hostname without being useful at all. Fixes bug 2979; bugfix on 0.1.2.2-alpha, which introduced the Unnamed nickname. Reported by tagnaq. - Fix an uncommon assertion failure when running with DNSPort under heavy load. Fixes bug 2933; bugfix on 0.2.0.1-alpha. - Avoid linkability based on cached hidden service descriptors: forget all hidden service descriptors cached as a client when processing a SIGNAL NEWNYM command. Fixes bug 3000; bugfix on 0.0.6. o Major features: - Export GeoIP information on bridge usage to controllers even if we have not yet been running for 24 hours. Now Vidalia bridge operators can get more accurate and immediate feedback about their contributions to the network. o Major features and bugfixes (node selection): - Revise and reconcile the meaning of the ExitNodes, EntryNodes, ExcludeEntryNodes, ExcludeExitNodes, ExcludeNodes, and StrictNodes options. Previously, we had been ambiguous in describing what counted as an "exit" node, and what operations exactly "StrictNodes 0" would permit. This created confusion when people saw nodes built through unexpected circuits, and made it hard to tell real bugs from surprises. Now the intended behavior is: . "Exit", in the context of ExitNodes and ExcludeExitNodes, means a node that delivers user traffic outside the Tor network. . "Entry", in the context of EntryNodes, means a node used as the first hop of a multihop circuit. It doesn't include direct connections to directory servers. . "ExcludeNodes" applies to all nodes. . "StrictNodes" changes the behavior of ExcludeNodes only. When StrictNodes is set, Tor should avoid all nodes listed in ExcludeNodes, even when it will make user requests fail. When StrictNodes is *not* set, then Tor should follow ExcludeNodes whenever it can, except when it must use an excluded node to perform self-tests, connect to a hidden service, provide a hidden service, fulfill a .exit request, upload directory information, or fetch directory information. Collectively, the changes to implement the behavior fix bug 1090. - ExcludeNodes now takes precedence over EntryNodes and ExitNodes: if a node is listed in both, it's treated as excluded. - ExcludeNodes now applies to directory nodes -- as a preference if StrictNodes is 0, or an absolute requirement if StrictNodes is 1. Don't exclude all the directory authorities and set StrictNodes to 1 unless you really want your Tor to break. - ExcludeNodes and ExcludeExitNodes now override exit enclaving. - ExcludeExitNodes now overrides .exit requests. - We don't use bridges listed in ExcludeNodes. - When StrictNodes is 1: . We now apply ExcludeNodes to hidden service introduction points and to rendezvous points selected by hidden service users. This can make your hidden service less reliable: use it with caution! . If we have used ExcludeNodes on ourself, do not try relay reachability self-tests. . If we have excluded all the directory authorities, we will not even try to upload our descriptor if we're a relay. . Do not honor .exit requests to an excluded node. - Remove a misfeature that caused us to ignore the Fast/Stable flags when ExitNodes is set. Bugfix on 0.2.2.7-alpha. - When the set of permitted nodes changes, we now remove any mappings introduced via TrackExitHosts to now-excluded nodes. Bugfix on 0.1.0.1-rc. - We never cannibalize a circuit that had excluded nodes on it, even if StrictNodes is 0. Bugfix on 0.1.0.1-rc. - Revert a change where we would be laxer about attaching streams to circuits than when building the circuits. This was meant to prevent a set of bugs where streams were never attachable, but our improved code here should make this unnecessary. Bugfix on 0.2.2.7-alpha. - Keep track of how many times we launch a new circuit to handle a given stream. Too many launches could indicate an inconsistency between our "launch a circuit to handle this stream" logic and our "attach this stream to one of the available circuits" logic. - Improve log messages related to excluded nodes. o Minor bugfixes: - Fix a spurious warning when moving from a short month to a long month on relays with month-based BandwidthAccounting. Bugfix on 0.2.2.17-alpha; fixes bug 3020. - When a client finds that an origin circuit has run out of 16-bit stream IDs, we now mark it as unusable for new streams. Previously, we would try to close the entire circuit. Bugfix on 0.0.6. - Add a forgotten cast that caused a compile warning on OS X 10.6. Bugfix on 0.2.2.24-alpha. - Be more careful about reporting the correct error from a failed connect() system call. Under some circumstances, it was possible to look at an incorrect value for errno when sending the end reason. Bugfix on 0.1.0.1-rc. - Correctly handle an "impossible" overflow cases in connection byte counting, where we write or read more than 4GB on an edge connection in a single second. Bugfix on 0.1.2.8-beta. - Correct the warning displayed when a rendezvous descriptor exceeds the maximum size. Fixes bug 2750; bugfix on 0.2.1.5-alpha. Found by John Brooks. - Clients and hidden services now use HSDir-flagged relays for hidden service descriptor downloads and uploads even if the relays have no DirPort set and the client has disabled TunnelDirConns. This will eventually allow us to give the HSDir flag to relays with no DirPort. Fixes bug 2722; bugfix on 0.2.1.6-alpha. - Downgrade "no current certificates known for authority" message from Notice to Info. Fixes bug 2899; bugfix on 0.2.0.10-alpha. - Make the SIGNAL DUMP control-port command work on FreeBSD. Fixes bug 2917. Bugfix on 0.1.1.1-alpha. - Only limit the lengths of single HS descriptors, even when multiple HS descriptors are published to an HSDir relay in a single POST operation. Fixes bug 2948; bugfix on 0.2.1.5-alpha. Found by hsdir. - Write the current time into the LastWritten line in our state file, rather than the time from the previous write attempt. Also, stop trying to use a time of -1 in our log statements. Fixes bug 3039; bugfix on 0.2.2.14-alpha. - Be more consistent in our treatment of file system paths. "~" should get expanded to the user's home directory in the Log config option. Fixes bug 2971; bugfix on 0.2.0.1-alpha, which introduced the feature for the -f and --DataDirectory options. o Minor features: - Make sure every relay writes a state file at least every 12 hours. Previously, a relay could go for weeks without writing its state file, and on a crash could lose its bandwidth history, capacity estimates, client country statistics, and so on. Addresses bug 3012. - Send END_STREAM_REASON_NOROUTE in response to EHOSTUNREACH errors. Clients before 0.2.1.27 didn't handle NOROUTE correctly, but such clients are already deprecated because of security bugs. - Don't allow v0 hidden service authorities to act as clients. Required by fix for bug 3000. - Ignore SIGNAL NEWNYM commands on relay-only Tor instances. Required by fix for bug 3000. - Ensure that no empty [dirreq-](read|write)-history lines are added to an extrainfo document. Implements ticket 2497. o Code simplification and refactoring: - Remove workaround code to handle directory responses from servers that had bug 539 (they would send HTTP status 503 responses _and_ send a body too). Since only server versions before 0.2.0.16-alpha/0.1.2.19 were affected, there is no longer reason to keep the workaround in place. - Remove the old 'fuzzy time' logic. It was supposed to be used for handling calculations where we have a known amount of clock skew and an allowed amount of unknown skew. But we only used it in three places, and we never adjusted the known/unknown skew values. This is still something we might want to do someday, but if we do, we'll want to do it differently. - Avoid signed/unsigned comparisons by making SIZE_T_CEILING unsigned. None of the cases where we did this before were wrong, but by making this change we avoid warnings. Fixes bug 2475; bugfix on 0.2.1.28. - Use GetTempDir to find the proper temporary directory location on Windows when generating temporary files for the unit tests. Patch by Gisle Vanem. Changes in version 0.2.2.24-alpha - 2011-04-08 Tor 0.2.2.24-alpha fixes a variety of bugs, including a big bug that prevented Tor clients from effectively using "multihomed" bridges, that is, bridges that listen on multiple ports or IP addresses so users can continue to use some of their addresses even if others get blocked. o Major bugfixes: - Fix a bug where bridge users who configure the non-canonical address of a bridge automatically switch to its canonical address. If a bridge listens at more than one address, it should be able to advertise those addresses independently and any non-blocked addresses should continue to work. Bugfix on Tor 0.2.0.x. Fixes bug 2510. - If you configured Tor to use bridge A, and then quit and configured Tor to use bridge B instead, it would happily continue to use bridge A if it's still reachable. While this behavior is a feature if your goal is connectivity, in some scenarios it's a dangerous bug. Bugfix on Tor 0.2.0.1-alpha; fixes bug 2511. - Directory authorities now use data collected from their own uptime observations when choosing whether to assign the HSDir flag to relays, instead of trusting the uptime value the relay reports in its descriptor. This change helps prevent an attack where a small set of nodes with frequently-changing identity keys can blackhole a hidden service. (Only authorities need upgrade; others will be fine once they do.) Bugfix on 0.2.0.10-alpha; fixes bug 2709. o Minor bugfixes: - When we restart our relay, we might get a successful connection from the outside before we've started our reachability tests, triggering a warning: "ORPort found reachable, but I have no routerinfo yet. Failing to inform controller of success." This bug was harmless unless Tor is running under a controller like Vidalia, in which case the controller would never get a REACHABILITY_SUCCEEDED status event. Bugfix on 0.1.2.6-alpha; fixes bug 1172. - Make directory authorities more accurate at recording when relays that have failed several reachability tests became unreachable, so we can provide more accuracy at assigning Stable, Guard, HSDir, etc flags. Bugfix on 0.2.0.6-alpha. Resolves bug 2716. - Fix an issue that prevented static linking of libevent on some platforms (notably Linux). Fixes bug 2698; bugfix on versions 0.2.1.23/0.2.2.8-alpha (the versions introducing the --with-static-libevent configure option). - We now ask the other side of a stream (the client or the exit) for more data on that stream when the amount of queued data on that stream dips low enough. Previously, we wouldn't ask the other side for more data until either it sent us more data (which it wasn't supposed to do if it had exhausted its window!) or we had completely flushed all our queued data. This flow control fix should improve throughput. Fixes bug 2756; bugfix on the earliest released versions of Tor (svn commit r152). - Avoid a double-mark-for-free warning when failing to attach a transparent proxy connection. (We thought we had fixed this in 0.2.2.23-alpha, but it turns out our fix was checking the wrong connection.) Fixes bug 2757; bugfix on 0.1.2.1-alpha (the original bug) and 0.2.2.23-alpha (the incorrect fix). - When warning about missing zlib development packages during compile, give the correct package names. Bugfix on 0.2.0.1-alpha. o Minor features: - Directory authorities now log the source of a rejected POSTed v3 networkstatus vote. - Make compilation with clang possible when using --enable-gcc-warnings by removing two warning options that clang hasn't implemented yet and by fixing a few warnings. Implements ticket 2696. - When expiring circuits, use microsecond timers rather than one-second timers. This can avoid an unpleasant situation where a circuit is launched near the end of one second and expired right near the beginning of the next, and prevent fluctuations in circuit timeout values. - Use computed circuit-build timeouts to decide when to launch parallel introduction circuits for hidden services. (Previously, we would retry after 15 seconds.) - Update to the April 1 2011 Maxmind GeoLite Country database. o Packaging fixes: - Create the /var/run/tor directory on startup on OpenSUSE if it is not already created. Patch from Andreas Stieger. Fixes bug 2573. o Documentation changes: - Modernize the doxygen configuration file slightly. Fixes bug 2707. - Resolve all doxygen warnings except those for missing documentation. Fixes bug 2705. - Add doxygen documentation for more functions, fields, and types. Changes in version 0.2.2.23-alpha - 2011-03-08 Tor 0.2.2.23-alpha lets relays record their bandwidth history so when they restart they don't lose their bandwidth capacity estimate. This release also fixes a diverse set of user-facing bugs, ranging from relays overrunning their rate limiting to clients falsely warning about clock skew to bridge descriptor leaks by our bridge directory authority. o Major bugfixes: - Stop sending a CLOCK_SKEW controller status event whenever we fetch directory information from a relay that has a wrong clock. Instead, only inform the controller when it's a trusted authority that claims our clock is wrong. Bugfix on 0.1.2.6-alpha; fixes the rest of bug 1074. - Fix an assert in parsing router descriptors containing IPv6 addresses. This one took down the directory authorities when somebody tried some experimental code. Bugfix on 0.2.1.3-alpha. - Make the bridge directory authority refuse to answer directory requests for "all" descriptors. It used to include bridge descriptors in its answer, which was a major information leak. Found by "piebeer". Bugfix on 0.2.0.3-alpha. - If relays set RelayBandwidthBurst but not RelayBandwidthRate, Tor would ignore their RelayBandwidthBurst setting, potentially using more bandwidth than expected. Bugfix on 0.2.0.1-alpha. Reported by Paul Wouters. Fixes bug 2470. - Ignore and warn if the user mistakenly sets "PublishServerDescriptor hidserv" in her torrc. The 'hidserv' argument never controlled publication of hidden service descriptors. Bugfix on 0.2.0.1-alpha. o Major features: - Relays now save observed peak bandwidth throughput rates to their state file (along with total usage, which was already saved) so that they can determine their correct estimated bandwidth on restart. Resolves bug 1863, where Tor relays would reset their estimated bandwidth to 0 after restarting. - Directory authorities now take changes in router IP address and ORPort into account when determining router stability. Previously, if a router changed its IP or ORPort, the authorities would not treat it as having any downtime for the purposes of stability calculation, whereas clients would experience downtime since the change could take a while to propagate to them. Resolves issue 1035. - Enable Address Space Layout Randomization (ASLR) and Data Execution Prevention (DEP) by default on Windows to make it harder for attackers to exploit vulnerabilities. Patch from John Brooks. o Minor bugfixes (on 0.2.1.x and earlier): - Fix a rare crash bug that could occur when a client was configured with a large number of bridges. Fixes bug 2629; bugfix on 0.2.1.2-alpha. Bugfix by trac user "shitlei". - Avoid a double mark-for-free warning when failing to attach a transparent proxy connection. Bugfix on 0.1.2.1-alpha. Fixes bug 2279. - Correctly detect failure to allocate an OpenSSL BIO. Fixes bug 2378; found by "cypherpunks". This bug was introduced before the first Tor release, in svn commit r110. - Country codes aren't supported in EntryNodes until 0.2.3.x, so don't mention them in the manpage. Fixes bug 2450; issue spotted by keb and G-Lo. - Fix a bug in bandwidth history state parsing that could have been triggered if a future version of Tor ever changed the timing granularity at which bandwidth history is measured. Bugfix on Tor 0.1.1.11-alpha. - When a relay decides that its DNS is too broken for it to serve as an exit server, it advertised itself as a non-exit, but continued to act as an exit. This could create accidental partitioning opportunities for users. Instead, if a relay is going to advertise reject *:* as its exit policy, it should really act with exit policy "reject *:*". Fixes bug 2366. Bugfix on Tor 0.1.2.5-alpha. Bugfix by user "postman" on trac. - In the special case where you configure a public exit relay as your bridge, Tor would be willing to use that exit relay as the last hop in your circuit as well. Now we fail that circuit instead. Bugfix on 0.2.0.12-alpha. Fixes bug 2403. Reported by "piebeer". - Fix a bug with our locking implementation on Windows that couldn't correctly detect when a file was already locked. Fixes bug 2504, bugfix on 0.2.1.6-alpha. - Fix IPv6-related connect() failures on some platforms (BSD, OS X). Bugfix on 0.2.0.3-alpha; fixes first part of bug 2660. Patch by "piebeer". - Set target port in get_interface_address6() correctly. Bugfix on 0.1.1.4-alpha and 0.2.0.3-alpha; fixes second part of bug 2660. - Directory authorities are now more robust to hops back in time when calculating router stability. Previously, if a run of uptime or downtime appeared to be negative, the calculation could give incorrect results. Bugfix on 0.2.0.6-alpha; noticed when fixing bug 1035. - Fix an assert that got triggered when using the TestingTorNetwork configuration option and then issuing a GETINFO config-text control command. Fixes bug 2250; bugfix on 0.2.1.2-alpha. o Minor bugfixes (on 0.2.2.x): - Clients should not weight BadExit nodes as Exits in their node selection. Similarly, directory authorities should not count BadExit bandwidth as Exit bandwidth when computing bandwidth-weights. Bugfix on 0.2.2.10-alpha; fixes bug 2203. - Correctly clear our dir_read/dir_write history when there is an error parsing any bw history value from the state file. Bugfix on Tor 0.2.2.15-alpha. - Resolve a bug in verifying signatures of directory objects with digests longer than SHA1. Bugfix on 0.2.2.20-alpha. Fixes bug 2409. Found by "piebeer". - Bridge authorities no longer crash on SIGHUP when they try to publish their relay descriptor to themselves. Fixes bug 2572. Bugfix on 0.2.2.22-alpha. o Minor features: - Log less aggressively about circuit timeout changes, and improve some other circuit timeout messages. Resolves bug 2004. - Log a little more clearly about the times at which we're no longer accepting new connections. Resolves bug 2181. - Reject attempts at the client side to open connections to private IP addresses (like 127.0.0.1, 10.0.0.1, and so on) with a randomly chosen exit node. Attempts to do so are always ill-defined, generally prevented by exit policies, and usually in error. This will also help to detect loops in transparent proxy configurations. You can disable this feature by setting "ClientRejectInternalAddresses 0" in your torrc. - Always treat failure to allocate an RSA key as an unrecoverable allocation error. - Update to the March 1 2011 Maxmind GeoLite Country database. o Minor features (log subsystem): - Add documentation for configuring logging at different severities in different log domains. We've had this feature since 0.2.1.1-alpha, but for some reason it never made it into the manpage. Fixes bug 2215. - Make it simpler to specify "All log domains except for A and B". Previously you needed to say "[*,~A,~B]". Now you can just say "[~A,~B]". - Add a "LogMessageDomains 1" option to include the domains of log messages along with the messages. Without this, there's no way to use log domains without reading the source or doing a lot of guessing. o Packaging changes: - Stop shipping the Tor specs files and development proposal documents in the tarball. They are now in a separate git repository at git://git.torproject.org/torspec.git Changes in version 0.2.1.30 - 2011-02-23 Tor 0.2.1.30 fixes a variety of less critical bugs. The main other change is a slight tweak to Tor's TLS handshake that makes relays and bridges that run this new version reachable from Iran again. We don't expect this tweak will win the arms race long-term, but it buys us time until we roll out a better solution. o Major bugfixes: - Stop sending a CLOCK_SKEW controller status event whenever we fetch directory information from a relay that has a wrong clock. Instead, only inform the controller when it's a trusted authority that claims our clock is wrong. Bugfix on 0.1.2.6-alpha; fixes the rest of bug 1074. - Fix a bounds-checking error that could allow an attacker to remotely crash a directory authority. Bugfix on 0.2.1.5-alpha. Found by "piebeer". - If relays set RelayBandwidthBurst but not RelayBandwidthRate, Tor would ignore their RelayBandwidthBurst setting, potentially using more bandwidth than expected. Bugfix on 0.2.0.1-alpha. Reported by Paul Wouters. Fixes bug 2470. - Ignore and warn if the user mistakenly sets "PublishServerDescriptor hidserv" in her torrc. The 'hidserv' argument never controlled publication of hidden service descriptors. Bugfix on 0.2.0.1-alpha. o Minor features: - Adjust our TLS Diffie-Hellman parameters to match those used by Apache's mod_ssl. - Update to the February 1 2011 Maxmind GeoLite Country database. o Minor bugfixes: - Check for and reject overly long directory certificates and directory tokens before they have a chance to hit any assertions. Bugfix on 0.2.1.28. Found by "doorss". - Bring the logic that gathers routerinfos and assesses the acceptability of circuits into line. This prevents a Tor OP from getting locked in a cycle of choosing its local OR as an exit for a path (due to a .exit request) and then rejecting the circuit because its OR is not listed yet. It also prevents Tor clients from using an OR running in the same instance as an exit (due to a .exit request) if the OR does not meet the same requirements expected of an OR running elsewhere. Fixes bug 1859; bugfix on 0.1.0.1-rc. o Packaging changes: - Stop shipping the Tor specs files and development proposal documents in the tarball. They are now in a separate git repository at git://git.torproject.org/torspec.git - Do not include Git version tags as though they are SVN tags when generating a tarball from inside a repository that has switched between branches. Bugfix on 0.2.1.15-rc; fixes bug 2402. Changes in version 0.2.2.22-alpha - 2011-01-25 Tor 0.2.2.22-alpha fixes a few more less-critical security issues. The main other change is a slight tweak to Tor's TLS handshake that makes relays and bridges that run this new version reachable from Iran again. We don't expect this tweak will win the arms race long-term, but it will buy us a bit more time until we roll out a better solution. o Major bugfixes: - Fix a bounds-checking error that could allow an attacker to remotely crash a directory authority. Bugfix on 0.2.1.5-alpha. Found by "piebeer". - Don't assert when changing from bridge to relay or vice versa via the controller. The assert happened because we didn't properly initialize our keys in this case. Bugfix on 0.2.2.18-alpha; fixes bug 2433. Reported by bastik. o Minor features: - Adjust our TLS Diffie-Hellman parameters to match those used by Apache's mod_ssl. - Provide a log message stating which geoip file we're parsing instead of just stating that we're parsing the geoip file. Implements ticket 2432. o Minor bugfixes: - Check for and reject overly long directory certificates and directory tokens before they have a chance to hit any assertions. Bugfix on 0.2.1.28 / 0.2.2.20-alpha. Found by "doorss". Changes in version 0.2.2.21-alpha - 2011-01-15 Tor 0.2.2.21-alpha includes all the patches from Tor 0.2.1.29, which continues our recent code security audit work. The main fix resolves a remote heap overflow vulnerability that can allow remote code execution (CVE-2011-0427). Other fixes address a variety of assert and crash bugs, most of which we think are hard to exploit remotely. o Major bugfixes (security), also included in 0.2.1.29: - Fix a heap overflow bug where an adversary could cause heap corruption. This bug probably allows remote code execution attacks. Reported by "debuger". Fixes CVE-2011-0427. Bugfix on 0.1.2.10-rc. - Prevent a denial-of-service attack by disallowing any zlib-compressed data whose compression factor is implausibly high. Fixes part of bug 2324; reported by "doorss". - Zero out a few more keys in memory before freeing them. Fixes bug 2384 and part of bug 2385. These key instances found by "cypherpunks", based on Andrew Case's report about being able to find sensitive data in Tor's memory space if you have enough permissions. Bugfix on 0.0.2pre9. o Major bugfixes (crashes), also included in 0.2.1.29: - Prevent calls to Libevent from inside Libevent log handlers. This had potential to cause a nasty set of crashes, especially if running Libevent with debug logging enabled, and running Tor with a controller watching for low-severity log messages. Bugfix on 0.1.0.2-rc. Fixes bug 2190. - Add a check for SIZE_T_MAX to tor_realloc() to try to avoid underflow errors there too. Fixes the other part of bug 2324. - Fix a bug where we would assert if we ever had a cached-descriptors.new file (or another file read directly into memory) of exactly SIZE_T_CEILING bytes. Fixes bug 2326; bugfix on 0.2.1.25. Found by doorss. - Fix some potential asserts and parsing issues with grossly malformed router caches. Fixes bug 2352; bugfix on Tor 0.2.1.27. Found by doorss. o Minor bugfixes (other), also included in 0.2.1.29: - Fix a bug with handling misformed replies to reverse DNS lookup requests in DNSPort. Bugfix on Tor 0.2.0.1-alpha. Related to a bug reported by doorss. - Fix compilation on mingw when a pthreads compatibility library has been installed. (We don't want to use it, so we shouldn't be including pthread.h.) Fixes bug 2313; bugfix on 0.1.0.1-rc. - Fix a bug where we would declare that we had run out of virtual addresses when the address space was only half-exhausted. Bugfix on 0.1.2.1-alpha. - Correctly handle the case where AutomapHostsOnResolve is set but no virtual addresses are available. Fixes bug 2328; bugfix on 0.1.2.1-alpha. Bug found by doorss. - Correctly handle wrapping around when we run out of virtual address space. Found by cypherpunks; bugfix on 0.2.0.5-alpha. o Minor features, also included in 0.2.1.29: - Update to the January 1 2011 Maxmind GeoLite Country database. - Introduce output size checks on all of our decryption functions. o Build changes, also included in 0.2.1.29: - Tor does not build packages correctly with Automake 1.6 and earlier; added a check to Makefile.am to make sure that we're building with Automake 1.7 or later. - The 0.2.1.28 tarball was missing src/common/OpenBSD_malloc_Linux.c because we built it with a too-old version of automake. Thus that release broke ./configure --enable-openbsd-malloc, which is popular among really fast exit relays on Linux. o Major bugfixes, new in 0.2.2.21-alpha: - Prevent crash/heap corruption when the cbtnummodes consensus parameter is set to 0 or large values. Fixes bug 2317; bugfix on 0.2.2.14-alpha. o Major features, new in 0.2.2.21-alpha: - Introduce minimum/maximum values that clients will believe from the consensus. Now we'll have a better chance to avoid crashes or worse when a consensus param has a weird value. o Minor features, new in 0.2.2.21-alpha: - Make sure to disable DirPort if running as a bridge. DirPorts aren't used on bridges, and it makes bridge scanning somewhat easier. - If writing the state file to disk fails, wait up to an hour before retrying again, rather than trying again each second. Fixes bug 2346; bugfix on Tor 0.1.1.3-alpha. - Make Libevent log messages get delivered to controllers later, and not from inside the Libevent log handler. This prevents unsafe reentrant Libevent calls while still letting the log messages get through. - Detect platforms that brokenly use a signed size_t, and refuse to build there. Found and analyzed by doorss and rransom. - Fix a bunch of compile warnings revealed by mingw with gcc 4.5. Resolves bug 2314. o Minor bugfixes, new in 0.2.2.21-alpha: - Handle SOCKS messages longer than 128 bytes long correctly, rather than waiting forever for them to finish. Fixes bug 2330; bugfix on 0.2.0.16-alpha. Found by doorss. - Add assertions to check for overflow in arguments to base32_encode() and base32_decode(); fix a signed-unsigned comparison there too. These bugs are not actually reachable in Tor, but it's good to prevent future errors too. Found by doorss. - Correctly detect failures to create DNS requests when using Libevent versions before v2. (Before Libevent 2, we used our own evdns implementation. Its return values for Libevent's evdns_resolve_*() functions are not consistent with those from Libevent.) Fixes bug 2363; bugfix on 0.2.2.6-alpha. Found by "lodger". o Documentation, new in 0.2.2.21-alpha: - Document the default socks host and port (127.0.0.1:9050) for tor-resolve. Changes in version 0.2.1.29 - 2011-01-15 Tor 0.2.1.29 continues our recent code security audit work. The main fix resolves a remote heap overflow vulnerability that can allow remote code execution. Other fixes address a variety of assert and crash bugs, most of which we think are hard to exploit remotely. o Major bugfixes (security): - Fix a heap overflow bug where an adversary could cause heap corruption. This bug probably allows remote code execution attacks. Reported by "debuger". Fixes CVE-2011-0427. Bugfix on 0.1.2.10-rc. - Prevent a denial-of-service attack by disallowing any zlib-compressed data whose compression factor is implausibly high. Fixes part of bug 2324; reported by "doorss". - Zero out a few more keys in memory before freeing them. Fixes bug 2384 and part of bug 2385. These key instances found by "cypherpunks", based on Andrew Case's report about being able to find sensitive data in Tor's memory space if you have enough permissions. Bugfix on 0.0.2pre9. o Major bugfixes (crashes): - Prevent calls to Libevent from inside Libevent log handlers. This had potential to cause a nasty set of crashes, especially if running Libevent with debug logging enabled, and running Tor with a controller watching for low-severity log messages. Bugfix on 0.1.0.2-rc. Fixes bug 2190. - Add a check for SIZE_T_MAX to tor_realloc() to try to avoid underflow errors there too. Fixes the other part of bug 2324. - Fix a bug where we would assert if we ever had a cached-descriptors.new file (or another file read directly into memory) of exactly SIZE_T_CEILING bytes. Fixes bug 2326; bugfix on 0.2.1.25. Found by doorss. - Fix some potential asserts and parsing issues with grossly malformed router caches. Fixes bug 2352; bugfix on Tor 0.2.1.27. Found by doorss. o Minor bugfixes (other): - Fix a bug with handling misformed replies to reverse DNS lookup requests in DNSPort. Bugfix on Tor 0.2.0.1-alpha. Related to a bug reported by doorss. - Fix compilation on mingw when a pthreads compatibility library has been installed. (We don't want to use it, so we shouldn't be including pthread.h.) Fixes bug 2313; bugfix on 0.1.0.1-rc. - Fix a bug where we would declare that we had run out of virtual addresses when the address space was only half-exhausted. Bugfix on 0.1.2.1-alpha. - Correctly handle the case where AutomapHostsOnResolve is set but no virtual addresses are available. Fixes bug 2328; bugfix on 0.1.2.1-alpha. Bug found by doorss. - Correctly handle wrapping around to when we run out of virtual address space. Found by cypherpunks, bugfix on 0.2.0.5-alpha. - The 0.2.1.28 tarball was missing src/common/OpenBSD_malloc_Linux.c because we built it with a too-old version of automake. Thus that release broke ./configure --enable-openbsd-malloc, which is popular among really fast exit relays on Linux. o Minor features: - Update to the January 1 2011 Maxmind GeoLite Country database. - Introduce output size checks on all of our decryption functions. o Build changes: - Tor does not build packages correctly with Automake 1.6 and earlier; added a check to Makefile.am to make sure that we're building with Automake 1.7 or later. Changes in version 0.2.2.20-alpha - 2010-12-17 Tor 0.2.2.20-alpha does some code cleanup to reduce the risk of remotely exploitable bugs. We also fix a variety of other significant bugs, change the IP address for one of our directory authorities, and update the minimum version that Tor relays must run to join the network. o Major bugfixes: - Fix a remotely exploitable bug that could be used to crash instances of Tor remotely by overflowing on the heap. Remote-code execution hasn't been confirmed, but can't be ruled out. Everyone should upgrade. Bugfix on the 0.1.1 series and later. - Fix a bug that could break accounting on 64-bit systems with large time_t values, making them hibernate for impossibly long intervals. Fixes bug 2146. Bugfix on 0.0.9pre6; fix by boboper. - Fix a logic error in directory_fetches_from_authorities() that would cause all _non_-exits refusing single-hop-like circuits to fetch from authorities, when we wanted to have _exits_ fetch from authorities. Fixes more of 2097. Bugfix on 0.2.2.16-alpha; fix by boboper. - Fix a stream fairness bug that would cause newer streams on a given circuit to get preference when reading bytes from the origin or destination. Fixes bug 2210. Fix by Mashael AlSabah. This bug was introduced before the first Tor release, in svn revision r152. o Directory authority changes: - Change IP address and ports for gabelmoo (v3 directory authority). o Minor bugfixes: - Avoid crashes when AccountingMax is set on clients. Fixes bug 2235. Bugfix on 0.2.2.18-alpha. Diagnosed by boboper. - Fix an off-by-one error in calculating some controller command argument lengths. Fortunately, this mistake is harmless since the controller code does redundant NUL termination too. Found by boboper. Bugfix on 0.1.1.1-alpha. - Do not dereference NULL if a bridge fails to build its extra-info descriptor. Found by an anonymous commenter on Trac. Bugfix on 0.2.2.19-alpha. o Minor features: - Update to the December 1 2010 Maxmind GeoLite Country database. - Directory authorities now reject relays running any versions of Tor between 0.2.1.3-alpha and 0.2.1.18 inclusive; they have known bugs that keep RELAY_EARLY cells from working on rendezvous circuits. Followup to fix for bug 2081. - Directory authorities now reject relays running any version of Tor older than 0.2.0.26-rc. That version is the earliest that fetches current directory information correctly. Fixes bug 2156. - Report only the top 10 ports in exit-port stats in order not to exceed the maximum extra-info descriptor length of 50 KB. Implements task 2196. Changes in version 0.2.1.28 - 2010-12-17 Tor 0.2.1.28 does some code cleanup to reduce the risk of remotely exploitable bugs. We also took this opportunity to change the IP address for one of our directory authorities, and to update the geoip database we ship. o Major bugfixes: - Fix a remotely exploitable bug that could be used to crash instances of Tor remotely by overflowing on the heap. Remote-code execution hasn't been confirmed, but can't be ruled out. Everyone should upgrade. Bugfix on the 0.1.1 series and later. o Directory authority changes: - Change IP address and ports for gabelmoo (v3 directory authority). o Minor features: - Update to the December 1 2010 Maxmind GeoLite Country database. Changes in version 0.2.1.27 - 2010-11-23 Yet another OpenSSL security patch broke its compatibility with Tor: Tor 0.2.1.27 makes relays work with openssl 0.9.8p and 1.0.0.b. We also took this opportunity to fix several crash bugs, integrate a new directory authority, and update the bundled GeoIP database. o Major bugfixes: - Resolve an incompatibility with OpenSSL 0.9.8p and OpenSSL 1.0.0b: No longer set the tlsext_host_name extension on server SSL objects; but continue to set it on client SSL objects. Our goal in setting it was to imitate a browser, not a vhosting server. Fixes bug 2204; bugfix on 0.2.1.1-alpha. - Do not log messages to the controller while shrinking buffer freelists. Doing so would sometimes make the controller connection try to allocate a buffer chunk, which would mess up the internals of the freelist and cause an assertion failure. Fixes bug 1125; fixed by Robert Ransom. Bugfix on 0.2.0.16-alpha. - Learn our external IP address when we're a relay or bridge, even if we set PublishServerDescriptor to 0. Bugfix on 0.2.0.3-alpha, where we introduced bridge relays that don't need to publish to be useful. Fixes bug 2050. - Do even more to reject (and not just ignore) annotations on router descriptors received anywhere but from the cache. Previously we would ignore such annotations at first, but cache them to disk anyway. Bugfix on 0.2.0.8-alpha. Found by piebeer. - When you're using bridges and your network goes away and your bridges get marked as down, recover when you attempt a new socks connection (if the network is back), rather than waiting up to an hour to try fetching new descriptors for your bridges. Bugfix on 0.2.0.3-alpha; fixes bug 1981. o Major features: - Move to the November 2010 Maxmind GeoLite country db (rather than the June 2009 ip-to-country GeoIP db) for our statistics that count how many users relays are seeing from each country. Now we'll have more accurate data, especially for many African countries. o New directory authorities: - Set up maatuska (run by Linus Nordberg) as the eighth v3 directory authority. o Minor bugfixes: - Fix an assertion failure that could occur in directory caches or bridge users when using a very short voting interval on a testing network. Diagnosed by Robert Hogan. Fixes bug 1141; bugfix on 0.2.0.8-alpha. - Enforce multiplicity rules when parsing annotations. Bugfix on 0.2.0.8-alpha. Found by piebeer. - Allow handshaking OR connections to take a full KeepalivePeriod seconds to handshake. Previously, we would close them after IDLE_OR_CONN_TIMEOUT (180) seconds, the same timeout as if they were open. Bugfix on 0.2.1.26; fixes bug 1840. Thanks to mingw-san for analysis help. - When building with --enable-gcc-warnings on OpenBSD, disable warnings in system headers. This makes --enable-gcc-warnings pass on OpenBSD 4.8. o Minor features: - Exit nodes didn't recognize EHOSTUNREACH as a plausible error code, and so sent back END_STREAM_REASON_MISC. Clients now recognize a new stream ending reason for this case: END_STREAM_REASON_NOROUTE. Servers can start sending this code when enough clients recognize it. Bugfix on 0.1.0.1-rc; fixes part of bug 1793. - Build correctly on mingw with more recent versions of OpenSSL 0.9.8. Patch from mingw-san. o Removed files: - Remove the old debian/ directory from the main Tor distribution. The official Tor-for-debian git repository lives at the URL https://git.torproject.org/debian/tor.git - Stop shipping the old doc/website/ directory in the tarball. We changed the website format in late 2010, and what we shipped in 0.2.1.26 really wasn't that useful anyway. Changes in version 0.2.2.19-alpha - 2010-11-22 Yet another OpenSSL security patch broke its compatibility with Tor: Tor 0.2.2.19-alpha makes relays work with OpenSSL 0.9.8p and 1.0.0.b. o Major bugfixes: - Resolve an incompatibility with OpenSSL 0.9.8p and OpenSSL 1.0.0b: No longer set the tlsext_host_name extension on server SSL objects; but continue to set it on client SSL objects. Our goal in setting it was to imitate a browser, not a vhosting server. Fixes bug 2204; bugfix on 0.2.1.1-alpha. o Minor bugfixes: - Try harder not to exceed the maximum length of 50 KB when writing statistics to extra-info descriptors. This bug was triggered by very fast relays reporting exit-port, entry, and dirreq statistics. Reported by Olaf Selke. Bugfix on 0.2.2.1-alpha. Fixes bug 2183. - Publish a router descriptor even if generating an extra-info descriptor fails. Previously we would not publish a router descriptor without an extra-info descriptor; this can cause fast exit relays collecting exit-port statistics to drop from the consensus. Bugfix on 0.1.2.9-rc; fixes bug 2195. Changes in version 0.2.2.18-alpha - 2010-11-16 Tor 0.2.2.18-alpha fixes several crash bugs that have been nagging us lately, makes unpublished bridge relays able to detect their IP address, and fixes a wide variety of other bugs to get us much closer to a stable release. o Major bugfixes: - Do even more to reject (and not just ignore) annotations on router descriptors received anywhere but from the cache. Previously we would ignore such annotations at first, but cache them to disk anyway. Bugfix on 0.2.0.8-alpha. Found by piebeer. - Do not log messages to the controller while shrinking buffer freelists. Doing so would sometimes make the controller connection try to allocate a buffer chunk, which would mess up the internals of the freelist and cause an assertion failure. Fixes bug 1125; fixed by Robert Ransom. Bugfix on 0.2.0.16-alpha. - Learn our external IP address when we're a relay or bridge, even if we set PublishServerDescriptor to 0. Bugfix on 0.2.0.3-alpha, where we introduced bridge relays that don't need to publish to be useful. Fixes bug 2050. - Maintain separate TLS contexts and certificates for incoming and outgoing connections in bridge relays. Previously we would use the same TLS contexts and certs for incoming and outgoing connections. Bugfix on 0.2.0.3-alpha; addresses bug 988. - Maintain separate identity keys for incoming and outgoing TLS contexts in bridge relays. Previously we would use the same identity keys for incoming and outgoing TLS contexts. Bugfix on 0.2.0.3-alpha; addresses the other half of bug 988. - Avoid an assertion failure when we as an authority receive a duplicate upload of a router descriptor that we already have, but which we previously considered an obsolete descriptor. Fixes another case of bug 1776. Bugfix on 0.2.2.16-alpha. - Avoid a crash bug triggered by looking at a dangling pointer while setting the network status consensus. Found by Robert Ransom. Bugfix on 0.2.2.17-alpha. Fixes bug 2097. - Fix a logic error where servers that _didn't_ act as exits would try to keep their server lists more aggressively up to date than exits, when it was supposed to be the other way around. Bugfix on 0.2.2.17-alpha. o Minor bugfixes (on Tor 0.2.1.x and earlier): - When we're trying to guess whether we know our IP address as a relay, we would log various ways that we failed to guess our address, but never log that we ended up guessing it successfully. Now add a log line to help confused and anxious relay operators. Bugfix on 0.1.2.1-alpha; fixes bug 1534. - Bring the logic that gathers routerinfos and assesses the acceptability of circuits into line. This prevents a Tor OP from getting locked in a cycle of choosing its local OR as an exit for a path (due to a .exit request) and then rejecting the circuit because its OR is not listed yet. It also prevents Tor clients from using an OR running in the same instance as an exit (due to a .exit request) if the OR does not meet the same requirements expected of an OR running elsewhere. Fixes bug 1859; bugfix on 0.1.0.1-rc. - Correctly describe errors that occur when generating a TLS object. Previously we would attribute them to a failure while generating a TLS context. Patch by Robert Ransom. Bugfix on 0.1.0.4-rc; fixes bug 1994. - Enforce multiplicity rules when parsing annotations. Bugfix on 0.2.0.8-alpha. Found by piebeer. - Fix warnings that newer versions of autoconf produced during ./autogen.sh. These warnings appear to be harmless in our case, but they were extremely verbose. Fixes bug 2020. o Minor bugfixes (on Tor 0.2.2.x): - Enable protection of small arrays whenever we build with gcc hardening features, not only when also building with warnings enabled. Fixes bug 2031; bugfix on 0.2.2.14-alpha. Reported by keb. o Minor features: - Make hidden services work better in private Tor networks by not requiring any uptime to join the hidden service descriptor DHT. Implements ticket 2088. - Rate-limit the "your application is giving Tor only an IP address" warning. Addresses bug 2000; bugfix on 0.0.8pre2. - When AllowSingleHopExits is set, print a warning to explain to the relay operator why most clients are avoiding her relay. - Update to the November 1 2010 Maxmind GeoLite Country database. o Code simplifications and refactoring: - When we fixed bug 1038 we had to put in a restriction not to send RELAY_EARLY cells on rend circuits. This was necessary as long as relays using Tor 0.2.1.3-alpha through 0.2.1.18-alpha were active. Now remove this obsolete check. Resolves bug 2081. - Some options used different conventions for uppercasing of acronyms when comparing manpage and source. Fix those in favor of the manpage, as it makes sense to capitalize acronyms. - Remove the torrc.complete file. It hasn't been kept up to date and users will have better luck checking out the manpage. - Remove the obsolete "NoPublish" option; it has been flagged as obsolete and has produced a warning since 0.1.1.18-rc. - Remove everything related to building the expert bundle for OS X. It has confused many users, doesn't work right on OS X 10.6, and is hard to get rid of once installed. Resolves bug 1274. Changes in version 0.2.2.17-alpha - 2010-09-30 Tor 0.2.2.17-alpha introduces a feature to make it harder for clients to use one-hop circuits (which can put the exit relays at higher risk, plus unbalance the network); fixes a big bug in bandwidth accounting for relays that want to limit their monthly bandwidth use; fixes a big pile of bugs in how clients tolerate temporary network failure; and makes our adaptive circuit build timeout feature (which improves client performance if your network is fast while not breaking things if your network is slow) better handle bad networks. o Major features: - Exit relays now try harder to block exit attempts from unknown relays, to make it harder for people to use them as one-hop proxies a la tortunnel. Controlled by the refuseunknownexits consensus parameter (currently enabled), or you can override it on your relay with the RefuseUnknownExits torrc option. Resolves bug 1751. o Major bugfixes (0.2.1.x and earlier): - Fix a bug in bandwidth accounting that could make us use twice the intended bandwidth when our interval start changes due to daylight saving time. Now we tolerate skew in stored vs computed interval starts: if the start of the period changes by no more than 50% of the period's duration, we remember bytes that we transferred in the old period. Fixes bug 1511; bugfix on 0.0.9pre5. - Always search the Windows system directory for system DLLs, and nowhere else. Bugfix on 0.1.1.23; fixes bug 1954. - When you're using bridges and your network goes away and your bridges get marked as down, recover when you attempt a new socks connection (if the network is back), rather than waiting up to an hour to try fetching new descriptors for your bridges. Bugfix on 0.2.0.3-alpha; fixes bug 1981. o Major bugfixes (on 0.2.2.x): - Fix compilation on Windows. Bugfix on 0.2.2.16-alpha; related to bug 1797. - Fix a segfault that could happen when operating a bridge relay with no GeoIP database set. Fixes bug 1964; bugfix on 0.2.2.15-alpha. - The consensus bandwidth-weights (used by clients to choose fast relays) entered an unexpected edge case in September where Exits were much scarcer than Guards, resulting in bad weight recommendations. Now we compute them using new constraints that should succeed in all cases. Also alter directory authorities to not include the bandwidth-weights line if they fail to produce valid values. Fixes bug 1952; bugfix on 0.2.2.10-alpha. - When weighting bridges during path selection, we used to trust the bandwidths they provided in their descriptor, only capping them at 10MB/s. This turned out to be problematic for two reasons: Bridges could claim to handle a lot more traffic then they actually would, thus making more clients pick them and have a pretty effective DoS attack. The other issue is that new bridges that might not have a good estimate for their bw capacity yet would not get used at all unless no other bridges are available to a client. Fixes bug 1912; bugfix on 0.2.2.7-alpha. o Major bugfixes (on the circuit build timeout feature, 0.2.2.x): - Ignore cannibalized circuits when recording circuit build times. This should provide for a minor performance improvement for hidden service users using 0.2.2.14-alpha, and should remove two spurious notice log messages. Bugfix on 0.2.2.14-alpha; fixes bug 1740. - Simplify the logic that causes us to decide if the network is unavailable for purposes of recording circuit build times. If we receive no cells whatsoever for the entire duration of a circuit's full measured lifetime, the network is probably down. Also ignore one-hop directory fetching circuit timeouts when calculating our circuit build times. These changes should hopefully reduce the cases where we see ridiculous circuit build timeouts for people with spotty wireless connections. Fixes part of bug 1772; bugfix on 0.2.2.2-alpha. - Prevent the circuit build timeout from becoming larger than the maximum build time we have ever seen. Also, prevent the time period for measurement circuits from becoming larger than twice that value. Fixes the other part of bug 1772; bugfix on 0.2.2.2-alpha. o Minor features: - When we run out of directory information such that we can't build circuits, but then get enough that we can build circuits, log when we actually construct a circuit, so the user has a better chance of knowing what's going on. Fixes bug 1362. - Be more generous with how much bandwidth we'd use up (with accounting enabled) before entering "soft hibernation". Previously, we'd refuse new connections and circuits once we'd used up 95% of our allotment. Now, we use up 95% of our allotment, AND make sure that we have no more than 500MB (or 3 hours of expected traffic, whichever is lower) remaining before we enter soft hibernation. - If we've configured EntryNodes and our network goes away and/or all our entrynodes get marked down, optimistically retry them all when a new socks application request appears. Fixes bug 1882. - Add some more defensive programming for architectures that can't handle unaligned integer accesses. We don't know of any actual bugs right now, but that's the best time to fix them. Fixes bug 1943. - Support line continuations in the torrc config file. If a line ends with a single backslash character, the newline is ignored, and the configuration value is treated as continuing on the next line. Resolves bug 1929. o Minor bugfixes (on 0.2.1.x and earlier): - For bandwidth accounting, calculate our expected bandwidth rate based on the time during which we were active and not in soft-hibernation during the last interval. Previously, we were also considering the time spent in soft-hibernation. If this was a long time, we would wind up underestimating our bandwidth by a lot, and skewing our wakeup time towards the start of the accounting interval. Fixes bug 1789. Bugfix on 0.0.9pre5. o Minor bugfixes (on 0.2.2.x): - Resume generating CIRC FAILED REASON=TIMEOUT control port messages, which were disabled by the circuit build timeout changes in 0.2.2.14-alpha. Bugfix on 0.2.2.14-alpha; fixes bug 1739. - Make sure we don't warn about missing bandwidth weights when choosing bridges or other relays not in the consensus. Bugfix on 0.2.2.10-alpha; fixes bug 1805. - In our logs, do not double-report signatures from unrecognized authorities both as "from unknown authority" and "not present". Fixes bug 1956, bugfix on 0.2.2.16-alpha. Changes in version 0.2.2.16-alpha - 2010-09-17 Tor 0.2.2.16-alpha fixes a variety of old stream fairness bugs (most evident at exit relays), and also continues to resolve all the little bugs that have been filling up trac lately. o Major bugfixes (stream-level fairness): - When receiving a circuit-level SENDME for a blocked circuit, try to package cells fairly from all the streams that had previously been blocked on that circuit. Previously, we had started with the oldest stream, and allowed each stream to potentially exhaust the circuit's package window. This gave older streams on any given circuit priority over newer ones. Fixes bug 1937. Detected originally by Camilo Viecco. This bug was introduced before the first Tor release, in svn commit r152: it is the new winner of the longest-lived bug prize. - When the exit relay got a circuit-level sendme cell, it started reading on the exit streams, even if had 500 cells queued in the circuit queue already, so the circuit queue just grew and grew in some cases. We fix this by not re-enabling reading on receipt of a sendme cell when the cell queue is blocked. Fixes bug 1653. Bugfix on 0.2.0.1-alpha. Detected by Mashael AlSabah. Original patch by "yetonetime". - Newly created streams were allowed to read cells onto circuits, even if the circuit's cell queue was blocked and waiting to drain. This created potential unfairness, as older streams would be blocked, but newer streams would gladly fill the queue completely. We add code to detect this situation and prevent any stream from getting more than one free cell. Bugfix on 0.2.0.1-alpha. Partially fixes bug 1298. o Minor features: - Update to the September 1 2010 Maxmind GeoLite Country database. - Warn when CookieAuthFileGroupReadable is set but CookieAuthFile is not. This would lead to a cookie that is still not group readable. Closes bug 1843. Suggested by katmagic. - When logging a rate-limited warning, we now mention how many messages got suppressed since the last warning. - Add new "perconnbwrate" and "perconnbwburst" consensus params to do individual connection-level rate limiting of clients. The torrc config options with the same names trump the consensus params, if both are present. Replaces the old "bwconnrate" and "bwconnburst" consensus params which were broken from 0.2.2.7-alpha through 0.2.2.14-alpha. Closes bug 1947. - When a router changes IP address or port, authorities now launch a new reachability test for it. Implements ticket 1899. - Make the formerly ugly "2 unknown, 7 missing key, 0 good, 0 bad, 2 no signature, 4 required" messages about consensus signatures easier to read, and make sure they get logged at the same severity as the messages explaining which keys are which. Fixes bug 1290. - Don't warn when we have a consensus that we can't verify because of missing certificates, unless those certificates are ones that we have been trying and failing to download. Fixes bug 1145. - If you configure your bridge with a known identity fingerprint, and the bridge authority is unreachable (as it is in at least one country now), fall back to directly requesting the descriptor from the bridge. Finishes the feature started in 0.2.0.10-alpha; closes bug 1138. - When building with --enable-gcc-warnings on OpenBSD, disable warnings in system headers. This makes --enable-gcc-warnings pass on OpenBSD 4.8. o Minor bugfixes (on 0.2.1.x and earlier): - Authorities will now attempt to download consensuses if their own efforts to make a live consensus have failed. This change means authorities that restart will fetch a valid consensus, and it means authorities that didn't agree with the current consensus will still fetch and serve it if it has enough signatures. Bugfix on 0.2.0.9-alpha; fixes bug 1300. - Ensure DNS requests launched by "RESOLVE" commands from the controller respect the __LeaveStreamsUnattached setconf options. The same goes for requests launched via DNSPort or transparent proxying. Bugfix on 0.2.0.1-alpha; fixes bug 1525. - Allow handshaking OR connections to take a full KeepalivePeriod seconds to handshake. Previously, we would close them after IDLE_OR_CONN_TIMEOUT (180) seconds, the same timeout as if they were open. Bugfix on 0.2.1.26; fixes bug 1840. Thanks to mingw-san for analysis help. - Rate-limit "Failed to hand off onionskin" warnings. - Never relay a cell for a circuit we have already destroyed. Between marking a circuit as closeable and finally closing it, it may have been possible for a few queued cells to get relayed, even though they would have been immediately dropped by the next OR in the circuit. Fixes bug 1184; bugfix on 0.2.0.1-alpha. - Never queue a cell for a circuit that's already been marked for close. - Never vote for a server as "Running" if we have a descriptor for it claiming to be hibernating, and that descriptor was published more recently than our last contact with the server. Bugfix on 0.2.0.3-alpha; fixes bug 911. - Squash a compile warning on OpenBSD. Reported by Tas; fixes bug 1848. o Minor bugfixes (on 0.2.2.x): - Fix a regression introduced in 0.2.2.7-alpha that marked relays down if a directory fetch fails and you've configured either bridges or EntryNodes. The intent was to mark the relay as down _unless_ you're using bridges or EntryNodes, since if you are then you could quickly run out of entry points. - Fix the Windows directory-listing code. A bug introduced in 0.2.2.14-alpha could make Windows directory servers forget to load some of their cached v2 networkstatus files. - Really allow clients to use relays as bridges. Fixes bug 1776; bugfix on 0.2.2.15-alpha. - Demote a warn to info that happens when the CellStatistics option was just enabled. Bugfix on 0.2.2.15-alpha; fixes bug 1921. Reported by Moritz Bartl. - On Windows, build correctly either with or without Unicode support. This is necessary so that Tor can support fringe platforms like Windows 98 (which has no Unicode), or Windows CE (which has no non-Unicode). Bugfix on 0.2.2.14-alpha; fixes bug 1797. o Testing - Add a unit test for cross-platform directory-listing code. Changes in version 0.2.2.15-alpha - 2010-08-18 Tor 0.2.2.15-alpha fixes a big bug in hidden service availability, fixes a variety of other bugs that were preventing performance experiments from moving forward, fixes several bothersome memory leaks, and generally closes a lot of smaller bugs that have been filling up trac lately. o Major bugfixes: - Stop assigning the HSDir flag to relays that disable their DirPort (and thus will refuse to answer directory requests). This fix should dramatically improve the reachability of hidden services: hidden services and hidden service clients pick six HSDir relays to store and retrieve the hidden service descriptor, and currently about half of the HSDir relays will refuse to work. Bugfix on 0.2.0.10-alpha; fixes part of bug 1693. - The PerConnBWRate and Burst config options, along with the bwconnrate and bwconnburst consensus params, initialized each conn's token bucket values only when the connection is established. Now we update them if the config options change, and update them every time we get a new consensus. Otherwise we can encounter an ugly edge case where we initialize an OR conn to client-level bandwidth, but then later the relay joins the consensus and we leave it throttled. Bugfix on 0.2.2.7-alpha; fixes bug 1830. - Fix a regression that caused Tor to rebind its ports if it receives SIGHUP while hibernating. Bugfix in 0.1.1.6-alpha; closes bug 919. o Major features: - Lower the maximum weighted-fractional-uptime cutoff to 98%. This should give us approximately 40-50% more Guard-flagged nodes, improving the anonymity the Tor network can provide and also decreasing the dropoff in throughput that relays experience when they first get the Guard flag. - Allow enabling or disabling the *Statistics config options while Tor is running. o Minor features: - Update to the August 1 2010 Maxmind GeoLite Country database. - Have the controller interface give a more useful message than "Internal Error" in response to failed GETINFO requests. - Warn when the same option is provided more than once in a torrc file, on the command line, or in a single SETCONF statement, and the option is one that only accepts a single line. Closes bug 1384. - Build correctly on mingw with more recent versions of OpenSSL 0.9.8. Patch from mingw-san. - Add support for the country code "{??}" in torrc options like ExcludeNodes, to indicate all routers of unknown country. Closes bug 1094. - Relays report the number of bytes spent on answering directory requests in extra-info descriptors similar to {read,write}-history. Implements enhancement 1790. o Minor bugfixes (on 0.2.1.x and earlier): - Complain if PublishServerDescriptor is given multiple arguments that include 0 or 1. This configuration will be rejected in the future. Bugfix on 0.2.0.1-alpha; closes bug 1107. - Disallow BridgeRelay 1 and ORPort 0 at once in the configuration. Bugfix on 0.2.0.13-alpha; closes bug 928. - Change "Application request when we're believed to be offline." notice to "Application request when we haven't used client functionality lately.", to clarify that it's not an error. Bugfix on 0.0.9.3; fixes bug 1222. - Fix a bug in the controller interface where "GETINFO ns/asdaskljkl" would return "551 Internal error" rather than "552 Unrecognized key ns/asdaskljkl". Bugfix on 0.1.2.3-alpha. - Users can't configure a regular relay to be their bridge. It didn't work because when Tor fetched the bridge descriptor, it found that it already had it, and didn't realize that the purpose of the descriptor had changed. Now we replace routers with a purpose other than bridge with bridge descriptors when fetching them. Bugfix on 0.1.1.9-alpha. Bug 1776 not yet fixed because now we immediately refetch the descriptor with router purpose 'general', disabling it as a bridge. - Fix a rare bug in rend_fn unit tests: we would fail a test when a randomly generated port is 0. Diagnosed by Matt Edman. Bugfix on 0.2.0.10-alpha; fixes bug 1808. - Exit nodes didn't recognize EHOSTUNREACH as a plausible error code, and so sent back END_STREAM_REASON_MISC. Clients now recognize a new stream ending reason for this case: END_STREAM_REASON_NOROUTE. Servers can start sending this code when enough clients recognize it. Also update the spec to reflect this new reason. Bugfix on 0.1.0.1-rc; fixes part of bug 1793. - Delay geoip stats collection by bridges for 6 hours, not 2 hours, when we switch from being a public relay to a bridge. Otherwise there will still be clients that see the relay in their consensus, and the stats will end up wrong. Bugfix on 0.2.1.15-rc; fixes bug 932 even more. - Instead of giving an assertion failure on an internal mismatch on estimated freelist size, just log a BUG warning and try later. Mitigates but does not fix bug 1125. - Fix an assertion failure that could occur in caches or bridge users when using a very short voting interval on a testing network. Diagnosed by Robert Hogan. Fixes bug 1141; bugfix on 0.2.0.8-alpha. o Minor bugfixes (on 0.2.2.x): - Alter directory authorities to always consider Exit-flagged nodes as potential Guard nodes in their votes. The actual decision to use Exits as Guards is done in the consensus bandwidth weights. Fixes bug 1294; bugfix on 0.2.2.10-alpha. - When the controller is reporting the purpose of circuits that didn't finish building before the circuit build timeout, it was printing UNKNOWN_13. Now print EXPIRED. Bugfix on 0.2.2.14-alpha. - Our libevent version parsing code couldn't handle versions like 1.4.14b-stable and incorrectly warned the user about using an old and broken version of libevent. Treat 1.4.14b-stable like 1.4.14-stable when parsing the version. Fixes bug 1731; bugfix on 0.2.2.1-alpha. - Don't use substitution references like $(VAR:MOD) when $(asciidoc_files) is empty -- make(1) on NetBSD transforms '$(:x)' to 'x' rather than the empty string. This bites us in doc/ when configured with --disable-asciidoc. Bugfix on 0.2.2.9-alpha; fixes bug 1773. - Remove a spurious hidden service server-side log notice about "Ancient non-dirty circuits". Bugfix on 0.2.2.14-alpha; fixes bug 1741. - Fix compilation with --with-dmalloc set. Bugfix on 0.2.2.6-alpha; fixes bug 1832. - Correctly report written bytes on linked connections. Found while implementing 1790. Bugfix on 0.2.2.4-alpha. - Fix three memory leaks: one in circuit_build_times_parse_state(), one in dirvote_add_signatures_to_pending_consensus(), and one every time we parse a v3 network consensus. Bugfixes on 0.2.2.14-alpha, 0.2.2.6-alpha, and 0.2.2.10-alpha respectively; fixes bug 1831. o Code simplifications and refactoring: - Take a first step towards making or.h smaller by splitting out function definitions for all source files in src/or/. Leave structures and defines in or.h for now. - Remove a bunch of unused function declarations as well as a block of #if 0'd code from the unit tests. Closes bug 1824. - New unit tests for exit-port history statistics; refactored exit statistics code to be more easily tested. - Remove the old debian/ directory from the main Tor distribution. The official Tor-for-debian git repository lives at the URL https://git.torproject.org/debian/tor.git Changes in version 0.2.2.14-alpha - 2010-07-12 Tor 0.2.2.14-alpha greatly improves client-side handling of circuit build timeouts, which are used to estimate speed and improve performance. We also move to a much better GeoIP database, port Tor to Windows CE, introduce new compile flags that improve code security, add an eighth v3 directory authority, and address a lot of more minor issues. o Major bugfixes: - Tor directory authorities no longer crash when started with a cached-microdesc-consensus file in their data directory. Bugfix on 0.2.2.6-alpha; fixes bug 1532. - Treat an unset $HOME like an empty $HOME rather than triggering an assert. Bugfix on 0.0.8pre1; fixes bug 1522. - Ignore negative and large circuit build timeout values that can happen during a suspend or hibernate. These values caused various asserts to fire. Bugfix on 0.2.2.2-alpha; fixes bug 1245. - Alter calculation of Pareto distribution parameter 'Xm' for Circuit Build Timeout learning to use the weighted average of the top N=3 modes (because we have three entry guards). Considering multiple modes should improve the timeout calculation in some cases, and prevent extremely high timeout values. Bugfix on 0.2.2.2-alpha; fixes bug 1335. - Alter calculation of Pareto distribution parameter 'Alpha' to use a right censored distribution model. This approach improves over the synthetic timeout generation approach that was producing insanely high timeout values. Now we calculate build timeouts using truncated times. Bugfix on 0.2.2.2-alpha; fixes bugs 1245 and 1335. - Do not close circuits that are under construction when they reach the circuit build timeout. Instead, leave them building (but do not use them) for up until the time corresponding to the 95th percentile on the Pareto CDF or 60 seconds, whichever is greater. This is done to provide better data for the new Pareto model. This percentile can be controlled by the consensus. o Major features: - Move to the June 2010 Maxmind GeoLite country db (rather than the June 2009 ip-to-country GeoIP db) for our statistics that count how many users relays are seeing from each country. Now we have more accurate data for many African countries. - Port Tor to build and run correctly on Windows CE systems, using the wcecompat library. Contributed by Valerio Lupi. - New "--enable-gcc-hardening" ./configure flag (off by default) to turn on gcc compile time hardening options. It ensures that signed ints have defined behavior (-fwrapv), enables -D_FORTIFY_SOURCE=2 (requiring -O2), adds stack smashing protection with canaries (-fstack-protector-all), turns on ASLR protection if supported by the kernel (-fPIE, -pie), and adds additional security related warnings. Verified to work on Mac OS X and Debian Lenny. - New "--enable-linker-hardening" ./configure flag (off by default) to turn on ELF specific hardening features (relro, now). This does not work with Mac OS X or any other non-ELF binary format. o New directory authorities: - Set up maatuska (run by Linus Nordberg) as the eighth v3 directory authority. o Minor features: - New config option "WarnUnsafeSocks 0" disables the warning that occurs whenever Tor receives a socks handshake using a version of the socks protocol that can only provide an IP address (rather than a hostname). Setups that do DNS locally over Tor are fine, and we shouldn't spam the logs in that case. - Convert the HACKING file to asciidoc, and add a few new sections to it, explaining how we use Git, how we make changelogs, and what should go in a patch. - Add a TIMEOUT_RATE keyword to the BUILDTIMEOUT_SET control port event, to give information on the current rate of circuit timeouts over our stored history. - Add ability to disable circuit build time learning via consensus parameter and via a LearnCircuitBuildTimeout config option. Also automatically disable circuit build time calculation if we are either a AuthoritativeDirectory, or if we fail to write our state file. Fixes bug 1296. - More gracefully handle corrupt state files, removing asserts in favor of saving a backup and resetting state. - Rename the "log.h" header to "torlog.h" so as to conflict with fewer system headers. o Minor bugfixes: - Build correctly on OSX with zlib 1.2.4 and higher with all warnings enabled. - When a2x fails, mention that the user could disable manpages instead of trying to fix their asciidoc installation. - Where available, use Libevent 2.0's periodic timers so that our once-per-second cleanup code gets called even more closely to once per second than it would otherwise. Fixes bug 943. - If you run a bridge that listens on multiple IP addresses, and some user configures a bridge address that uses a different IP address than your bridge writes in its router descriptor, and the user doesn't specify an identity key, their Tor would discard the descriptor because "it isn't one of our configured bridges", and fail to bootstrap. Now believe the descriptor and bootstrap anyway. Bugfix on 0.2.0.3-alpha. - If OpenSSL fails to make a duplicate of a private or public key, log an error message and try to exit cleanly. May help with debugging if bug 1209 ever remanifests. - Save a couple bytes in memory allocation every time we escape certain characters in a string. Patch from Florian Zumbiehl. - Make it explicit that we don't cannibalize one-hop circuits. This happens in the wild, but doesn't turn out to be a problem because we fortunately don't use those circuits. Many thanks to outofwords for the initial analysis and to swissknife who confirmed that two-hop circuits are actually created. - Make directory mirrors report non-zero dirreq-v[23]-shares again. Fixes bug 1564; bugfix on 0.2.2.9-alpha. - Eliminate a case where a circuit build time warning was displayed after network connectivity resumed. Bugfix on 0.2.2.2-alpha. Changes in version 0.2.1.26 - 2010-05-02 Tor 0.2.1.26 addresses the recent connection and memory overload problems we've been seeing on relays, especially relays with their DirPort open. If your relay has been crashing, or you turned it off because it used too many resources, give this release a try. This release also fixes yet another instance of broken OpenSSL libraries that was causing some relays to drop out of the consensus. o Major bugfixes: - Teach relays to defend themselves from connection overload. Relays now close idle circuits early if it looks like they were intended for directory fetches. Relays are also more aggressive about closing TLS connections that have no circuits on them. Such circuits are unlikely to be re-used, and tens of thousands of them were piling up at the fast relays, causing the relays to run out of sockets and memory. Bugfix on 0.2.0.22-rc (where clients started tunneling their directory fetches over TLS). - Fix SSL renegotiation behavior on OpenSSL versions like on Centos that claim to be earlier than 0.9.8m, but which have in reality backported huge swaths of 0.9.8m or 0.9.8n renegotiation behavior. Possible fix for some cases of bug 1346. - Directory mirrors were fetching relay descriptors only from v2 directory authorities, rather than v3 authorities like they should. Only 2 v2 authorities remain (compared to 7 v3 authorities), leading to a serious bottleneck. Bugfix on 0.2.0.9-alpha. Fixes bug 1324. o Minor bugfixes: - Finally get rid of the deprecated and now harmful notion of "clique mode", where directory authorities maintain TLS connections to every other relay. o Testsuite fixes: - In the util/threads test, no longer free the test_mutex before all worker threads have finished. Bugfix on 0.2.1.6-alpha. - The master thread could starve the worker threads quite badly on certain systems, causing them to run only partially in the allowed window. This resulted in test failures. Now the master thread sleeps occasionally for a few microseconds while the two worker-threads compete for the mutex. Bugfix on 0.2.0.1-alpha. Changes in version 0.2.2.13-alpha - 2010-04-24 Tor 0.2.2.13-alpha addresses the recent connection and memory overload problems we've been seeing on relays, especially relays with their DirPort open. If your relay has been crashing, or you turned it off because it used too many resources, give this release a try. o Major bugfixes: - Teach relays to defend themselves from connection overload. Relays now close idle circuits early if it looks like they were intended for directory fetches. Relays are also more aggressive about closing TLS connections that have no circuits on them. Such circuits are unlikely to be re-used, and tens of thousands of them were piling up at the fast relays, causing the relays to run out of sockets and memory. Bugfix on 0.2.0.22-rc (where clients started tunneling their directory fetches over TLS). o Minor features: - Finally get rid of the deprecated and now harmful notion of "clique mode", where directory authorities maintain TLS connections to every other relay. - Directory authorities now do an immediate reachability check as soon as they hear about a new relay. This change should slightly reduce the time between setting up a relay and getting listed as running in the consensus. It should also improve the time between setting up a bridge and seeing use by bridge users. - Directory authorities no longer launch a TLS connection to every relay as they startup. Now that we have 2k+ descriptors cached, the resulting network hiccup is becoming a burden. Besides, authorities already avoid voting about Running for the first half hour of their uptime. Changes in version 0.2.2.12-alpha - 2010-04-20 Tor 0.2.2.12-alpha fixes a critical bug in how directory authorities handle and vote on descriptors. It was causing relays to drop out of the consensus. o Major bugfixes: - Many relays have been falling out of the consensus lately because not enough authorities know about their descriptor for them to get a majority of votes. When we deprecated the v2 directory protocol, we got rid of the only way that v3 authorities can hear from each other about other descriptors. Now authorities examine every v3 vote for new descriptors, and fetch them from that authority. Bugfix on 0.2.1.23. - Fix two typos in tor_vasprintf() that broke the compile on Windows, and a warning in or.h related to bandwidth_weight_rule_t that prevented clean compile on OS X. Fixes bug 1363; bugfix on 0.2.2.11-alpha. - Fix a segfault on relays when DirReqStatistics is enabled and 24 hours pass. Bug found by keb. Fixes bug 1365; bugfix on 0.2.2.11-alpha. o Minor bugfixes: - Demote a confusing TLS warning that relay operators might get when someone tries to talk to their OrPort. It is neither the operator's fault nor can they do anything about it. Fixes bug 1364; bugfix on 0.2.0.14-alpha. Changes in version 0.2.2.11-alpha - 2010-04-15 Tor 0.2.2.11-alpha fixes yet another instance of broken OpenSSL libraries that was causing some relays to drop out of the consensus. o Major bugfixes: - Directory mirrors were fetching relay descriptors only from v2 directory authorities, rather than v3 authorities like they should. Only 2 v2 authorities remain (compared to 7 v3 authorities), leading to a serious bottleneck. Bugfix on 0.2.0.9-alpha. Fixes bug 1324. - Fix a parsing error that made every possible value of CircPriorityHalflifeMsec get treated as "1 msec". Bugfix on 0.2.2.7-alpha. Rename CircPriorityHalflifeMsec to CircuitPriorityHalflifeMsec, so authorities can tell newer relays about the option without breaking older ones. - Fix SSL renegotiation behavior on OpenSSL versions like on Centos that claim to be earlier than 0.9.8m, but which have in reality backported huge swaths of 0.9.8m or 0.9.8n renegotiation behavior. Possible fix for some cases of bug 1346. o Minor features: - Experiment with a more aggressive approach to preventing clients from making one-hop exit streams. Exit relays who want to try it out can set "RefuseUnknownExits 1" in their torrc, and then look for "Attempt by %s to open a stream" log messages. Let us know how it goes! - Add support for statically linking zlib by specifying --enable-static-zlib, to go with our support for statically linking openssl and libevent. Resolves bug 1358. o Minor bugfixes: - Fix a segfault that happens whenever a Tor client that is using libevent2's bufferevents gets a hup signal. Bugfix on 0.2.2.5-alpha; fixes bug 1341. - When we cleaned up the contrib/tor-exit-notice.html file, we left out the first line. Fixes bug 1295. - When building the manpage from a tarball, we required asciidoc, but the asciidoc -> roff/html conversion was already done for the tarball. Make 'make' complain only when we need asciidoc (either because we're compiling directly from git, or because we altered the asciidoc manpage in the tarball). Bugfix on 0.2.2.9-alpha. - When none of the directory authorities vote on any params, Tor segfaulted when trying to make the consensus from the votes. We didn't trigger the bug in practice, because authorities do include params in their votes. Bugfix on 0.2.2.10-alpha; fixes bug 1322. o Testsuite fixes: - In the util/threads test, no longer free the test_mutex before all worker threads have finished. Bugfix on 0.2.1.6-alpha. - The master thread could starve the worker threads quite badly on certain systems, causing them to run only partially in the allowed window. This resulted in test failures. Now the master thread sleeps occasionally for a few microseconds while the two worker-threads compete for the mutex. Bugfix on 0.2.0.1-alpha. Changes in version 0.2.2.10-alpha - 2010-03-07 Tor 0.2.2.10-alpha fixes a regression introduced in 0.2.2.9-alpha that could prevent relays from guessing their IP address correctly. It also starts the groundwork for another client-side performance boost, since currently we're not making efficient use of relays that have both the Guard flag and the Exit flag. o Major bugfixes: - Fix a regression from our patch for bug 1244 that caused relays to guess their IP address incorrectly if they didn't set Address in their torrc and/or their address fails to resolve. Bugfix on 0.2.2.9-alpha; fixes bug 1269. o Major features (performance): - Directory authorities now compute consensus weightings that instruct clients how to weight relays flagged as Guard, Exit, Guard+Exit, and no flag. Clients that use these weightings will distribute network load more evenly across these different relay types. The weightings are in the consensus so we can change them globally in the future. Extra thanks to "outofwords" for finding some nasty security bugs in the first implementation of this feature. o Minor features (performance): - Always perform router selections using weighted relay bandwidth, even if we don't need a high capacity circuit at the time. Non-fast circuits now only differ from fast ones in that they can use relays not marked with the Fast flag. This "feature" could turn out to be a horrible bug; we should investigate more before it goes into a stable release. o Minor features: - Allow disabling building of the manpages. Skipping the manpage speeds up the build considerably. o Minor bugfixes (on 0.2.2.x): - Fix a memleak in the EXTENDCIRCUIT logic. Spotted by coverity. Bugfix on 0.2.2.9-alpha. - Disallow values larger than INT32_MAX for PerConnBWRate|Burst config option. Bugfix on 0.2.2.7-alpha. - Ship the asciidoc-helper file in the tarball, so that people can build from source if they want to, and touching the .1.txt files doesn't break the build. Bugfix on 0.2.2.9-alpha. o Minor bugfixes (on 0.2.1.x or earlier): - Fix a dereference-then-NULL-check sequence when publishing descriptors. Bugfix on 0.2.1.5-alpha. Discovered by ekir; fixes bug 1255. - Fix another dereference-then-NULL-check sequence. Bugfix on 0.2.1.14-rc. Discovered by ekir; fixes bug 1256. - Make sure we treat potentially not NUL-terminated strings correctly. Bugfix on 0.1.1.13-alpha. Discovered by rieo; fixes bug 1257. o Code simplifications and refactoring: - Fix some urls in the exit notice file and make it XHTML1.1 strict compliant. Based on a patch from Christian Kujau. - Don't use sed in asciidoc-helper anymore. - Make the build process fail if asciidoc cannot be found and building with asciidoc isn't disabled. Changes in version 0.2.2.9-alpha - 2010-02-22 Tor 0.2.2.9-alpha makes Tor work again on the latest OS X, updates the location of a directory authority, and cleans up a bunch of small bugs. o Directory authority changes: - Change IP address for dannenberg (v3 directory authority), and remove moria2 (obsolete v1, v2 directory authority and v0 hidden service directory authority) from the list. o Major bugfixes: - Make Tor work again on the latest OS X: when deciding whether to use strange flags to turn TLS renegotiation on, detect the OpenSSL version at run-time, not compile time. We need to do this because Apple doesn't update its dev-tools headers when it updates its libraries in a security patch. - Fix a potential buffer overflow in lookup_last_hid_serv_request() that could happen on 32-bit platforms with 64-bit time_t. Also fix a memory leak when requesting a hidden service descriptor we've requested before. Fixes bug 1242, bugfix on 0.2.0.18-alpha. Found by aakova. - Authorities could be tricked into giving out the Exit flag to relays that didn't allow exiting to any ports. This bug could screw with load balancing and stats. Bugfix on 0.1.1.6-alpha; fixes bug 1238. Bug discovered by Martin Kowalczyk. - When freeing a session key, zero it out completely. We only zeroed the first ptrsize bytes. Bugfix on 0.0.2pre8. Discovered and patched by ekir. Fixes bug 1254. o Minor bugfixes: - Fix static compilation by listing the openssl libraries in the right order. Bugfix on Tor 0.2.2.8-alpha; fixes bug 1237. - Resume handling .exit hostnames in a special way: originally we stripped the .exit part and used the requested exit relay. In 0.2.2.1-alpha we stopped treating them in any special way, meaning if you use a .exit address then Tor will pass it on to the exit relay. Now we reject the .exit stream outright, since that behavior might be more expected by the user. Found and diagnosed by Scott Bennett and Downie on or-talk. - Don't spam the controller with events when we have no file descriptors available. Bugfix on 0.2.1.5-alpha. (Rate-limiting for log messages was already solved from bug 748.) - Avoid a bogus overlapped memcpy in tor_addr_copy(). Reported by "memcpyfail". - Make the DNSPort option work with libevent 2.x. Don't alter the behaviour for libevent 1.x. Fixes bug 1143. Found by SwissTorExit. - Emit a GUARD DROPPED controller event for a case we missed. - Make more fields in the controller protocol case-insensitive, since control-spec.txt said they were. - Refactor resolve_my_address() to not use gethostbyname() anymore. Fixes bug 1244; bugfix on 0.0.2pre25. Reported by Mike Mestnik. - Fix a spec conformance issue: the network-status-version token must be the first token in a v3 consensus or vote. Discovered by parakeep. Bugfix on 0.2.0.3-alpha. o Code simplifications and refactoring: - Generate our manpage and HTML documentation using Asciidoc. This change should make it easier to maintain the documentation, and produce nicer HTML. - Remove the --enable-iphone option. According to reports from Marco Bonetti, Tor builds fine without any special tweaking on recent iPhone SDK versions. - Removed some unnecessary files from the source distribution. The AUTHORS file has now been merged into the people page on the website. The roadmaps and design doc can now be found in the projects directory in svn. - Enabled various circuit build timeout constants to be controlled by consensus parameters. Also set better defaults for these parameters based on experimentation on broadband and simulated high latency links. o Minor features: - The 'EXTENDCIRCUIT' control port command can now be used with a circ id of 0 and no path. This feature will cause Tor to build a new 'fast' general purpose circuit using its own path selection algorithms. - Added a BUILDTIMEOUT_SET controller event to describe changes to the circuit build timeout. - Future-proof the controller protocol a bit by ignoring keyword arguments we do not recognize. - Expand homedirs passed to tor-checkkey. This should silence a coverity complaint about passing a user-supplied string into open() without checking it. Changes in version 0.2.1.25 - 2010-03-16 Tor 0.2.1.25 fixes a regression introduced in 0.2.1.23 that could prevent relays from guessing their IP address correctly. It also fixes several minor potential security bugs. o Major bugfixes: - Fix a regression from our patch for bug 1244 that caused relays to guess their IP address incorrectly if they didn't set Address in their torrc and/or their address fails to resolve. Bugfix on 0.2.1.23; fixes bug 1269. - When freeing a session key, zero it out completely. We only zeroed the first ptrsize bytes. Bugfix on 0.0.2pre8. Discovered and patched by ekir. Fixes bug 1254. o Minor bugfixes: - Fix a dereference-then-NULL-check sequence when publishing descriptors. Bugfix on 0.2.1.5-alpha. Discovered by ekir; fixes bug 1255. - Fix another dereference-then-NULL-check sequence. Bugfix on 0.2.1.14-rc. Discovered by ekir; fixes bug 1256. - Make sure we treat potentially not NUL-terminated strings correctly. Bugfix on 0.1.1.13-alpha. Discovered by rieo; fixes bug 1257. Changes in version 0.2.1.24 - 2010-02-21 Tor 0.2.1.24 makes Tor work again on the latest OS X -- this time for sure! o Minor bugfixes: - Work correctly out-of-the-box with even more vendor-patched versions of OpenSSL. In particular, make it so Debian and OS X don't need customized patches to run/build. Changes in version 0.2.1.23 - 2010-02-13 Tor 0.2.1.23 fixes a huge client-side performance bug, makes Tor work again on the latest OS X, and updates the location of a directory authority. o Major bugfixes (performance): - We were selecting our guards uniformly at random, and then weighting which of our guards we'd use uniformly at random. This imbalance meant that Tor clients were severely limited on throughput (and probably latency too) by the first hop in their circuit. Now we select guards weighted by currently advertised bandwidth. We also automatically discard guards picked using the old algorithm. Fixes bug 1217; bugfix on 0.2.1.3-alpha. Found by Mike Perry. o Major bugfixes: - Make Tor work again on the latest OS X: when deciding whether to use strange flags to turn TLS renegotiation on, detect the OpenSSL version at run-time, not compile time. We need to do this because Apple doesn't update its dev-tools headers when it updates its libraries in a security patch. - Fix a potential buffer overflow in lookup_last_hid_serv_request() that could happen on 32-bit platforms with 64-bit time_t. Also fix a memory leak when requesting a hidden service descriptor we've requested before. Fixes bug 1242, bugfix on 0.2.0.18-alpha. Found by aakova. o Directory authority changes: - Change IP address for dannenberg (v3 directory authority), and remove moria2 (obsolete v1, v2 directory authority and v0 hidden service directory authority) from the list. o Minor bugfixes: - Refactor resolve_my_address() to not use gethostbyname() anymore. Fixes bug 1244; bugfix on 0.0.2pre25. Reported by Mike Mestnik. o Minor features: - Avoid a mad rush at the beginning of each month when each client rotates half of its guards. Instead we spread the rotation out throughout the month, but we still avoid leaving a precise timestamp in the state file about when we first picked the guard. Improves over the behavior introduced in 0.1.2.17. Changes in version 0.2.2.8-alpha - 2010-01-26 Tor 0.2.2.8-alpha fixes a crash bug in 0.2.2.7-alpha that has been causing bridge relays to disappear. If you're running a bridge, please upgrade. o Major bugfixes: - Fix a memory corruption bug on bridges that occured during the inclusion of stats data in extra-info descriptors. Also fix the interface for geoip_get_bridge_stats* to prevent similar bugs in the future. Diagnosis by Tas, patch by Karsten and Sebastian. Fixes bug 1208; bugfix on 0.2.2.7-alpha. o Minor bugfixes: - Ignore OutboundBindAddress when connecting to localhost. Connections to localhost need to come _from_ localhost, or else local servers (like DNS and outgoing HTTP/SOCKS proxies) will often refuse to listen. Changes in version 0.2.2.7-alpha - 2010-01-19 Tor 0.2.2.7-alpha fixes a huge client-side performance bug, as well as laying the groundwork for further relay-side performance fixes. It also starts cleaning up client behavior with respect to the EntryNodes, ExitNodes, and StrictNodes config options. This release also rotates two directory authority keys, due to a security breach of some of the Torproject servers. o Directory authority changes: - Rotate keys (both v3 identity and relay identity) for moria1 and gabelmoo. o Major features (performance): - We were selecting our guards uniformly at random, and then weighting which of our guards we'd use uniformly at random. This imbalance meant that Tor clients were severely limited on throughput (and probably latency too) by the first hop in their circuit. Now we select guards weighted by currently advertised bandwidth. We also automatically discard guards picked using the old algorithm. Fixes bug 1217; bugfix on 0.2.1.3-alpha. Found by Mike Perry. - When choosing which cells to relay first, relays can now favor circuits that have been quiet recently, to provide lower latency for low-volume circuits. By default, relays enable or disable this feature based on a setting in the consensus. You can override this default by using the new "CircuitPriorityHalflife" config option. Design and code by Ian Goldberg, Can Tang, and Chris Alexander. - Add separate per-conn write limiting to go with the per-conn read limiting. We added a global write limit in Tor 0.1.2.5-alpha, but never per-conn write limits. - New consensus params "bwconnrate" and "bwconnburst" to let us rate-limit client connections as they enter the network. It's controlled in the consensus so we can turn it on and off for experiments. It's starting out off. Based on proposal 163. o Major features (relay selection options): - Switch to a StrictNodes config option, rather than the previous "StrictEntryNodes" / "StrictExitNodes" separation that was missing a "StrictExcludeNodes" option. - If EntryNodes, ExitNodes, ExcludeNodes, or ExcludeExitNodes change during a config reload, mark and discard all our origin circuits. This fix should address edge cases where we change the config options and but then choose a circuit that we created before the change. - If EntryNodes or ExitNodes are set, be more willing to use an unsuitable (e.g. slow or unstable) circuit. The user asked for it, they get it. - Make EntryNodes config option much more aggressive even when StrictNodes is not set. Before it would prepend your requested entrynodes to your list of guard nodes, but feel free to use others after that. Now it chooses only from your EntryNodes if any of those are available, and only falls back to others if a) they're all down and b) StrictNodes is not set. - Now we refresh your entry guards from EntryNodes at each consensus fetch -- rather than just at startup and then they slowly rot as the network changes. o Major bugfixes: - Stop bridge directory authorities from answering dbg-stability.txt directory queries, which would let people fetch a list of all bridge identities they track. Bugfix on 0.2.1.6-alpha. o Minor features: - Log a notice when we get a new control connection. Now it's easier for security-conscious users to recognize when a local application is knocking on their controller door. Suggested by bug 1196. - New config option "CircuitStreamTimeout" to override our internal timeout schedule for how many seconds until we detach a stream from a circuit and try a new circuit. If your network is particularly slow, you might want to set this to a number like 60. - New controller command "getinfo config-text". It returns the contents that Tor would write if you send it a SAVECONF command, so the controller can write the file to disk itself. - New options for SafeLogging to allow scrubbing only log messages generated while acting as a relay. - Ship the bridges spec file in the tarball too. - Avoid a mad rush at the beginning of each month when each client rotates half of its guards. Instead we spread the rotation out throughout the month, but we still avoid leaving a precise timestamp in the state file about when we first picked the guard. Improves over the behavior introduced in 0.1.2.17. o Minor bugfixes (compiling): - Fix compilation on OS X 10.3, which has a stub mlockall() but hides it. Bugfix on 0.2.2.6-alpha. - Fix compilation on Solaris by removing support for the DisableAllSwap config option. Solaris doesn't have an rlimit for mlockall, so we cannot use it safely. Fixes bug 1198; bugfix on 0.2.2.6-alpha. o Minor bugfixes (crashes): - Do not segfault when writing buffer stats when we haven't observed a single circuit to report about. Found by Fabian Lanze. Bugfix on 0.2.2.1-alpha. - If we're in the pathological case where there's no exit bandwidth but there is non-exit bandwidth, or no guard bandwidth but there is non-guard bandwidth, don't crash during path selection. Bugfix on 0.2.0.3-alpha. - Fix an impossible-to-actually-trigger buffer overflow in relay descriptor generation. Bugfix on 0.1.0.15. o Minor bugfixes (privacy): - Fix an instance where a Tor directory mirror might accidentally log the IP address of a misbehaving Tor client. Bugfix on 0.1.0.1-rc. - Don't list Windows capabilities in relay descriptors. We never made use of them, and maybe it's a bad idea to publish them. Bugfix on 0.1.1.8-alpha. o Minor bugfixes (other): - Resolve an edge case in path weighting that could make us misweight our relay selection. Fixes bug 1203; bugfix on 0.0.8rc1. - Fix statistics on client numbers by country as seen by bridges that were broken in 0.2.2.1-alpha. Also switch to reporting full 24-hour intervals instead of variable 12-to-48-hour intervals. - After we free an internal connection structure, overwrite it with a different memory value than we use for overwriting a freed internal circuit structure. Should help with debugging. Suggested by bug 1055. - Update our OpenSSL 0.9.8l fix so that it works with OpenSSL 0.9.8m too. o Removed features: - Remove the HSAuthorityRecordStats option that version 0 hidden service authorities could have used to track statistics of overall hidden service usage. Changes in version 0.2.1.22 - 2010-01-19 Tor 0.2.1.22 fixes a critical privacy problem in bridge directory authorities -- it would tell you its whole history of bridge descriptors if you make the right directory request. This stable update also rotates two of the seven v3 directory authority keys and locations. o Directory authority changes: - Rotate keys (both v3 identity and relay identity) for moria1 and gabelmoo. o Major bugfixes: - Stop bridge directory authorities from answering dbg-stability.txt directory queries, which would let people fetch a list of all bridge identities they track. Bugfix on 0.2.1.6-alpha. Changes in version 0.2.1.21 - 2009-12-21 Tor 0.2.1.21 fixes an incompatibility with the most recent OpenSSL library. If you use Tor on Linux / Unix and you're getting SSL renegotiation errors, upgrading should help. We also recommend an upgrade if you're an exit relay. o Major bugfixes: - Work around a security feature in OpenSSL 0.9.8l that prevents our handshake from working unless we explicitly tell OpenSSL that we are using SSL renegotiation safely. We are, of course, but OpenSSL 0.9.8l won't work unless we say we are. - Avoid crashing if the client is trying to upload many bytes and the circuit gets torn down at the same time, or if the flip side happens on the exit relay. Bugfix on 0.2.0.1-alpha; fixes bug 1150. o Minor bugfixes: - Do not refuse to learn about authority certs and v2 networkstatus documents that are older than the latest consensus. This bug might have degraded client bootstrapping. Bugfix on 0.2.0.10-alpha. Spotted and fixed by xmux. - Fix a couple of very-hard-to-trigger memory leaks, and one hard-to- trigger platform-specific option misparsing case found by Coverity Scan. - Fix a compilation warning on Fedora 12 by removing an impossible-to- trigger assert. Fixes bug 1173. Changes in version 0.2.2.6-alpha - 2009-11-19 Tor 0.2.2.6-alpha lays the groundwork for many upcoming features: support for the new lower-footprint "microdescriptor" directory design, future-proofing our consensus format against new hash functions or other changes, and an Android port. It also makes Tor compatible with the upcoming OpenSSL 0.9.8l release, and fixes a variety of bugs. o Major features: - Directory authorities can now create, vote on, and serve multiple parallel formats of directory data as part of their voting process. Partially implements Proposal 162: "Publish the consensus in multiple flavors". - Directory authorities can now agree on and publish small summaries of router information that clients can use in place of regular server descriptors. This transition will eventually allow clients to use far less bandwidth for downloading information about the network. Begins the implementation of Proposal 158: "Clients download consensus + microdescriptors". - The directory voting system is now extensible to use multiple hash algorithms for signatures and resource selection. Newer formats are signed with SHA256, with a possibility for moving to a better hash algorithm in the future. - New DisableAllSwap option. If set to 1, Tor will attempt to lock all current and future memory pages via mlockall(). On supported platforms (modern Linux and probably BSD but not Windows or OS X), this should effectively disable any and all attempts to page out memory. This option requires that you start your Tor as root -- if you use DisableAllSwap, please consider using the User option to properly reduce the privileges of your Tor. - Numerous changes, bugfixes, and workarounds from Nathan Freitas to help Tor build correctly for Android phones. o Major bugfixes: - Work around a security feature in OpenSSL 0.9.8l that prevents our handshake from working unless we explicitly tell OpenSSL that we are using SSL renegotiation safely. We are, but OpenSSL 0.9.8l won't work unless we say we are. o Minor bugfixes: - Fix a crash bug when trying to initialize the evdns module in Libevent 2. Bugfix on 0.2.1.16-rc. - Stop logging at severity 'warn' when some other Tor client tries to establish a circuit with us using weak DH keys. It's a protocol violation, but that doesn't mean ordinary users need to hear about it. Fixes the bug part of bug 1114. Bugfix on 0.1.0.13. - Do not refuse to learn about authority certs and v2 networkstatus documents that are older than the latest consensus. This bug might have degraded client bootstrapping. Bugfix on 0.2.0.10-alpha. Spotted and fixed by xmux. - Fix numerous small code-flaws found by Coverity Scan Rung 3. - If all authorities restart at once right before a consensus vote, nobody will vote about "Running", and clients will get a consensus with no usable relays. Instead, authorities refuse to build a consensus if this happens. Bugfix on 0.2.0.10-alpha; fixes bug 1066. - If your relay can't keep up with the number of incoming create cells, it would log one warning per failure into your logs. Limit warnings to 1 per minute. Bugfix on 0.0.2pre10; fixes bug 1042. - Bridges now use "reject *:*" as their default exit policy. Bugfix on 0.2.0.3-alpha; fixes bug 1113. - Fix a memory leak on directory authorities during voting that was introduced in 0.2.2.1-alpha. Found via valgrind. Changes in version 0.2.1.20 - 2009-10-15 Tor 0.2.1.20 fixes a crash bug when you're accessing many hidden services at once, prepares for more performance improvements, and fixes a bunch of smaller bugs. The Windows and OS X bundles also include a more recent Vidalia, and switch from Privoxy to Polipo. The OS X installers are now drag and drop. It's best to un-install Tor/Vidalia and then install this new bundle, rather than upgrade. If you want to upgrade, you'll need to update the paths for Tor and Polipo in the Vidalia Settings window. o Major bugfixes: - Send circuit or stream sendme cells when our window has decreased by 100 cells, not when it has decreased by 101 cells. Bug uncovered by Karsten when testing the "reduce circuit window" performance patch. Bugfix on the 54th commit on Tor -- from July 2002, before the release of Tor 0.0.0. This is the new winner of the oldest-bug prize. - Fix a remotely triggerable memory leak when a consensus document contains more than one signature from the same voter. Bugfix on 0.2.0.3-alpha. - Avoid segfault in rare cases when finishing an introduction circuit as a client and finding out that we don't have an introduction key for it. Fixes bug 1073. Reported by Aaron Swartz. o Major features: - Tor now reads the "circwindow" parameter out of the consensus, and uses that value for its circuit package window rather than the default of 1000 cells. Begins the implementation of proposal 168. o New directory authorities: - Set up urras (run by Jacob Appelbaum) as the seventh v3 directory authority. - Move moria1 and tonga to alternate IP addresses. o Minor bugfixes: - Fix a signed/unsigned compile warning in 0.2.1.19. - Fix possible segmentation fault on directory authorities. Bugfix on 0.2.1.14-rc. - Fix an extremely rare infinite recursion bug that could occur if we tried to log a message after shutting down the log subsystem. Found by Matt Edman. Bugfix on 0.2.0.16-alpha. - Fix an obscure bug where hidden services on 64-bit big-endian systems might mis-read the timestamp in v3 introduce cells, and refuse to connect back to the client. Discovered by "rotor". Bugfix on 0.2.1.6-alpha. - We were triggering a CLOCK_SKEW controller status event whenever we connect via the v2 connection protocol to any relay that has a wrong clock. Instead, we should only inform the controller when it's a trusted authority that claims our clock is wrong. Bugfix on 0.2.0.20-rc; starts to fix bug 1074. Reported by SwissTorExit. - We were telling the controller about CHECKING_REACHABILITY and REACHABILITY_FAILED status events whenever we launch a testing circuit or notice that one has failed. Instead, only tell the controller when we want to inform the user of overall success or overall failure. Bugfix on 0.1.2.6-alpha. Fixes bug 1075. Reported by SwissTorExit. - Don't warn when we're using a circuit that ends with a node excluded in ExcludeExitNodes, but the circuit is not used to access the outside world. This should help fix bug 1090. Bugfix on 0.2.1.6-alpha. - Work around a small memory leak in some versions of OpenSSL that stopped the memory used by the hostname TLS extension from being freed. o Minor features: - Add a "getinfo status/accepted-server-descriptor" controller command, which is the recommended way for controllers to learn whether our server descriptor has been successfully received by at least on directory authority. Un-recommend good-server-descriptor getinfo and status events until we have a better design for them. Changes in version 0.2.2.5-alpha - 2009-10-11 Tor 0.2.2.5-alpha fixes a few compile problems in 0.2.2.4-alpha. o Major bugfixes: - Make the tarball compile again. Oops. Bugfix on 0.2.2.4-alpha. o Directory authorities: - Temporarily (just for this release) move dizum to an alternate IP address. Changes in version 0.2.2.4-alpha - 2009-10-10 Tor 0.2.2.4-alpha fixes more crash bugs in 0.2.2.2-alpha. It also introduces a new unit test framework, shifts directry authority addresses around to reduce the impact from recent blocking events, and fixes a few smaller bugs. o Major bugfixes: - Fix several more asserts in the circuit_build_times code, for example one that causes Tor to fail to start once we have accumulated 5000 build times in the state file. Bugfixes on 0.2.2.2-alpha; fixes bug 1108. o New directory authorities: - Move moria1 and Tonga to alternate IP addresses. o Minor features: - Log SSL state transitions at debug level during handshake, and include SSL states in error messages. This may help debug future SSL handshake issues. - Add a new "Handshake" log domain for activities that happen during the TLS handshake. - Revert to the "June 3 2009" ip-to-country file. The September one seems to have removed most US IP addresses. - Directory authorities now reject Tor relays with versions less than 0.1.2.14. This step cuts out four relays from the current network, none of which are very big. o Minor bugfixes: - Fix a couple of smaller issues with gathering statistics. Bugfixes on 0.2.2.1-alpha. - Fix two memory leaks in the error case of circuit_build_times_parse_state(). Bugfix on 0.2.2.2-alpha. - Don't count one-hop circuits when we're estimating how long it takes circuits to build on average. Otherwise we'll set our circuit build timeout lower than we should. Bugfix on 0.2.2.2-alpha. - Directory authorities no longer change their opinion of, or vote on, whether a router is Running, unless they have themselves been online long enough to have some idea. Bugfix on 0.2.0.6-alpha. Fixes bug 1023. o Code simplifications and refactoring: - Revise our unit tests to use the "tinytest" framework, so we can run tests in their own processes, have smarter setup/teardown code, and so on. The unit test code has moved to its own subdirectory, and has been split into multiple modules. Changes in version 0.2.2.3-alpha - 2009-09-23 Tor 0.2.2.3-alpha fixes a few crash bugs in 0.2.2.2-alpha. o Major bugfixes: - Fix an overzealous assert in our new circuit build timeout code. Bugfix on 0.2.2.2-alpha; fixes bug 1103. o Minor bugfixes: - If the networkstatus consensus tells us that we should use a negative circuit package window, ignore it. Otherwise we'll believe it and then trigger an assert. Bugfix on 0.2.2.2-alpha. Changes in version 0.2.2.2-alpha - 2009-09-21 Tor 0.2.2.2-alpha introduces our latest performance improvement for clients: Tor tracks the average time it takes to build a circuit, and avoids using circuits that take too long to build. For fast connections, this feature can cut your expected latency in half. For slow or flaky connections, it could ruin your Tor experience. Let us know if it does! o Major features: - Tor now tracks how long it takes to build client-side circuits over time, and adapts its timeout to local network performance. Since a circuit that takes a long time to build will also provide bad performance, we get significant latency improvements by discarding the slowest 20% of circuits. Specifically, Tor creates circuits more aggressively than usual until it has enough data points for a good timeout estimate. Implements proposal 151. We are especially looking for reports (good and bad) from users with both EDGE and broadband connections that can move from broadband to EDGE and find out if the build-time data in the .tor/state gets reset without loss of Tor usability. You should also see a notice log message telling you that Tor has reset its timeout. - Directory authorities can now vote on arbitary integer values as part of the consensus process. This is designed to help set network-wide parameters. Implements proposal 167. - Tor now reads the "circwindow" parameter out of the consensus, and uses that value for its circuit package window rather than the default of 1000 cells. Begins the implementation of proposal 168. o Major bugfixes: - Fix a remotely triggerable memory leak when a consensus document contains more than one signature from the same voter. Bugfix on 0.2.0.3-alpha. o Minor bugfixes: - Fix an extremely rare infinite recursion bug that could occur if we tried to log a message after shutting down the log subsystem. Found by Matt Edman. Bugfix on 0.2.0.16-alpha. - Fix parsing for memory or time units given without a space between the number and the unit. Bugfix on 0.2.2.1-alpha; fixes bug 1076. - A networkstatus vote must contain exactly one signature. Spec conformance issue. Bugfix on 0.2.0.3-alpha. - Fix an obscure bug where hidden services on 64-bit big-endian systems might mis-read the timestamp in v3 introduce cells, and refuse to connect back to the client. Discovered by "rotor". Bugfix on 0.2.1.6-alpha. - We were triggering a CLOCK_SKEW controller status event whenever we connect via the v2 connection protocol to any relay that has a wrong clock. Instead, we should only inform the controller when it's a trusted authority that claims our clock is wrong. Bugfix on 0.2.0.20-rc; starts to fix bug 1074. Reported by SwissTorExit. - We were telling the controller about CHECKING_REACHABILITY and REACHABILITY_FAILED status events whenever we launch a testing circuit or notice that one has failed. Instead, only tell the controller when we want to inform the user of overall success or overall failure. Bugfix on 0.1.2.6-alpha. Fixes bug 1075. Reported by SwissTorExit. - Don't warn when we're using a circuit that ends with a node excluded in ExcludeExitNodes, but the circuit is not used to access the outside world. This should help fix bug 1090, but more problems remain. Bugfix on 0.2.1.6-alpha. - Work around a small memory leak in some versions of OpenSSL that stopped the memory used by the hostname TLS extension from being freed. - Make our 'torify' script more portable; if we have only one of 'torsocks' or 'tsocks' installed, don't complain to the user; and explain our warning about tsocks better. o Minor features: - Add a "getinfo status/accepted-server-descriptor" controller command, which is the recommended way for controllers to learn whether our server descriptor has been successfully received by at least on directory authority. Un-recommend good-server-descriptor getinfo and status events until we have a better design for them. - Update to the "September 4 2009" ip-to-country file. Changes in version 0.2.2.1-alpha - 2009-08-26 Tor 0.2.2.1-alpha disables ".exit" address notation by default, allows Tor clients to bootstrap on networks where only port 80 is reachable, makes it more straightforward to support hardware crypto accelerators, and starts the groundwork for gathering stats safely at relays. o Security fixes: - Start the process of disabling ".exit" address notation, since it can be used for a variety of esoteric application-level attacks on users. To reenable it, set "AllowDotExit 1" in your torrc. Fix on 0.0.9rc5. o New directory authorities: - Set up urras (run by Jacob Appelbaum) as the seventh v3 directory authority. o Major features: - New AccelName and AccelDir options add support for dynamic OpenSSL hardware crypto acceleration engines. - Tor now supports tunneling all of its outgoing connections over a SOCKS proxy, using the SOCKS4Proxy and/or SOCKS5Proxy configuration options. Code by Christopher Davis. o Major bugfixes: - Send circuit or stream sendme cells when our window has decreased by 100 cells, not when it has decreased by 101 cells. Bug uncovered by Karsten when testing the "reduce circuit window" performance patch. Bugfix on the 54th commit on Tor -- from July 2002, before the release of Tor 0.0.0. This is the new winner of the oldest-bug prize. o New options for gathering stats safely: - Directory mirrors that set "DirReqStatistics 1" write statistics about directory requests to disk every 24 hours. As compared to the --enable-geoip-stats flag in 0.2.1.x, there are a few improvements: 1) stats are written to disk exactly every 24 hours; 2) estimated shares of v2 and v3 requests are determined as mean values, not at the end of a measurement period; 3) unresolved requests are listed with country code '??'; 4) directories also measure download times. - Exit nodes that set "ExitPortStatistics 1" write statistics on the number of exit streams and transferred bytes per port to disk every 24 hours. - Relays that set "CellStatistics 1" write statistics on how long cells spend in their circuit queues to disk every 24 hours. - Entry nodes that set "EntryStatistics 1" write statistics on the rough number and origins of connecting clients to disk every 24 hours. - Relays that write any of the above statistics to disk and set "ExtraInfoStatistics 1" include the past 24 hours of statistics in their extra-info documents. o Minor features: - New --digests command-line switch to output the digests of the source files Tor was built with. - The "torify" script now uses torsocks where available. - The memarea code now uses a sentinel value at the end of each area to make sure nothing writes beyond the end of an area. This might help debug some conceivable causes of bug 930. - Time and memory units in the configuration file can now be set to fractional units. For example, "2.5 GB" is now a valid value for AccountingMax. - Certain Tor clients (such as those behind check.torproject.org) may want to fetch the consensus in an extra early manner. To enable this a user may now set FetchDirInfoExtraEarly to 1. This also depends on setting FetchDirInfoEarly to 1. Previous behavior will stay the same as only certain clients who must have this information sooner should set this option. - Instead of adding the svn revision to the Tor version string, report the git commit (when we're building from a git checkout). o Minor bugfixes: - If any of the v3 certs we download are unparseable, we should actually notice the failure so we don't retry indefinitely. Bugfix on 0.2.0.x; reported by "rotator". - If the cached cert file is unparseable, warn but don't exit. - Fix possible segmentation fault on directory authorities. Bugfix on 0.2.1.14-rc. - When Tor fails to parse a descriptor of any kind, dump it to disk. Might help diagnosing bug 1051. o Deprecated and removed features: - The controller no longer accepts the old obsolete "addr-mappings/" or "unregistered-servers-" GETINFO values. - Hidden services no longer publish version 0 descriptors, and clients do not request or use version 0 descriptors. However, the old hidden service authorities still accept and serve version 0 descriptors when contacted by older hidden services/clients. - The EXTENDED_EVENTS and VERBOSE_NAMES controller features are now always on; using them is necessary for correct forward-compatible controllers. - Remove support for .noconnect style addresses. Nobody was using them, and they provided another avenue for detecting Tor users via application-level web tricks. o Packaging changes: - Upgrade Vidalia from 0.1.15 to 0.2.3 in the Windows and OS X installer bundles. See https://trac.vidalia-project.net/browser/vidalia/tags/vidalia-0.2.3/CHANGELOG for details of what's new in Vidalia 0.2.3. - Windows Vidalia Bundle: update Privoxy from 3.0.6 to 3.0.14-beta. - OS X Vidalia Bundle: move to Polipo 1.0.4 with Tor specific configuration file, rather than the old Privoxy. - OS X Vidalia Bundle: Vidalia, Tor, and Polipo are compiled as x86-only for better compatibility with OS X 10.6, aka Snow Leopard. - OS X Tor Expert Bundle: Tor is compiled as x86-only for better compatibility with OS X 10.6, aka Snow Leopard. - OS X Vidalia Bundle: The multi-package installer is now replaced by a simple drag and drop to the /Applications folder. This change occurred with the upgrade to Vidalia 0.2.3. Changes in version 0.2.1.19 - 2009-07-28 Tor 0.2.1.19 fixes a major bug with accessing and providing hidden services on Tor 0.2.1.3-alpha through 0.2.1.18. o Major bugfixes: - Make accessing hidden services on 0.2.1.x work right again. Bugfix on 0.2.1.3-alpha; workaround for bug 1038. Diagnosis and part of patch provided by "optimist". o Minor features: - When a relay/bridge is writing out its identity key fingerprint to the "fingerprint" file and to its logs, write it without spaces. Now it will look like the fingerprints in our bridges documentation, and confuse fewer users. o Minor bugfixes: - Relays no longer publish a new server descriptor if they change their MaxAdvertisedBandwidth config option but it doesn't end up changing their advertised bandwidth numbers. Bugfix on 0.2.0.28-rc; fixes bug 1026. Patch from Sebastian. - Avoid leaking memory every time we get a create cell but we have so many already queued that we refuse it. Bugfix on 0.2.0.19-alpha; fixes bug 1034. Reported by BarkerJr. Changes in version 0.2.1.18 - 2009-07-24 Tor 0.2.1.18 lays the foundations for performance improvements, adds status events to help users diagnose bootstrap problems, adds optional authentication/authorization for hidden services, fixes a variety of potential anonymity problems, and includes a huge pile of other features and bug fixes. o Build fixes: - Add LIBS=-lrt to Makefile.am so the Tor RPMs use a static libevent. Changes in version 0.2.1.17-rc - 2009-07-07 Tor 0.2.1.17-rc marks the fourth -- and hopefully last -- release candidate for the 0.2.1.x series. It lays the groundwork for further client performance improvements, and also fixes a big bug with directory authorities that were causing them to assign Guard and Stable flags poorly. The Windows bundles also finally include the geoip database that we thought we'd been shipping since 0.2.0.x (oops), and the OS X bundles should actually install Torbutton rather than giving you a cryptic failure message (oops). o Major features: - Clients now use the bandwidth values in the consensus, rather than the bandwidth values in each relay descriptor. This approach opens the door to more accurate bandwidth estimates once the directory authorities start doing active measurements. Implements more of proposal 141. o Major bugfixes: - When Tor clients restart after 1-5 days, they discard all their cached descriptors as too old, but they still use the cached consensus document. This approach is good for robustness, but bad for performance: since they don't know any bandwidths, they end up choosing at random rather than weighting their choice by speed. Fixed by the above feature of putting bandwidths in the consensus. Bugfix on 0.2.0.x. - Directory authorities were neglecting to mark relays down in their internal histories if the relays fall off the routerlist without ever being found unreachable. So there were relays in the histories that haven't been seen for eight months, and are listed as being up for eight months. This wreaked havoc on the "median wfu" and "median mtbf" calculations, in turn making Guard and Stable flags very wrong, hurting network performance. Fixes bugs 696 and 969. Bugfix on 0.2.0.6-alpha. o Minor bugfixes: - Serve the DirPortFrontPage page even when we have been approaching our quotas recently. Fixes bug 1013; bugfix on 0.2.1.8-alpha. - The control port would close the connection before flushing long replies, such as the network consensus, if a QUIT command was issued before the reply had completed. Now, the control port flushes all pending replies before closing the connection. Also fixed a spurious warning when a QUIT command is issued after a malformed or rejected AUTHENTICATE command, but before the connection was closed. Patch by Marcus Griep. Bugfix on 0.2.0.x; fixes bugs 1015 and 1016. - When we can't find an intro key for a v2 hidden service descriptor, fall back to the v0 hidden service descriptor and log a bug message. Workaround for bug 1024. - Fix a log message that did not respect the SafeLogging option. Resolves bug 1027. o Minor features: - If we're a relay and we change our IP address, be more verbose about the reason that made us change. Should help track down further bugs for relays on dynamic IP addresses. Changes in version 0.2.0.35 - 2009-06-24 o Security fix: - Avoid crashing in the presence of certain malformed descriptors. Found by lark, and by automated fuzzing. - Fix an edge case where a malicious exit relay could convince a controller that the client's DNS question resolves to an internal IP address. Bug found and fixed by "optimist"; bugfix on 0.1.2.8-beta. o Major bugfixes: - Finally fix the bug where dynamic-IP relays disappear when their IP address changes: directory mirrors were mistakenly telling them their old address if they asked via begin_dir, so they never got an accurate answer about their new address, so they just vanished after a day. For belt-and-suspenders, relays that don't set Address in their config now avoid using begin_dir for all direct connections. Should fix bugs 827, 883, and 900. - Fix a timing-dependent, allocator-dependent, DNS-related crash bug that would occur on some exit nodes when DNS failures and timeouts occurred in certain patterns. Fix for bug 957. o Minor bugfixes: - When starting with a cache over a few days old, do not leak memory for the obsolete router descriptors in it. Bugfix on 0.2.0.33; fixes bug 672. - Hidden service clients didn't use a cached service descriptor that was older than 15 minutes, but wouldn't fetch a new one either, because there was already one in the cache. Now, fetch a v2 descriptor unless the same descriptor was added to the cache within the last 15 minutes. Fixes bug 997; reported by Marcus Griep. Changes in version 0.2.1.16-rc - 2009-06-20 Tor 0.2.1.16-rc speeds up performance for fast exit relays, and fixes a bunch of minor bugs. o Security fixes: - Fix an edge case where a malicious exit relay could convince a controller that the client's DNS question resolves to an internal IP address. Bug found and fixed by "optimist"; bugfix on 0.1.2.8-beta. o Major performance improvements (on 0.2.0.x): - Disable and refactor some debugging checks that forced a linear scan over the whole server-side DNS cache. These accounted for over 50% of CPU time on a relatively busy exit node's gprof profile. Found by Jacob. - Disable some debugging checks that appeared in exit node profile data. o Minor features: - Update to the "June 3 2009" ip-to-country file. - Do not have tor-resolve automatically refuse all .onion addresses; if AutomapHostsOnResolve is set in your torrc, this will work fine. o Minor bugfixes (on 0.2.0.x): - Log correct error messages for DNS-related network errors on Windows. - Fix a race condition that could cause crashes or memory corruption when running as a server with a controller listening for log messages. - Avoid crashing when we have a policy specified in a DirPolicy or SocksPolicy or ReachableAddresses option with ports set on it, and we re-load the policy. May fix bug 996. - Hidden service clients didn't use a cached service descriptor that was older than 15 minutes, but wouldn't fetch a new one either, because there was already one in the cache. Now, fetch a v2 descriptor unless the same descriptor was added to the cache within the last 15 minutes. Fixes bug 997; reported by Marcus Griep. o Minor bugfixes (on 0.2.1.x): - Don't warn users about low port and hibernation mix when they provide a *ListenAddress directive to fix that. Bugfix on 0.2.1.15-rc. - When switching back and forth between bridge mode, do not start gathering GeoIP data until two hours have passed. - Do not complain that the user has requested an excluded node as an exit when the node is not really an exit. This could happen because the circuit was for testing, or an introduction point. Fix for bug 984. Changes in version 0.2.1.15-rc - 2009-05-25 Tor 0.2.1.15-rc marks the second release candidate for the 0.2.1.x series. It fixes a major bug on fast exit relays, as well as a variety of more minor bugs. o Major bugfixes (on 0.2.0.x): - Fix a timing-dependent, allocator-dependent, DNS-related crash bug that would occur on some exit nodes when DNS failures and timeouts occurred in certain patterns. Fix for bug 957. o Minor bugfixes (on 0.2.0.x): - Actually return -1 in the error case for read_bandwidth_usage(). Harmless bug, since we currently don't care about the return value anywhere. Bugfix on 0.2.0.9-alpha. - Provide a more useful log message if bug 977 (related to buffer freelists) ever reappears, and do not crash right away. - Fix an assertion failure on 64-bit platforms when we allocated memory right up to the end of a memarea, then realigned the memory one step beyond the end. Fixes a possible cause of bug 930. - Protect the count of open sockets with a mutex, so we can't corrupt it when two threads are closing or opening sockets at once. Fix for bug 939. Bugfix on 0.2.0.1-alpha. - Don't allow a bridge to publish its router descriptor to a non-bridge directory authority. Fixes part of bug 932. - When we change to or from being a bridge, reset our counts of client usage by country. Fixes bug 932. - Fix a bug that made stream bandwidth get misreported to the controller. - Stop using malloc_usable_size() to use more area than we had actually allocated: it was safe, but made valgrind really unhappy. - Fix a memory leak when v3 directory authorities load their keys and cert from disk. Bugfix on 0.2.0.1-alpha. o Minor bugfixes (on 0.2.1.x): - Fix use of freed memory when deciding to mark a non-addable descriptor as never-downloadable. Bugfix on 0.2.1.9-alpha. Changes in version 0.2.1.14-rc - 2009-04-12 Tor 0.2.1.14-rc marks the first release candidate for the 0.2.1.x series. It begins fixing some major performance problems, and also finally addresses the bug that was causing relays on dynamic IP addresses to fall out of the directory. o Major features: - Clients replace entry guards that were chosen more than a few months ago. This change should significantly improve client performance, especially once more people upgrade, since relays that have been a guard for a long time are currently overloaded. o Major bugfixes (on 0.2.0): - Finally fix the bug where dynamic-IP relays disappear when their IP address changes: directory mirrors were mistakenly telling them their old address if they asked via begin_dir, so they never got an accurate answer about their new address, so they just vanished after a day. For belt-and-suspenders, relays that don't set Address in their config now avoid using begin_dir for all direct connections. Should fix bugs 827, 883, and 900. - Relays were falling out of the networkstatus consensus for part of a day if they changed their local config but the authorities discarded their new descriptor as "not sufficiently different". Now directory authorities accept a descriptor as changed if bandwidthrate or bandwidthburst changed. Partial fix for bug 962; patch by Sebastian. - Avoid crashing in the presence of certain malformed descriptors. Found by lark, and by automated fuzzing. o Minor features: - When generating circuit events with verbose nicknames for controllers, try harder to look up nicknames for routers on a circuit. (Previously, we would look in the router descriptors we had for nicknames, but not in the consensus.) Partial fix for bug 941. - If the bridge config line doesn't specify a port, assume 443. This makes bridge lines a bit smaller and easier for users to understand. - Raise the minimum bandwidth to be a relay from 20000 bytes to 20480 bytes (aka 20KB/s), to match our documentation. Also update directory authorities so they always assign the Fast flag to relays with 20KB/s of capacity. Now people running relays won't suddenly find themselves not seeing any use, if the network gets faster on average. - Update to the "April 3 2009" ip-to-country file. o Minor bugfixes: - Avoid trying to print raw memory to the logs when we decide to give up on downloading a given relay descriptor. Bugfix on 0.2.1.9-alpha. - In tor-resolve, when the Tor client to use is specified by :, actually use the specified port rather than defaulting to 9050. Bugfix on 0.2.1.6-alpha. - Make directory usage recording work again. Bugfix on 0.2.1.6-alpha. - When starting with a cache over a few days old, do not leak memory for the obsolete router descriptors in it. Bugfix on 0.2.0.33. - Avoid double-free on list of successfully uploaded hidden service discriptors. Fix for bug 948. Bugfix on 0.2.1.6-alpha. - Change memarea_strndup() implementation to work even when duplicating a string at the end of a page. This bug was harmless for now, but could have meant crashes later. Fix by lark. Bugfix on 0.2.1.1-alpha. - Limit uploaded directory documents to be 16M rather than 500K. The directory authorities were refusing v3 consensus votes from other authorities, since the votes are now 504K. Fixes bug 959; bugfix on 0.0.2pre17 (where we raised it from 50K to 500K ;). - Directory authorities should never send a 503 "busy" response to requests for votes or keys. Bugfix on 0.2.0.8-alpha; exposed by bug 959. Changes in version 0.2.1.13-alpha - 2009-03-09 Tor 0.2.1.13-alpha includes another big pile of minor bugfixes and cleanups. We're finally getting close to a release candidate. o Major bugfixes: - Correctly update the list of which countries we exclude as exits, when the GeoIP file is loaded or reloaded. Diagnosed by lark. Bugfix on 0.2.1.6-alpha. o Minor bugfixes (on 0.2.0.x and earlier): - Automatically detect MacOSX versions earlier than 10.4.0, and disable kqueue from inside Tor when running with these versions. We previously did this from the startup script, but that was no help to people who didn't use the startup script. Resolves bug 863. - When we had picked an exit node for a connection, but marked it as "optional", and it turned out we had no onion key for the exit, stop wanting that exit and try again. This situation may not be possible now, but will probably become feasible with proposal 158. Spotted by rovv. Fixes another case of bug 752. - Clients no longer cache certificates for authorities they do not recognize. Bugfix on 0.2.0.9-alpha. - When we can't transmit a DNS request due to a network error, retry it after a while, and eventually transmit a failing response to the RESOLVED cell. Bugfix on 0.1.2.5-alpha. - If the controller claimed responsibility for a stream, but that stream never finished making its connection, it would live forever in circuit_wait state. Now we close it after SocksTimeout seconds. Bugfix on 0.1.2.7-alpha; reported by Mike Perry. - Drop begin cells to a hidden service if they come from the middle of a circuit. Patch from lark. - When we erroneously receive two EXTEND cells for the same circuit ID on the same connection, drop the second. Patch from lark. - Fix a crash that occurs on exit nodes when a nameserver request timed out. Bugfix on 0.1.2.1-alpha; our CLEAR debugging code had been suppressing the bug since 0.1.2.10-alpha. Partial fix for bug 929. - Do not assume that a stack-allocated character array will be 64-bit aligned on platforms that demand that uint64_t access is aligned. Possible fix for bug 604. - Parse dates and IPv4 addresses in a locale- and libc-independent manner, to avoid platform-dependent behavior on malformed input. - Build correctly when configured to build outside the main source path. Patch from Michael Gold. - We were already rejecting relay begin cells with destination port of 0. Now also reject extend cells with destination port or address of 0. Suggested by lark. o Minor bugfixes (on 0.2.1.x): - Don't re-extend introduction circuits if we ran out of RELAY_EARLY cells. Bugfix on 0.2.1.3-alpha. Fixes more of bug 878. - If we're an exit node, scrub the IP address to which we are exiting in the logs. Bugfix on 0.2.1.8-alpha. o Minor features: - On Linux, use the prctl call to re-enable core dumps when the user is option is set. - New controller event NEWCONSENSUS that lists the networkstatus lines for every recommended relay. Now controllers like Torflow can keep up-to-date on which relays they should be using. - Update to the "February 26 2009" ip-to-country file. Changes in version 0.2.0.34 - 2009-02-08 Tor 0.2.0.34 features several more security-related fixes. You should upgrade, especially if you run an exit relay (remote crash) or a directory authority (remote infinite loop), or you're on an older (pre-XP) or not-recently-patched Windows (remote exploit). This release marks end-of-life for Tor 0.1.2.x. Those Tor versions have many known flaws, and nobody should be using them. You should upgrade. If you're using a Linux or BSD and its packages are obsolete, stop using those packages and upgrade anyway. o Security fixes: - Fix an infinite-loop bug on handling corrupt votes under certain circumstances. Bugfix on 0.2.0.8-alpha. - Fix a temporary DoS vulnerability that could be performed by a directory mirror. Bugfix on 0.2.0.9-alpha; reported by lark. - Avoid a potential crash on exit nodes when processing malformed input. Remote DoS opportunity. Bugfix on 0.2.0.33. - Do not accept incomplete ipv4 addresses (like 192.168.0) as valid. Spec conformance issue. Bugfix on Tor 0.0.2pre27. o Minor bugfixes: - Fix compilation on systems where time_t is a 64-bit integer. Patch from Matthias Drochner. - Don't consider expiring already-closed client connections. Fixes bug 893. Bugfix on 0.0.2pre20. Changes in version 0.2.1.12-alpha - 2009-02-08 Tor 0.2.1.12-alpha features several more security-related fixes. You should upgrade, especially if you run an exit relay (remote crash) or a directory authority (remote infinite loop), or you're on an older (pre-XP) or not-recently-patched Windows (remote exploit). It also includes a big pile of minor bugfixes and cleanups. o Security fixes: - Fix an infinite-loop bug on handling corrupt votes under certain circumstances. Bugfix on 0.2.0.8-alpha. - Fix a temporary DoS vulnerability that could be performed by a directory mirror. Bugfix on 0.2.0.9-alpha; reported by lark. - Avoid a potential crash on exit nodes when processing malformed input. Remote DoS opportunity. Bugfix on 0.2.1.7-alpha. o Minor bugfixes: - Let controllers actually ask for the "clients_seen" event for getting usage summaries on bridge relays. Bugfix on 0.2.1.10-alpha; reported by Matt Edman. - Fix a compile warning on OSX Panther. Fixes bug 913; bugfix against 0.2.1.11-alpha. - Fix a bug in address parsing that was preventing bridges or hidden service targets from being at IPv6 addresses. - Solve a bug that kept hardware crypto acceleration from getting enabled when accounting was turned on. Fixes bug 907. Bugfix on 0.0.9pre6. - Remove a bash-ism from configure.in to build properly on non-Linux platforms. Bugfix on 0.2.1.1-alpha. - Fix code so authorities _actually_ send back X-Descriptor-Not-New headers. Bugfix on 0.2.0.10-alpha. - Don't consider expiring already-closed client connections. Fixes bug 893. Bugfix on 0.0.2pre20. - Fix another interesting corner-case of bug 891 spotted by rovv: Previously, if two hosts had different amounts of clock drift, and one of them created a new connection with just the wrong timing, the other might decide to deprecate the new connection erroneously. Bugfix on 0.1.1.13-alpha. - Resolve a very rare crash bug that could occur when the user forced a nameserver reconfiguration during the middle of a nameserver probe. Fixes bug 526. Bugfix on 0.1.2.1-alpha. - Support changing value of ServerDNSRandomizeCase during SIGHUP. Bugfix on 0.2.1.7-alpha. - If we're using bridges and our network goes away, be more willing to forgive our bridges and try again when we get an application request. Bugfix on 0.2.0.x. o Minor features: - Support platforms where time_t is 64 bits long. (Congratulations, NetBSD!) Patch from Matthias Drochner. - Add a 'getinfo status/clients-seen' controller command, in case controllers want to hear clients_seen events but connect late. o Build changes: - Disable GCC's strict alias optimization by default, to avoid the likelihood of its introducing subtle bugs whenever our code violates the letter of C99's alias rules. Changes in version 0.2.0.33 - 2009-01-21 Tor 0.2.0.33 fixes a variety of bugs that were making relays less useful to users. It also finally fixes a bug where a relay or client that's been off for many days would take a long time to bootstrap. This update also fixes an important security-related bug reported by Ilja van Sprundel. You should upgrade. (We'll send out more details about the bug once people have had some time to upgrade.) o Security fixes: - Fix a heap-corruption bug that may be remotely triggerable on some platforms. Reported by Ilja van Sprundel. o Major bugfixes: - When a stream at an exit relay is in state "resolving" or "connecting" and it receives an "end" relay cell, the exit relay would silently ignore the end cell and not close the stream. If the client never closes the circuit, then the exit relay never closes the TCP connection. Bug introduced in Tor 0.1.2.1-alpha; reported by "wood". - When sending CREATED cells back for a given circuit, use a 64-bit connection ID to find the right connection, rather than an addr:port combination. Now that we can have multiple OR connections between the same ORs, it is no longer possible to use addr:port to uniquely identify a connection. - Bridge relays that had DirPort set to 0 would stop fetching descriptors shortly after startup, and then briefly resume after a new bandwidth test and/or after publishing a new bridge descriptor. Bridge users that try to bootstrap from them would get a recent networkstatus but would get descriptors from up to 18 hours earlier, meaning most of the descriptors were obsolete already. Reported by Tas; bugfix on 0.2.0.13-alpha. - Prevent bridge relays from serving their 'extrainfo' document to anybody who asks, now that extrainfo docs include potentially sensitive aggregated client geoip summaries. Bugfix on 0.2.0.13-alpha. - If the cached networkstatus consensus is more than five days old, discard it rather than trying to use it. In theory it could be useful because it lists alternate directory mirrors, but in practice it just means we spend many minutes trying directory mirrors that are long gone from the network. Also discard router descriptors as we load them if they are more than five days old, since the onion key is probably wrong by now. Bugfix on 0.2.0.x. Fixes bug 887. o Minor bugfixes: - Do not mark smartlist_bsearch_idx() function as ATTR_PURE. This bug could make gcc generate non-functional binary search code. Bugfix on 0.2.0.10-alpha. - Build correctly on platforms without socklen_t. - Compile without warnings on solaris. - Avoid potential crash on internal error during signature collection. Fixes bug 864. Patch from rovv. - Correct handling of possible malformed authority signing key certificates with internal signature types. Fixes bug 880. Bugfix on 0.2.0.3-alpha. - Fix a hard-to-trigger resource leak when logging credential status. CID 349. - When we can't initialize DNS because the network is down, do not automatically stop Tor from starting. Instead, we retry failed dns_init() every 10 minutes, and change the exit policy to reject *:* until one succeeds. Fixes bug 691. - Use 64 bits instead of 32 bits for connection identifiers used with the controller protocol, to greatly reduce risk of identifier reuse. - When we're choosing an exit node for a circuit, and we have no pending streams, choose a good general exit rather than one that supports "all the pending streams". Bugfix on 0.1.1.x. Fix by rovv. - Fix another case of assuming, when a specific exit is requested, that we know more than the user about what hosts it allows. Fixes one case of bug 752. Patch from rovv. - Clip the MaxCircuitDirtiness config option to a minimum of 10 seconds. Warn the user if lower values are given in the configuration. Bugfix on 0.1.0.1-rc. Patch by Sebastian. - Clip the CircuitBuildTimeout to a minimum of 30 seconds. Warn the user if lower values are given in the configuration. Bugfix on 0.1.1.17-rc. Patch by Sebastian. - Fix a memory leak when we decline to add a v2 rendezvous descriptor to the cache because we already had a v0 descriptor with the same ID. Bugfix on 0.2.0.18-alpha. - Fix a race condition when freeing keys shared between main thread and CPU workers that could result in a memory leak. Bugfix on 0.1.0.1-rc. Fixes bug 889. - Send a valid END cell back when a client tries to connect to a nonexistent hidden service port. Bugfix on 0.1.2.15. Fixes bug 840. Patch from rovv. - Check which hops rendezvous stream cells are associated with to prevent possible guess-the-streamid injection attacks from intermediate hops. Fixes another case of bug 446. Based on patch from rovv. - If a broken client asks a non-exit router to connect somewhere, do not even do the DNS lookup before rejecting the connection. Fixes another case of bug 619. Patch from rovv. - When a relay gets a create cell it can't decrypt (e.g. because it's using the wrong onion key), we were dropping it and letting the client time out. Now actually answer with a destroy cell. Fixes bug 904. Bugfix on 0.0.2pre8. o Minor bugfixes (hidden services): - Do not throw away existing introduction points on SIGHUP. Bugfix on 0.0.6pre1. Patch by Karsten. Fixes bug 874. o Minor features: - Report the case where all signatures in a detached set are rejected differently than the case where there is an error handling the detached set. - When we realize that another process has modified our cached descriptors, print out a more useful error message rather than triggering an assertion. Fixes bug 885. Patch from Karsten. - Implement the 0x20 hack to better resist DNS poisoning: set the case on outgoing DNS requests randomly, and reject responses that do not match the case correctly. This logic can be disabled with the ServerDNSRamdomizeCase setting, if you are using one of the 0.3% of servers that do not reliably preserve case in replies. See "Increased DNS Forgery Resistance through 0x20-Bit Encoding" for more info. - Check DNS replies for more matching fields to better resist DNS poisoning. - Never use OpenSSL compression: it wastes RAM and CPU trying to compress cells, which are basically all encrypted, compressed, or both. Changes in version 0.2.1.11-alpha - 2009-01-20 Tor 0.2.1.11-alpha finishes fixing the "if your Tor is off for a week it will take a long time to bootstrap again" bug. It also fixes an important security-related bug reported by Ilja van Sprundel. You should upgrade. (We'll send out more details about the bug once people have had some time to upgrade.) o Security fixes: - Fix a heap-corruption bug that may be remotely triggerable on some platforms. Reported by Ilja van Sprundel. o Major bugfixes: - Discard router descriptors as we load them if they are more than five days old. Otherwise if Tor is off for a long time and then starts with cached descriptors, it will try to use the onion keys in those obsolete descriptors when building circuits. Bugfix on 0.2.0.x. Fixes bug 887. o Minor features: - Try to make sure that the version of Libevent we're running with is binary-compatible with the one we built with. May address bug 897 and others. - Make setting ServerDNSRandomizeCase to 0 actually work. Bugfix for bug 905. Bugfix on 0.2.1.7-alpha. - Add a new --enable-local-appdata configuration switch to change the default location of the datadir on win32 from APPDATA to LOCAL_APPDATA. In the future, we should migrate to LOCAL_APPDATA entirely. Patch from coderman. o Minor bugfixes: - Make outbound DNS packets respect the OutboundBindAddress setting. Fixes the bug part of bug 798. Bugfix on 0.1.2.2-alpha. - When our circuit fails at the first hop (e.g. we get a destroy cell back), avoid using that OR connection anymore, and also tell all the one-hop directory requests waiting for it that they should fail. Bugfix on 0.2.1.3-alpha. - In the torify(1) manpage, mention that tsocks will leak your DNS requests. Changes in version 0.2.1.10-alpha - 2009-01-06 Tor 0.2.1.10-alpha fixes two major bugs in bridge relays (one that would make the bridge relay not so useful if it had DirPort set to 0, and one that could let an attacker learn a little bit of information about the bridge's users), and a bug that would cause your Tor relay to ignore a circuit create request it can't decrypt (rather than reply with an error). It also fixes a wide variety of other bugs. o Major bugfixes: - If the cached networkstatus consensus is more than five days old, discard it rather than trying to use it. In theory it could be useful because it lists alternate directory mirrors, but in practice it just means we spend many minutes trying directory mirrors that are long gone from the network. Helps bug 887 a bit; bugfix on 0.2.0.x. - Bridge relays that had DirPort set to 0 would stop fetching descriptors shortly after startup, and then briefly resume after a new bandwidth test and/or after publishing a new bridge descriptor. Bridge users that try to bootstrap from them would get a recent networkstatus but would get descriptors from up to 18 hours earlier, meaning most of the descriptors were obsolete already. Reported by Tas; bugfix on 0.2.0.13-alpha. - Prevent bridge relays from serving their 'extrainfo' document to anybody who asks, now that extrainfo docs include potentially sensitive aggregated client geoip summaries. Bugfix on 0.2.0.13-alpha. o Minor features: - New controller event "clients_seen" to report a geoip-based summary of which countries we've seen clients from recently. Now controllers like Vidalia can show bridge operators that they're actually making a difference. - Build correctly against versions of OpenSSL 0.9.8 or later built without support for deprecated functions. - Update to the "December 19 2008" ip-to-country file. o Minor bugfixes (on 0.2.0.x): - Authorities now vote for the Stable flag for any router whose weighted MTBF is at least 5 days, regardless of the mean MTBF. - Do not remove routers as too old if we do not have any consensus document. Bugfix on 0.2.0.7-alpha. - Do not accept incomplete ipv4 addresses (like 192.168.0) as valid. Spec conformance issue. Bugfix on Tor 0.0.2pre27. - When an exit relay resolves a stream address to a local IP address, do not just keep retrying that same exit relay over and over. Instead, just close the stream. Addresses bug 872. Bugfix on 0.2.0.32. Patch from rovv. - If a hidden service sends us an END cell, do not consider retrying the connection; just close it. Patch from rovv. - When we made bridge authorities stop serving bridge descriptors over unencrypted links, we also broke DirPort reachability testing for bridges. So bridges with a non-zero DirPort were printing spurious warns to their logs. Bugfix on 0.2.0.16-alpha. Fixes bug 709. - When a relay gets a create cell it can't decrypt (e.g. because it's using the wrong onion key), we were dropping it and letting the client time out. Now actually answer with a destroy cell. Fixes bug 904. Bugfix on 0.0.2pre8. - Squeeze 2-5% out of client performance (according to oprofile) by improving the implementation of some policy-manipulation functions. o Minor bugfixes (on 0.2.1.x): - Make get_interface_address() function work properly again; stop guessing the wrong parts of our address as our address. - Do not cannibalize a circuit if we're out of RELAY_EARLY cells to send on that circuit. Otherwise we might violate the proposal-110 limit. Bugfix on 0.2.1.3-alpha. Partial fix for bug 878. Diagnosis thanks to Karsten. - When we're sending non-EXTEND cells to the first hop in a circuit, for example to use an encrypted directory connection, we don't need to use RELAY_EARLY cells: the first hop knows what kind of cell it is, and nobody else can even see the cell type. Conserving RELAY_EARLY cells makes it easier to cannibalize circuits like this later. - Stop logging nameserver addresses in reverse order. - If we are retrying a directory download slowly over and over, do not automatically give up after the 254th failure. Bugfix on 0.2.1.9-alpha. - Resume reporting accurate "stream end" reasons to the local control port. They were lost in the changes for Proposal 148. Bugfix on 0.2.1.9-alpha. o Deprecated and removed features: - The old "tor --version --version" command, which would print out the subversion "Id" of most of the source files, is now removed. It turned out to be less useful than we'd expected, and harder to maintain. o Code simplifications and refactoring: - Change our header file guard macros to be less likely to conflict with system headers. Adam Langley noticed that we were conflicting with log.h on Android. - Tool-assisted documentation cleanup. Nearly every function or static variable in Tor should have its own documentation now. Changes in version 0.2.1.9-alpha - 2008-12-25 Tor 0.2.1.9-alpha fixes many more bugs, some of them security-related. o New directory authorities: - gabelmoo (the authority run by Karsten Loesing) now has a new IP address. o Security fixes: - Never use a connection with a mismatched address to extend a circuit, unless that connection is canonical. A canonical connection is one whose address is authenticated by the router's identity key, either in a NETINFO cell or in a router descriptor. - Avoid a possible memory corruption bug when receiving hidden service descriptors. Bugfix on 0.2.1.6-alpha. o Major bugfixes: - Fix a logic error that would automatically reject all but the first configured DNS server. Bugfix on 0.2.1.5-alpha. Possible fix for part of bug 813/868. Bug spotted by coderman. - When a stream at an exit relay is in state "resolving" or "connecting" and it receives an "end" relay cell, the exit relay would silently ignore the end cell and not close the stream. If the client never closes the circuit, then the exit relay never closes the TCP connection. Bug introduced in 0.1.2.1-alpha; reported by "wood". - When we can't initialize DNS because the network is down, do not automatically stop Tor from starting. Instead, retry failed dns_init() every 10 minutes, and change the exit policy to reject *:* until one succeeds. Fixes bug 691. o Minor features: - Give a better error message when an overzealous init script says "sudo -u username tor --user username". Makes Bug 882 easier for users to diagnose. - When a directory authority gives us a new guess for our IP address, log which authority we used. Hopefully this will help us debug the recent complaints about bad IP address guesses. - Detect svn revision properly when we're using git-svn. - Try not to open more than one descriptor-downloading connection to an authority at once. This should reduce load on directory authorities. Fixes bug 366. - Add cross-certification to newly generated certificates, so that a signing key is enough information to look up a certificate. Partial implementation of proposal 157. - Start serving certificates by pairs. Partial implementation of proposal 157. - Clients now never report any stream end reason except 'MISC'. Implements proposal 148. - On platforms with a maximum syslog string length, truncate syslog messages to that length ourselves, rather than relying on the system to do it for us. - Optimize out calls to time(NULL) that occur for every IO operation, or for every cell. On systems where time() is a slow syscall, this fix will be slightly helpful. - Exit servers can now answer resolve requests for ip6.arpa addresses. - When we download a descriptor that we then immediately (as a directory authority) reject, do not retry downloading it right away. Should save some bandwidth on authorities. Fix for bug 888. Patch by Sebastian Hahn. - When a download gets us zero good descriptors, do not notify Tor that new directory information has arrived. - Avoid some nasty corner cases in the logic for marking connections as too old or obsolete or noncanonical for circuits. Partial bugfix on bug 891. o Minor features (controller): - New CONSENSUS_ARRIVED event to note when a new consensus has been fetched and validated. - When we realize that another process has modified our cached descriptors file, print out a more useful error message rather than triggering an assertion. Fixes bug 885. Patch from Karsten. - Add an internal-use-only __ReloadTorrcOnSIGHUP option for controllers to prevent SIGHUP from reloading the configuration. Fixes bug 856. o Minor bugfixes: - Resume using the correct "REASON=" stream when telling the controller why we closed a stream. Bugfix in 0.2.1.1-alpha. - When a canonical connection appears later in our internal list than a noncanonical one for a given OR ID, always use the canonical one. Bugfix on 0.2.0.12-alpha. Fixes bug 805. Spotted by rovv. - Clip the MaxCircuitDirtiness config option to a minimum of 10 seconds. Warn the user if lower values are given in the configuration. Bugfix on 0.1.0.1-rc. Patch by Sebastian. - Clip the CircuitBuildTimeout to a minimum of 30 seconds. Warn the user if lower values are given in the configuration. Bugfix on 0.1.1.17-rc. Patch by Sebastian. - Fix a race condition when freeing keys shared between main thread and CPU workers that could result in a memory leak. Bugfix on 0.1.0.1-rc. Fixes bug 889. o Minor bugfixes (hidden services): - Do not throw away existing introduction points on SIGHUP (bugfix on 0.0.6pre1); also, do not stall hidden services because we're throwing away introduction points; bugfix on 0.2.1.7-alpha. Spotted by John Brooks. Patch by Karsten. Fixes bug 874. - Fix a memory leak when we decline to add a v2 rendezvous descriptor to the cache because we already had a v0 descriptor with the same ID. Bugfix on 0.2.0.18-alpha. o Deprecated and removed features: - RedirectExits has been removed. It was deprecated since 0.2.0.3-alpha. - Finally remove deprecated "EXTENDED_FORMAT" controller feature. It has been called EXTENDED_EVENTS since 0.1.2.4-alpha. - Cell pools are now always enabled; --disable-cell-pools is ignored. o Code simplifications and refactoring: - Rename the confusing or_is_obsolete field to the more appropriate is_bad_for_new_circs, and move it to or_connection_t where it belongs. - Move edge-only flags from connection_t to edge_connection_t: not only is this better coding, but on machines of plausible alignment, it should save 4-8 bytes per connection_t. "Every little bit helps." - Rename ServerDNSAllowBrokenResolvConf to ServerDNSAllowBrokenConfig for consistency; keep old option working for backward compatibility. - Simplify the code for finding connections to use for a circuit. Changes in version 0.2.1.8-alpha - 2008-12-08 Tor 0.2.1.8-alpha fixes some crash bugs in earlier alpha releases, builds better on unusual platforms like Solaris and old OS X, and fixes a variety of other issues. o Major features: - New DirPortFrontPage option that takes an html file and publishes it as "/" on the DirPort. Now relay operators can provide a disclaimer without needing to set up a separate webserver. There's a sample disclaimer in contrib/tor-exit-notice.html. o Security fixes: - When the client is choosing entry guards, now it selects at most one guard from a given relay family. Otherwise we could end up with all of our entry points into the network run by the same operator. Suggested by Camilo Viecco. Fix on 0.1.1.11-alpha. o Major bugfixes: - Fix a DOS opportunity during the voting signature collection process at directory authorities. Spotted by rovv. Bugfix on 0.2.0.x. - Fix a possible segfault when establishing an exit connection. Bugfix on 0.2.1.5-alpha. o Minor bugfixes: - Get file locking working on win32. Bugfix on 0.2.1.6-alpha. Fixes bug 859. - Made Tor a little less aggressive about deleting expired certificates. Partial fix for bug 854. - Stop doing unaligned memory access that generated bus errors on sparc64. Bugfix on 0.2.0.10-alpha. Fix for bug 862. - Fix a crash bug when changing EntryNodes from the controller. Bugfix on 0.2.1.6-alpha. Fix for bug 867. Patched by Sebastian. - Make USR2 log-level switch take effect immediately. Bugfix on 0.1.2.8-beta. - If one win32 nameserver fails to get added, continue adding the rest, and don't automatically fail. - Use fcntl() for locking when flock() is not available. Should fix compilation on Solaris. Should fix Bug 873. Bugfix on 0.2.1.6-alpha. - Do not mark smartlist_bsearch_idx() function as ATTR_PURE. This bug could make gcc generate non-functional binary search code. Bugfix on 0.2.0.10-alpha. - Build correctly on platforms without socklen_t. - Avoid potential crash on internal error during signature collection. Fixes bug 864. Patch from rovv. - Do not use C's stdio library for writing to log files. This will improve logging performance by a minute amount, and will stop leaking fds when our disk is full. Fixes bug 861. - Stop erroneous use of O_APPEND in cases where we did not in fact want to re-seek to the end of a file before every last write(). - Correct handling of possible malformed authority signing key certificates with internal signature types. Fixes bug 880. Bugfix on 0.2.0.3-alpha. - Fix a hard-to-trigger resource leak when logging credential status. CID 349. o Minor features: - Directory mirrors no longer fetch the v1 directory or running-routers files. They are obsolete, and nobody asks for them anymore. This is the first step to making v1 authorities obsolete. o Minor features (controller): - Return circuit purposes in response to GETINFO circuit-status. Fixes bug 858. Changes in version 0.2.0.32 - 2008-11-20 Tor 0.2.0.32 fixes a major security problem in Debian and Ubuntu packages (and maybe other packages) noticed by Theo de Raadt, fixes a smaller security flaw that might allow an attacker to access local services, further improves hidden service performance, and fixes a variety of other issues. o Security fixes: - The "User" and "Group" config options did not clear the supplementary group entries for the Tor process. The "User" option is now more robust, and we now set the groups to the specified user's primary group. The "Group" option is now ignored. For more detailed logging on credential switching, set CREDENTIAL_LOG_LEVEL in common/compat.c to LOG_NOTICE or higher. Patch by Jacob Appelbaum and Steven Murdoch. Bugfix on 0.0.2pre14. Fixes bug 848 and 857. - The "ClientDNSRejectInternalAddresses" config option wasn't being consistently obeyed: if an exit relay refuses a stream because its exit policy doesn't allow it, we would remember what IP address the relay said the destination address resolves to, even if it's an internal IP address. Bugfix on 0.2.0.7-alpha; patch by rovv. o Major bugfixes: - Fix a DOS opportunity during the voting signature collection process at directory authorities. Spotted by rovv. Bugfix on 0.2.0.x. o Major bugfixes (hidden services): - When fetching v0 and v2 rendezvous service descriptors in parallel, we were failing the whole hidden service request when the v0 descriptor fetch fails, even if the v2 fetch is still pending and might succeed. Similarly, if the last v2 fetch fails, we were failing the whole hidden service request even if a v0 fetch is still pending. Fixes bug 814. Bugfix on 0.2.0.10-alpha. - When extending a circuit to a hidden service directory to upload a rendezvous descriptor using a BEGIN_DIR cell, almost 1/6 of all requests failed, because the router descriptor has not been downloaded yet. In these cases, do not attempt to upload the rendezvous descriptor, but wait until the router descriptor is downloaded and retry. Likewise, do not attempt to fetch a rendezvous descriptor from a hidden service directory for which the router descriptor has not yet been downloaded. Fixes bug 767. Bugfix on 0.2.0.10-alpha. o Minor bugfixes: - Fix several infrequent memory leaks spotted by Coverity. - When testing for libevent functions, set the LDFLAGS variable correctly. Found by Riastradh. - Avoid a bug where the FastFirstHopPK 0 option would keep Tor from bootstrapping with tunneled directory connections. Bugfix on 0.1.2.5-alpha. Fixes bug 797. Found by Erwin Lam. - When asked to connect to A.B.exit:80, if we don't know the IP for A and we know that server B rejects most-but-not all connections to port 80, we would previously reject the connection. Now, we assume the user knows what they were asking for. Fixes bug 752. Bugfix on 0.0.9rc5. Diagnosed by BarkerJr. - If we overrun our per-second write limits a little, count this as having used up our write allocation for the second, and choke outgoing directory writes. Previously, we had only counted this when we had met our limits precisely. Fixes bug 824. Patch from by rovv. Bugfix on 0.2.0.x (??). - Remove the old v2 directory authority 'lefkada' from the default list. It has been gone for many months. - Stop doing unaligned memory access that generated bus errors on sparc64. Bugfix on 0.2.0.10-alpha. Fixes bug 862. - Make USR2 log-level switch take effect immediately. Bugfix on 0.1.2.8-beta. o Minor bugfixes (controller): - Make DNS resolved events into "CLOSED", not "FAILED". Bugfix on 0.1.2.5-alpha. Fix by Robert Hogan. Resolves bug 807. Changes in version 0.2.1.7-alpha - 2008-11-08 Tor 0.2.1.7-alpha fixes a major security problem in Debian and Ubuntu packages (and maybe other packages) noticed by Theo de Raadt, fixes a smaller security flaw that might allow an attacker to access local services, adds better defense against DNS poisoning attacks on exit relays, further improves hidden service performance, and fixes a variety of other issues. o Security fixes: - The "ClientDNSRejectInternalAddresses" config option wasn't being consistently obeyed: if an exit relay refuses a stream because its exit policy doesn't allow it, we would remember what IP address the relay said the destination address resolves to, even if it's an internal IP address. Bugfix on 0.2.0.7-alpha; patch by rovv. - The "User" and "Group" config options did not clear the supplementary group entries for the Tor process. The "User" option is now more robust, and we now set the groups to the specified user's primary group. The "Group" option is now ignored. For more detailed logging on credential switching, set CREDENTIAL_LOG_LEVEL in common/compat.c to LOG_NOTICE or higher. Patch by Jacob Appelbaum and Steven Murdoch. Bugfix on 0.0.2pre14. Fixes bug 848. - Do not use or believe expired v3 authority certificates. Patch from Karsten. Bugfix in 0.2.0.x. Fixes bug 851. o Minor features: - Now NodeFamily and MyFamily config options allow spaces in identity fingerprints, so it's easier to paste them in. Suggested by Lucky Green. - Implement the 0x20 hack to better resist DNS poisoning: set the case on outgoing DNS requests randomly, and reject responses that do not match the case correctly. This logic can be disabled with the ServerDNSRandomizeCase setting, if you are using one of the 0.3% of servers that do not reliably preserve case in replies. See "Increased DNS Forgery Resistance through 0x20-Bit Encoding" for more info. - Preserve case in replies to DNSPort requests in order to support the 0x20 hack for resisting DNS poisoning attacks. o Hidden service performance improvements: - When the client launches an introduction circuit, retry with a new circuit after 30 seconds rather than 60 seconds. - Launch a second client-side introduction circuit in parallel after a delay of 15 seconds (based on work by Christian Wilms). - Hidden services start out building five intro circuits rather than three, and when the first three finish they publish a service descriptor using those. Now we publish our service descriptor much faster after restart. o Minor bugfixes: - Minor fix in the warning messages when you're having problems bootstrapping; also, be more forgiving of bootstrap problems when we're still making incremental progress on a given bootstrap phase. - When we're choosing an exit node for a circuit, and we have no pending streams, choose a good general exit rather than one that supports "all the pending streams". Bugfix on 0.1.1.x. Fix by rovv. - Send a valid END cell back when a client tries to connect to a nonexistent hidden service port. Bugfix on 0.1.2.15. Fixes bug 840. Patch from rovv. - If a broken client asks a non-exit router to connect somewhere, do not even do the DNS lookup before rejecting the connection. Fixes another case of bug 619. Patch from rovv. - Fix another case of assuming, when a specific exit is requested, that we know more than the user about what hosts it allows. Fixes another case of bug 752. Patch from rovv. - Check which hops rendezvous stream cells are associated with to prevent possible guess-the-streamid injection attacks from intermediate hops. Fixes another case of bug 446. Based on patch from rovv. - Avoid using a negative right-shift when comparing 32-bit addresses. Possible fix for bug 845 and bug 811. - Make the assert_circuit_ok() function work correctly on circuits that have already been marked for close. - Fix read-off-the-end-of-string error in unit tests when decoding introduction points. - Fix uninitialized size field for memory area allocation: may improve memory performance during directory parsing. - Treat duplicate certificate fetches as failures, so that we do not try to re-fetch an expired certificate over and over and over. - Do not say we're fetching a certificate when we'll in fact skip it because of a pending download. Changes in version 0.2.1.6-alpha - 2008-09-30 Tor 0.2.1.6-alpha further improves performance and robustness of hidden services, starts work on supporting per-country relay selection, and fixes a variety of smaller issues. o Major features: - Implement proposal 121: make it possible to build hidden services that only certain clients are allowed to connect to. This is enforced at several points, so that unauthorized clients are unable to send INTRODUCE cells to the service, or even (depending on the type of authentication) to learn introduction points. This feature raises the bar for certain kinds of active attacks against hidden services. Code by Karsten Loesing. - Relays now store and serve v2 hidden service descriptors by default, i.e., the new default value for HidServDirectoryV2 is 1. This is the last step in proposal 114, which aims to make hidden service lookups more reliable. - Start work to allow node restrictions to include country codes. The syntax to exclude nodes in a country with country code XX is "ExcludeNodes {XX}". Patch from Robert Hogan. It still needs some refinement to decide what config options should take priority if you ask to both use a particular node and exclude it. - Allow ExitNodes list to include IP ranges and country codes, just like the Exclude*Nodes lists. Patch from Robert Hogan. o Major bugfixes: - Fix a bug when parsing ports in tor_addr_port_parse() that caused Tor to fail to start if you had it configured to use a bridge relay. Fixes bug 809. Bugfix on 0.2.1.5-alpha. - When extending a circuit to a hidden service directory to upload a rendezvous descriptor using a BEGIN_DIR cell, almost 1/6 of all requests failed, because the router descriptor had not been downloaded yet. In these cases, we now wait until the router descriptor is downloaded, and then retry. Likewise, clients now skip over a hidden service directory if they don't yet have its router descriptor, rather than futilely requesting it and putting mysterious complaints in the logs. Fixes bug 767. Bugfix on 0.2.0.10-alpha. - When fetching v0 and v2 rendezvous service descriptors in parallel, we were failing the whole hidden service request when the v0 descriptor fetch fails, even if the v2 fetch is still pending and might succeed. Similarly, if the last v2 fetch fails, we were failing the whole hidden service request even if a v0 fetch is still pending. Fixes bug 814. Bugfix on 0.2.0.10-alpha. - DNS replies need to have names matching their requests, but these names should be in the questions section, not necessarily in the answers section. Fixes bug 823. Bugfix on 0.2.1.5-alpha. o Minor features: - Update to the "September 1 2008" ip-to-country file. - Allow ports 465 and 587 in the default exit policy again. We had rejected them in 0.1.0.15, because back in 2005 they were commonly misconfigured and ended up as spam targets. We hear they are better locked down these days. - Use a lockfile to make sure that two Tor processes are not simultaneously running with the same datadir. - Serve the latest v3 networkstatus consensus via the control port. Use "getinfo dir/status-vote/current/consensus" to fetch it. - Better logging about stability/reliability calculations on directory servers. - Drop the requirement to have an open dir port for storing and serving v2 hidden service descriptors. - Directory authorities now serve a /tor/dbg-stability.txt URL to help debug WFU and MTBF calculations. - Implement most of Proposal 152: allow specialized servers to permit single-hop circuits, and clients to use those servers to build single-hop circuits when using a specialized controller. Patch from Josh Albrecht. Resolves feature request 768. - Add a -p option to tor-resolve for specifying the SOCKS port: some people find host:port too confusing. - Make TrackHostExit mappings expire a while after their last use, not after their creation. Patch from Robert Hogan. - Provide circuit purposes along with circuit events to the controller. o Minor bugfixes: - Fix compile on OpenBSD 4.4-current. Bugfix on 0.2.1.5-alpha. Reported by Tas. - Fixed some memory leaks -- some quite frequent, some almost impossible to trigger -- based on results from Coverity. - When testing for libevent functions, set the LDFLAGS variable correctly. Found by Riastradh. - Fix an assertion bug in parsing policy-related options; possible fix for bug 811. - Catch and report a few more bootstrapping failure cases when Tor fails to establish a TCP connection. Cleanup on 0.2.1.x. - Avoid a bug where the FastFirstHopPK 0 option would keep Tor from bootstrapping with tunneled directory connections. Bugfix on 0.1.2.5-alpha. Fixes bug 797. Found by Erwin Lam. - When asked to connect to A.B.exit:80, if we don't know the IP for A and we know that server B rejects most-but-not all connections to port 80, we would previously reject the connection. Now, we assume the user knows what they were asking for. Fixes bug 752. Bugfix on 0.0.9rc5. Diagnosed by BarkerJr. - If we are not using BEGIN_DIR cells, don't attempt to contact hidden service directories if they have no advertised dir port. Bugfix on 0.2.0.10-alpha. - If we overrun our per-second write limits a little, count this as having used up our write allocation for the second, and choke outgoing directory writes. Previously, we had only counted this when we had met our limits precisely. Fixes bug 824. Patch by rovv. Bugfix on 0.2.0.x (??). - Avoid a "0 divided by 0" calculation when calculating router uptime at directory authorities. Bugfix on 0.2.0.8-alpha. - Make DNS resolved controller events into "CLOSED", not "FAILED". Bugfix on 0.1.2.5-alpha. Fix by Robert Hogan. Resolves bug 807. - Fix a bug where an unreachable relay would establish enough reachability testing circuits to do a bandwidth test -- if we already have a connection to the middle hop of the testing circuit, then it could establish the last hop by using the existing connection. Bugfix on 0.1.2.2-alpha, exposed when we made testing circuits no longer use entry guards in 0.2.1.3-alpha. - If we have correct permissions on $datadir, we complain to stdout and fail to start. But dangerous permissions on $datadir/cached-status/ would cause us to open a log and complain there. Now complain to stdout and fail to start in both cases. Fixes bug 820, reported by seeess. - Remove the old v2 directory authority 'lefkada' from the default list. It has been gone for many months. o Code simplifications and refactoring: - Revise the connection_new functions so that a more typesafe variant exists. This will work better with Coverity, and let us find any actual mistakes we're making here. - Refactor unit testing logic so that dmalloc can be used sensibly with unit tests to check for memory leaks. - Move all hidden-service related fields from connection and circuit structure to substructures: this way they won't eat so much memory. Changes in version 0.2.0.31 - 2008-09-03 Tor 0.2.0.31 addresses two potential anonymity issues, starts to fix a big bug we're seeing where in rare cases traffic from one Tor stream gets mixed into another stream, and fixes a variety of smaller issues. o Major bugfixes: - Make sure that two circuits can never exist on the same connection with the same circuit ID, even if one is marked for close. This is conceivably a bugfix for bug 779. Bugfix on 0.1.0.4-rc. - Relays now reject risky extend cells: if the extend cell includes a digest of all zeroes, or asks to extend back to the relay that sent the extend cell, tear down the circuit. Ideas suggested by rovv. - If not enough of our entry guards are available so we add a new one, we might use the new one even if it overlapped with the current circuit's exit relay (or its family). Anonymity bugfix pointed out by rovv. o Minor bugfixes: - Recover 3-7 bytes that were wasted per memory chunk. Fixes bug 794; bug spotted by rovv. Bugfix on 0.2.0.1-alpha. - Correctly detect the presence of the linux/netfilter_ipv4.h header when building against recent kernels. Bugfix on 0.1.2.1-alpha. - Pick size of default geoip filename string correctly on windows. Fixes bug 806. Bugfix on 0.2.0.30. - Make the autoconf script accept the obsolete --with-ssl-dir option as an alias for the actually-working --with-openssl-dir option. Fix the help documentation to recommend --with-openssl-dir. Based on a patch by "Dave". Bugfix on 0.2.0.1-alpha. - When using the TransPort option on OpenBSD, and using the User option to change UID and drop privileges, make sure to open /dev/pf before dropping privileges. Fixes bug 782. Patch from Christopher Davis. Bugfix on 0.1.2.1-alpha. - Try to attach connections immediately upon receiving a RENDEZVOUS2 or RENDEZVOUS_ESTABLISHED cell. This can save a second or two on the client side when connecting to a hidden service. Bugfix on 0.0.6pre1. Found and fixed by Christian Wilms; resolves bug 743. - When closing an application-side connection because its circuit is getting torn down, generate the stream event correctly. Bugfix on 0.1.2.x. Anonymous patch. Changes in version 0.2.1.5-alpha - 2008-08-31 Tor 0.2.1.5-alpha moves us closer to handling IPv6 destinations, puts in a lot of the infrastructure for adding authorization to hidden services, lays the groundwork for having clients read their load balancing information out of the networkstatus consensus rather than the individual router descriptors, addresses two potential anonymity issues, and fixes a variety of smaller issues. o Major features: - Convert many internal address representations to optionally hold IPv6 addresses. - Generate and accept IPv6 addresses in many protocol elements. - Make resolver code handle nameservers located at ipv6 addresses. - Begin implementation of proposal 121 ("Client authorization for hidden services"): configure hidden services with client authorization, publish descriptors for them, and configure authorization data for hidden services at clients. The next step is to actually access hidden services that perform client authorization. - More progress toward proposal 141: Network status consensus documents and votes now contain bandwidth information for each router and a summary of that router's exit policy. Eventually this will be used by clients so that they do not have to download every known descriptor before building circuits. o Major bugfixes (on 0.2.0.x and before): - When sending CREATED cells back for a given circuit, use a 64-bit connection ID to find the right connection, rather than an addr:port combination. Now that we can have multiple OR connections between the same ORs, it is no longer possible to use addr:port to uniquely identify a connection. - Relays now reject risky extend cells: if the extend cell includes a digest of all zeroes, or asks to extend back to the relay that sent the extend cell, tear down the circuit. Ideas suggested by rovv. - If not enough of our entry guards are available so we add a new one, we might use the new one even if it overlapped with the current circuit's exit relay (or its family). Anonymity bugfix pointed out by rovv. o Minor bugfixes: - Recover 3-7 bytes that were wasted per memory chunk. Fixes bug 794; bug spotted by rovv. Bugfix on 0.2.0.1-alpha. - When using the TransPort option on OpenBSD, and using the User option to change UID and drop privileges, make sure to open /dev/pf before dropping privileges. Fixes bug 782. Patch from Christopher Davis. Bugfix on 0.1.2.1-alpha. - Correctly detect the presence of the linux/netfilter_ipv4.h header when building against recent kernels. Bugfix on 0.1.2.1-alpha. - Add a missing safe_str() call for a debug log message. - Use 64 bits instead of 32 bits for connection identifiers used with the controller protocol, to greatly reduce risk of identifier reuse. - Make the autoconf script accept the obsolete --with-ssl-dir option as an alias for the actually-working --with-openssl-dir option. Fix the help documentation to recommend --with-openssl-dir. Based on a patch by "Dave". Bugfix on 0.2.0.1-alpha. o Minor features: - Rate-limit too-many-sockets messages: when they happen, they happen a lot. Resolves bug 748. - Resist DNS poisoning a little better by making sure that names in answer sections match. - Print the SOCKS5 error message string as well as the error code when a tor-resolve request fails. Patch from Jacob. Changes in version 0.2.1.4-alpha - 2008-08-04 Tor 0.2.1.4-alpha fixes a pair of crash bugs in 0.2.1.3-alpha. o Major bugfixes: - The address part of exit policies was not correctly written to router descriptors. This generated router descriptors that failed their self-checks. Noticed by phobos, fixed by Karsten. Bugfix on 0.2.1.3-alpha. - Tor triggered a false assert when extending a circuit to a relay but we already have a connection open to that relay. Noticed by phobos, fixed by Karsten. Bugfix on 0.2.1.3-alpha. o Minor bugfixes: - Fix a hidden service logging bug: in some edge cases, the router descriptor of a previously picked introduction point becomes obsolete and we need to give up on it rather than continually complaining that it has become obsolete. Observed by xiando. Bugfix on 0.2.1.3-alpha. o Removed features: - Take out the TestVia config option, since it was a workaround for a bug that was fixed in Tor 0.1.1.21. Changes in version 0.2.1.3-alpha - 2008-08-03 Tor 0.2.1.3-alpha implements most of the pieces to prevent infinite-length circuit attacks (see proposal 110); fixes a bug that might cause exit relays to corrupt streams they send back; allows address patterns (e.g. 255.128.0.0/16) to appear in ExcludeNodes and ExcludeExitNodes config options; and fixes a big pile of bugs. o Bootstrapping bugfixes (on 0.2.1.x-alpha): - Send a bootstrap problem "warn" event on the first problem if the reason is NO_ROUTE (that is, our network is down). o Major features: - Implement most of proposal 110: The first K cells to be sent along a circuit are marked as special "early" cells; only K "early" cells will be allowed. Once this code is universal, we can block certain kinds of DOS attack by requiring that EXTEND commands must be sent using an "early" cell. o Major bugfixes: - Try to attach connections immediately upon receiving a RENDEZVOUS2 or RENDEZVOUS_ESTABLISHED cell. This can save a second or two on the client side when connecting to a hidden service. Bugfix on 0.0.6pre1. Found and fixed by Christian Wilms; resolves bug 743. - Ensure that two circuits can never exist on the same connection with the same circuit ID, even if one is marked for close. This is conceivably a bugfix for bug 779; fixes a bug on 0.1.0.4-rc. o Minor features: - When relays do their initial bandwidth measurement, don't limit to just our entry guards for the test circuits. Otherwise we tend to have multiple test circuits going through a single entry guard, which makes our bandwidth test less accurate. Fixes part of bug 654; patch contributed by Josh Albrecht. - Add an ExcludeExitNodes option so users can list a set of nodes that should be be excluded from the exit node position, but allowed elsewhere. Implements proposal 151. - Allow address patterns (e.g., 255.128.0.0/16) to appear in ExcludeNodes and ExcludeExitNodes lists. - Change the implementation of ExcludeNodes and ExcludeExitNodes to be more efficient. Formerly it was quadratic in the number of servers; now it should be linear. Fixes bug 509. - Save 16-22 bytes per open circuit by moving the n_addr, n_port, and n_conn_id_digest fields into a separate structure that's only needed when the circuit has not yet attached to an n_conn. o Minor bugfixes: - Change the contrib/tor.logrotate script so it makes the new logs as "_tor:_tor" rather than the default, which is generally "root:wheel". Fixes bug 676, reported by Serge Koksharov. - Stop using __attribute__((nonnull)) with GCC: it can give us useful warnings (occasionally), but it can also cause the compiler to eliminate error-checking code. Suggested by Peter Gutmann. - When a hidden service is giving up on an introduction point candidate that was not included in the last published rendezvous descriptor, don't reschedule publication of the next descriptor. Fixes bug 763. Bugfix on 0.0.9.3. - Mark RendNodes, RendExcludeNodes, HiddenServiceNodes, and HiddenServiceExcludeNodes as obsolete: they never worked properly, and nobody claims to be using them. Fixes bug 754. Bugfix on 0.1.0.1-rc. Patch from Christian Wilms. - Fix a small alignment and memory-wasting bug on buffer chunks. Spotted by rovv. o Minor bugfixes (controller): - When closing an application-side connection because its circuit is getting torn down, generate the stream event correctly. Bugfix on 0.1.2.x. Anonymous patch. o Removed features: - Remove all backward-compatibility code to support relays running versions of Tor so old that they no longer work at all on the Tor network. Changes in version 0.2.0.30 - 2008-07-15 o Minor bugfixes: - Stop using __attribute__((nonnull)) with GCC: it can give us useful warnings (occasionally), but it can also cause the compiler to eliminate error-checking code. Suggested by Peter Gutmann. Changes in version 0.2.0.29-rc - 2008-07-08 Tor 0.2.0.29-rc fixes two big bugs with using bridges, fixes more hidden-service performance bugs, and fixes a bunch of smaller bugs. o Major bugfixes: - If you have more than one bridge but don't know their keys, you would only launch a request for the descriptor of the first one on your list. (Tor considered launching requests for the others, but found that it already had a connection on the way for $0000...0000 so it didn't open another.) Bugfix on 0.2.0.x. - If you have more than one bridge but don't know their keys, and the connection to one of the bridges failed, you would cancel all pending bridge connections. (After all, they all have the same digest.) Bugfix on 0.2.0.x. - When a hidden service was trying to establish an introduction point, and Tor had built circuits preemptively for such purposes, we were ignoring all the preemptive circuits and launching a new one instead. Bugfix on 0.2.0.14-alpha. - When a hidden service was trying to establish an introduction point, and Tor *did* manage to reuse one of the preemptively built circuits, it didn't correctly remember which one it used, so it asked for another one soon after, until there were no more preemptive circuits, at which point it launched one from scratch. Bugfix on 0.0.9.x. - Make directory servers include the X-Your-Address-Is: http header in their responses even for begin_dir conns. Now clients who only ever use begin_dir connections still have a way to learn their IP address. Fixes bug 737; bugfix on 0.2.0.22-rc. Reported by goldy. o Minor bugfixes: - Fix a macro/CPP interaction that was confusing some compilers: some GCCs don't like #if/#endif pairs inside macro arguments. Fixes bug 707. - Fix macro collision between OpenSSL 0.9.8h and Windows headers. Fixes bug 704; fix from Steven Murdoch. - When opening /dev/null in finish_daemonize(), do not pass the O_CREAT flag. Fortify was complaining, and correctly so. Fixes bug 742; fix from Michael Scherer. Bugfix on 0.0.2pre19. - Correctly detect transparent proxy support on Linux hosts that require in.h to be included before netfilter_ipv4.h. Patch from coderman. - Disallow session resumption attempts during the renegotiation stage of the v2 handshake protocol. Clients should never be trying session resumption at this point, but apparently some did, in ways that caused the handshake to fail. Bugfix on 0.2.0.20-rc. Bug found by Geoff Goodell. Changes in version 0.2.1.2-alpha - 2008-06-20 Tor 0.2.1.2-alpha includes a new "TestingTorNetwork" config option to make it easier to set up your own private Tor network; fixes several big bugs with using more than one bridge relay; fixes a big bug with offering hidden services quickly after Tor starts; and uses a better API for reporting potential bootstrapping problems to the controller. o Major features: - New TestingTorNetwork config option to allow adjustment of previously constant values that, while reasonable, could slow bootstrapping. Implements proposal 135. Patch from Karsten. o Major bugfixes: - If you have more than one bridge but don't know their digests, you would only learn a request for the descriptor of the first one on your list. (Tor considered launching requests for the others, but found that it already had a connection on the way for $0000...0000 so it didn't open another.) Bugfix on 0.2.0.x. - If you have more than one bridge but don't know their digests, and the connection to one of the bridges failed, you would cancel all pending bridge connections. (After all, they all have the same digest.) Bugfix on 0.2.0.x. - When establishing a hidden service, introduction points that originate from cannibalized circuits are completely ignored and not included in rendezvous service descriptors. This might be another reason for delay in making a hidden service available. Bugfix from long ago (0.0.9.x?) o Minor features: - Allow OpenSSL to use dynamic locks if it wants. - When building a consensus, do not include routers that are down. This will cut down 30% to 40% on consensus size. Implements proposal 138. - In directory authorities' approved-routers files, allow fingerprints with or without space. - Add a "GETINFO /status/bootstrap-phase" controller option, so the controller can query our current bootstrap state in case it attaches partway through and wants to catch up. - Send an initial "Starting" bootstrap status event, so we have a state to start out in. o Minor bugfixes: - Asking for a conditional consensus at .../consensus/ would crash a dirserver if it did not already have a consensus. Bugfix on 0.2.1.1-alpha. - Clean up some macro/CPP interactions: some GCC versions don't like #if/#endif pairs inside macro arguments. Fixes bug 707. Bugfix on 0.2.0.x. o Bootstrapping bugfixes (on 0.2.1.1-alpha): - Directory authorities shouldn't complain about bootstrapping problems just because they do a lot of reachability testing and some of the connection attempts fail. - Start sending "count" and "recommendation" key/value pairs in bootstrap problem status events, so the controller can hear about problems even before Tor decides they're worth reporting for sure. - If you're using bridges, generate "bootstrap problem" warnings as soon as you run out of working bridges, rather than waiting for ten failures -- which will never happen if you have less than ten bridges. - If we close our OR connection because there's been a circuit pending on it for too long, we were telling our bootstrap status events "REASON=NONE". Now tell them "REASON=TIMEOUT". Changes in version 0.2.1.1-alpha - 2008-06-13 Tor 0.2.1.1-alpha fixes a lot of memory fragmentation problems that were making the Tor process bloat especially on Linux; makes our TLS handshake blend in better; sends "bootstrap phase" status events to the controller, so it can keep the user informed of progress (and problems) fetching directory information and establishing circuits; and adds a variety of smaller features. o Major features: - More work on making our TLS handshake blend in: modify the list of ciphers advertised by OpenSSL in client mode to even more closely resemble a common web browser. We cheat a little so that we can advertise ciphers that the locally installed OpenSSL doesn't know about. - Start sending "bootstrap phase" status events to the controller, so it can keep the user informed of progress fetching directory information and establishing circuits. Also inform the controller if we think we're stuck at a particular bootstrap phase. Implements proposal 137. - Resume using OpenSSL's RAND_poll() for better (and more portable) cross-platform entropy collection again. We used to use it, then stopped using it because of a bug that could crash systems that called RAND_poll when they had a lot of fds open. It looks like the bug got fixed in late 2006. Our new behavior is to call RAND_poll() at startup, and to call RAND_poll() when we reseed later only if we have a non-buggy OpenSSL version. o Major bugfixes: - When we choose to abandon a new entry guard because we think our older ones might be better, close any circuits pending on that new entry guard connection. This fix should make us recover much faster when our network is down and then comes back. Bugfix on 0.1.2.8-beta; found by lodger. o Memory fixes and improvements: - Add a malloc_good_size implementation to OpenBSD_malloc_linux.c, to avoid unused RAM in buffer chunks and memory pools. - Speed up parsing and cut down on memory fragmentation by using stack-style allocations for parsing directory objects. Previously, this accounted for over 40% of allocations from within Tor's code on a typical directory cache. - Use a Bloom filter rather than a digest-based set to track which descriptors we need to keep around when we're cleaning out old router descriptors. This speeds up the computation significantly, and may reduce fragmentation. - Reduce the default smartlist size from 32 to 16; it turns out that most smartlists hold around 8-12 elements tops. - Make dumpstats() log the fullness and size of openssl-internal buffers. - If the user has applied the experimental SSL_MODE_RELEASE_BUFFERS patch to their OpenSSL, turn it on to save memory on servers. This patch will (with any luck) get included in a mainline distribution before too long. - Never use OpenSSL compression: it wastes RAM and CPU trying to compress cells, which are basically all encrypted, compressed, or both. o Minor bugfixes: - Stop reloading the router list from disk for no reason when we run out of reachable directory mirrors. Once upon a time reloading it would set the 'is_running' flag back to 1 for them. It hasn't done that for a long time. - In very rare situations new hidden service descriptors were published earlier than 30 seconds after the last change to the service. (We currently think that a hidden service descriptor that's been stable for 30 seconds is worth publishing.) o Minor features: - Allow separate log levels to be configured for different logging domains. For example, this allows one to log all notices, warnings, or errors, plus all memory management messages of level debug or higher, with: Log [MM] debug-err [*] notice-err file /var/log/tor. - Add a couple of extra warnings to --enable-gcc-warnings for GCC 4.3, and stop using a warning that had become unfixably verbose under GCC 4.3. - New --hush command-line option similar to --quiet. While --quiet disables all logging to the console on startup, --hush limits the output to messages of warning and error severity. - Servers support a new URL scheme for consensus downloads that allows the client to specify which authorities are trusted. The server then only sends the consensus if the client will trust it. Otherwise a 404 error is sent back. Clients use this new scheme when the server supports it (meaning it's running 0.2.1.1-alpha or later). Implements proposal 134. - New configure/torrc options (--enable-geoip-stats, DirRecordUsageByCountry) to record how many IPs we've served directory info to in each country code, how many status documents total we've sent to each country code, and what share of the total directory requests we should expect to see. - Use the TLS1 hostname extension to more closely resemble browser behavior. - Lots of new unit tests. - Add a macro to implement the common pattern of iterating through two parallel lists in lockstep. Changes in version 0.2.0.28-rc - 2008-06-13 Tor 0.2.0.28-rc fixes an anonymity-related bug, fixes a hidden-service performance bug, and fixes a bunch of smaller bugs. o Anonymity fixes: - Fix a bug where, when we were choosing the 'end stream reason' to put in our relay end cell that we send to the exit relay, Tor clients on Windows were sometimes sending the wrong 'reason'. The anonymity problem is that exit relays may be able to guess whether the client is running Windows, thus helping partition the anonymity set. Down the road we should stop sending reasons to exit relays, or otherwise prevent future versions of this bug. o Major bugfixes: - While setting up a hidden service, some valid introduction circuits were overlooked and abandoned. This might be the reason for the long delay in making a hidden service available. Bugfix on 0.2.0.14-alpha. o Minor features: - Update to the "June 9 2008" ip-to-country file. - Run 'make test' as part of 'make dist', so we stop releasing so many development snapshots that fail their unit tests. o Minor bugfixes: - When we're checking if we have enough dir info for each relay to begin establishing circuits, make sure that we actually have the descriptor listed in the consensus, not just any descriptor. Bugfix on 0.1.2.x. - Bridge relays no longer print "xx=0" in their extrainfo document for every single country code in the geoip db. Bugfix on 0.2.0.27-rc. - Only warn when we fail to load the geoip file if we were planning to include geoip stats in our extrainfo document. Bugfix on 0.2.0.27-rc. - If we change our MaxAdvertisedBandwidth and then reload torrc, Tor won't realize it should publish a new relay descriptor. Fixes bug 688, reported by mfr. Bugfix on 0.1.2.x. - When we haven't had any application requests lately, don't bother logging that we have expired a bunch of descriptors. Bugfix on 0.1.2.x. - Make relay cells written on a connection count as non-padding when tracking how long a connection has been in use. Bugfix on 0.2.0.1-alpha. Spotted by lodger. - Fix unit tests in 0.2.0.27-rc. - Fix compile on Windows. Changes in version 0.2.0.27-rc - 2008-06-03 Tor 0.2.0.27-rc adds a few features we left out of the earlier release candidates. In particular, we now include an IP-to-country GeoIP database, so controllers can easily look up what country a given relay is in, and so bridge relays can give us some sanitized summaries about which countries are making use of bridges. (See proposal 126-geoip-fetching.txt for details.) o Major features: - Include an IP-to-country GeoIP file in the tarball, so bridge relays can report sanitized summaries of the usage they're seeing. o Minor features: - Add a "PURPOSE=" argument to "STREAM NEW" events, as suggested by Robert Hogan. Fixes the first part of bug 681. - Make bridge authorities never serve extrainfo docs. - Add support to detect Libevent versions in the 1.4.x series on mingw. - Fix build on gcc 4.3 with --enable-gcc-warnings set. - Include a new contrib/tor-exit-notice.html file that exit relay operators can put on their website to help reduce abuse queries. o Minor bugfixes: - When tunneling an encrypted directory connection, and its first circuit fails, do not leave it unattached and ask the controller to deal. Fixes the second part of bug 681. - Make bridge authorities correctly expire old extrainfo documents from time to time. Changes in version 0.2.0.26-rc - 2008-05-13 Tor 0.2.0.26-rc fixes a major security vulnerability caused by a bug in Debian's OpenSSL packages. All users running any 0.2.0.x version should upgrade, whether they're running Debian or not. o Major security fixes: - Use new V3 directory authority keys on the tor26, gabelmoo, and moria1 V3 directory authorities. The old keys were generated with a vulnerable version of Debian's OpenSSL package, and must be considered compromised. Other authorities' keys were not generated with an affected version of OpenSSL. o Major bugfixes: - List authority signatures as "unrecognized" based on DirServer lines, not on cert cache. Bugfix on 0.2.0.x. o Minor features: - Add a new V3AuthUseLegacyKey option to make it easier for authorities to change their identity keys if they have to. Changes in version 0.2.0.25-rc - 2008-04-23 Tor 0.2.0.25-rc makes Tor work again on OS X and certain BSDs. o Major bugfixes: - Remember to initialize threading before initializing logging. Otherwise, many BSD-family implementations will crash hard on startup. Fixes bug 671. Bugfix on 0.2.0.24-rc. o Minor bugfixes: - Authorities correctly free policies on bad servers on exit. Fixes bug 672. Bugfix on 0.2.0.x. Changes in version 0.2.0.24-rc - 2008-04-22 Tor 0.2.0.24-rc adds dizum (run by Alex de Joode) as the new sixth v3 directory authority, makes relays with dynamic IP addresses and no DirPort notice more quickly when their IP address changes, fixes a few rare crashes and memory leaks, and fixes a few other miscellaneous bugs. o New directory authorities: - Take lefkada out of the list of v3 directory authorities, since it has been down for months. - Set up dizum (run by Alex de Joode) as the new sixth v3 directory authority. o Major bugfixes: - Detect address changes more quickly on non-directory mirror relays. Bugfix on 0.2.0.18-alpha; fixes bug 652. o Minor features (security): - Reject requests for reverse-dns lookup of names that are in a private address space. Patch from lodger. - Non-exit relays no longer allow DNS requests. Fixes bug 619. Patch from lodger. o Minor bugfixes (crashes): - Avoid a rare assert that can trigger when Tor doesn't have much directory information yet and it tries to fetch a v2 hidden service descriptor. Fixes bug 651, reported by nwf. - Initialize log mutex before initializing dmalloc. Otherwise, running with dmalloc would crash. Bugfix on 0.2.0.x-alpha. - Use recursive pthread mutexes in order to avoid deadlock when logging debug-level messages to a controller. Bug spotted by nwf, bugfix on 0.2.0.16-alpha. o Minor bugfixes (resource management): - Keep address policies from leaking memory: start their refcount at 1, not 2. Bugfix on 0.2.0.16-alpha. - Free authority certificates on exit, so they don't look like memory leaks. Bugfix on 0.2.0.19-alpha. - Free static hashtables for policy maps and for TLS connections on shutdown, so they don't look like memory leaks. Bugfix on 0.2.0.x. - Avoid allocating extra space when computing consensuses on 64-bit platforms. Bug spotted by aakova. o Minor bugfixes (misc): - Do not read the configuration file when we've only been told to generate a password hash. Fixes bug 643. Bugfix on 0.0.9pre5. Fix based on patch from Sebastian Hahn. - Exit relays that are used as a client can now reach themselves using the .exit notation, rather than just launching an infinite pile of circuits. Fixes bug 641. Reported by Sebastian Hahn. - When attempting to open a logfile fails, tell us why. - Fix a dumb bug that was preventing us from knowing that we should preemptively build circuits to handle expected directory requests. Fixes bug 660. Bugfix on 0.1.2.x. - Warn less verbosely about clock skew from netinfo cells from untrusted sources. Fixes bug 663. - Make controller stream events for DNS requests more consistent, by adding "new stream" events for DNS requests, and removing spurious "stream closed" events" for cached reverse resolves. Patch from mwenge. Fixes bug 646. - Correctly notify one-hop connections when a circuit build has failed. Possible fix for bug 669. Found by lodger. Changes in version 0.2.0.23-rc - 2008-03-24 Tor 0.2.0.23-rc is the fourth release candidate for the 0.2.0 series. It makes bootstrapping faster if the first directory mirror you contact is down. The bundles also include the new Vidalia 0.1.2 release. o Major bugfixes: - When a tunneled directory request is made to a directory server that's down, notice after 30 seconds rather than 120 seconds. Also, fail any begindir streams that are pending on it, so they can retry elsewhere. This was causing multi-minute delays on bootstrap. Changes in version 0.2.0.22-rc - 2008-03-18 Tor 0.2.0.22-rc is the third release candidate for the 0.2.0 series. It enables encrypted directory connections by default for non-relays, fixes some broken TLS behavior we added in 0.2.0.20-rc, and resolves many other bugs. The bundles also include Vidalia 0.1.1 and Torbutton 1.1.17. o Major features: - Enable encrypted directory connections by default for non-relays, so censor tools that block Tor directory connections based on their plaintext patterns will no longer work. This means Tor works in certain censored countries by default again. o Major bugfixes: - Make sure servers always request certificates from clients during TLS renegotiation. Reported by lodger; bugfix on 0.2.0.20-rc. - Do not enter a CPU-eating loop when a connection is closed in the middle of client-side TLS renegotiation. Fixes bug 622. Bug diagnosed by lodger; bugfix on 0.2.0.20-rc. - Fix assertion failure that could occur when a blocked circuit became unblocked, and it had pending client DNS requests. Bugfix on 0.2.0.1-alpha. Fixes bug 632. o Minor bugfixes (on 0.1.2.x): - Generate "STATUS_SERVER" events rather than misspelled "STATUS_SEVER" events. Caught by mwenge. - When counting the number of bytes written on a TLS connection, look at the BIO actually used for writing to the network, not at the BIO used (sometimes) to buffer data for the network. Looking at different BIOs could result in write counts on the order of ULONG_MAX. Fixes bug 614. - On Windows, correctly detect errors when listing the contents of a directory. Fix from lodger. o Minor bugfixes (on 0.2.0.x): - Downgrade "sslv3 alert handshake failure" message to INFO. - If we set RelayBandwidthRate and RelayBandwidthBurst very high but left BandwidthRate and BandwidthBurst at the default, we would be silently limited by those defaults. Now raise them to match the RelayBandwidth* values. - Fix the SVK version detection logic to work correctly on a branch. - Make --enable-openbsd-malloc work correctly on Linux with alpha CPUs. Fixes bug 625. - Logging functions now check that the passed severity is sane. - Use proper log levels in the testsuite call of get_interface_address6(). - When using a nonstandard malloc, do not use the platform values for HAVE_MALLOC_GOOD_SIZE or HAVE_MALLOC_USABLE_SIZE. - Make the openbsd malloc code use 8k pages on alpha CPUs and 16k pages on ia64. - Detect mismatched page sizes when using --enable-openbsd-malloc. - Avoid double-marked-for-close warning when certain kinds of invalid .in-addr.arpa addresses are passed to the DNSPort. Part of a fix for bug 617. Bugfix on 0.2.0.1-alpha. - Make sure that the "NULL-means-reject *:*" convention is followed by all the policy manipulation functions, avoiding some possible crash bugs. Bug found by lodger. Bugfix on 0.2.0.16-alpha. - Fix the implementation of ClientDNSRejectInternalAddresses so that it actually works, and doesn't warn about every single reverse lookup. Fixes the other part of bug 617. Bugfix on 0.2.0.1-alpha. o Minor features: - Only log guard node status when guard node status has changed. - Downgrade the 3 most common "INFO" messages to "DEBUG". This will make "INFO" 75% less verbose. Changes in version 0.2.0.21-rc - 2008-03-02 Tor 0.2.0.21-rc is the second release candidate for the 0.2.0 series. It makes Tor work well with Vidalia again, fixes a rare assert bug, and fixes a pair of more minor bugs. The bundles also include Vidalia 0.1.0 and Torbutton 1.1.16. o Major bugfixes: - The control port should declare that it requires password auth when HashedControlSessionPassword is set too. Patch from Matt Edman; bugfix on 0.2.0.20-rc. Fixes bug 615. - Downgrade assert in connection_buckets_decrement() to a log message. This may help us solve bug 614, and in any case will make its symptoms less severe. Bugfix on 0.2.0.20-rc. Reported by fredzupy. - We were sometimes miscounting the number of bytes read from the network, causing our rate limiting to not be followed exactly. Bugfix on 0.2.0.16-alpha. Reported by lodger. o Minor bugfixes: - Fix compilation with OpenSSL 0.9.8 and 0.9.8a. All other supported OpenSSL versions should have been working fine. Diagnosis and patch from lodger, Karsten Loesing, and Sebastian Hahn. Fixes bug 616. Bugfix on 0.2.0.20-rc. Changes in version 0.2.0.20-rc - 2008-02-24 Tor 0.2.0.20-rc is the first release candidate for the 0.2.0 series. It makes more progress towards normalizing Tor's TLS handshake, makes hidden services work better again, helps relays bootstrap if they don't know their IP address, adds optional support for linking in openbsd's allocator or tcmalloc, allows really fast relays to scale past 15000 sockets, and fixes a bunch of minor bugs reported by Veracode. o Major features: - Enable the revised TLS handshake based on the one designed by Steven Murdoch in proposal 124, as revised in proposal 130. It includes version negotiation for OR connections as described in proposal 105. The new handshake is meant to be harder for censors to fingerprint, and it adds the ability to detect certain kinds of man-in-the-middle traffic analysis attacks. The version negotiation feature will allow us to improve Tor's link protocol more safely in the future. - Choose which bridge to use proportional to its advertised bandwidth, rather than uniformly at random. This should speed up Tor for bridge users. Also do this for people who set StrictEntryNodes. - When a TrackHostExits-chosen exit fails too many times in a row, stop using it. Bugfix on 0.1.2.x; fixes bug 437. o Major bugfixes: - Resolved problems with (re-)fetching hidden service descriptors. Patch from Karsten Loesing; fixes problems with 0.2.0.18-alpha and 0.2.0.19-alpha. - If we only ever used Tor for hidden service lookups or posts, we would stop building circuits and start refusing connections after 24 hours, since we falsely believed that Tor was dormant. Reported by nwf; bugfix on 0.1.2.x. - Servers that don't know their own IP address should go to the authorities for their first directory fetch, even if their DirPort is off or if they don't know they're reachable yet. This will help them bootstrap better. Bugfix on 0.2.0.18-alpha; fixes bug 609. - When counting the number of open sockets, count not only the number of sockets we have received from the socket() call, but also the number we've gotten from accept() and socketpair(). This bug made us fail to count all sockets that we were using for incoming connections. Bugfix on 0.2.0.x. - Fix code used to find strings within buffers, when those strings are not in the first chunk of the buffer. Bugfix on 0.2.0.x. - Fix potential segfault when parsing HTTP headers. Bugfix on 0.2.0.x. - Add a new __HashedControlSessionPassword option for controllers to use for one-off session password hashes that shouldn't get saved to disk by SAVECONF --- Vidalia users were accumulating a pile of HashedControlPassword lines in their torrc files, one for each time they had restarted Tor and then clicked Save. Make Tor automatically convert "HashedControlPassword" to this new option but only when it's given on the command line. Partial fix for bug 586. o Minor features (performance): - Tune parameters for cell pool allocation to minimize amount of RAM overhead used. - Add OpenBSD malloc code from phk as an optional malloc replacement on Linux: some glibc libraries do very poorly with Tor's memory allocation patterns. Pass --enable-openbsd-malloc to get the replacement malloc code. - Add a --with-tcmalloc option to the configure script to link against tcmalloc (if present). Does not yet search for non-system include paths. - Stop imposing an arbitrary maximum on the number of file descriptors used for busy servers. Bug reported by Olaf Selke; patch from Sebastian Hahn. o Minor features (other): - When SafeLogging is disabled, log addresses along with all TLS errors. - When building with --enable-gcc-warnings, check for whether Apple's warning "-Wshorten-64-to-32" is available. - Add a --passphrase-fd argument to the tor-gencert command for scriptability. o Minor bugfixes (memory leaks and code problems): - We were leaking a file descriptor if Tor started with a zero-length cached-descriptors file. Patch by freddy77; bugfix on 0.1.2. - Detect size overflow in zlib code. Reported by Justin Ferguson and Dan Kaminsky. - We were comparing the raw BridgePassword entry with a base64'ed version of it, when handling a "/tor/networkstatus-bridges" directory request. Now compare correctly. Noticed by Veracode. - Recover from bad tracked-since value in MTBF-history file. Should fix bug 537. - Alter the code that tries to recover from unhandled write errors, to not try to flush onto a socket that's given us unhandled errors. Bugfix on 0.1.2.x. - Make Unix controlsockets work correctly on OpenBSD. Patch from tup. Bugfix on 0.2.0.3-alpha. o Minor bugfixes (other): - If we have an extra-info document for our server, always make it available on the control port, even if we haven't gotten a copy of it from an authority yet. Patch from mwenge. - Log the correct memory chunk sizes for empty RAM chunks in mempool.c. - Directory mirrors no longer include a guess at the client's IP address if the connection appears to be coming from the same /24 network; it was producing too many wrong guesses. - Make the new hidden service code respect the SafeLogging setting. Bugfix on 0.2.0.x. Patch from Karsten. - When starting as an authority, do not overwrite all certificates cached from other authorities. Bugfix on 0.2.0.x. Fixes bug 606. - If we're trying to flush the last bytes on a connection (for example, when answering a directory request), reset the time-to-give-up timeout every time we manage to write something on the socket. Bugfix on 0.1.2.x. - Change the behavior of "getinfo status/good-server-descriptor" so it doesn't return failure when any authority disappears. - Even though the man page said that "TrackHostExits ." should work, nobody had ever implemented it. Bugfix on 0.1.0.x. - Report TLS "zero return" case as a "clean close" and "IO error" as a "close". Stop calling closes "unexpected closes": existing Tors don't use SSL_close(), so having a connection close without the TLS shutdown handshake is hardly unexpected. - Send NAMESERVER_STATUS messages for a single failed nameserver correctly. o Code simplifications and refactoring: - Remove the tor_strpartition function: its logic was confused, and it was only used for one thing that could be implemented far more easily. Changes in version 0.2.0.19-alpha - 2008-02-09 Tor 0.2.0.19-alpha makes more progress towards normalizing Tor's TLS handshake, makes path selection for relays more secure and IP address guessing more robust, and generally fixes a lot of bugs in preparation for calling the 0.2.0 branch stable. o Major features: - Do not include recognizeable strings in the commonname part of Tor's x509 certificates. o Major bugfixes: - If we're a relay, avoid picking ourselves as an introduction point, a rendezvous point, or as the final hop for internal circuits. Bug reported by taranis and lodger. Bugfix on 0.1.2.x. - Patch from "Andrew S. Lists" to catch when we contact a directory mirror at IP address X and he says we look like we're coming from IP address X. Bugfix on 0.1.2.x. o Minor features (security): - Be more paranoid about overwriting sensitive memory on free(), as a defensive programming tactic to ensure forward secrecy. o Minor features (directory authority): - Actually validate the options passed to AuthDirReject, AuthDirInvalid, AuthDirBadDir, and AuthDirBadExit. - Reject router descriptors with out-of-range bandwidthcapacity or bandwidthburst values. o Minor features (controller): - Reject controller commands over 1MB in length. This keeps rogue processes from running us out of memory. o Minor features (misc): - Give more descriptive well-formedness errors for out-of-range hidden service descriptor/protocol versions. - Make memory debugging information describe more about history of cell allocation, so we can help reduce our memory use. o Deprecated features (controller): - The status/version/num-versioning and status/version/num-concurring GETINFO options are no longer useful in the v3 directory protocol: treat them as deprecated, and warn when they're used. o Minor bugfixes: - When our consensus networkstatus has been expired for a while, stop being willing to build circuits using it. Fixes bug 401. Bugfix on 0.1.2.x. - Directory caches now fetch certificates from all authorities listed in a networkstatus consensus, even when they do not recognize them. Fixes bug 571. Bugfix on 0.2.0.x. - When connecting to a bridge without specifying its key, insert the connection into the identity-to-connection map as soon as a key is learned. Fixes bug 574. Bugfix on 0.2.0.x. - Detect versions of OS X where malloc_good_size() is present in the library but never actually declared. Resolves bug 587. Bugfix on 0.2.0.x. - Stop incorrectly truncating zlib responses to directory authority signature download requests. Fixes bug 593. Bugfix on 0.2.0.x. - Stop recommending that every server operator send mail to tor-ops. Resolves bug 597. Bugfix on 0.1.2.x. - Don't trigger an assert if we start a directory authority with a private IP address (like 127.0.0.1). - Avoid possible failures when generating a directory with routers with over-long versions strings, or too many flags set. Bugfix on 0.1.2.x. - If an attempt to launch a DNS resolve request over the control port fails because we have overrun the limit on the number of connections, tell the controller that the request has failed. - Avoid using too little bandwidth when our clock skips a few seconds. Bugfix on 0.1.2.x. - Fix shell error when warning about missing packages in configure script, on Fedora or Red Hat machines. Bugfix on 0.2.0.x. - Do not become confused when receiving a spurious VERSIONS-like cell from a confused v1 client. Bugfix on 0.2.0.x. - Re-fetch v2 (as well as v0) rendezvous descriptors when all introduction points for a hidden service have failed. Patch from Karsten Loesing. Bugfix on 0.2.0.x. o Code simplifications and refactoring: - Remove some needless generality from cpuworker code, for improved type-safety. - Stop overloading the circuit_t.onionskin field for both "onionskin from a CREATE cell that we are waiting for a cpuworker to be assigned" and "onionskin from an EXTEND cell that we are going to send to an OR as soon as we are connected". Might help with bug 600. - Add an in-place version of aes_crypt() so that we can avoid doing a needless memcpy() call on each cell payload. Changes in version 0.2.0.18-alpha - 2008-01-25 Tor 0.2.0.18-alpha adds a sixth v3 directory authority run by CCC, fixes a big memory leak in 0.2.0.17-alpha, and adds new config options that can warn or reject connections to ports generally associated with vulnerable-plaintext protocols. o New directory authorities: - Set up dannenberg (run by CCC) as the sixth v3 directory authority. o Major bugfixes: - Fix a major memory leak when attempting to use the v2 TLS handshake code. Bugfix on 0.2.0.x; fixes bug 589. - We accidentally enabled the under-development v2 TLS handshake code, which was causing log entries like "TLS error while renegotiating handshake". Disable it again. Resolves bug 590. - We were computing the wrong Content-Length: header for directory responses that need to be compressed on the fly, causing clients asking for those items to always fail. Bugfix on 0.2.0.x; partially fixes bug 593. o Major features: - Avoid going directly to the directory authorities even if you're a relay, if you haven't found yourself reachable yet or if you've decided not to advertise your dirport yet. Addresses bug 556. - If we've gone 12 hours since our last bandwidth check, and we estimate we have less than 50KB bandwidth capacity but we could handle more, do another bandwidth test. - New config options WarnPlaintextPorts and RejectPlaintextPorts so Tor can warn and/or refuse connections to ports commonly used with vulnerable-plaintext protocols. Currently we warn on ports 23, 109, 110, and 143, but we don't reject any. o Minor bugfixes: - When we setconf ClientOnly to 1, close any current OR and Dir listeners. Reported by mwenge. - When we get a consensus that's been signed by more people than we expect, don't log about it; it's not a big deal. Reported by Kyle Williams. o Minor features: - Don't answer "/tor/networkstatus-bridges" directory requests if the request isn't encrypted. - Make "ClientOnly 1" config option disable directory ports too. - Patches from Karsten Loesing to make v2 hidden services more robust: work even when there aren't enough HSDir relays available; retry when a v2 rend desc fetch fails; but don't retry if we already have a usable v0 rend desc. Changes in version 0.2.0.17-alpha - 2008-01-17 Tor 0.2.0.17-alpha makes the tarball build cleanly again (whoops). o Compile fixes: - Make the tor-gencert man page get included correctly in the tarball. Changes in version 0.2.0.16-alpha - 2008-01-17 Tor 0.2.0.16-alpha adds a fifth v3 directory authority run by Karsten Loesing, and generally cleans up a lot of features and minor bugs. o New directory authorities: - Set up gabelmoo (run by Karsten Loesing) as the fifth v3 directory authority. o Major performance improvements: - Switch our old ring buffer implementation for one more like that used by free Unix kernels. The wasted space in a buffer with 1mb of data will now be more like 8k than 1mb. The new implementation also avoids realloc();realloc(); patterns that can contribute to memory fragmentation. o Minor features: - Configuration files now accept C-style strings as values. This helps encode characters not allowed in the current configuration file format, such as newline or #. Addresses bug 557. - Although we fixed bug 539 (where servers would send HTTP status 503 responses _and_ send a body too), there are still servers out there that haven't upgraded. Therefore, make clients parse such bodies when they receive them. - When we're not serving v2 directory information, there is no reason to actually keep any around. Remove the obsolete files and directory on startup if they are very old and we aren't going to serve them. o Minor performance improvements: - Reference-count and share copies of address policy entries; only 5% of them were actually distinct. - Never walk through the list of logs if we know that no log is interested in a given message. o Minor bugfixes: - When an authority has not signed a consensus, do not try to download a nonexistent "certificate with key 00000000". Bugfix on 0.2.0.x. Fixes bug 569. - Fix a rare assert error when we're closing one of our threads: use a mutex to protect the list of logs, so we never write to the list as it's being freed. Bugfix on 0.1.2.x. Fixes the very rare bug 575, which is kind of the revenge of bug 222. - Patch from Karsten Loesing to complain less at both the client and the relay when a relay used to have the HSDir flag but doesn't anymore, and we try to upload a hidden service descriptor. - Stop leaking one cert per TLS context. Fixes bug 582. Bugfix on 0.2.0.15-alpha. - Do not try to download missing certificates until we have tried to check our fallback consensus. Fixes bug 583. - Make bridges round reported GeoIP stats info up to the nearest estimate, not down. Now we can distinguish between "0 people from this country" and "1 person from this country". - Avoid a spurious free on base64 failure. Bugfix on 0.1.2. - Avoid possible segfault if key generation fails in crypto_pk_hybrid_encrypt. Bugfix on 0.2.0. - Avoid segfault in the case where a badly behaved v2 versioning directory sends a signed networkstatus with missing client-versions. Bugfix on 0.1.2. - Avoid segfaults on certain complex invocations of router_get_by_hexdigest(). Bugfix on 0.1.2. - Correct bad index on array access in parse_http_time(). Bugfix on 0.2.0. - Fix possible bug in vote generation when server versions are present but client versions are not. - Fix rare bug on REDIRECTSTREAM control command when called with no port set: it could erroneously report an error when none had happened. - Avoid bogus crash-prone, leak-prone tor_realloc when we're compressing large objects and find ourselves with more than 4k left over. Bugfix on 0.2.0. - Fix a small memory leak when setting up a hidden service. - Fix a few memory leaks that could in theory happen under bizarre error conditions. - Fix an assert if we post a general-purpose descriptor via the control port but that descriptor isn't mentioned in our current network consensus. Bug reported by Jon McLachlan; bugfix on 0.2.0.9-alpha. o Minor features (controller): - Get NS events working again. Patch from tup. - The GETCONF command now escapes and quotes configuration values that don't otherwise fit into the torrc file. - The SETCONF command now handles quoted values correctly. o Minor features (directory authorities): - New configuration options to override default maximum number of servers allowed on a single IP address. This is important for running a test network on a single host. - Actually implement the -s option to tor-gencert. - Add a manual page for tor-gencert. o Minor features (bridges): - Bridge authorities no longer serve bridge descriptors over unencrypted connections. o Minor features (other): - Add hidden services and DNSPorts to the list of things that make Tor accept that it has running ports. Change starting Tor with no ports from a fatal error to a warning; we might change it back if this turns out to confuse anybody. Fixes bug 579. Changes in version 0.1.2.19 - 2008-01-17 Tor 0.1.2.19 fixes a huge memory leak on exit relays, makes the default exit policy a little bit more conservative so it's safer to run an exit relay on a home system, and fixes a variety of smaller issues. o Security fixes: - Exit policies now reject connections that are addressed to a relay's public (external) IP address too, unless ExitPolicyRejectPrivate is turned off. We do this because too many relays are running nearby to services that trust them based on network address. o Major bugfixes: - When the clock jumps forward a lot, do not allow the bandwidth buckets to become negative. Fixes bug 544. - Fix a memory leak on exit relays; we were leaking a cached_resolve_t on every successful resolve. Reported by Mike Perry. - Purge old entries from the "rephist" database and the hidden service descriptor database even when DirPort is zero. - Stop thinking that 0.1.2.x directory servers can handle "begin_dir" requests. Should ease bugs 406 and 419 where 0.1.2.x relays are crashing or mis-answering these requests. - When we decide to send a 503 response to a request for servers, do not then also send the server descriptors: this defeats the whole purpose. Fixes bug 539. o Minor bugfixes: - Changing the ExitPolicyRejectPrivate setting should cause us to rebuild our server descriptor. - Fix handling of hex nicknames when answering controller requests for networkstatus by name, or when deciding whether to warn about unknown routers in a config option. (Patch from mwenge.) - Fix a couple of hard-to-trigger autoconf problems that could result in really weird results on platforms whose sys/types.h files define nonstandard integer types. - Don't try to create the datadir when running --verify-config or --hash-password. Resolves bug 540. - If we were having problems getting a particular descriptor from the directory caches, and then we learned about a new descriptor for that router, we weren't resetting our failure count. Reported by lodger. - Although we fixed bug 539 (where servers would send HTTP status 503 responses _and_ send a body too), there are still servers out there that haven't upgraded. Therefore, make clients parse such bodies when they receive them. - Run correctly on systems where rlim_t is larger than unsigned long. This includes some 64-bit systems. - Run correctly on platforms (like some versions of OS X 10.5) where the real limit for number of open files is OPEN_FILES, not rlim_max from getrlimit(RLIMIT_NOFILES). - Avoid a spurious free on base64 failure. - Avoid segfaults on certain complex invocations of router_get_by_hexdigest(). - Fix rare bug on REDIRECTSTREAM control command when called with no port set: it could erroneously report an error when none had happened. Changes in version 0.2.0.15-alpha - 2007-12-25 Tor 0.2.0.14-alpha and 0.2.0.15-alpha fix a bunch of bugs with the features added in 0.2.0.13-alpha. o Major bugfixes: - Fix several remotely triggerable asserts based on DirPort requests for a v2 or v3 networkstatus object before we were prepared. This was particularly bad for 0.2.0.13 and later bridge relays, who would never have a v2 networkstatus and would thus always crash when used. Bugfixes on 0.2.0.x. - Estimate the v3 networkstatus size more accurately, rather than estimating it at zero bytes and giving it artificially high priority compared to other directory requests. Bugfix on 0.2.0.x. o Minor bugfixes: - Fix configure.in logic for cross-compilation. - When we load a bridge descriptor from the cache, and it was previously unreachable, mark it as retriable so we won't just ignore it. Also, try fetching a new copy immediately. Bugfixes on 0.2.0.13-alpha. - The bridge GeoIP stats were counting other relays, for example self-reachability and authority-reachability tests. o Minor features: - Support compilation to target iPhone; patch from cjacker huang. To build for iPhone, pass the --enable-iphone option to configure. Changes in version 0.2.0.14-alpha - 2007-12-23 o Major bugfixes: - Fix a crash on startup if you install Tor 0.2.0.13-alpha fresh without a datadirectory from a previous Tor install. Reported by Zax. - Fix a crash when we fetch a descriptor that turns out to be unexpected (it used to be in our networkstatus when we started fetching it, but it isn't in our current networkstatus), and we aren't using bridges. Bugfix on 0.2.0.x. - Fix a crash when accessing hidden services: it would work the first time you use a given introduction point for your service, but on subsequent requests we'd be using garbage memory. Fixed by Karsten Loesing. Bugfix on 0.2.0.13-alpha. - Fix a crash when we load a bridge descriptor from disk but we don't currently have a Bridge line for it in our torrc. Bugfix on 0.2.0.13-alpha. o Major features: - If bridge authorities set BridgePassword, they will serve a snapshot of known bridge routerstatuses from their DirPort to anybody who knows that password. Unset by default. o Minor bugfixes: - Make the unit tests build again. - Make "GETINFO/desc-annotations/id/" actually work. - Make PublishServerDescriptor default to 1, so the default doesn't have to change as we invent new directory protocol versions. - Fix test for rlim_t on OSX 10.3: sys/resource.h doesn't want to be included unless sys/time.h is already included. Fixes bug 553. Bugfix on 0.2.0.x. - If we receive a general-purpose descriptor and then receive an identical bridge-purpose descriptor soon after, don't discard the next one as a duplicate. o Minor features: - If BridgeRelay is set to 1, then the default for PublishServerDescriptor is now "bridge" rather than "v2,v3". - If the user sets RelayBandwidthRate but doesn't set RelayBandwidthBurst, then make them equal rather than erroring out. Changes in version 0.2.0.13-alpha - 2007-12-21 Tor 0.2.0.13-alpha adds a fourth v3 directory authority run by Geoff Goodell, fixes many more bugs, and adds a lot of infrastructure for upcoming features. o New directory authorities: - Set up lefkada (run by Geoff Goodell) as the fourth v3 directory authority. o Major bugfixes: - Only update guard status (usable / not usable) once we have enough directory information. This was causing us to always pick two new guards on startup (bugfix on 0.2.0.9-alpha), and it was causing us to discard all our guards on startup if we hadn't been running for a few weeks (bugfix on 0.1.2.x). Fixes bug 448. - Purge old entries from the "rephist" database and the hidden service descriptor databases even when DirPort is zero. Bugfix on 0.1.2.x. - We were ignoring our RelayBandwidthRate for the first 30 seconds after opening a circuit -- even a relayed circuit. Bugfix on 0.2.0.3-alpha. - Stop thinking that 0.1.2.x directory servers can handle "begin_dir" requests. Should ease bugs 406 and 419 where 0.1.2.x relays are crashing or mis-answering these types of requests. - Relays were publishing their server descriptor to v1 and v2 directory authorities, but they didn't try publishing to v3-only authorities. Fix this; and also stop publishing to v1 authorities. Bugfix on 0.2.0.x. - When we were reading router descriptors from cache, we were ignoring the annotations -- so for example we were reading in bridge-purpose descriptors as general-purpose descriptors. Bugfix on 0.2.0.8-alpha. - When we decided to send a 503 response to a request for servers, we were then also sending the server descriptors: this defeats the whole purpose. Fixes bug 539; bugfix on 0.1.2.x. o Major features: - Bridge relays now behave like clients with respect to time intervals for downloading new consensus documents -- otherwise they stand out. Bridge users now wait until the end of the interval, so their bridge relay will be sure to have a new consensus document. - Three new config options (AlternateDirAuthority, AlternateBridgeAuthority, and AlternateHSAuthority) that let the user selectively replace the default directory authorities by type, rather than the all-or-nothing replacement that DirServer offers. - Tor can now be configured to read a GeoIP file from disk in one of two formats. This can be used by controllers to map IP addresses to countries. Eventually, it may support exit-by-country. - When possible, bridge relays remember which countries users are coming from, and report aggregate information in their extra-info documents, so that the bridge authorities can learn where Tor is blocked. - Bridge directory authorities now do reachability testing on the bridges they know. They provide router status summaries to the controller via "getinfo ns/purpose/bridge", and also dump summaries to a file periodically. - Stop fetching directory info so aggressively if your DirPort is on but your ORPort is off; stop fetching v2 dir info entirely. You can override these choices with the new FetchDirInfoEarly config option. o Minor bugfixes: - The fix in 0.2.0.12-alpha cleared the "hsdir" flag in v3 network consensus documents when there are too many relays at a single IP address. Now clear it in v2 network status documents too, and also clear it in routerinfo_t when the relay is no longer listed in the relevant networkstatus document. - Don't crash if we get an unexpected value for the PublishServerDescriptor config option. Reported by Matt Edman; bugfix on 0.2.0.9-alpha. - Our new v2 hidden service descriptor format allows descriptors that have no introduction points. But Tor crashed when we tried to build a descriptor with no intro points (and it would have crashed if we had tried to parse one). Bugfix on 0.2.0.x; patch by Karsten Loesing. - Fix building with dmalloc 5.5.2 with glibc. - Reject uploaded descriptors and extrainfo documents if they're huge. Otherwise we'll cache them all over the network and it'll clog everything up. Reported by Aljosha Judmayer. - Check for presence of s6_addr16 and s6_addr32 fields in in6_addr via autoconf. Should fix compile on solaris. Bugfix on 0.2.0.x. - When the DANGEROUS_VERSION controller status event told us we're running an obsolete version, it used the string "OLD" to describe it. Yet the "getinfo" interface used the string "OBSOLETE". Now use "OBSOLETE" in both cases. Bugfix on 0.1.2.x. - If we can't expand our list of entry guards (e.g. because we're using bridges or we have StrictEntryNodes set), don't mark relays down when they fail a directory request. Otherwise we're too quick to mark all our entry points down. Bugfix on 0.1.2.x. - Fix handling of hex nicknames when answering controller requests for networkstatus by name, or when deciding whether to warn about unknown routers in a config option. Bugfix on 0.1.2.x. (Patch from mwenge.) - Fix a couple of hard-to-trigger autoconf problems that could result in really weird results on platforms whose sys/types.h files define nonstandard integer types. Bugfix on 0.1.2.x. - Fix compilation with --disable-threads set. Bugfix on 0.2.0.x. - Don't crash on name lookup when we have no current consensus. Fixes bug 538; bugfix on 0.2.0.x. - Only Tors that want to mirror the v2 directory info should create the "cached-status" directory in their datadir. (All Tors used to create it.) Bugfix on 0.2.0.9-alpha. - Directory authorities should only automatically download Extra Info documents if they're v1, v2, or v3 authorities. Bugfix on 0.1.2.x. o Minor features: - On the USR1 signal, when dmalloc is in use, log the top 10 memory consumers. (We already do this on HUP.) - Authorities and caches fetch the v2 networkstatus documents less often, now that v3 is encouraged. - Add a new config option BridgeRelay that specifies you want to be a bridge relay. Right now the only difference is that it makes you answer begin_dir requests, and it makes you cache dir info, even if your DirPort isn't on. - Add "GETINFO/desc-annotations/id/" so controllers can ask about source, timestamp of arrival, purpose, etc. We need something like this to help Vidalia not do GeoIP lookups on bridge addresses. - Allow multiple HashedControlPassword config lines, to support multiple controller passwords. - Authorities now decide whether they're authoritative for a given router based on the router's purpose. - New config options AuthDirBadDir and AuthDirListBadDirs for authorities to mark certain relays as "bad directories" in the networkstatus documents. Also supports the "!baddir" directive in the approved-routers file. Changes in version 0.2.0.12-alpha - 2007-11-16 This twelfth development snapshot fixes some more build problems as well as a few minor bugs. o Compile fixes: - Make it build on OpenBSD again. Patch from tup. - Substitute BINDIR and LOCALSTATEDIR in scripts. Fixes package-building for Red Hat, OS X, etc. o Minor bugfixes (on 0.1.2.x): - Changing the ExitPolicyRejectPrivate setting should cause us to rebuild our server descriptor. o Minor bugfixes (on 0.2.0.x): - When we're lacking a consensus, don't try to perform rendezvous operations. Reported by Karsten Loesing. - Fix a small memory leak whenever we decide against using a newly picked entry guard. Reported by Mike Perry. - When authorities detected more than two relays running on the same IP address, they were clearing all the status flags but forgetting to clear the "hsdir" flag. So clients were being told that a given relay was the right choice for a v2 hsdir lookup, yet they never had its descriptor because it was marked as 'not running' in the consensus. - If we're trying to fetch a bridge descriptor and there's no way the bridge authority could help us (for example, we don't know a digest, or there is no bridge authority), don't be so eager to fall back to asking the bridge authority. - If we're using bridges or have strictentrynodes set, and our chosen exit is in the same family as all our bridges/entry guards, then be flexible about families. o Minor features: - When we negotiate a v2 link-layer connection (not yet implemented), accept RELAY_EARLY cells and turn them into RELAY cells if we've negotiated a v1 connection for their next step. Initial code for proposal 110. Changes in version 0.2.0.11-alpha - 2007-11-12 This eleventh development snapshot fixes some build problems with the previous snapshot. It also includes a more secure-by-default exit policy for relays, fixes an enormous memory leak for exit relays, and fixes another bug where servers were falling out of the directory list. o Security fixes: - Exit policies now reject connections that are addressed to a relay's public (external) IP address too, unless ExitPolicyRejectPrivate is turned off. We do this because too many relays are running nearby to services that trust them based on network address. Bugfix on 0.1.2.x. o Major bugfixes: - Fix a memory leak on exit relays; we were leaking a cached_resolve_t on every successful resolve. Reported by Mike Perry; bugfix on 0.1.2.x. - On authorities, never downgrade to old router descriptors simply because they're listed in the consensus. This created a catch-22 where we wouldn't list a new descriptor because there was an old one in the consensus, and we couldn't get the new one in the consensus because we wouldn't list it. Possible fix for bug 548. Also, this might cause bug 543 to appear on authorities; if so, we'll need a band-aid for that. Bugfix on 0.2.0.9-alpha. o Packaging fixes on 0.2.0.10-alpha: - We were including instructions about what to do with the src/config/fallback-consensus file, but we weren't actually including it in the tarball. Disable all of that for now. o Minor features: - Allow people to say PreferTunnelledDirConns rather than PreferTunneledDirConns, for those alternate-spellers out there. o Minor bugfixes: - Don't reevaluate all the information from our consensus document just because we've downloaded a v2 networkstatus that we intend to cache. Fixes bug 545; bugfix on 0.2.0.x. Changes in version 0.2.0.10-alpha - 2007-11-10 This tenth development snapshot adds a third v3 directory authority run by Mike Perry, adds most of Karsten Loesing's new hidden service descriptor format, fixes a bad crash bug and new bridge bugs introduced in 0.2.0.9-alpha, fixes many bugs with the v3 directory implementation, fixes some minor memory leaks in previous 0.2.0.x snapshots, and addresses many more minor issues. o New directory authorities: - Set up ides (run by Mike Perry) as the third v3 directory authority. o Major features: - Allow tunnelled directory connections to ask for an encrypted "begin_dir" connection or an anonymized "uses a full Tor circuit" connection independently. Now we can make anonymized begin_dir connections for (e.g.) more secure hidden service posting and fetching. - More progress on proposal 114: code from Karsten Loesing to implement new hidden service descriptor format. - Raise the default BandwidthRate/BandwidthBurst to 5MB/10MB, to accommodate the growing number of servers that use the default and are reaching it. - Directory authorities use a new formula for selecting which nodes to advertise as Guards: they must be in the top 7/8 in terms of how long we have known about them, and above the median of those nodes in terms of weighted fractional uptime. - Make "not enough dir info yet" warnings describe *why* Tor feels it doesn't have enough directory info yet. o Major bugfixes: - Stop servers from crashing if they set a Family option (or maybe in other situations too). Bugfix on 0.2.0.9-alpha; reported by Fabian Keil. - Make bridge users work again -- the move to v3 directories in 0.2.0.9-alpha had introduced a number of bugs that made bridges no longer work for clients. - When the clock jumps forward a lot, do not allow the bandwidth buckets to become negative. Bugfix on 0.1.2.x; fixes bug 544. o Major bugfixes (v3 dir, bugfixes on 0.2.0.9-alpha): - When the consensus lists a router descriptor that we previously were mirroring, but that we considered non-canonical, reload the descriptor as canonical. This fixes bug 543 where Tor servers would start complaining after a few days that they don't have enough directory information to build a circuit. - Consider replacing the current consensus when certificates arrive that make the pending consensus valid. Previously, we were only considering replacement when the new certs _didn't_ help. - Fix an assert error on startup if we didn't already have the consensus and certs cached in our datadirectory: we were caching the consensus in consensus_waiting_for_certs but then free'ing it right after. - Avoid sending a request for "keys/fp" (for which we'll get a 400 Bad Request) if we need more v3 certs but we've already got pending requests for all of them. - Correctly back off from failing certificate downloads. Fixes bug 546. - Authorities don't vote on the Running flag if they have been running for less than 30 minutes themselves. Fixes bug 547, where a newly started authority would vote that everyone was down. o New requirements: - Drop support for OpenSSL version 0.9.6. Just about nobody was using it, it had no AES, and it hasn't seen any security patches since 2004. o Minor features: - Clients now hold circuitless TLS connections open for 1.5 times MaxCircuitDirtiness (15 minutes), since it is likely that they'll rebuild a new circuit over them within that timeframe. Previously, they held them open only for KeepalivePeriod (5 minutes). - Use "If-Modified-Since" to avoid retrieving consensus networkstatuses that we already have. - When we have no consensus, check FallbackNetworkstatusFile (defaults to $PREFIX/share/tor/fallback-consensus) for a consensus. This way we start knowing some directory caches. - When we receive a consensus from the future, warn about skew. - Improve skew reporting: try to give the user a better log message about how skewed they are, and how much this matters. - When we have a certificate for an authority, believe that certificate's claims about the authority's IP address. - New --quiet command-line option to suppress the default console log. Good in combination with --hash-password. - Authorities send back an X-Descriptor-Not-New header in response to an accepted-but-discarded descriptor upload. Partially implements fix for bug 535. - Make the log message for "tls error. breaking." more useful. - Better log messages about certificate downloads, to attempt to track down the second incarnation of bug 546. o Minor features (bridges): - If bridge users set UpdateBridgesFromAuthority, but the digest they ask for is a 404 from the bridge authority, they now fall back to trying the bridge directly. - Bridges now use begin_dir to publish their server descriptor to the bridge authority, even when they haven't set TunnelDirConns. o Minor features (controller): - When reporting clock skew, and we know that the clock is _at least as skewed_ as some value, but we don't know the actual value, report the value as a "minimum skew." o Utilities: - Update linux-tor-prio.sh script to allow QoS based on the uid of the Tor process. Patch from Marco Bonetti with tweaks from Mike Perry. o Minor bugfixes: - Refuse to start if both ORPort and UseBridges are set. Bugfix on 0.2.0.x, suggested by Matt Edman. - Don't stop fetching descriptors when FetchUselessDescriptors is set, even if we stop asking for circuits. Bugfix on 0.1.2.x; reported by tup and ioerror. - Better log message on vote from unknown authority. - Don't log "Launching 0 request for 0 router" message. o Minor bugfixes (memory leaks): - Stop leaking memory every time we parse a v3 certificate. Bugfix on 0.2.0.1-alpha. - Stop leaking memory every time we load a v3 certificate. Bugfix on 0.2.0.1-alpha. Fixes bug 536. - Stop leaking a cached networkstatus on exit. Bugfix on 0.2.0.3-alpha. - Stop leaking voter information every time we free a consensus. Bugfix on 0.2.0.3-alpha. - Stop leaking signed data every time we check a voter signature. Bugfix on 0.2.0.3-alpha. - Stop leaking a signature every time we fail to parse a consensus or a vote. Bugfix on 0.2.0.3-alpha. - Stop leaking v2_download_status_map on shutdown. Bugfix on 0.2.0.9-alpha. - Stop leaking conn->nickname every time we make a connection to a Tor relay without knowing its expected identity digest (e.g. when using bridges). Bugfix on 0.2.0.3-alpha. - Minor bugfixes (portability): - Run correctly on platforms where rlim_t is larger than unsigned long, and/or where the real limit for number of open files is OPEN_FILES, not rlim_max from getrlimit(RLIMIT_NOFILES). In particular, these may be needed for OS X 10.5. Changes in version 0.1.2.18 - 2007-10-28 Tor 0.1.2.18 fixes many problems including crash bugs, problems with hidden service introduction that were causing huge delays, and a big bug that was causing some servers to disappear from the network status lists for a few hours each day. o Major bugfixes (crashes): - If a connection is shut down abruptly because of something that happened inside connection_flushed_some(), do not call connection_finished_flushing(). Should fix bug 451: "connection_stop_writing: Assertion conn->write_event failed" Bugfix on 0.1.2.7-alpha. - Fix possible segfaults in functions called from rend_process_relay_cell(). o Major bugfixes (hidden services): - Hidden services were choosing introduction points uniquely by hexdigest, but when constructing the hidden service descriptor they merely wrote the (potentially ambiguous) nickname. - Clients now use the v2 intro format for hidden service connections: they specify their chosen rendezvous point by identity digest rather than by (potentially ambiguous) nickname. These changes could speed up hidden service connections dramatically. o Major bugfixes (other): - Stop publishing a new server descriptor just because we get a HUP signal. This led (in a roundabout way) to some servers getting dropped from the networkstatus lists for a few hours each day. - When looking for a circuit to cannibalize, consider family as well as identity. Fixes bug 438. Bugfix on 0.1.0.x (which introduced circuit cannibalization). - When a router wasn't listed in a new networkstatus, we were leaving the flags for that router alone -- meaning it remained Named, Running, etc -- even though absence from the networkstatus means that it shouldn't be considered to exist at all anymore. Now we clear all the flags for routers that fall out of the networkstatus consensus. Fixes bug 529. o Minor bugfixes: - Don't try to access (or alter) the state file when running --list-fingerprint or --verify-config or --hash-password. Resolves bug 499. - When generating information telling us how to extend to a given router, do not try to include the nickname if it is absent. Resolves bug 467. - Fix a user-triggerable segfault in expand_filename(). (There isn't a way to trigger this remotely.) - When sending a status event to the controller telling it that an OR address is reachable, set the port correctly. (Previously we were reporting the dir port.) - Fix a minor memory leak whenever a controller sends the PROTOCOLINFO command. Bugfix on 0.1.2.17. - When loading bandwidth history, do not believe any information in the future. Fixes bug 434. - When loading entry guard information, do not believe any information in the future. - When we have our clock set far in the future and generate an onion key, then re-set our clock to be correct, we should not stop the onion key from getting rotated. - On some platforms, accept() can return a broken address. Detect this more quietly, and deal accordingly. Fixes bug 483. - It's not actually an error to find a non-pending entry in the DNS cache when canceling a pending resolve. Don't log unless stuff is fishy. Resolves bug 463. - Don't reset trusted dir server list when we set a configuration option. Patch from Robert Hogan. - Don't try to create the datadir when running --verify-config or --hash-password. Resolves bug 540. Changes in version 0.2.0.9-alpha - 2007-10-24 This ninth development snapshot switches clients to the new v3 directory system; allows servers to be listed in the network status even when they have the same nickname as a registered server; and fixes many other bugs including a big one that was causing some servers to disappear from the network status lists for a few hours each day. o Major features (directory system): - Clients now download v3 consensus networkstatus documents instead of v2 networkstatus documents. Clients and caches now base their opinions about routers on these consensus documents. Clients only download router descriptors listed in the consensus. - Authorities now list servers who have the same nickname as a different named server, but list them with a new flag, "Unnamed". Now we can list servers that happen to pick the same nickname as a server that registered two years ago and then disappeared. Partially implements proposal 122. - If the consensus lists a router as "Unnamed", the name is assigned to a different router: do not identify the router by that name. Partially implements proposal 122. - Authorities can now come to a consensus on which method to use to compute the consensus. This gives us forward compatibility. o Major bugfixes: - Stop publishing a new server descriptor just because we HUP or when we find our DirPort to be reachable but won't actually publish it. New descriptors without any real changes are dropped by the authorities, and can screw up our "publish every 18 hours" schedule. Bugfix on 0.1.2.x. - When a router wasn't listed in a new networkstatus, we were leaving the flags for that router alone -- meaning it remained Named, Running, etc -- even though absence from the networkstatus means that it shouldn't be considered to exist at all anymore. Now we clear all the flags for routers that fall out of the networkstatus consensus. Fixes bug 529; bugfix on 0.1.2.x. - Fix awful behavior in DownloadExtraInfo option where we'd fetch extrainfo documents and then discard them immediately for not matching the latest router. Bugfix on 0.2.0.1-alpha. o Minor features (v3 directory protocol): - Allow tor-gencert to generate a new certificate without replacing the signing key. - Allow certificates to include an address. - When we change our directory-cache settings, reschedule all voting and download operations. - Reattempt certificate downloads immediately on failure, as long as we haven't failed a threshold number of times yet. - Delay retrying consensus downloads while we're downloading certificates to verify the one we just got. Also, count getting a consensus that we already have (or one that isn't valid) as a failure, and count failing to get the certificates after 20 minutes as a failure. - Build circuits and download descriptors even if our consensus is a little expired. (This feature will go away once authorities are more reliable.) o Minor features (router descriptor cache): - If we find a cached-routers file that's been sitting around for more than 28 days unmodified, then most likely it's a leftover from when we upgraded to 0.2.0.8-alpha. Remove it. It has no good routers anyway. - When we (as a cache) download a descriptor because it was listed in a consensus, remember when the consensus was supposed to expire, and don't expire the descriptor until then. o Minor features (performance): - Call routerlist_remove_old_routers() much less often. This should speed startup, especially on directory caches. - Don't try to launch new descriptor downloads quite so often when we already have enough directory information to build circuits. - Base64 decoding was actually showing up on our profile when parsing the initial descriptor file; switch to an in-process all-at-once implementation that's about 3.5x times faster than calling out to OpenSSL. o Minor features (compilation): - Detect non-ASCII platforms (if any still exist) and refuse to build there: some of our code assumes that 'A' is 65 and so on. o Minor bugfixes (v3 directory authorities, bugfixes on 0.2.0.x): - Make the "next period" votes into "current period" votes immediately after publishing the consensus; avoid a heisenbug that made them stick around indefinitely. - When we discard a vote as a duplicate, do not report this as an error. - Treat missing v3 keys or certificates as an error when running as a v3 directory authority. - When we're configured to be a v3 authority, but we're only listed as a non-v3 authority in our DirServer line for ourself, correct the listing. - If an authority doesn't have a qualified hostname, just put its address in the vote. This fixes the problem where we referred to "moria on moria:9031." - Distinguish between detached signatures for the wrong period, and detached signatures for a divergent vote. - Fix a small memory leak when computing a consensus. - When there's no concensus, we were forming a vote every 30 minutes, but writing the "valid-after" line in our vote based on our configured V3AuthVotingInterval: so unless the intervals matched up, we immediately rejected our own vote because it didn't start at the voting interval that caused us to construct a vote. o Minor bugfixes (v3 directory protocol, bugfixes on 0.2.0.x): - Delete unverified-consensus when the real consensus is set. - Consider retrying a consensus networkstatus fetch immediately after one fails: don't wait 60 seconds to notice. - When fetching a consensus as a cache, wait until a newer consensus should exist before trying to replace the current one. - Use a more forgiving schedule for retrying failed consensus downloads than for other types. o Minor bugfixes (other directory issues): - Correct the implementation of "download votes by digest." Bugfix on 0.2.0.8-alpha. - Authorities no longer send back "400 you're unreachable please fix it" errors to Tor servers that aren't online all the time. We're supposed to tolerate these servers now. Bugfix on 0.1.2.x. o Minor bugfixes (controller): - Don't reset trusted dir server list when we set a configuration option. Patch from Robert Hogan; bugfix on 0.1.2.x. - Respond to INT and TERM SIGNAL commands before we execute the signal, in case the signal shuts us down. We had a patch in 0.1.2.1-alpha that tried to do this by queueing the response on the connection's buffer before shutting down, but that really isn't the same thing at all. Bug located by Matt Edman. o Minor bugfixes (misc): - Correctly check for bad options to the "PublishServerDescriptor" config option. Bugfix on 0.2.0.1-alpha; reported by Matt Edman. - Stop leaking memory on failing case of base32_decode, and make it accept upper-case letters. Bugfixes on 0.2.0.7-alpha. - Don't try to download extrainfo documents when we're trying to fetch enough directory info to build a circuit: having enough info should get priority. Bugfix on 0.2.0.x. - Don't complain that "your server has not managed to confirm that its ports are reachable" if we haven't been able to build any circuits yet. Bug found by spending four hours without a v3 consensus. Bugfix on 0.1.2.x. - Detect the reason for failing to mmap a descriptor file we just wrote, and give a more useful log message. Fixes bug 533. Bugfix on 0.1.2.x. o Code simplifications and refactoring: - Remove support for the old bw_accounting file: we've been storing bandwidth accounting information in the state file since 0.1.2.5-alpha. This may result in bandwidth accounting errors if you try to upgrade from 0.1.1.x or earlier, or if you try to downgrade to 0.1.1.x or earlier. - New convenience code to locate a file within the DataDirectory. - Move non-authority functionality out of dirvote.c. - Refactor the arguments for router_pick_{directory_|trusteddir}server so that they all take the same named flags. o Utilities - Include the "tor-ctrl.sh" bash script by Stefan Behte to provide Unix users an easy way to script their Tor process (e.g. by adjusting bandwidth based on the time of the day). Changes in version 0.2.0.8-alpha - 2007-10-12 This eighth development snapshot fixes a crash bug that's been bothering us since February 2007, lets bridge authorities store a list of bridge descriptors they've seen, gets v3 directory voting closer to working, starts caching v3 directory consensus documents on directory mirrors, and fixes a variety of smaller issues including some minor memory leaks. o Major features (router descriptor cache): - Store routers in a file called cached-descriptors instead of in cached-routers. Initialize cached-descriptors from cached-routers if the old format is around. The new format allows us to store annotations along with descriptors. - Use annotations to record the time we received each descriptor, its source, and its purpose. - Disable the SETROUTERPURPOSE controller command: it is now obsolete. - Controllers should now specify cache=no or cache=yes when using the +POSTDESCRIPTOR command. - Bridge authorities now write bridge descriptors to disk, meaning we can export them to other programs and begin distributing them to blocked users. o Major features (directory authorities): - When a v3 authority is missing votes or signatures, it now tries to fetch them. - Directory authorities track weighted fractional uptime as well as weighted mean-time-between failures. WFU is suitable for deciding whether a node is "usually up", while MTBF is suitable for deciding whether a node is "likely to stay up." We need both, because "usually up" is a good requirement for guards, while "likely to stay up" is a good requirement for long-lived connections. o Major features (v3 directory system): - Caches now download v3 network status documents as needed, and download the descriptors listed in them. - All hosts now attempt to download and keep fresh v3 authority certificates, and re-attempt after failures. - More internal-consistency checks for vote parsing. o Major bugfixes (crashes): - If a connection is shut down abruptly because of something that happened inside connection_flushed_some(), do not call connection_finished_flushing(). Should fix bug 451. Bugfix on 0.1.2.7-alpha. o Major bugfixes (performance): - Fix really bad O(n^2) performance when parsing a long list of routers: Instead of searching the entire list for an "extra-info " string which usually wasn't there, once for every routerinfo we read, just scan lines forward until we find one we like. Bugfix on 0.2.0.1. - When we add data to a write buffer in response to the data on that write buffer getting low because of a flush, do not consider the newly added data as a candidate for immediate flushing, but rather make it wait until the next round of writing. Otherwise, we flush and refill recursively, and a single greedy TLS connection can eat all of our bandwidth. Bugfix on 0.1.2.7-alpha. o Minor features (v3 authority system): - Add more ways for tools to download the votes that lead to the current consensus. - Send a 503 when low on bandwidth and a vote, consensus, or certificate is requested. - If-modified-since is now implemented properly for all kinds of certificate requests. o Minor bugfixes (network statuses): - Tweak the implementation of proposal 109 slightly: allow at most two Tor servers on the same IP address, except if it's the location of a directory authority, in which case allow five. Bugfix on 0.2.0.3-alpha. o Minor bugfixes (controller): - When sending a status event to the controller telling it that an OR address is reachable, set the port correctly. (Previously we were reporting the dir port.) Bugfix on 0.1.2.x. o Minor bugfixes (v3 directory system): - Fix logic to look up a cert by its signing key digest. Bugfix on 0.2.0.7-alpha. - Only change the reply to a vote to "OK" if it's not already set. This gets rid of annoying "400 OK" log messages, which may have been masking some deeper issue. Bugfix on 0.2.0.7-alpha. - When we get a valid consensus, recompute the voting schedule. - Base the valid-after time of a vote on the consensus voting schedule, not on our preferred schedule. - Make the return values and messages from signature uploads and downloads more sensible. - Fix a memory leak when serving votes and consensus documents, and another when serving certificates. o Minor bugfixes (performance): - Use a slightly simpler string hashing algorithm (copying Python's instead of Java's) and optimize our digest hashing algorithm to take advantage of 64-bit platforms and to remove some possibly-costly voodoo. - Fix a minor memory leak whenever we parse guards from our state file. Bugfix on 0.2.0.7-alpha. - Fix a minor memory leak whenever we write out a file. Bugfix on 0.2.0.7-alpha. - Fix a minor memory leak whenever a controller sends the PROTOCOLINFO command. Bugfix on 0.2.0.5-alpha. o Minor bugfixes (portability): - On some platforms, accept() can return a broken address. Detect this more quietly, and deal accordingly. Fixes bug 483. - Stop calling tor_strlower() on uninitialized memory in some cases. Bugfix in 0.2.0.7-alpha. o Minor bugfixes (usability): - Treat some 403 responses from directory servers as INFO rather than WARN-severity events. - It's not actually an error to find a non-pending entry in the DNS cache when canceling a pending resolve. Don't log unless stuff is fishy. Resolves bug 463. o Minor bugfixes (anonymity): - Never report that we've used more bandwidth than we're willing to relay: it leaks how much non-relay traffic we're using. Resolves bug 516. - When looking for a circuit to cannibalize, consider family as well as identity. Fixes bug 438. Bugfix on 0.1.0.x (which introduced circuit cannibalization). o Code simplifications and refactoring: - Make a bunch of functions static. Remove some dead code. - Pull out about a third of the really big routerlist.c; put it in a new module, networkstatus.c. - Merge the extra fields in local_routerstatus_t back into routerstatus_t: we used to need one routerstatus_t for each authority's opinion, plus a local_routerstatus_t for the locally computed consensus opinion. To save space, we put the locally modified fields into local_routerstatus_t, and only the common stuff into routerstatus_t. But once v3 directories are in use, clients and caches will no longer need to hold authority opinions; thus, the rationale for keeping the types separate is now gone. - Make the code used to reschedule and reattempt downloads more uniform. - Turn all 'Are we a directory server/mirror?' logic into a call to dirserver_mode(). - Remove the code to generate the oldest (v1) directory format. The code has been disabled since 0.2.0.5-alpha. Changes in version 0.2.0.7-alpha - 2007-09-21 This seventh development snapshot makes bridges work again, makes bridge authorities work for the first time, fixes two huge performance flaws in hidden services, and fixes a variety of minor issues. o New directory authorities: - Set up moria1 and tor26 as the first v3 directory authorities. See doc/spec/dir-spec.txt for details on the new directory design. o Major bugfixes (crashes): - Fix possible segfaults in functions called from rend_process_relay_cell(). Bugfix on 0.1.2.x. o Major bugfixes (bridges): - Fix a bug that made servers send a "404 Not found" in response to attempts to fetch their server descriptor. This caused Tor servers to take many minutes to establish reachability for their DirPort, and it totally crippled bridges. Bugfix on 0.2.0.5-alpha. - Make "UpdateBridgesFromAuthority" torrc option work: when bridge users configure that and specify a bridge with an identity fingerprint, now they will lookup the bridge descriptor at the default bridge authority via a one-hop tunnel, but once circuits are established they will switch to a three-hop tunnel for later connections to the bridge authority. Bugfix in 0.2.0.3-alpha. o Major bugfixes (hidden services): - Hidden services were choosing introduction points uniquely by hexdigest, but when constructing the hidden service descriptor they merely wrote the (potentially ambiguous) nickname. - Clients now use the v2 intro format for hidden service connections: they specify their chosen rendezvous point by identity digest rather than by (potentially ambiguous) nickname. Both are bugfixes on 0.1.2.x, and they could speed up hidden service connections dramatically. Thanks to Karsten Loesing. o Minor features (security): - As a client, do not believe any server that tells us that an address maps to an internal address space. - Make it possible to enable HashedControlPassword and CookieAuthentication at the same time. o Minor features (guard nodes): - Tag every guard node in our state file with the version that we believe added it, or with our own version if we add it. This way, if a user temporarily runs an old version of Tor and then switches back to a new one, she doesn't automatically lose her guards. o Minor features (speed): - When implementing AES counter mode, update only the portions of the counter buffer that need to change, and don't keep separate network-order and host-order counters when they are the same (i.e., on big-endian hosts.) o Minor features (controller): - Accept LF instead of CRLF on controller, since some software has a hard time generating real Internet newlines. - Add GETINFO values for the server status events "REACHABILITY_SUCCEEDED" and "GOOD_SERVER_DESCRIPTOR". Patch from Robert Hogan. o Removed features: - Routers no longer include bandwidth-history lines in their descriptors; this information is already available in extra-info documents, and including it in router descriptors took up 60% (!) of compressed router descriptor downloads. Completes implementation of proposal 104. - Remove the contrib scripts ExerciseServer.py, PathDemo.py, and TorControl.py, as they use the old v0 controller protocol, and are obsoleted by TorFlow anyway. - Drop support for v1 rendezvous descriptors, since we never used them anyway, and the code has probably rotted by now. Based on patch from Karsten Loesing. - On OSX, stop warning the user that kqueue support in libevent is "experimental", since it seems to have worked fine for ages. o Minor bugfixes: - When generating information telling us how to extend to a given router, do not try to include the nickname if it is absent. Fixes bug 467. Bugfix on 0.2.0.3-alpha. - Fix a user-triggerable (but not remotely-triggerable) segfault in expand_filename(). Bugfix on 0.1.2.x. - Fix a memory leak when freeing incomplete requests from DNSPort. Found by Niels Provos with valgrind. Bugfix on 0.2.0.1-alpha. - Don't try to access (or alter) the state file when running --list-fingerprint or --verify-config or --hash-password. (Resolves bug 499.) Bugfix on 0.1.2.x. - Servers used to decline to publish their DirPort if their BandwidthRate, RelayBandwidthRate, or MaxAdvertisedBandwidth were below a threshold. Now they only look at BandwidthRate and RelayBandwidthRate. Bugfix on 0.1.2.x. - Remove an optimization in the AES counter-mode code that assumed that the counter never exceeded 2^68. When the counter can be set arbitrarily as an IV (as it is by Karsten's new hidden services code), this assumption no longer holds. Bugfix on 0.1.2.x. - Resume listing "AUTHORITY" flag for authorities in network status. Bugfix on 0.2.0.3-alpha; reported by Alex de Joode. o Code simplifications and refactoring: - Revamp file-writing logic so we don't need to have the entire contents of a file in memory at once before we write to disk. Tor, meet stdio. - Turn "descriptor store" into a full-fledged type. - Move all NT services code into a separate source file. - Unify all code that computes medians, percentile elements, etc. - Get rid of a needless malloc when parsing address policies. Changes in version 0.1.2.17 - 2007-08-30 Tor 0.1.2.17 features a new Vidalia version in the Windows and OS X bundles. Vidalia 0.0.14 makes authentication required for the ControlPort in the default configuration, which addresses important security risks. Everybody who uses Vidalia (or another controller) should upgrade. In addition, this Tor update fixes major load balancing problems with path selection, which should speed things up a lot once many people have upgraded. o Major bugfixes (security): - We removed support for the old (v0) control protocol. It has been deprecated since Tor 0.1.1.1-alpha, and keeping it secure has become more of a headache than it's worth. o Major bugfixes (load balancing): - When choosing nodes for non-guard positions, weight guards proportionally less, since they already have enough load. Patch from Mike Perry. - Raise the "max believable bandwidth" from 1.5MB/s to 10MB/s. This will allow fast Tor servers to get more attention. - When we're upgrading from an old Tor version, forget our current guards and pick new ones according to the new weightings. These three load balancing patches could raise effective network capacity by a factor of four. Thanks to Mike Perry for measurements. o Major bugfixes (stream expiration): - Expire not-yet-successful application streams in all cases if they've been around longer than SocksTimeout. Right now there are some cases where the stream will live forever, demanding a new circuit every 15 seconds. Fixes bug 454; reported by lodger. o Minor features (controller): - Add a PROTOCOLINFO controller command. Like AUTHENTICATE, it is valid before any authentication has been received. It tells a controller what kind of authentication is expected, and what protocol is spoken. Implements proposal 119. o Minor bugfixes (performance): - Save on most routerlist_assert_ok() calls in routerlist.c, thus greatly speeding up loading cached-routers from disk on startup. - Disable sentinel-based debugging for buffer code: we squashed all the bugs that this was supposed to detect a long time ago, and now its only effect is to change our buffer sizes from nice powers of two (which platform mallocs tend to like) to values slightly over powers of two (which make some platform mallocs sad). o Minor bugfixes (misc): - If exit bandwidth ever exceeds one third of total bandwidth, then use the correct formula to weight exit nodes when choosing paths. Based on patch from Mike Perry. - Choose perfectly fairly among routers when choosing by bandwidth and weighting by fraction of bandwidth provided by exits. Previously, we would choose with only approximate fairness, and correct ourselves if we ran off the end of the list. - If we require CookieAuthentication but we fail to write the cookie file, we would warn but not exit, and end up in a state where no controller could authenticate. Now we exit. - If we require CookieAuthentication, stop generating a new cookie every time we change any piece of our config. - Refuse to start with certain directory authority keys, and encourage people using them to stop. - Terminate multi-line control events properly. Original patch from tup. - Fix a minor memory leak when we fail to find enough suitable servers to choose a circuit. - Stop leaking part of the descriptor when we run into a particularly unparseable piece of it. Changes in version 0.2.0.6-alpha - 2007-08-26 This sixth development snapshot features a new Vidalia version in the Windows and OS X bundles. Vidalia 0.0.14 makes authentication required for the ControlPort in the default configuration, which addresses important security risks. In addition, this snapshot fixes major load balancing problems with path selection, which should speed things up a lot once many people have upgraded. The directory authorities also use a new mean-time-between-failure approach to tracking which servers are stable, rather than just looking at the most recent uptime. o New directory authorities: - Set up Tonga as the default bridge directory authority. o Major features: - Directory authorities now track servers by weighted mean-times-between-failures. When we have 4 or more days of data, use measured MTBF rather than declared uptime to decide whether to call a router Stable. Implements proposal 108. o Major bugfixes (load balancing): - When choosing nodes for non-guard positions, weight guards proportionally less, since they already have enough load. Patch from Mike Perry. - Raise the "max believable bandwidth" from 1.5MB/s to 10MB/s. This will allow fast Tor servers to get more attention. - When we're upgrading from an old Tor version, forget our current guards and pick new ones according to the new weightings. These three load balancing patches could raise effective network capacity by a factor of four. Thanks to Mike Perry for measurements. o Major bugfixes (descriptor parsing): - Handle unexpected whitespace better in malformed descriptors. Bug found using Benedikt Boss's new Tor fuzzer! Bugfix on 0.2.0.x. o Minor features: - There is now an ugly, temporary "desc/all-recent-extrainfo-hack" GETINFO for Torstat to use until it can switch to using extrainfos. - Optionally (if built with -DEXPORTMALLINFO) export the output of mallinfo via http, as tor/mallinfo.txt. Only accessible from localhost. o Minor bugfixes: - Do not intermix bridge routers with controller-added routers. (Bugfix on 0.2.0.x) - Do not fail with an assert when accept() returns an unexpected address family. Addresses but does not wholly fix bug 483. (Bugfix on 0.2.0.x) - Let directory authorities startup even when they can't generate a descriptor immediately, e.g. because they don't know their address. - Stop putting the authentication cookie in a file called "0" in your working directory if you don't specify anything for the new CookieAuthFile option. Reported by Matt Edman. - Make it possible to read the PROTOCOLINFO response in a way that conforms to our control-spec. Reported by Matt Edman. - Fix a minor memory leak when we fail to find enough suitable servers to choose a circuit. Bugfix on 0.1.2.x. - Stop leaking part of the descriptor when we run into a particularly unparseable piece of it. Bugfix on 0.1.2.x. - Unmap the extrainfo cache file on exit. Changes in version 0.2.0.5-alpha - 2007-08-19 This fifth development snapshot fixes compilation on Windows again; fixes an obnoxious client-side bug that slowed things down and put extra load on the network; gets us closer to using the v3 directory voting scheme; makes it easier for Tor controllers to use cookie-based authentication; and fixes a variety of other bugs. o Removed features: - Version 1 directories are no longer generated in full. Instead, authorities generate and serve "stub" v1 directories that list no servers. This will stop Tor versions 0.1.0.x and earlier from working, but (for security reasons) nobody should be running those versions anyway. o Major bugfixes (compilation, 0.2.0.x): - Try to fix Win32 compilation again: improve checking for IPv6 types. - Try to fix MSVC compilation: build correctly on platforms that do not define s6_addr16 or s6_addr32. - Fix compile on platforms without getaddrinfo: bug found by Li-Hui Zhou. o Major bugfixes (stream expiration): - Expire not-yet-successful application streams in all cases if they've been around longer than SocksTimeout. Right now there are some cases where the stream will live forever, demanding a new circuit every 15 seconds. Bugfix on 0.1.2.7-alpha; fixes bug 454; reported by lodger. o Minor features (directory servers): - When somebody requests a list of statuses or servers, and we have none of those, return a 404 rather than an empty 200. o Minor features (directory voting): - Store v3 consensus status consensuses on disk, and reload them on startup. o Minor features (security): - Warn about unsafe ControlPort configurations. - Refuse to start with certain directory authority keys, and encourage people using them to stop. o Minor features (controller): - Add a PROTOCOLINFO controller command. Like AUTHENTICATE, it is valid before any authentication has been received. It tells a controller what kind of authentication is expected, and what protocol is spoken. Implements proposal 119. - New config option CookieAuthFile to choose a new location for the cookie authentication file, and config option CookieAuthFileGroupReadable to make it group-readable. o Minor features (unit testing): - Add command-line arguments to unit-test executable so that we can invoke any chosen test from the command line rather than having to run the whole test suite at once; and so that we can turn on logging for the unit tests. o Minor bugfixes (on 0.1.2.x): - If we require CookieAuthentication but we fail to write the cookie file, we would warn but not exit, and end up in a state where no controller could authenticate. Now we exit. - If we require CookieAuthentication, stop generating a new cookie every time we change any piece of our config. - When loading bandwidth history, do not believe any information in the future. Fixes bug 434. - When loading entry guard information, do not believe any information in the future. - When we have our clock set far in the future and generate an onion key, then re-set our clock to be correct, we should not stop the onion key from getting rotated. - Clean up torrc sample config file. - Do not automatically run configure from autogen.sh. This non-standard behavior tended to annoy people who have built other programs. o Minor bugfixes (on 0.2.0.x): - Fix a bug with AutomapHostsOnResolve that would always cause the second request to fail. Bug reported by Kate. Bugfix on 0.2.0.3-alpha. - Fix a bug in ADDRMAP controller replies that would sometimes try to print a NULL. Patch from tup. - Read v3 directory authority keys from the right location. - Numerous bugfixes to directory voting code. Changes in version 0.1.2.16 - 2007-08-01 Tor 0.1.2.16 fixes a critical security vulnerability that allows a remote attacker in certain situations to rewrite the user's torrc configuration file. This can completely compromise anonymity of users in most configurations, including those running the Vidalia bundles, TorK, etc. Or worse. o Major security fixes: - Close immediately after missing authentication on control port; do not allow multiple authentication attempts. Changes in version 0.2.0.4-alpha - 2007-08-01 This fourth development snapshot fixes a critical security vulnerability for most users, specifically those running Vidalia, TorK, etc. Everybody should upgrade to either 0.1.2.16 or 0.2.0.4-alpha. o Major security fixes: - Close immediately after missing authentication on control port; do not allow multiple authentication attempts. o Major bugfixes (compilation): - Fix win32 compilation: apparently IN_ADDR and IN6_ADDR are already defined there. o Minor features (performance): - Be even more aggressive about releasing RAM from small empty buffers. Thanks to our free-list code, this shouldn't be too performance-intensive. - Disable sentinel-based debugging for buffer code: we squashed all the bugs that this was supposed to detect a long time ago, and now its only effect is to change our buffer sizes from nice powers of two (which platform mallocs tend to like) to values slightly over powers of two (which make some platform mallocs sad). - Log malloc statistics from mallinfo() on platforms where it exists. Changes in version 0.2.0.3-alpha - 2007-07-29 This third development snapshot introduces new experimental blocking-resistance features and a preliminary version of the v3 directory voting design, and includes many other smaller features and bugfixes. o Major features: - The first pieces of our "bridge" design for blocking-resistance are implemented. People can run bridge directory authorities; people can run bridges; and people can configure their Tor clients with a set of bridges to use as the first hop into the Tor network. See http://archives.seul.org/or/talk/Jul-2007/msg00249.html for details. - Create listener connections before we setuid to the configured User and Group. Now non-Windows users can choose port values under 1024, start Tor as root, and have Tor bind those ports before it changes to another UID. (Windows users could already pick these ports.) - Added a new ConstrainedSockets config option to set SO_SNDBUF and SO_RCVBUF on TCP sockets. Hopefully useful for Tor servers running on "vserver" accounts. (Patch from coderman.) - Be even more aggressive about separating local traffic from relayed traffic when RelayBandwidthRate is set. (Refines proposal 111.) o Major features (experimental): - First cut of code for "v3 dir voting": directory authorities will vote on a common network status document rather than each publishing their own opinion. This code needs more testing and more corner-case handling before it's ready for use. o Security fixes: - Directory authorities now call routers Fast if their bandwidth is at least 100KB/s, and consider their bandwidth adequate to be a Guard if it is at least 250KB/s, no matter the medians. This fix complements proposal 107. [Bugfix on 0.1.2.x] - Directory authorities now never mark more than 3 servers per IP as Valid and Running. (Implements proposal 109, by Kevin Bauer and Damon McCoy.) - Minor change to organizationName and commonName generation procedures in TLS certificates during Tor handshakes, to invalidate some earlier censorware approaches. This is not a long-term solution, but applying it will give us a bit of time to look into the epidemiology of countermeasures as they spread. o Major bugfixes (directory): - Rewrite directory tokenization code to never run off the end of a string. Fixes bug 455. Patch from croup. [Bugfix on 0.1.2.x] o Minor features (controller): - Add a SOURCE_ADDR field to STREAM NEW events so that controllers can match requests to applications. (Patch from Robert Hogan.) - Report address and port correctly on connections to DNSPort. (Patch from Robert Hogan.) - Add a RESOLVE command to launch hostname lookups. (Original patch from Robert Hogan.) - Add GETINFO status/enough-dir-info to let controllers tell whether Tor has downloaded sufficient directory information. (Patch from Tup.) - You can now use the ControlSocket option to tell Tor to listen for controller connections on Unix domain sockets on systems that support them. (Patch from Peter Palfrader.) - STREAM NEW events are generated for DNSPort requests and for tunneled directory connections. (Patch from Robert Hogan.) - New "GETINFO address-mappings/*" command to get address mappings with expiry information. "addr-mappings/*" is now deprecated. (Patch from Tup.) o Minor features (misc): - Merge in some (as-yet-unused) IPv6 address manipulation code. (Patch from croup.) - The tor-gencert tool for v3 directory authorities now creates all files as readable to the file creator only, and write-protects the authority identity key. - When dumping memory usage, list bytes used in buffer memory free-lists. - When running with dmalloc, dump more stats on hup and on exit. - Directory authorities now fail quickly and (relatively) harmlessly if they generate a network status document that is somehow malformed. o Traffic load balancing improvements: - If exit bandwidth ever exceeds one third of total bandwidth, then use the correct formula to weight exit nodes when choosing paths. (Based on patch from Mike Perry.) - Choose perfectly fairly among routers when choosing by bandwidth and weighting by fraction of bandwidth provided by exits. Previously, we would choose with only approximate fairness, and correct ourselves if we ran off the end of the list. [Bugfix on 0.1.2.x] o Performance improvements: - Be more aggressive with freeing buffer RAM or putting it on the memory free lists. - Use Critical Sections rather than Mutexes for synchronizing threads on win32; Mutexes are heavier-weight, and designed for synchronizing between processes. o Deprecated and removed features: - RedirectExits is now deprecated. - Stop allowing address masks that do not correspond to bit prefixes. We have warned about these for a really long time; now it's time to reject them. (Patch from croup.) o Minor bugfixes (directory): - Fix another crash bug related to extra-info caching. (Bug found by Peter Palfrader.) [Bugfix on 0.2.0.2-alpha] - Directories no longer return a "304 not modified" when they don't have the networkstatus the client asked for. Also fix a memory leak when returning 304 not modified. [Bugfixes on 0.2.0.2-alpha] - We had accidentally labelled 0.1.2.x directory servers as not suitable for begin_dir requests, and had labelled no directory servers as suitable for uploading extra-info documents. [Bugfix on 0.2.0.1-alpha] o Minor bugfixes (dns): - Fix a crash when DNSPort is set more than once. (Patch from Robert Hogan.) [Bugfix on 0.2.0.2-alpha] - Add DNSPort connections to the global connection list, so that we can time them out correctly. (Bug found by Robert Hogan.) [Bugfix on 0.2.0.2-alpha] - Fix a dangling reference that could lead to a crash when DNSPort is changed or closed (Patch from Robert Hogan.) [Bugfix on 0.2.0.2-alpha] o Minor bugfixes (controller): - Provide DNS expiry times in GMT, not in local time. For backward compatibility, ADDRMAP events only provide GMT expiry in an extended field. "GETINFO address-mappings" always does the right thing. - Use CRLF line endings properly in NS events. - Terminate multi-line control events properly. (Original patch from tup.) [Bugfix on 0.1.2.x-alpha] - Do not include spaces in SOURCE_ADDR fields in STREAM events. Resolves bug 472. [Bugfix on 0.2.0.x-alpha] Changes in version 0.1.2.15 - 2007-07-17 Tor 0.1.2.15 fixes several crash bugs, fixes some anonymity-related problems, fixes compilation on BSD, and fixes a variety of other bugs. Everybody should upgrade. o Major bugfixes (compilation): - Fix compile on FreeBSD/NetBSD/OpenBSD. Oops. o Major bugfixes (crashes): - Try even harder not to dereference the first character after an mmap(). Reported by lodger. - Fix a crash bug in directory authorities when we re-number the routerlist while inserting a new router. - When the cached-routers file is an even multiple of the page size, don't run off the end and crash. (Fixes bug 455; based on idea from croup.) - Fix eventdns.c behavior on Solaris: It is critical to include orconfig.h _before_ sys/types.h, so that we can get the expected definition of _FILE_OFFSET_BITS. o Major bugfixes (security): - Fix a possible buffer overrun when using BSD natd support. Bug found by croup. - When sending destroy cells from a circuit's origin, don't include the reason for tearing down the circuit. The spec says we didn't, and now we actually don't. Reported by lodger. - Keep streamids from different exits on a circuit separate. This bug may have allowed other routers on a given circuit to inject cells into streams. Reported by lodger; fixes bug 446. - If there's a never-before-connected-to guard node in our list, never choose any guards past it. This way we don't expand our guard list unless we need to. o Minor bugfixes (guard nodes): - Weight guard selection by bandwidth, so that low-bandwidth nodes don't get overused as guards. o Minor bugfixes (directory): - Correctly count the number of authorities that recommend each version. Previously, we were under-counting by 1. - Fix a potential crash bug when we load many server descriptors at once and some of them make others of them obsolete. Fixes bug 458. o Minor bugfixes (hidden services): - Stop tearing down the whole circuit when the user asks for a connection to a port that the hidden service didn't configure. Resolves bug 444. o Minor bugfixes (misc): - On Windows, we were preventing other processes from reading cached-routers while Tor was running. Reported by janbar. - Fix a possible (but very unlikely) bug in picking routers by bandwidth. Add a log message to confirm that it is in fact unlikely. Patch from lodger. - Backport a couple of memory leak fixes. - Backport miscellaneous cosmetic bugfixes. Changes in version 0.2.0.2-alpha - 2007-06-02 o Major bugfixes on 0.2.0.1-alpha: - Fix an assertion failure related to servers without extra-info digests. Resolves bugs 441 and 442. o Minor features (directory): - Support "If-Modified-Since" when answering HTTP requests for directories, running-routers documents, and network-status documents. (There's no need to support it for router descriptors, since those are downloaded by descriptor digest.) o Minor build issues: - Clear up some MIPSPro compiler warnings. - When building from a tarball on a machine that happens to have SVK installed, report the micro-revision as whatever version existed in the tarball, not as "x". Changes in version 0.2.0.1-alpha - 2007-06-01 This early development snapshot provides new features for people running Tor as both a client and a server (check out the new RelayBandwidth config options); lets Tor run as a DNS proxy; and generally moves us forward on a lot of fronts. o Major features, server usability: - New config options RelayBandwidthRate and RelayBandwidthBurst: a separate set of token buckets for relayed traffic. Right now relayed traffic is defined as answers to directory requests, and OR connections that don't have any local circuits on them. o Major features, client usability: - A client-side DNS proxy feature to replace the need for dns-proxy-tor: Just set "DNSPort 9999", and Tor will now listen for DNS requests on port 9999, use the Tor network to resolve them anonymously, and send the reply back like a regular DNS server. The code still only implements a subset of DNS. - Make PreferTunneledDirConns and TunnelDirConns work even when we have no cached directory info. This means Tor clients can now do all of their connections protected by TLS. o Major features, performance and efficiency: - Directory authorities accept and serve "extra info" documents for routers. These documents contain fields from router descriptors that aren't usually needed, and that use a lot of excess bandwidth. Once these fields are removed from router descriptors, the bandwidth savings should be about 60%. [Partially implements proposal 104.] - Servers upload extra-info documents to any authority that accepts them. Authorities (and caches that have been configured to download extra-info documents) download them as needed. [Partially implements proposal 104.] - Change the way that Tor buffers data that it is waiting to write. Instead of queueing data cells in an enormous ring buffer for each client->OR or OR->OR connection, we now queue cells on a separate queue for each circuit. This lets us use less slack memory, and will eventually let us be smarter about prioritizing different kinds of traffic. - Use memory pools to allocate cells with better speed and memory efficiency, especially on platforms where malloc() is inefficient. - Stop reading on edge connections when their corresponding circuit buffers are full; start again as the circuits empty out. o Major features, other: - Add an HSAuthorityRecordStats option that hidden service authorities can use to track statistics of overall hidden service usage without logging information that would be very useful to an attacker. - Start work implementing multi-level keys for directory authorities: Add a standalone tool to generate key certificates. (Proposal 103.) o Security fixes: - Directory authorities now call routers Stable if they have an uptime of at least 30 days, even if that's not the median uptime in the network. Implements proposal 107, suggested by Kevin Bauer and Damon McCoy. o Minor fixes (resource management): - Count the number of open sockets separately from the number of active connection_t objects. This will let us avoid underusing our allocated connection limit. - We no longer use socket pairs to link an edge connection to an anonymous directory connection or a DirPort test connection. Instead, we track the link internally and transfer the data in-process. This saves two sockets per "linked" connection (at the client and at the server), and avoids the nasty Windows socketpair() workaround. - Keep unused 4k and 16k buffers on free lists, rather than wasting 8k for every single inactive connection_t. Free items from the 4k/16k-buffer free lists when they haven't been used for a while. o Minor features (build): - Make autoconf search for libevent, openssl, and zlib consistently. - Update deprecated macros in configure.in. - When warning about missing headers, tell the user to let us know if the compile succeeds anyway, so we can downgrade the warning. - Include the current subversion revision as part of the version string: either fetch it directly if we're in an SVN checkout, do some magic to guess it if we're in an SVK checkout, or use the last-detected version if we're building from a .tar.gz. Use this version consistently in log messages. o Minor features (logging): - Always prepend "Bug: " to any log message about a bug. - Put a platform string (e.g. "Linux i686") in the startup log message, so when people paste just their logs, we know if it's OpenBSD or Windows or what. - When logging memory usage, break down memory used in buffers by buffer type. o Minor features (directory system): - New config option V2AuthoritativeDirectory that all directory authorities should set. This will let future authorities choose not to serve V2 directory information. - Directory authorities allow multiple router descriptors and/or extra info documents to be uploaded in a single go. This will make implementing proposal 104 simpler. o Minor features (controller): - Add a new config option __DisablePredictedCircuits designed for use by the controller, when we don't want Tor to build any circuits preemptively. - Let the controller specify HOP=%d as an argument to ATTACHSTREAM, so we can exit from the middle of the circuit. - Implement "getinfo status/circuit-established". - Implement "getinfo status/version/..." so a controller can tell whether the current version is recommended, and whether any versions are good, and how many authorities agree. (Patch from shibz.) o Minor features (hidden services): - Allow multiple HiddenServicePort directives with the same virtual port; when they occur, the user is sent round-robin to one of the target ports chosen at random. Partially fixes bug 393 by adding limited ad-hoc round-robining. o Minor features (other): - More unit tests. - Add a new AutomapHostsOnResolve option: when it is enabled, any resolve request for hosts matching a given pattern causes Tor to generate an internal virtual address mapping for that host. This allows DNSPort to work sensibly with hidden service users. By default, .exit and .onion addresses are remapped; the list of patterns can be reconfigured with AutomapHostsSuffixes. - Add an "-F" option to tor-resolve to force a resolve for a .onion address. Thanks to the AutomapHostsOnResolve option, this is no longer a completely silly thing to do. - If Tor is invoked from something that isn't a shell (e.g. Vidalia), now we expand "-f ~/.tor/torrc" correctly. Suggested by Matt Edman. - Treat "2gb" when given in torrc for a bandwidth as meaning 2gb, minus 1 byte: the actual maximum declared bandwidth. o Removed features: - Removed support for the old binary "version 0" controller protocol. This has been deprecated since 0.1.1, and warnings have been issued since 0.1.2. When we encounter a v0 control message, we now send back an error and close the connection. - Remove the old "dns worker" server DNS code: it hasn't been default since 0.1.2.2-alpha, and all the servers seem to be using the new eventdns code. o Minor bugfixes (portability): - Even though Windows is equally happy with / and \ as path separators, try to use \ consistently on Windows and / consistently on Unix: it makes the log messages nicer. - Correctly report platform name on Windows 95 OSR2 and Windows 98 SE. - Read resolv.conf files correctly on platforms where read() returns partial results on small file reads. o Minor bugfixes (directory): - Correctly enforce that elements of directory objects do not appear more often than they are allowed to appear. - When we are reporting the DirServer line we just parsed, we were logging the second stanza of the key fingerprint, not the first. o Minor bugfixes (logging): - When we hit an EOF on a log (probably because we're shutting down), don't try to remove the log from the list: just mark it as unusable. (Bulletproofs against bug 222.) o Minor bugfixes (other): - In the exitlist script, only consider the most recently published server descriptor for each server. Also, when the user requests a list of servers that _reject_ connections to a given address, explicitly exclude the IPs that also have servers that accept connections to that address. (Resolves bug 405.) - Stop allowing hibernating servers to be "stable" or "fast". - On Windows, we were preventing other processes from reading cached-routers while Tor was running. (Reported by janbar) - Make the NodeFamilies config option work. (Reported by lodger -- it has never actually worked, even though we added it in Oct 2004.) - Check return values from pthread_mutex functions. - Don't save non-general-purpose router descriptors to the disk cache, because we have no way of remembering what their purpose was when we restart. - Add even more asserts to hunt down bug 417. - Build without verbose warnings even on (not-yet-released) gcc 4.2. - Fix a possible (but very unlikely) bug in picking routers by bandwidth. Add a log message to confirm that it is in fact unlikely. o Minor bugfixes (controller): - Make 'getinfo fingerprint' return a 551 error if we're not a server, so we match what the control spec claims we do. Reported by daejees. - Fix a typo in an error message when extendcircuit fails that caused us to not follow the \r\n-based delimiter protocol. Reported by daejees. o Code simplifications and refactoring: - Stop passing around circuit_t and crypt_path_t pointers that are implicit in other procedure arguments. - Drop the old code to choke directory connections when the corresponding OR connections got full: thanks to the cell queue feature, OR conns don't get full any more. - Make dns_resolve() handle attaching connections to circuits properly, so the caller doesn't have to. - Rename wants_to_read and wants_to_write to read/write_blocked_on_bw. - Keep the connection array as a dynamic smartlist_t, rather than as a fixed-sized array. This is important, as the number of connections is becoming increasingly decoupled from the number of sockets. Changes in version 0.1.2.14 - 2007-05-25 Tor 0.1.2.14 changes the addresses of two directory authorities (this change especially affects those who serve or use hidden services), and fixes several other crash- and security-related bugs. o Directory authority changes: - Two directory authorities (moria1 and moria2) just moved to new IP addresses. This change will particularly affect those who serve or use hidden services. o Major bugfixes (crashes): - If a directory server runs out of space in the connection table as it's processing a begin_dir request, it will free the exit stream but leave it attached to the circuit, leading to unpredictable behavior. (Reported by seeess, fixes bug 425.) - Fix a bug in dirserv_remove_invalid() that would cause authorities to corrupt memory under some really unlikely scenarios. - Tighten router parsing rules. (Bugs reported by Benedikt Boss.) - Avoid segfaults when reading from mmaped descriptor file. (Reported by lodger.) o Major bugfixes (security): - When choosing an entry guard for a circuit, avoid using guards that are in the same family as the chosen exit -- not just guards that are exactly the chosen exit. (Reported by lodger.) o Major bugfixes (resource management): - If a directory authority is down, skip it when deciding where to get networkstatus objects or descriptors. Otherwise we keep asking every 10 seconds forever. Fixes bug 384. - Count it as a failure if we fetch a valid network-status but we don't want to keep it. Otherwise we'll keep fetching it and keep not wanting to keep it. Fixes part of bug 422. - If all of our dirservers have given us bad or no networkstatuses lately, then stop hammering them once per minute even when we think they're failed. Fixes another part of bug 422. o Minor bugfixes: - Actually set the purpose correctly for descriptors inserted with purpose=controller. - When we have k non-v2 authorities in our DirServer config, we ignored the last k authorities in the list when updating our network-statuses. - Correctly back-off from requesting router descriptors that we are having a hard time downloading. - Read resolv.conf files correctly on platforms where read() returns partial results on small file reads. - Don't rebuild the entire router store every time we get 32K of routers: rebuild it when the journal gets very large, or when the gaps in the store get very large. o Minor features: - When routers publish SVN revisions in their router descriptors, authorities now include those versions correctly in networkstatus documents. - Warn when using a version of libevent before 1.3b to run a server on OSX or BSD: these versions interact badly with userspace threads. Changes in version 0.1.2.13 - 2007-04-24 This release features some major anonymity fixes, such as safer path selection; better client performance; faster bootstrapping, better address detection, and better DNS support for servers; write limiting as well as read limiting to make servers easier to run; and a huge pile of other features and bug fixes. The bundles also ship with Vidalia 0.0.11. Tor 0.1.2.13 is released in memory of Rob Levin (1955-2006), aka lilo of the Freenode IRC network, remembering his patience and vision for free speech on the Internet. o Minor fixes: - Fix a memory leak when we ask for "all" networkstatuses and we get one we don't recognize. - Add more asserts to hunt down bug 417. - Disable kqueue on OS X 10.3 and earlier, to fix bug 371. Changes in version 0.1.2.12-rc - 2007-03-16 o Major bugfixes: - Fix an infinite loop introduced in 0.1.2.7-alpha when we serve directory information requested inside Tor connections (i.e. via begin_dir cells). It only triggered when the same connection was serving other data at the same time. Reported by seeess. o Minor bugfixes: - When creating a circuit via the controller, send a 'launched' event when we're done, so we follow the spec better. Changes in version 0.1.2.11-rc - 2007-03-15 o Minor bugfixes (controller), reported by daejees: - Correct the control spec to match how the code actually responds to 'getinfo addr-mappings/*'. - The control spec described a GUARDS event, but the code implemented a GUARD event. Standardize on GUARD, but let people ask for GUARDS too. Changes in version 0.1.2.10-rc - 2007-03-07 o Major bugfixes (Windows): - Do not load the NT services library functions (which may not exist) just to detect if we're a service trying to shut down. Now we run on Win98 and friends again. o Minor bugfixes (other): - Clarify a couple of log messages. - Fix a misleading socks5 error number. Changes in version 0.1.2.9-rc - 2007-03-02 o Major bugfixes (Windows): - On MinGW, use "%I64u" to printf/scanf 64-bit integers, instead of the usual GCC "%llu". This prevents a bug when saving 64-bit int configuration values: the high-order 32 bits would get truncated. In particular, we were being bitten by the default MaxAdvertisedBandwidth of 128 TB turning into 0. (Fixes bug 400 and maybe also bug 397.) o Minor bugfixes (performance): - Use OpenSSL's AES implementation on platforms where it's faster. This could save us as much as 10% CPU usage. o Minor bugfixes (server): - Do not rotate onion key immediately after setting it for the first time. o Minor bugfixes (directory authorities): - Stop calling servers that have been hibernating for a long time "stable". Also, stop letting hibernating or obsolete servers affect uptime and bandwidth cutoffs. - Stop listing hibernating servers in the v1 directory. o Minor bugfixes (hidden services): - Upload hidden service descriptors slightly less often, to reduce load on authorities. o Minor bugfixes (other): - Fix an assert that could trigger if a controller quickly set then cleared EntryNodes. Bug found by Udo van den Heuvel. - On architectures where sizeof(int)>4, still clamp declarable bandwidth to INT32_MAX. - Fix a potential race condition in the rpm installer. Found by Stefan Nordhausen. - Try to fix eventdns warnings once and for all: do not treat a dns rcode of 2 as indicating that the server is completely bad; it sometimes means that the server is just bad for the request in question. (may fix the last of bug 326.) - Disable encrypted directory connections when we don't have a server descriptor for the destination. We'll get this working again in the 0.2.0 branch. Changes in version 0.1.2.8-beta - 2007-02-26 o Major bugfixes (crashes): - Stop crashing when the controller asks us to resetconf more than one config option at once. (Vidalia 0.0.11 does this.) - Fix a crash that happened on Win98 when we're given command-line arguments: don't try to load NT service functions from advapi32.dll except when we need them. (Bug introduced in 0.1.2.7-alpha; resolves bug 389.) - Fix a longstanding obscure crash bug that could occur when we run out of DNS worker processes. (Resolves bug 390.) o Major bugfixes (hidden services): - Correctly detect whether hidden service descriptor downloads are in-progress. (Suggested by Karsten Loesing; fixes bug 399.) o Major bugfixes (accounting): - When we start during an accounting interval before it's time to wake up, remember to wake up at the correct time. (May fix bug 342.) o Minor bugfixes (controller): - Give the controller END_STREAM_REASON_DESTROY events _before_ we clear the corresponding on_circuit variable, and remember later that we don't need to send a redundant CLOSED event. Resolves part 3 of bug 367. - Report events where a resolve succeeded or where we got a socks protocol error correctly, rather than calling both of them "INTERNAL". - Change reported stream target addresses to IP consistently when we finally get the IP from an exit node. - Send log messages to the controller even if they happen to be very long. o Minor bugfixes (other): - Display correct results when reporting which versions are recommended, and how recommended they are. (Resolves bug 383.) - Improve our estimates for directory bandwidth to be less random: guess that an unrecognized directory will have the average bandwidth from all known directories, not that it will have the average bandwidth from those directories earlier than it on the list. - If we start a server with ClientOnly 1, then set ClientOnly to 0 and hup, stop triggering an assert based on an empty onion_key. - On platforms with no working mmap() equivalent, don't warn the user when cached-routers doesn't exist. - Warn the user when mmap() [or its equivalent] fails for some reason other than file-not-found. - Don't warn the user when cached-routers.new doesn't exist: that's perfectly fine when starting up for the first time. - When EntryNodes are configured, rebuild the guard list to contain, in order: the EntryNodes that were guards before; the rest of the EntryNodes; the nodes that were guards before. - Mask out all signals in sub-threads; only the libevent signal handler should be processing them. This should prevent some crashes on some machines using pthreads. (Patch from coderman.) - Fix switched arguments on memset in the implementation of tor_munmap() for systems with no mmap() call. - When Tor receives a router descriptor that it asked for, but no longer wants (because it has received fresh networkstatuses in the meantime), do not warn the user. Cache the descriptor if we're a cache; drop it if we aren't. - Make earlier entry guards _really_ get retried when the network comes back online. - On a malformed DNS reply, always give an error to the corresponding DNS request. - Build with recent libevents on platforms that do not define the nonstandard types "u_int8_t" and friends. o Minor features (controller): - Warn the user when an application uses the obsolete binary v0 control protocol. We're planning to remove support for it during the next development series, so it's good to give people some advance warning. - Add STREAM_BW events to report per-entry-stream bandwidth use. (Patch from Robert Hogan.) - Rate-limit SIGNEWNYM signals in response to controllers that impolitely generate them for every single stream. (Patch from mwenge; closes bug 394.) - Make REMAP stream events have a SOURCE (cache or exit), and make them generated in every case where we get a successful connected or resolved cell. o Minor bugfixes (performance): - Call router_have_min_dir_info half as often. (This is showing up in some profiles, but not others.) - When using GCC, make log_debug never get called at all, and its arguments never get evaluated, when no debug logs are configured. (This is showing up in some profiles, but not others.) o Minor features: - Remove some never-implemented options. Mark PathlenCoinWeight as obsolete. - Implement proposal 106: Stop requiring clients to have well-formed certificates; stop checking nicknames in certificates. (Clients have certificates so that they can look like Tor servers, but in the future we might want to allow them to look like regular TLS clients instead. Nicknames in certificates serve no purpose other than making our protocol easier to recognize on the wire.) - Revise messages on handshake failure again to be even more clear about which are incoming connections and which are outgoing. - Discard any v1 directory info that's over 1 month old (for directories) or over 1 week old (for running-routers lists). - Do not warn when individual nodes in the configuration's EntryNodes, ExitNodes, etc are down: warn only when all possible nodes are down. (Fixes bug 348.) - Always remove expired routers and networkstatus docs before checking whether we have enough information to build circuits. (Fixes bug 373.) - Put a lower-bound on MaxAdvertisedBandwidth. Changes in version 0.1.2.7-alpha - 2007-02-06 o Major bugfixes (rate limiting): - Servers decline directory requests much more aggressively when they're low on bandwidth. Otherwise they end up queueing more and more directory responses, which can't be good for latency. - But never refuse directory requests from local addresses. - Fix a memory leak when sending a 503 response for a networkstatus request. - Be willing to read or write on local connections (e.g. controller connections) even when the global rate limiting buckets are empty. - If our system clock jumps back in time, don't publish a negative uptime in the descriptor. Also, don't let the global rate limiting buckets go absurdly negative. - Flush local controller connection buffers periodically as we're writing to them, so we avoid queueing 4+ megabytes of data before trying to flush. o Major bugfixes (NT services): - Install as NT_AUTHORITY\LocalService rather than as SYSTEM; add a command-line flag so that admins can override the default by saying "tor --service install --user "SomeUser"". This will not affect existing installed services. Also, warn the user that the service will look for its configuration file in the service user's %appdata% directory. (We can't do the 'hardwire the user's appdata directory' trick any more, since we may not have read access to that directory.) o Major bugfixes (other): - Previously, we would cache up to 16 old networkstatus documents indefinitely, if they came from nontrusted authorities. Now we discard them if they are more than 10 days old. - Fix a crash bug in the presence of DNS hijacking (reported by Andrew Del Vecchio). - Detect and reject malformed DNS responses containing circular pointer loops. - If exits are rare enough that we're not marking exits as guards, ignore exit bandwidth when we're deciding the required bandwidth to become a guard. - When we're handling a directory connection tunneled over Tor, don't fill up internal memory buffers with all the data we want to tunnel; instead, only add it if the OR connection that will eventually receive it has some room for it. (This can lead to slowdowns in tunneled dir connections; a better solution will have to wait for 0.2.0.) o Minor bugfixes (dns): - Add some defensive programming to eventdns.c in an attempt to catch possible memory-stomping bugs. - Detect and reject DNS replies containing IPv4 or IPv6 records with an incorrect number of bytes. (Previously, we would ignore the extra bytes.) - Fix as-yet-unused reverse IPv6 lookup code so it sends nybbles in the correct order, and doesn't crash. - Free memory held in recently-completed DNS lookup attempts on exit. This was not a memory leak, but may have been hiding memory leaks. - Handle TTL values correctly on reverse DNS lookups. - Treat failure to parse resolv.conf as an error. o Minor bugfixes (other): - Fix crash with "tor --list-fingerprint" (reported by seeess). - When computing clock skew from directory HTTP headers, consider what time it was when we finished asking for the directory, not what time it is now. - Expire socks connections if they spend too long waiting for the handshake to finish. Previously we would let them sit around for days, if the connecting application didn't close them either. - And if the socks handshake hasn't started, don't send a "DNS resolve socks failed" handshake reply; just close it. - Stop using C functions that OpenBSD's linker doesn't like. - Don't launch requests for descriptors unless we have networkstatuses from at least half of the authorities. This delays the first download slightly under pathological circumstances, but can prevent us from downloading a bunch of descriptors we don't need. - Do not log IPs with TLS failures for incoming TLS connections. (Fixes bug 382.) - If the user asks to use invalid exit nodes, be willing to use unstable ones. - Stop using the reserved ac_cv namespace in our configure script. - Call stat() slightly less often; use fstat() when possible. - Refactor the way we handle pending circuits when an OR connection completes or fails, in an attempt to fix a rare crash bug. - Only rewrite a conn's address based on X-Forwarded-For: headers if it's a parseable public IP address; and stop adding extra quotes to the resulting address. o Major features: - Weight directory requests by advertised bandwidth. Now we can let servers enable write limiting but still allow most clients to succeed at their directory requests. (We still ignore weights when choosing a directory authority; I hope this is a feature.) o Minor features: - Create a new file ReleaseNotes which was the old ChangeLog. The new ChangeLog file now includes the summaries for all development versions too. - Check for addresses with invalid characters at the exit as well as at the client, and warn less verbosely when they fail. You can override this by setting ServerDNSAllowNonRFC953Addresses to 1. - Adapt a patch from goodell to let the contrib/exitlist script take arguments rather than require direct editing. - Inform the server operator when we decide not to advertise a DirPort due to AccountingMax enabled or a low BandwidthRate. It was confusing Zax, so now we're hopefully more helpful. - Bring us one step closer to being able to establish an encrypted directory tunnel without knowing a descriptor first. Still not ready yet. As part of the change, now assume we can use a create_fast cell if we don't know anything about a router. - Allow exit nodes to use nameservers running on ports other than 53. - Servers now cache reverse DNS replies. - Add an --ignore-missing-torrc command-line option so that we can get the "use sensible defaults if the configuration file doesn't exist" behavior even when specifying a torrc location on the command line. o Minor features (controller): - Track reasons for OR connection failure; make these reasons available via the controller interface. (Patch from Mike Perry.) - Add a SOCKS_BAD_HOSTNAME client status event so controllers can learn when clients are sending malformed hostnames to Tor. - Clean up documentation for controller status events. - Add a REMAP status to stream events to note that a stream's address has changed because of a cached address or a MapAddress directive. Changes in version 0.1.2.6-alpha - 2007-01-09 o Major bugfixes: - Fix an assert error introduced in 0.1.2.5-alpha: if a single TLS connection handles more than 4 gigs in either direction, we crash. - Fix an assert error introduced in 0.1.2.5-alpha: if we're an advertised exit node, somebody might try to exit from us when we're bootstrapping and before we've built our descriptor yet. Refuse the connection rather than crashing. o Minor bugfixes: - Warn if we (as a server) find that we've resolved an address that we weren't planning to resolve. - Warn that using select() on any libevent version before 1.1 will be unnecessarily slow (even for select()). - Flush ERR-level controller status events just like we currently flush ERR-level log events, so that a Tor shutdown doesn't prevent the controller from learning about current events. o Minor features (more controller status events): - Implement EXTERNAL_ADDRESS server status event so controllers can learn when our address changes. - Implement BAD_SERVER_DESCRIPTOR server status event so controllers can learn when directories reject our descriptor. - Implement SOCKS_UNKNOWN_PROTOCOL client status event so controllers can learn when a client application is speaking a non-socks protocol to our SocksPort. - Implement DANGEROUS_SOCKS client status event so controllers can learn when a client application is leaking DNS addresses. - Implement BUG general status event so controllers can learn when Tor is unhappy about its internal invariants. - Implement CLOCK_SKEW general status event so controllers can learn when Tor thinks the system clock is set incorrectly. - Implement GOOD_SERVER_DESCRIPTOR and ACCEPTED_SERVER_DESCRIPTOR server status events so controllers can learn when their descriptors are accepted by a directory. - Implement CHECKING_REACHABILITY and REACHABILITY_{SUCCEEDED|FAILED} server status events so controllers can learn about Tor's progress in deciding whether it's reachable from the outside. - Implement BAD_LIBEVENT general status event so controllers can learn when we have a version/method combination in libevent that needs to be changed. - Implement NAMESERVER_STATUS, NAMESERVER_ALL_DOWN, DNS_HIJACKED, and DNS_USELESS server status events so controllers can learn about changes to DNS server status. o Minor features (directory): - Authorities no longer recommend exits as guards if this would shift too much load to the exit nodes. Changes in version 0.1.2.5-alpha - 2007-01-06 o Major features: - Enable write limiting as well as read limiting. Now we sacrifice capacity if we're pushing out lots of directory traffic, rather than overrunning the user's intended bandwidth limits. - Include TLS overhead when counting bandwidth usage; previously, we would count only the bytes sent over TLS, but not the bytes used to send them. - Support running the Tor service with a torrc not in the same directory as tor.exe and default to using the torrc located in the %appdata%\Tor\ of the user who installed the service. Patch from Matt Edman. - Servers now check for the case when common DNS requests are going to wildcarded addresses (i.e. all getting the same answer), and change their exit policy to reject *:* if it's happening. - Implement BEGIN_DIR cells, so we can connect to the directory server via TLS to do encrypted directory requests rather than plaintext. Enable via the TunnelDirConns and PreferTunneledDirConns config options if you like. o Minor features (config and docs): - Start using the state file to store bandwidth accounting data: the bw_accounting file is now obsolete. We'll keep generating it for a while for people who are still using 0.1.2.4-alpha. - Try to batch changes to the state file so that we do as few disk writes as possible while still storing important things in a timely fashion. - The state file and the bw_accounting file get saved less often when the AvoidDiskWrites config option is set. - Make PIDFile work on Windows (untested). - Add internal descriptions for a bunch of configuration options: accessible via controller interface and in comments in saved options files. - Reject *:563 (NNTPS) in the default exit policy. We already reject NNTP by default, so this seems like a sensible addition. - Clients now reject hostnames with invalid characters. This should avoid some inadvertent info leaks. Add an option AllowNonRFC953Hostnames to disable this behavior, in case somebody is running a private network with hosts called @, !, and #. - Add a maintainer script to tell us which options are missing documentation: "make check-docs". - Add a new address-spec.txt document to describe our special-case addresses: .exit, .onion, and .noconnnect. o Minor features (DNS): - Ongoing work on eventdns infrastructure: now it has dns server and ipv6 support. One day Tor will make use of it. - Add client-side caching for reverse DNS lookups. - Add support to tor-resolve tool for reverse lookups and SOCKS5. - When we change nameservers or IP addresses, reset and re-launch our tests for DNS hijacking. o Minor features (directory): - Authorities now specify server versions in networkstatus. This adds about 2% to the size of compressed networkstatus docs, and allows clients to tell which servers support BEGIN_DIR and which don't. The implementation is forward-compatible with a proposed future protocol version scheme not tied to Tor versions. - DirServer configuration lines now have an orport= option so clients can open encrypted tunnels to the authorities without having downloaded their descriptors yet. Enabled for moria1, moria2, tor26, and lefkada now in the default configuration. - Directory servers are more willing to send a 503 "busy" if they are near their write limit, especially for v1 directory requests. Now they can use their limited bandwidth for actual Tor traffic. - Clients track responses with status 503 from dirservers. After a dirserver has given us a 503, we try not to use it until an hour has gone by, or until we have no dirservers that haven't given us a 503. - When we get a 503 from a directory, and we're not a server, we don't count the failure against the total number of failures allowed for the thing we're trying to download. - Report X-Your-Address-Is correctly from tunneled directory connections; don't report X-Your-Address-Is when it's an internal address; and never believe reported remote addresses when they're internal. - Protect against an unlikely DoS attack on directory servers. - Add a BadDirectory flag to network status docs so that authorities can (eventually) tell clients about caches they believe to be broken. o Minor features (controller): - Have GETINFO dir/status/* work on hosts with DirPort disabled. - Reimplement GETINFO so that info/names stays in sync with the actual keys. - Implement "GETINFO fingerprint". - Implement "SETEVENTS GUARD" so controllers can get updates on entry guard status as it changes. o Minor features (clean up obsolete pieces): - Remove some options that have been deprecated since at least 0.1.0.x: AccountingMaxKB, LogFile, DebugLogFile, LogLevel, and SysLog. Use AccountingMax instead of AccountingMaxKB, and use Log to set log options. - We no longer look for identity and onion keys in "identity.key" and "onion.key" -- these were replaced by secret_id_key and secret_onion_key in 0.0.8pre1. - We no longer require unrecognized directory entries to be preceded by "opt". o Major bugfixes (security): - Stop sending the HttpProxyAuthenticator string to directory servers when directory connections are tunnelled through Tor. - Clients no longer store bandwidth history in the state file. - Do not log introduction points for hidden services if SafeLogging is set. - When generating bandwidth history, round down to the nearest 1k. When storing accounting data, round up to the nearest 1k. - When we're running as a server, remember when we last rotated onion keys, so that we will rotate keys once they're a week old even if we never stay up for a week ourselves. o Major bugfixes (other): - Fix a longstanding bug in eventdns that prevented the count of timed-out resolves from ever being reset. This bug caused us to give up on a nameserver the third time it timed out, and try it 10 seconds later... and to give up on it every time it timed out after that. - Take out the '5 second' timeout from the connection retry schedule. Now the first connect attempt will wait a full 10 seconds before switching to a new circuit. Perhaps this will help a lot. Based on observations from Mike Perry. - Fix a bug on the Windows implementation of tor_mmap_file() that would prevent the cached-routers file from ever loading. Reported by John Kimble. o Minor bugfixes: - Fix an assert failure when a directory authority sets AuthDirRejectUnlisted and then receives a descriptor from an unlisted router. Reported by seeess. - Avoid a double-free when parsing malformed DirServer lines. - Fix a bug when a BSD-style PF socket is first used. Patch from Fabian Keil. - Fix a bug in 0.1.2.2-alpha that prevented clients from asking to resolve an address at a given exit node even when they ask for it by name. - Servers no longer ever list themselves in their "family" line, even if configured to do so. This makes it easier to configure family lists conveniently. - When running as a server, don't fall back to 127.0.0.1 when no nameservers are configured in /etc/resolv.conf; instead, make the user fix resolv.conf or specify nameservers explicitly. (Resolves bug 363.) - Stop accepting certain malformed ports in configured exit policies. - Don't re-write the fingerprint file every restart, unless it has changed. - Stop warning when a single nameserver fails: only warn when _all_ of our nameservers have failed. Also, when we only have one nameserver, raise the threshold for deciding that the nameserver is dead. - Directory authorities now only decide that routers are reachable if their identity keys are as expected. - When the user uses bad syntax in the Log config line, stop suggesting other bad syntax as a replacement. - Correctly detect ipv6 DNS capability on OpenBSD. o Minor bugfixes (controller): - Report the circuit number correctly in STREAM CLOSED events. Bug reported by Mike Perry. - Do not report bizarre values for results of accounting GETINFOs when the last second's write or read exceeds the allotted bandwidth. - Report "unrecognized key" rather than an empty string when the controller tries to fetch a networkstatus that doesn't exist. Changes in version 0.1.1.26 - 2006-12-14 o Security bugfixes: - Stop sending the HttpProxyAuthenticator string to directory servers when directory connections are tunnelled through Tor. - Clients no longer store bandwidth history in the state file. - Do not log introduction points for hidden services if SafeLogging is set. o Minor bugfixes: - Fix an assert failure when a directory authority sets AuthDirRejectUnlisted and then receives a descriptor from an unlisted router (reported by seeess). Changes in version 0.1.2.4-alpha - 2006-12-03 o Major features: - Add support for using natd; this allows FreeBSDs earlier than 5.1.2 to have ipfw send connections through Tor without using SOCKS. (Patch from Zajcev Evgeny with tweaks from tup.) o Minor features: - Make all connections to addresses of the form ".noconnect" immediately get closed. This lets application/controller combos successfully test whether they're talking to the same Tor by watching for STREAM events. - Make cross.sh cross-compilation script work even when autogen.sh hasn't been run. (Patch from Michael Mohr.) - Statistics dumped by -USR2 now include a breakdown of public key operations, for profiling. o Major bugfixes: - Fix a major leak when directory authorities parse their approved-routers list, a minor memory leak when we fail to pick an exit node, and a few rare leaks on errors. - Handle TransPort connections even when the server sends data before the client sends data. Previously, the connection would just hang until the client sent data. (Patch from tup based on patch from Zajcev Evgeny.) - Avoid assert failure when our cached-routers file is empty on startup. o Minor bugfixes: - Don't log spurious warnings when we see a circuit close reason we don't recognize; it's probably just from a newer version of Tor. - Have directory authorities allow larger amounts of drift in uptime without replacing the server descriptor: previously, a server that restarted every 30 minutes could have 48 "interesting" descriptors per day. - Start linking to the Tor specification and Tor reference manual correctly in the Windows installer. - Add Vidalia to the OS X uninstaller script, so when we uninstall Tor/Privoxy we also uninstall Vidalia. - Resume building on Irix64, and fix a lot of warnings from its MIPSpro C compiler. - Don't corrupt last_guessed_ip in router_new_address_suggestion() when we're running as a client. Changes in version 0.1.1.25 - 2006-11-04 o Major bugfixes: - When a client asks us to resolve (rather than connect to) an address, and we have a cached answer, give them the cached answer. Previously, we would give them no answer at all. - We were building exactly the wrong circuits when we predict hidden service requirements, meaning Tor would have to build all its circuits on demand. - If none of our live entry guards have a high uptime, but we require a guard with a high uptime, try adding a new guard before we give up on the requirement. This patch should make long-lived connections more stable on average. - When testing reachability of our DirPort, don't launch new tests when there's already one in progress -- unreachable servers were stacking up dozens of testing streams. o Security bugfixes: - When the user sends a NEWNYM signal, clear the client-side DNS cache too. Otherwise we continue to act on previous information. o Minor bugfixes: - Avoid a memory corruption bug when creating a hash table for the first time. - Avoid possibility of controller-triggered crash when misusing certain commands from a v0 controller on platforms that do not handle printf("%s",NULL) gracefully. - Avoid infinite loop on unexpected controller input. - Don't log spurious warnings when we see a circuit close reason we don't recognize; it's probably just from a newer version of Tor. - Add Vidalia to the OS X uninstaller script, so when we uninstall Tor/Privoxy we also uninstall Vidalia. Changes in version 0.1.2.3-alpha - 2006-10-29 o Minor features: - Prepare for servers to publish descriptors less often: never discard a descriptor simply for being too old until either it is recommended by no authorities, or until we get a better one for the same router. Make caches consider retaining old recommended routers for even longer. - If most authorities set a BadExit flag for a server, clients don't think of it as a general-purpose exit. Clients only consider authorities that advertise themselves as listing bad exits. - Directory servers now provide 'Pragma: no-cache' and 'Expires' headers for content, so that we can work better in the presence of caching HTTP proxies. - Allow authorities to list nodes as bad exits by fingerprint or by address. o Minor features, controller: - Add a REASON field to CIRC events; for backward compatibility, this field is sent only to controllers that have enabled the extended event format. Also, add additional reason codes to explain why a given circuit has been destroyed or truncated. (Patches from Mike Perry) - Add a REMOTE_REASON field to extended CIRC events to tell the controller about why a remote OR told us to close a circuit. - Stream events also now have REASON and REMOTE_REASON fields, working much like those for circuit events. - There's now a GETINFO ns/... field so that controllers can ask Tor about the current status of a router. - A new event type "NS" to inform a controller when our opinion of a router's status has changed. - Add a GETINFO events/names and GETINFO features/names so controllers can tell which events and features are supported. - A new CLEARDNSCACHE signal to allow controllers to clear the client-side DNS cache without expiring circuits. o Security bugfixes: - When the user sends a NEWNYM signal, clear the client-side DNS cache too. Otherwise we continue to act on previous information. o Minor bugfixes: - Avoid sending junk to controllers or segfaulting when a controller uses EVENT_NEW_DESC with verbose nicknames. - Stop triggering asserts if the controller tries to extend hidden service circuits (reported by mwenge). - Avoid infinite loop on unexpected controller input. - When the controller does a "GETINFO network-status", tell it about even those routers whose descriptors are very old, and use long nicknames where appropriate. - Change NT service functions to be loaded on demand. This lets us build with MinGW without breaking Tor for Windows 98 users. - Do DirPort reachability tests less often, since a single test chews through many circuits before giving up. - In the hidden service example in torrc.sample, stop recommending esoteric and discouraged hidden service options. - When stopping an NT service, wait up to 10 sec for it to actually stop. Patch from Matt Edman; resolves bug 295. - Fix handling of verbose nicknames with ORCONN controller events: make them show up exactly when requested, rather than exactly when not requested. - When reporting verbose nicknames in entry_guards_getinfo(), avoid printing a duplicate "$" in the keys we send (reported by mwenge). - Correctly set maximum connection limit on Cygwin. (This time for sure!) - Try to detect Windows correctly when cross-compiling. - Detect the size of the routers file correctly even if it is corrupted (on systems without mmap) or not page-aligned (on systems with mmap). This bug was harmless. - Sometimes we didn't bother sending a RELAY_END cell when an attempt to open a stream fails; now we do in more cases. This should make clients able to find a good exit faster in some cases, since unhandleable requests will now get an error rather than timing out. - Resolve two memory leaks when rebuilding the on-disk router cache (reported by fookoowa). - Clean up minor code warnings suggested by the MIPSpro C compiler, and reported by some Centos users. - Controller signals now work on non-Unix platforms that don't define SIGUSR1 and SIGUSR2 the way we expect. - Patch from Michael Mohr to contrib/cross.sh, so it checks more values before failing, and always enables eventdns. - Libevent-1.2 exports, but does not define in its headers, strlcpy. Try to fix this in configure.in by checking for most functions before we check for libevent. Changes in version 0.1.2.2-alpha - 2006-10-07 o Major features: - Make our async eventdns library on-by-default for Tor servers, and plan to deprecate the separate dnsworker threads. - Add server-side support for "reverse" DNS lookups (using PTR records so clients can determine the canonical hostname for a given IPv4 address). Only supported by servers using eventdns; servers now announce in their descriptors whether they support eventdns. - Specify and implement client-side SOCKS5 interface for reverse DNS lookups (see doc/socks-extensions.txt). - Add a BEGIN_DIR relay cell type for an easier in-protocol way to connect to directory servers through Tor. Previously, clients needed to find Tor exits to make private connections to directory servers. - Avoid choosing Exit nodes for entry or middle hops when the total bandwidth available from non-Exit nodes is much higher than the total bandwidth available from Exit nodes. - Workaround for name servers (like Earthlink's) that hijack failing DNS requests and replace the no-such-server answer with a "helpful" redirect to an advertising-driven search portal. Also work around DNS hijackers who "helpfully" decline to hijack known-invalid RFC2606 addresses. Config option "ServerDNSDetectHijacking 0" lets you turn it off. - Send out a burst of long-range padding cells once we've established that we're reachable. Spread them over 4 circuits, so hopefully a few will be fast. This exercises our bandwidth and bootstraps us into the directory more quickly. o New/improved config options: - Add new config option "ResolvConf" to let the server operator choose an alternate resolve.conf file when using eventdns. - Add an "EnforceDistinctSubnets" option to control our "exclude servers on the same /16" behavior. It's still on by default; this is mostly for people who want to operate private test networks with all the machines on the same subnet. - If one of our entry guards is on the ExcludeNodes list, or the directory authorities don't think it's a good guard, treat it as if it were unlisted: stop using it as a guard, and throw it off the guards list if it stays that way for a long time. - Allow directory authorities to be marked separately as authorities for the v1 directory protocol, the v2 directory protocol, and as hidden service directories, to make it easier to retire old authorities. V1 authorities should set "HSAuthoritativeDir 1" to continue being hidden service authorities too. - Remove 8888 as a LongLivedPort, and add 6697 (IRCS). o Minor features, controller: - Fix CIRC controller events so that controllers can learn the identity digests of non-Named servers used in circuit paths. - Let controllers ask for more useful identifiers for servers. Instead of learning identity digests for un-Named servers and nicknames for Named servers, the new identifiers include digest, nickname, and indication of Named status. Off by default; see control-spec.txt for more information. - Add a "getinfo address" controller command so it can display Tor's best guess to the user. - New controller event to alert the controller when our server descriptor has changed. - Give more meaningful errors on controller authentication failure. o Minor features, other: - When asked to resolve a hostname, don't use non-exit servers unless requested to do so. This allows servers with broken DNS to be useful to the network. - Divide eventdns log messages into warn and info messages. - Reserve the nickname "Unnamed" for routers that can't pick a hostname: any router can call itself Unnamed; directory authorities will never allocate Unnamed to any particular router; clients won't believe that any router is the canonical Unnamed. - Only include function names in log messages for info/debug messages. For notice/warn/err, the content of the message should be clear on its own, and printing the function name only confuses users. - Avoid some false positives during reachability testing: don't try to test via a server that's on the same /24 as us. - If we fail to build a circuit to an intended enclave, and it's not mandatory that we use that enclave, stop wanting it. - When eventdns is enabled, allow multithreaded builds on NetBSD and OpenBSD. (We had previously disabled threads on these platforms because they didn't have working thread-safe resolver functions.) o Major bugfixes, anonymity/security: - If a client asked for a server by name, and there's a named server in our network-status but we don't have its descriptor yet, we could return an unnamed server instead. - Fix NetBSD bug that could allow someone to force uninitialized RAM to be sent to a server's DNS resolver. This only affects NetBSD and other platforms that do not bounds-check tolower(). - Reject (most) attempts to use Tor circuits with length one. (If many people start using Tor as a one-hop proxy, exit nodes become a more attractive target for compromise.) - Just because your DirPort is open doesn't mean people should be able to remotely teach you about hidden service descriptors. Now only accept rendezvous posts if you've got HSAuthoritativeDir set. o Major bugfixes, other: - Don't crash on race condition in dns.c: tor_assert(!resolve->expire) - When a client asks the server to resolve (not connect to) an address, and it has a cached answer, give them the cached answer. Previously, the server would give them no answer at all. - Allow really slow clients to not hang up five minutes into their directory downloads (suggested by Adam J. Richter). - We were building exactly the wrong circuits when we anticipated hidden service requirements, meaning Tor would have to build all its circuits on demand. - Avoid crashing when we mmap a router cache file of size 0. - When testing reachability of our DirPort, don't launch new tests when there's already one in progress -- unreachable servers were stacking up dozens of testing streams. o Minor bugfixes, correctness: - If we're a directory mirror and we ask for "all" network status documents, we would discard status documents from authorities we don't recognize. - Avoid a memory corruption bug when creating a hash table for the first time. - Avoid controller-triggered crash when misusing certain commands from a v0 controller on platforms that do not handle printf("%s",NULL) gracefully. - Don't crash when a controller sends a third argument to an "extendcircuit" request. - Controller protocol fixes: fix encoding in "getinfo addr-mappings" response; fix error code when "getinfo dir/status/" fails. - Avoid crash when telling controller stream-status and a stream is detached. - Patch from Adam Langley to fix assert() in eventdns.c. - Fix a debug log message in eventdns to say "X resolved to Y" instead of "X resolved to X". - Make eventdns give strings for DNS errors, not just error numbers. - Track unreachable entry guards correctly: don't conflate 'unreachable by us right now' with 'listed as down by the directory authorities'. With the old code, if a guard was unreachable by us but listed as running, it would clog our guard list forever. - Behave correctly in case we ever have a network with more than 2GB/s total advertised capacity. - Make TrackExitHosts case-insensitive, and fix the behavior of ".suffix" TrackExitHosts items to avoid matching in the middle of an address. - Finally fix the openssl warnings from newer gccs that believe that ignoring a return value is okay, but casting a return value and then ignoring it is a sign of madness. - Prevent the contrib/exitlist script from printing the same result more than once. - Patch from Steve Hildrey: Generate network status correctly on non-versioning dirservers. - Don't listen to the X-Your-Address-Is hint if you did the lookup via Tor; otherwise you'll think you're the exit node's IP address. o Minor bugfixes, performance: - Two small performance improvements on parsing descriptors. - Major performance improvement on inserting descriptors: change algorithm from O(n^2) to O(n). - Make the common memory allocation path faster on machines where malloc(0) returns a pointer. - Start remembering X-Your-Address-Is directory hints even if you're a client, so you can become a server more smoothly. - Avoid duplicate entries on MyFamily line in server descriptor. o Packaging, features: - Remove architecture from OS X builds. The official builds are now universal binaries. - The Debian package now uses --verify-config when (re)starting, to distinguish configuration errors from other errors. - Update RPMs to require libevent 1.1b. o Packaging, bugfixes: - Patches so Tor builds with MinGW on Windows. - Patches so Tor might run on Cygwin again. - Resume building on non-gcc compilers and ancient gcc. Resume building with the -O0 compile flag. Resume building cleanly on Debian woody. - Run correctly on OS X platforms with case-sensitive filesystems. - Correct includes for net/if.h and net/pfvar.h on OpenBSD (from Tup). - Add autoconf checks so Tor can build on Solaris x86 again. o Documentation - Documented (and renamed) ServerDNSSearchDomains and ServerDNSResolvConfFile options. - Be clearer that the *ListenAddress directives can be repeated multiple times. Changes in version 0.1.1.24 - 2006-09-29 o Major bugfixes: - Allow really slow clients to not hang up five minutes into their directory downloads (suggested by Adam J. Richter). - Fix major performance regression from 0.1.0.x: instead of checking whether we have enough directory information every time we want to do something, only check when the directory information has changed. This should improve client CPU usage by 25-50%. - Don't crash if, after a server has been running for a while, it can't resolve its hostname. o Minor bugfixes: - Allow Tor to start when RunAsDaemon is set but no logs are set. - Don't crash when the controller receives a third argument to an "extendcircuit" request. - Controller protocol fixes: fix encoding in "getinfo addr-mappings" response; fix error code when "getinfo dir/status/" fails. - Fix configure.in to not produce broken configure files with more recent versions of autoconf. Thanks to Clint for his auto* voodoo. - Fix security bug on NetBSD that could allow someone to force uninitialized RAM to be sent to a server's DNS resolver. This only affects NetBSD and other platforms that do not bounds-check tolower(). - Warn user when using libevent 1.1a or earlier with win32 or kqueue methods: these are known to be buggy. - If we're a directory mirror and we ask for "all" network status documents, we would discard status documents from authorities we don't recognize. Changes in version 0.1.2.1-alpha - 2006-08-27 o Major features: - Add "eventdns" async dns library from Adam Langley, tweaked to build on OSX and Windows. Only enabled if you pass the --enable-eventdns argument to configure. - Allow servers with no hostname or IP address to learn their IP address by asking the directory authorities. This code only kicks in when you would normally have exited with a "no address" error. Nothing's authenticated, so use with care. - Rather than waiting a fixed amount of time between retrying application connections, we wait only 5 seconds for the first, 10 seconds for the second, and 15 seconds for each retry after that. Hopefully this will improve the expected user experience. - Patch from Tup to add support for transparent AP connections: this basically bundles the functionality of trans-proxy-tor into the Tor mainline. Now hosts with compliant pf/netfilter implementations can redirect TCP connections straight to Tor without diverting through SOCKS. Needs docs. - Busy directory servers save lots of memory by spooling server descriptors, v1 directories, and v2 networkstatus docs to buffers as needed rather than en masse. Also mmap the cached-routers files, so we don't need to keep the whole thing in memory too. - Automatically avoid picking more than one node from the same /16 network when constructing a circuit. - Revise and clean up the torrc.sample that we ship with; add a section for BandwidthRate and BandwidthBurst. o Minor features: - Split circuit_t into origin_circuit_t and or_circuit_t, and split connection_t into edge, or, dir, control, and base structs. These will save quite a bit of memory on busy servers, and they'll also help us track down bugs in the code and bugs in the spec. - Experimentally re-enable kqueue on OSX when using libevent 1.1b or later. Log when we are doing this, so we can diagnose it when it fails. (Also, recommend libevent 1.1b for kqueue and win32 methods; deprecate libevent 1.0b harder; make libevent recommendation system saner.) - Start being able to build universal binaries on OS X (thanks to Phobos). - Export the default exit policy via the control port, so controllers don't need to guess what it is / will be later. - Add a man page entry for ProtocolWarnings. - Add TestVia config option to the man page. - Remove even more protocol-related warnings from Tor server logs, such as bad TLS handshakes and malformed begin cells. - Stop fetching descriptors if you're not a dir mirror and you haven't tried to establish any circuits lately. [This currently causes some dangerous behavior, because when you start up again you'll use your ancient server descriptors.] - New DirPort behavior: if you have your dirport set, you download descriptors aggressively like a directory mirror, whether or not your ORPort is set. - Get rid of the router_retry_connections notion. Now routers no longer try to rebuild long-term connections to directory authorities, and directory authorities no longer try to rebuild long-term connections to all servers. We still don't hang up connections in these two cases though -- we need to look at it more carefully to avoid flapping, and we likely need to wait til 0.1.1.x is obsolete. - Drop compatibility with obsolete Tors that permit create cells to have the wrong circ_id_type. - Re-enable per-connection rate limiting. Get rid of the "OP bandwidth" concept. Lay groundwork for "bandwidth classes" -- separate global buckets that apply depending on what sort of conn it is. - Start publishing one minute or so after we find our ORPort to be reachable. This will help reduce the number of descriptors we have for ourselves floating around, since it's quite likely other things (e.g. DirPort) will change during that minute too. - Fork the v1 directory protocol into its own spec document, and mark dir-spec.txt as the currently correct (v2) spec. o Major bugfixes: - When we find our DirPort to be reachable, publish a new descriptor so we'll tell the world (reported by pnx). - Publish a new descriptor after we hup/reload. This is important if our config has changed such that we'll want to start advertising our DirPort now, etc. - Allow Tor to start when RunAsDaemon is set but no logs are set. - When we have a state file we cannot parse, tell the user and move it aside. Now we avoid situations where the user starts Tor in 1904, Tor writes a state file with that timestamp in it, the user fixes her clock, and Tor refuses to start. - Fix configure.in to not produce broken configure files with more recent versions of autoconf. Thanks to Clint for his auto* voodoo. - "tor --verify-config" now exits with -1(255) or 0 depending on whether the config options are bad or good. - Resolve bug 321 when using dnsworkers: append a period to every address we resolve at the exit node, so that we do not accidentally pick up local addresses, and so that failing searches are retried in the resolver search domains. (This is already solved for eventdns.) (This breaks Blossom servers for now.) - If we are using an exit enclave and we can't connect, e.g. because its webserver is misconfigured to not listen on localhost, then back off and try connecting from somewhere else before we fail. o Minor bugfixes: - Start compiling on MinGW on Windows (patches from Mike Chiussi). - Start compiling on MSVC6 on Windows (patches from Frediano Ziglio). - Fix bug 314: Tor clients issued "unsafe socks" warnings even when the IP address is mapped through MapAddress to a hostname. - Start passing "ipv4" hints to getaddrinfo(), so servers don't do useless IPv6 DNS resolves. - Patch suggested by Karsten Loesing: respond to SIGNAL command before we execute the signal, in case the signal shuts us down. - Clean up AllowInvalidNodes man page entry. - Claim a commonname of Tor, rather than TOR, in TLS handshakes. - Add more asserts to track down an assert error on a windows Tor server with connection_add being called with socket == -1. - Handle reporting OR_CONN_EVENT_NEW events to the controller. - Fix misleading log messages: an entry guard that is "unlisted", as well as not known to be "down" (because we've never heard of it), is not therefore "up". - Remove code to special-case "-cvs" ending, since it has not actually mattered since 0.0.9. - Make our socks5 handling more robust to broken socks clients: throw out everything waiting on the buffer in between socks handshake phases, since they can't possibly (so the theory goes) have predicted what we plan to respond to them. Changes in version 0.1.1.23 - 2006-07-30 o Major bugfixes: - Fast Tor servers, especially exit nodes, were triggering asserts due to a bug in handling the list of pending DNS resolves. Some bugs still remain here; we're hunting them. - Entry guards could crash clients by sending unexpected input. - More fixes on reachability testing: if you find yourself reachable, then don't ever make any client requests (so you stop predicting circuits), then hup or have your clock jump, then later your IP changes, you won't think circuits are working, so you won't try to test reachability, so you won't publish. o Minor bugfixes: - Avoid a crash if the controller does a resetconf firewallports and then a setconf fascistfirewall=1. - Avoid an integer underflow when the dir authority decides whether a router is stable: we might wrongly label it stable, and compute a slightly wrong median stability, when a descriptor is published later than now. - Fix a place where we might trigger an assert if we can't build our own server descriptor yet. Changes in version 0.1.1.22 - 2006-07-05 o Major bugfixes: - Fix a big bug that was causing servers to not find themselves reachable if they changed IP addresses. Since only 0.1.1.22+ servers can do reachability testing correctly, now we automatically make sure to test via one of these. - Fix to allow clients and mirrors to learn directory info from descriptor downloads that get cut off partway through. - Directory authorities had a bug in deciding if a newly published descriptor was novel enough to make everybody want a copy -- a few servers seem to be publishing new descriptors many times a minute. o Minor bugfixes: - Fix a rare bug that was causing some servers to complain about "closing wedged cpuworkers" and skip some circuit create requests. - Make the Exit flag in directory status documents actually work. Changes in version 0.1.1.21 - 2006-06-10 o Crash and assert fixes from 0.1.1.20: - Fix a rare crash on Tor servers that have enabled hibernation. - Fix a seg fault on startup for Tor networks that use only one directory authority. - Fix an assert from a race condition that occurs on Tor servers while exiting, where various threads are trying to log that they're exiting, and delete the logs, at the same time. - Make our unit tests pass again on certain obscure platforms. o Other fixes: - Add support for building SUSE RPM packages. - Speed up initial bootstrapping for clients: if we are making our first ever connection to any entry guard, then don't mark it down right after that. - When only one Tor server in the network is labelled as a guard, and we've already picked him, we would cycle endlessly picking him again, being unhappy about it, etc. Now we specifically exclude current guards when picking a new guard. - Servers send create cells more reliably after the TLS connection is established: we were sometimes forgetting to send half of them when we had more than one pending. - If we get a create cell that asks us to extend somewhere, but the Tor server there doesn't match the expected digest, we now send a destroy cell back, rather than silently doing nothing. - Make options->RedirectExit work again. - Make cookie authentication for the controller work again. - Stop being picky about unusual characters in the arguments to mapaddress. It's none of our business. - Add a new config option "TestVia" that lets you specify preferred middle hops to use for test circuits. Perhaps this will let me debug the reachability problems better. o Log / documentation fixes: - If we're a server and some peer has a broken TLS certificate, don't log about it unless ProtocolWarnings is set, i.e., we want to hear about protocol violations by others. - Fix spelling of VirtualAddrNetwork in man page. - Add a better explanation at the top of the autogenerated torrc file about what happened to our old torrc. Changes in version 0.1.1.20 - 2006-05-23 o Bugfixes: - Downgrade a log severity where servers complain that they're invalid. - Avoid a compile warning on FreeBSD. - Remove string size limit on NEWDESC messages; solve bug 291. - Correct the RunAsDaemon entry in the man page; ignore RunAsDaemon more thoroughly when we're running on windows. Changes in version 0.1.1.19-rc - 2006-05-03 o Minor bugs: - Regenerate our local descriptor if it's dirty and we try to use it locally (e.g. if it changes during reachability detection). - If we setconf our ORPort to 0, we continued to listen on the old ORPort and receive connections. - Avoid a second warning about machine/limits.h on Debian GNU/kFreeBSD. - Be willing to add our own routerinfo into the routerlist. Now authorities will include themselves in their directories and network-statuses. - Stop trying to upload rendezvous descriptors to every directory authority: only try the v1 authorities. - Servers no longer complain when they think they're not registered with the directory authorities. There were too many false positives. - Backport dist-rpm changes so rpms can be built without errors. o Features: - Implement an option, VirtualAddrMask, to set which addresses get handed out in response to mapaddress requests. This works around a bug in tsocks where 127.0.0.0/8 is never socksified. Changes in version 0.1.1.18-rc - 2006-04-10 o Major fixes: - Work harder to download live network-statuses from all the directory authorities we know about. Improve the threshold decision logic so we're more robust to edge cases. - When fetching rendezvous descriptors, we were willing to ask v2 authorities too, which would always return 404. o Minor fixes: - Stop listing down or invalid nodes in the v1 directory. This will reduce its bulk by about 1/3, and reduce load on directory mirrors. - When deciding whether a router is Fast or Guard-worthy, consider his advertised BandwidthRate and not just the BandwidthCapacity. - No longer ship INSTALL and README files -- they are useless now. - Force rpmbuild to behave and honor target_cpu. - Avoid warnings about machine/limits.h on Debian GNU/kFreeBSD. - Start to include translated versions of the tor-doc-*.html files, along with the screenshots. Still needs more work. - Start sending back 512 and 451 errors if mapaddress fails, rather than not sending anything back at all. - When we fail to bind or listen on an incoming or outgoing socket, we should close it before failing. otherwise we just leak it. (thanks to weasel for finding.) - Allow "getinfo dir/status/foo" to work, as long as your DirPort is enabled. (This is a hack, and will be fixed in 0.1.2.x.) - Make NoPublish (even though deprecated) work again. - Fix a minor security flaw where a versioning auth dirserver could list a recommended version many times in a row to make clients more convinced that it's recommended. - Fix crash bug if there are two unregistered servers running with the same nickname, one of them is down, and you ask for them by nickname in your EntryNodes or ExitNodes. Also, try to pick the one that's running rather than an arbitrary one. - Fix an infinite loop we could hit if we go offline for too long. - Complain when we hit WSAENOBUFS on recv() or write() too. Perhaps this will help us hunt the bug. - If you're not a versioning dirserver, don't put the string "client-versions \nserver-versions \n" in your network-status. - Lower the minimum required number of file descriptors to 1000, so we can have some overhead for Valgrind on Linux, where the default ulimit -n is 1024. o New features: - Add tor.dizum.com as the fifth authoritative directory server. - Add a new config option FetchUselessDescriptors, off by default, for when you plan to run "exitlist" on your client and you want to know about even the non-running descriptors. Changes in version 0.1.1.17-rc - 2006-03-28 o Major fixes: - Clients and servers since 0.1.1.10-alpha have been expiring connections whenever they are idle for 5 minutes and they *do* have circuits on them. Oops. With this new version, clients will discard their previous entry guard choices and avoid choosing entry guards running these flawed versions. - Fix memory leak when uncompressing concatenated zlib streams. This was causing substantial leaks over time on Tor servers. - The v1 directory was including servers as much as 48 hours old, because that's how the new routerlist->routers works. Now only include them if they're 20 hours old or less. o Minor fixes: - Resume building on irix64, netbsd 2.0, etc. - On non-gcc compilers (e.g. solaris), use "-g -O" instead of "-Wall -g -O2". - Stop writing the "router.desc" file, ever. Nothing uses it anymore, and it is confusing some users. - Mirrors stop caching the v1 directory so often. - Make the max number of old descriptors that a cache will hold rise with the number of directory authorities, so we can scale. - Change our win32 uname() hack to be more forgiving about what win32 versions it thinks it's found. o New features: - Add lefkada.eecs.harvard.edu as a fourth authoritative directory server. - When the controller's *setconf commands fail, collect an error message in a string and hand it back to the controller. - Make the v2 dir's "Fast" flag based on relative capacity, just like "Stable" is based on median uptime. Name everything in the top 7/8 Fast, and only the top 1/2 gets to be a Guard. - Log server fingerprint on startup, so new server operators don't have to go hunting around their filesystem for it. - Return a robots.txt on our dirport to discourage google indexing. - Let the controller ask for GETINFO dir/status/foo so it can ask directly rather than connecting to the dir port. Only works when dirport is set for now. o New config options rather than constants in the code: - SocksTimeout: How long do we let a socks connection wait unattached before we fail it? - CircuitBuildTimeout: Cull non-open circuits that were born at least this many seconds ago. - CircuitIdleTimeout: Cull open clean circuits that were born at least this many seconds ago. Changes in version 0.1.1.16-rc - 2006-03-18 o Bugfixes on 0.1.1.15-rc: - Fix assert when the controller asks to attachstream a connect-wait or resolve-wait stream. - Now do address rewriting when the controller asks us to attach to a particular circuit too. This will let Blossom specify "moria2.exit" without having to learn what moria2's IP address is. - Make the "tor --verify-config" command-line work again, so people can automatically check if their torrc will parse. - Authoritative dirservers no longer require an open connection from a server to consider him "reachable". We need this change because when we add new auth dirservers, old servers won't know not to hang up on them. - Let Tor build on Sun CC again. - Fix an off-by-one buffer size in dirserv.c that magically never hit our three authorities but broke sjmurdoch's own tor network. - If we as a directory mirror don't know of any v1 directory authorities, then don't try to cache any v1 directories. - Stop warning about unknown servers in our family when they are given as hex digests. - Stop complaining as quickly to the server operator that he hasn't registered his nickname/key binding. - Various cleanups so we can add new V2 Auth Dirservers. - Change "AllowUnverifiedNodes" to "AllowInvalidNodes", to reflect the updated flags in our v2 dir protocol. - Resume allowing non-printable characters for exit streams (both for connecting and for resolving). Now we tolerate applications that don't follow the RFCs. But continue to block malformed names at the socks side. o Bugfixes on 0.1.0.x: - Fix assert bug in close_logs(): when we close and delete logs, remove them all from the global "logfiles" list. - Fix minor integer overflow in calculating when we expect to use up our bandwidth allocation before hibernating. - Fix a couple of bugs in OpenSSL detection. Also, deal better when there are multiple SSLs installed with different versions. - When we try to be a server and Address is not explicitly set and our hostname resolves to a private IP address, try to use an interface address if it has a public address. Now Windows machines that think of themselves as localhost can work by default. o New features: - Let the controller ask for GETINFO dir/server/foo so it can ask directly rather than connecting to the dir port. - Let the controller tell us about certain router descriptors that it doesn't want Tor to use in circuits. Implement SETROUTERPURPOSE and modify +POSTDESCRIPTOR to do this. - New config option SafeSocks to reject all application connections using unsafe socks protocols. Defaults to off. Changes in version 0.1.1.15-rc - 2006-03-11 o Bugfixes and cleanups: - When we're printing strings from the network, don't try to print non-printable characters. This protects us against shell escape sequence exploits, and also against attacks to fool humans into misreading their logs. - Fix a bug where Tor would fail to establish any connections if you left it off for 24 hours and then started it: we were happy with the obsolete network statuses, but they all referred to router descriptors that were too old to fetch, so we ended up with no valid router descriptors. - Fix a seg fault in the controller's "getinfo orconn-status" command while listing status on incoming handshaking connections. Introduce a status name "NEW" for these connections. - If we get a linelist or linelist_s config option from the torrc (e.g. ExitPolicy) and it has no value, warn and skip rather than silently resetting it to its default. - Don't abandon entry guards until they've been down or gone for a whole month. - Cleaner and quieter log messages. o New features: - New controller signal NEWNYM that makes new application requests use clean circuits. - Add a new circuit purpose 'controller' to let the controller ask for a circuit that Tor won't try to use. Extend the EXTENDCIRCUIT controller command to let you specify the purpose if you're starting a new circuit. Add a new SETCIRCUITPURPOSE controller command to let you change a circuit's purpose after it's been created. - Accept "private:*" in routerdesc exit policies; not generated yet because older Tors do not understand it. - Add BSD-style contributed startup script "rc.subr" from Peter Thoenen. Changes in version 0.1.1.14-alpha - 2006-02-20 o Bugfixes on 0.1.1.x: - Don't die if we ask for a stdout or stderr log (even implicitly) and we're set to RunAsDaemon -- just warn. - We still had a few bugs in the OR connection rotation code that caused directory servers to slowly aggregate connections to other fast Tor servers. This time for sure! - Make log entries on Win32 include the name of the function again. - We were treating a pair of exit policies if they were equal even if one said accept and the other said reject -- causing us to not always publish a new descriptor since we thought nothing had changed. - Retry pending server downloads as well as pending networkstatus downloads when we unexpectedly get a socks request. - We were ignoring the IS_FAST flag in the directory status, meaning we were willing to pick trivial-bandwidth nodes for "fast" connections. - If the controller's SAVECONF command fails (e.g. due to file permissions), let the controller know that it failed. o Features: - If we're trying to be a Tor server and running Windows 95/98/ME as a server, explain that we'll likely crash. - When we're a server, a client asks for an old-style directory, and our write bucket is empty, don't give it to him. This way small servers can continue to serve the directory *sometimes*, without getting overloaded. - Compress exit policies even more -- look for duplicate lines and remove them. - Clients now honor the "guard" flag in the router status when picking entry guards, rather than looking at is_fast or is_stable. - Retain unrecognized lines in $DATADIR/state file, so that we can be forward-compatible. - Generate 18.0.0.0/8 address policy format in descs when we can; warn when the mask is not reducible to a bit-prefix. - Let the user set ControlListenAddress in the torrc. This can be dangerous, but there are some cases (like a secured LAN) where it makes sense. - Split ReachableAddresses into ReachableDirAddresses and ReachableORAddresses, so we can restrict Dir conns to port 80 and OR conns to port 443. - Now we can target arch and OS in rpm builds (contributed by Phobos). Also make the resulting dist-rpm filename match the target arch. - New config options to help controllers: FetchServerDescriptors and FetchHidServDescriptors for whether to fetch server info and hidserv info or let the controller do it, and PublishServerDescriptor and PublishHidServDescriptors. - Also let the controller set the __AllDirActionsPrivate config option if you want all directory fetches/publishes to happen via Tor (it assumes your controller bootstraps your circuits). Changes in version 0.1.0.17 - 2006-02-17 o Crash bugfixes on 0.1.0.x: - When servers with a non-zero DirPort came out of hibernation, sometimes they would trigger an assert. o Other important bugfixes: - On platforms that don't have getrlimit (like Windows), we were artificially constraining ourselves to a max of 1024 connections. Now just assume that we can handle as many as 15000 connections. Hopefully this won't cause other problems. o Backported features: - When we're a server, a client asks for an old-style directory, and our write bucket is empty, don't give it to him. This way small servers can continue to serve the directory *sometimes*, without getting overloaded. - Whenever you get a 503 in response to a directory fetch, try once more. This will become important once servers start sending 503's whenever they feel busy. - Fetch a new directory every 120 minutes, not every 40 minutes. Now that we have hundreds of thousands of users running the old directory algorithm, it's starting to hurt a lot. - Bump up the period for forcing a hidden service descriptor upload from 20 minutes to 1 hour. Changes in version 0.1.1.13-alpha - 2006-02-09 o Crashes in 0.1.1.x: - When you tried to setconf ORPort via the controller, Tor would crash. So people using TorCP to become a server were sad. - Solve (I hope) the stack-smashing bug that we were seeing on fast servers. The problem appears to be something do with OpenSSL's random number generation, or how we call it, or something. Let me know if the crashes continue. - Turn crypto hardware acceleration off by default, until we find somebody smart who can test it for us. (It appears to produce seg faults in at least some cases.) - Fix a rare assert error when we've tried all intro points for a hidden service and we try fetching the service descriptor again: "Assertion conn->state != AP_CONN_STATE_RENDDESC_WAIT failed" o Major fixes: - Fix a major load balance bug: we were round-robining in 16 KB chunks, and servers with bandwidthrate of 20 KB, while downloading a 600 KB directory, would starve their other connections. Now we try to be a bit more fair. - Dir authorities and mirrors were never expiring the newest descriptor for each server, causing memory and directory bloat. - Fix memory-bloating and connection-bloating bug on servers: We were never closing any connection that had ever had a circuit on it, because we were checking conn->n_circuits == 0, yet we had a bug that let it go negative. - Make Tor work using squid as your http proxy again -- squid returns an error if you ask for a URL that's too long, and it uses a really generic error message. Plus, many people are behind a transparent squid so they don't even realize it. - On platforms that don't have getrlimit (like Windows), we were artificially constraining ourselves to a max of 1024 connections. Now just assume that we can handle as many as 15000 connections. Hopefully this won't cause other problems. - Add a new config option ExitPolicyRejectPrivate which defaults to 1. This means all exit policies will begin with rejecting private addresses, unless the server operator explicitly turns it off. o Major features: - Clients no longer download descriptors for non-running descriptors. - Before we add new directory authorities, we should make it clear that only v1 authorities should receive/publish hidden service descriptors. o Minor features: - As soon as we've fetched some more directory info, immediately try to download more server descriptors. This way we don't have a 10 second pause during initial bootstrapping. - Remove even more loud log messages that the server operator can't do anything about. - When we're running an obsolete or un-recommended version, make the log message more clear about what the problem is and what versions *are* still recommended. - Provide a more useful warn message when our onion queue gets full: the CPU is too slow or the exit policy is too liberal. - Don't warn when we receive a 503 from a dirserver/cache -- this will pave the way for them being able to refuse if they're busy. - When we fail to bind a listener, try to provide a more useful log message: e.g., "Is Tor already running?" - Adjust tor-spec to parameterize cell and key lengths. Now Ian Goldberg can prove things about our handshake protocol more easily. - MaxConn has been obsolete for a while now. Document the ConnLimit config option, which is a *minimum* number of file descriptors that must be available else Tor refuses to start. - Apply Matt Ghali's --with-syslog-facility patch to ./configure if you log to syslog and want something other than LOG_DAEMON. - Make dirservers generate a separate "guard" flag to mean, "would make a good entry guard". Make clients parse it and vote on it. Not used by clients yet. - Implement --with-libevent-dir option to ./configure. Also, improve search techniques to find libevent, and use those for openssl too. - Bump the default bandwidthrate to 3 MB, and burst to 6 MB - Only start testing reachability once we've established a circuit. This will make startup on dirservers less noisy. - Don't try to upload hidden service descriptors until we have established a circuit. - Fix the controller's "attachstream 0" command to treat conn like it just connected, doing address remapping, handling .exit and .onion idioms, and so on. Now we're more uniform in making sure that the controller hears about new and closing connections. Changes in version 0.1.1.12-alpha - 2006-01-11 o Bugfixes on 0.1.1.x: - The fix to close duplicate server connections was closing all Tor client connections if they didn't establish a circuit quickly enough. Oops. - Fix minor memory issue (double-free) that happened on exit. o Bugfixes on 0.1.0.x: - Tor didn't warn when it failed to open a log file. Changes in version 0.1.1.11-alpha - 2006-01-10 o Crashes in 0.1.1.x: - Include all the assert/crash fixes from 0.1.0.16. - If you start Tor and then quit very quickly, there were some races that tried to free things that weren't allocated yet. - Fix a rare memory stomp if you're running hidden services. - Fix segfault when specifying DirServer in config without nickname. - Fix a seg fault when you finish connecting to a server but at that moment you dump his server descriptor. - Extendcircuit and Attachstream controller commands would assert/crash if you don't give them enough arguments. - Fix an assert error when we're out of space in the connection_list and we try to post a hidden service descriptor (reported by weasel). - If you specify a relative torrc path and you set RunAsDaemon in your torrc, then it chdir()'s to the new directory. If you HUP, it tries to load the new torrc location, fails, and exits. The fix: no longer allow a relative path to torrc using -f. o Major features: - Implement "entry guards": automatically choose a handful of entry nodes and stick with them for all circuits. Only pick new guards when the ones you have are unsuitable, and if the old guards become suitable again, switch back. This will increase security dramatically against certain end-point attacks. The EntryNodes config option now provides some hints about which entry guards you want to use most; and StrictEntryNodes means to only use those. - New directory logic: download by descriptor digest, not by fingerprint. Caches try to download all listed digests from authorities; clients try to download "best" digests from caches. This avoids partitioning and isolating attacks better. - Make the "stable" router flag in network-status be the median of the uptimes of running valid servers, and make clients pay attention to the network-status flags. Thus the cutoff adapts to the stability of the network as a whole, making IRC, IM, etc connections more reliable. o Major fixes: - Tor servers with dynamic IP addresses were needing to wait 18 hours before they could start doing reachability testing using the new IP address and ports. This is because they were using the internal descriptor to learn what to test, yet they were only rebuilding the descriptor once they decided they were reachable. - Tor 0.1.1.9 and 0.1.1.10 had a serious bug that caused clients to download certain server descriptors, throw them away, and then fetch them again after 30 minutes. Now mirrors throw away these server descriptors so clients can't get them. - We were leaving duplicate connections to other ORs open for a week, rather than closing them once we detect a duplicate. This only really affected authdirservers, but it affected them a lot. - Spread the authdirservers' reachability testing over the entire testing interval, so we don't try to do 500 TLS's at once every 20 minutes. o Minor fixes: - If the network is down, and we try to connect to a conn because we have a circuit in mind, and we timeout (30 seconds) because the network never answers, we were expiring the circuit, but we weren't obsoleting the connection or telling the entry_guards functions. - Some Tor servers process billions of cells per day. These statistics need to be uint64_t's. - Check for integer overflows in more places, when adding elements to smartlists. This could possibly prevent a buffer overflow on malicious huge inputs. I don't see any, but I haven't looked carefully. - ReachableAddresses kept growing new "reject *:*" lines on every setconf/reload. - When you "setconf log" via the controller, it should remove all logs. We were automatically adding back in a "log notice stdout". - Newly bootstrapped Tor networks couldn't establish hidden service circuits until they had nodes with high uptime. Be more tolerant. - We were marking servers down when they could not answer every piece of the directory request we sent them. This was far too harsh. - Fix the torify (tsocks) config file to not use Tor for localhost connections. - Directory authorities now go to the proper authority when asking for a networkstatus, even when they want a compressed one. - Fix a harmless bug that was causing Tor servers to log "Got an end because of misc error, but we're not an AP. Closing." - Authorities were treating their own descriptor changes as cosmetic, meaning the descriptor available in the network-status and the descriptor that clients downloaded were different. - The OS X installer was adding a symlink for tor_resolve but the binary was called tor-resolve (reported by Thomas Hardly). - Workaround a problem with some http proxies where they refuse GET requests that specify "Content-Length: 0" (reported by Adrian). - Fix wrong log message when you add a "HiddenServiceNodes" config line without any HiddenServiceDir line (reported by Chris Thomas). o Minor features: - Write the TorVersion into the state file so we have a prayer of keeping forward and backward compatibility. - Revive the FascistFirewall config option rather than eliminating it: now it's a synonym for ReachableAddresses *:80,*:443. - Clients choose directory servers from the network status lists, not from their internal list of router descriptors. Now they can go to caches directly rather than needing to go to authorities to bootstrap. - Directory authorities ignore router descriptors that have only cosmetic differences: do this for 0.1.0.x servers now too. - Add a new flag to network-status indicating whether the server can answer v2 directory requests too. - Authdirs now stop whining so loudly about bad descriptors that they fetch from other dirservers. So when there's a log complaint, it's for sure from a freshly uploaded descriptor. - Reduce memory requirements in our structs by changing the order of fields. - There used to be two ways to specify your listening ports in a server descriptor: on the "router" line and with a separate "ports" line. Remove support for the "ports" line. - New config option "AuthDirRejectUnlisted" for auth dirservers as a panic button: if we get flooded with unusable servers we can revert to only listing servers in the approved-routers file. - Auth dir servers can now mark a fingerprint as "!reject" or "!invalid" in the approved-routers file (as its nickname), to refuse descriptors outright or include them but marked as invalid. - Servers store bandwidth history across restarts/crashes. - Add reasons to DESTROY and RELAY_TRUNCATED cells, so clients can get a better idea of why their circuits failed. Not used yet. - Directory mirrors now cache up to 16 unrecognized network-status docs. Now we can add new authdirservers and they'll be cached too. - When picking a random directory, prefer non-authorities if any are known. - New controller option "getinfo desc/all-recent" to fetch the latest server descriptor for every router that Tor knows about. Changes in version 0.1.0.16 - 2006-01-02 o Crash bugfixes on 0.1.0.x: - On Windows, build with a libevent patch from "I-M Weasel" to avoid corrupting the heap, losing FDs, or crashing when we need to resize the fd_sets. (This affects the Win32 binaries, not Tor's sources.) - It turns out sparc64 platforms crash on unaligned memory access too -- so detect and avoid this. - Handle truncated compressed data correctly (by detecting it and giving an error). - Fix possible-but-unlikely free(NULL) in control.c. - When we were closing connections, there was a rare case that stomped on memory, triggering seg faults and asserts. - Avoid potential infinite recursion when building a descriptor. (We don't know that it ever happened, but better to fix it anyway.) - We were neglecting to unlink marked circuits from soon-to-close OR connections, which caused some rare scribbling on freed memory. - Fix a memory stomping race bug when closing the joining point of two rendezvous circuits. - Fix an assert in time parsing found by Steven Murdoch. o Other bugfixes on 0.1.0.x: - When we're doing reachability testing, provide more useful log messages so the operator knows what to expect. - Do not check whether DirPort is reachable when we are suppressing advertising it because of hibernation. - When building with -static or on Solaris, we sometimes needed -ldl. - When we're deciding whether a stream has enough circuits around that can handle it, count the freshly dirty ones and not the ones that are so dirty they won't be able to handle it. - When we're expiring old circuits, we had a logic error that caused us to close new rendezvous circuits rather than old ones. - Give a more helpful log message when you try to change ORPort via the controller: you should upgrade Tor if you want that to work. - We were failing to parse Tor versions that start with "Tor ". - Tolerate faulty streams better: when a stream fails for reason exitpolicy, stop assuming that the router is lying about his exit policy. When a stream fails for reason misc, allow it to retry just as if it was resolvefailed. When a stream has failed three times, reset its failure count so we can try again and get all three tries. Changes in version 0.1.1.10-alpha - 2005-12-11 o Correctness bugfixes on 0.1.0.x: - On Windows, build with a libevent patch from "I-M Weasel" to avoid corrupting the heap, losing FDs, or crashing when we need to resize the fd_sets. (This affects the Win32 binaries, not Tor's sources.) - Stop doing the complex voodoo overkill checking for insecure Diffie-Hellman keys. Just check if it's in [2,p-2] and be happy. - When we were closing connections, there was a rare case that stomped on memory, triggering seg faults and asserts. - We were neglecting to unlink marked circuits from soon-to-close OR connections, which caused some rare scribbling on freed memory. - When we're deciding whether a stream has enough circuits around that can handle it, count the freshly dirty ones and not the ones that are so dirty they won't be able to handle it. - Recover better from TCP connections to Tor servers that are broken but don't tell you (it happens!); and rotate TLS connections once a week. - When we're expiring old circuits, we had a logic error that caused us to close new rendezvous circuits rather than old ones. - Fix a scary-looking but apparently harmless bug where circuits would sometimes start out in state CIRCUIT_STATE_OR_WAIT at servers, and never switch to state CIRCUIT_STATE_OPEN. - When building with -static or on Solaris, we sometimes needed to build with -ldl. - Give a useful message when people run Tor as the wrong user, rather than telling them to start chowning random directories. - We were failing to inform the controller about new .onion streams. o Security bugfixes on 0.1.0.x: - Refuse server descriptors if the fingerprint line doesn't match the included identity key. Tor doesn't care, but other apps (and humans) might actually be trusting the fingerprint line. - We used to kill the circuit when we receive a relay command we don't recognize. Now we just drop it. - Start obeying our firewall options more rigorously: . If we can't get to a dirserver directly, try going via Tor. . Don't ever try to connect (as a client) to a place our firewall options forbid. . If we specify a proxy and also firewall options, obey the firewall options even when we're using the proxy: some proxies can only proxy to certain destinations. - Fix a bug found by Lasse Overlier: when we were making internal circuits (intended to be cannibalized later for rendezvous and introduction circuits), we were picking them so that they had useful exit nodes. There was no need for this, and it actually aids some statistical attacks. - Start treating internal circuits and exit circuits separately. It's important to keep them separate because internal circuits have their last hops picked like middle hops, rather than like exit hops. So exiting on them will break the user's expectations. o Bugfixes on 0.1.1.x: - Take out the mis-feature where we tried to detect IP address flapping for people with DynDNS, and chose not to upload a new server descriptor sometimes. - Try to be compatible with OpenSSL 0.9.6 again. - Log fix: when the controller is logging about .onion addresses, sometimes it didn't include the ".onion" part of the address. - Don't try to modify options->DirServers internally -- if the user didn't specify any, just add the default ones directly to the trusted dirserver list. This fixes a bug where people running controllers would use SETCONF on some totally unrelated config option, and Tor would start yelling at them about changing their DirServer lines. - Let the controller's redirectstream command specify a port, in case the controller wants to change that too. - When we requested a pile of server descriptors, we sometimes accidentally launched a duplicate request for the first one. - Bugfix for trackhostexits: write down the fingerprint of the chosen exit, not its nickname, because the chosen exit might not be verified. - When parsing foo.exit, if foo is unknown, and we are leaving circuits unattached, set the chosen_exit field and leave the address empty. This matters because controllers got confused otherwise. - Directory authorities no longer try to download server descriptors that they know they will reject. o Features and updates: - Replace balanced trees with hash tables: this should make stuff significantly faster. - Resume using the AES counter-mode implementation that we ship, rather than OpenSSL's. Ours is significantly faster. - Many other CPU and memory improvements. - Add a new config option FastFirstHopPK (on by default) so clients do a trivial crypto handshake for their first hop, since TLS has already taken care of confidentiality and authentication. - Add a new config option TestSocks so people can see if their applications are using socks4, socks4a, socks5-with-ip, or socks5-with-hostname. This way they don't have to keep mucking with tcpdump and wondering if something got cached somewhere. - Warn when listening on a public address for socks. I suspect a lot of people are setting themselves up as open socks proxies, and they have no idea that jerks on the Internet are using them, since they simply proxy the traffic into the Tor network. - Add "private:*" as an alias in configuration for policies. Now you can simplify your exit policy rather than needing to list every single internal or nonroutable network space. - Add a new controller event type that allows controllers to get all server descriptors that were uploaded to a router in its role as authoritative dirserver. - Start shipping socks-extensions.txt, tor-doc-unix.html, tor-doc-server.html, and stylesheet.css in the tarball. - Stop shipping tor-doc.html in the tarball. Changes in version 0.1.1.9-alpha - 2005-11-15 o Usability improvements: - Start calling it FooListenAddress rather than FooBindAddress, since few of our users know what it means to bind an address or port. - Reduce clutter in server logs. We're going to try to make them actually usable now. New config option ProtocolWarnings that lets you hear about how _other Tors_ are breaking the protocol. Off by default. - Divide log messages into logging domains. Once we put some sort of interface on this, it will let people looking at more verbose log levels specify the topics they want to hear more about. - Make directory servers return better http 404 error messages instead of a generic "Servers unavailable". - Check for even more Windows version flags when writing the platform string in server descriptors, and note any we don't recognize. - Clean up more of the OpenSSL memory when exiting, so we can detect memory leaks better. - Make directory authorities be non-versioning, non-naming by default. Now we can add new directory servers without requiring their operators to pay close attention. - When logging via syslog, include the pid whenever we provide a log entry. Suggested by Todd Fries. o Performance improvements: - Directory servers now silently throw away new descriptors that haven't changed much if the timestamps are similar. We do this to tolerate older Tor servers that upload a new descriptor every 15 minutes. (It seemed like a good idea at the time.) - Inline bottleneck smartlist functions; use fast versions by default. - Add a "Map from digest to void*" abstraction digestmap_t so we can do less hex encoding/decoding. Use it in router_get_by_digest() to resolve a performance bottleneck. - Allow tor_gzip_uncompress to extract as much as possible from truncated compressed data. Try to extract as many descriptors as possible from truncated http responses (when DIR_PURPOSE_FETCH_ROUTERDESC). - Make circ->onionskin a pointer, not a static array. moria2 was using 125000 circuit_t's after it had been up for a few weeks, which translates to 20+ megs of wasted space. - The private half of our EDH handshake keys are now chosen out of 320 bits, not 1024 bits. (Suggested by Ian Goldberg.) o Security improvements: - Start making directory caches retain old routerinfos, so soon clients can start asking by digest of descriptor rather than by fingerprint of server. - Add half our entropy from RAND_poll in OpenSSL. This knows how to use egd (if present), openbsd weirdness (if present), vms/os2 weirdness (if we ever port there), and more in the future. o Bugfixes on 0.1.0.x: - Do round-robin writes of at most 16 kB per write. This might be more fair on loaded Tor servers, and it might resolve our Windows crash bug. It might also slow things down. - Our TLS handshakes were generating a single public/private keypair for the TLS context, rather than making a new one for each new connections. Oops. (But we were still rotating them periodically, so it's not so bad.) - When we were cannibalizing a circuit with a particular exit node in mind, we weren't checking to see if that exit node was already present earlier in the circuit. Oops. - When a Tor server's IP changes (e.g. from a dyndns address), upload a new descriptor so clients will learn too. - Really busy servers were keeping enough circuits open on stable connections that they were wrapping around the circuit_id space. (It's only two bytes.) This exposed a bug where we would feel free to reuse a circuit_id even if it still exists but has been marked for close. Try to fix this bug. Some bug remains. - If we would close a stream early (e.g. it asks for a .exit that we know would refuse it) but the LeaveStreamsUnattached config option is set by the controller, then don't close it. o Bugfixes on 0.1.1.8-alpha: - Fix a big pile of memory leaks, some of them serious. - Do not try to download a routerdesc if we would immediately reject it as obsolete. - Resume inserting a newline between all router descriptors when generating (old style) signed directories, since our spec says we do. - When providing content-type application/octet-stream for server descriptors using .z, we were leaving out the content-encoding header. Oops. (Everything tolerated this just fine, but that doesn't mean we need to be part of the problem.) - Fix a potential seg fault in getconf and getinfo using version 1 of the controller protocol. - Avoid crash: do not check whether DirPort is reachable when we are suppressing it because of hibernation. - Make --hash-password not crash on exit. Changes in version 0.1.1.8-alpha - 2005-10-07 o New features (major): - Clients don't download or use the directory anymore. Now they download and use network-statuses from the trusted dirservers, and fetch individual server descriptors as needed from mirrors. See dir-spec.txt for all the gory details. - Be more conservative about whether to advertise our DirPort. The main change is to not advertise if we're running at capacity and either a) we could hibernate or b) our capacity is low and we're using a default DirPort. - Use OpenSSL's AES when OpenSSL has version 0.9.7 or later. o New features (minor): - Try to be smart about when to retry network-status and server-descriptor fetches. Still needs some tuning. - Stop parsing, storing, or using running-routers output (but mirrors still cache and serve it). - Consider a threshold of versioning dirservers (dirservers who have an opinion about which Tor versions are still recommended) before deciding whether to warn the user that he's obsolete. - Dirservers can now reject/invalidate by key and IP, with the config options "AuthDirInvalid" and "AuthDirReject". This is useful since currently we automatically list servers as running and usable even if we know they're jerks. - Provide dire warnings to any users who set DirServer; move it out of torrc.sample and into torrc.complete. - Add MyFamily to torrc.sample in the server section. - Add nicknames to the DirServer line, so we can refer to them without requiring all our users to memorize their IP addresses. - When we get an EOF or a timeout on a directory connection, note how many bytes of serverdesc we are dropping. This will help us determine whether it is smart to parse incomplete serverdesc responses. - Add a new function to "change pseudonyms" -- that is, to stop using any currently-dirty circuits for new streams, so we don't link new actions to old actions. Currently it's only called on HUP (or SIGNAL RELOAD). - On sighup, if UseHelperNodes changed to 1, use new circuits. - Start using RAND_bytes rather than RAND_pseudo_bytes from OpenSSL. Also, reseed our entropy every hour, not just at startup. And entropy in 512-bit chunks, not 160-bit chunks. o Fixes on 0.1.1.7-alpha: - Nobody ever implemented EVENT_ADDRMAP for control protocol version 0, so don't let version 0 controllers ask for it. - If you requested something with too many newlines via the v1 controller protocol, you could crash tor. - Fix a number of memory leaks, including some pretty serious ones. - Re-enable DirPort testing again, so Tor servers will be willing to advertise their DirPort if it's reachable. - On TLS handshake, only check the other router's nickname against its expected nickname if is_named is set. o Fixes forward-ported from 0.1.0.15: - Don't crash when we don't have any spare file descriptors and we try to spawn a dns or cpu worker. - Make the numbers in read-history and write-history into uint64s, so they don't overflow and publish negatives in the descriptor. o Fixes on 0.1.0.x: - For the OS X package's modified privoxy config file, comment out the "logfile" line so we don't log everything passed through privoxy. - We were whining about using socks4 or socks5-with-local-lookup even when it's an IP in the "virtual" range we designed exactly for this case. - We were leaking some memory every time the client changes IPs. - Never call free() on tor_malloc()d memory. This will help us use dmalloc to detect memory leaks. - Check for named servers when looking them up by nickname; warn when we'recalling a non-named server by its nickname; don't warn twice about the same name. - Try to list MyFamily elements by key, not by nickname, and warn if we've not heard of the server. - Make windows platform detection (uname equivalent) smarter. - It turns out sparc64 doesn't like unaligned access either. Changes in version 0.1.0.15 - 2005-09-23 o Bugfixes on 0.1.0.x: - Reject ports 465 and 587 (spam targets) in default exit policy. - Don't crash when we don't have any spare file descriptors and we try to spawn a dns or cpu worker. - Get rid of IgnoreVersion undocumented config option, and make us only warn, never exit, when we're running an obsolete version. - Don't try to print a null string when your server finds itself to be unreachable and the Address config option is empty. - Make the numbers in read-history and write-history into uint64s, so they don't overflow and publish negatives in the descriptor. - Fix a minor memory leak in smartlist_string_remove(). - We were only allowing ourselves to upload a server descriptor at most every 20 minutes, even if it changed earlier than that. - Clean up log entries that pointed to old URLs. Changes in version 0.1.1.7-alpha - 2005-09-14 o Fixes on 0.1.1.6-alpha: - Exit servers were crashing when people asked them to make a connection to an address not in their exit policy. - Looking up a non-existent stream for a v1 control connection would cause a segfault. - Fix a seg fault if we ask a dirserver for a descriptor by fingerprint but he doesn't know about him. - SETCONF was appending items to linelists, not clearing them. - SETCONF SocksBindAddress killed Tor if it fails to bind. Now back out and refuse the setconf if it would fail. - Downgrade the dirserver log messages when whining about unreachability. o New features: - Add Peter Palfrader's check-tor script to tor/contrib/ It lets you easily check whether a given server (referenced by nickname) is reachable by you. - Numerous changes to move towards client-side v2 directories. Not enabled yet. o Fixes on 0.1.0.x: - If the user gave tor an odd number of command-line arguments, we were silently ignoring the last one. Now we complain and fail. [This wins the oldest-bug prize -- this bug has been present since November 2002, as released in Tor 0.0.0.] - Do not use unaligned memory access on alpha, mips, or mipsel. It *works*, but is very slow, so we treat them as if it doesn't. - Retry directory requests if we fail to get an answer we like from a given dirserver (we were retrying before, but only if we fail to connect). - When writing the RecommendedVersions line, sort them first. - When the client asked for a rendezvous port that the hidden service didn't want to provide, we were sending an IP address back along with the end cell. Fortunately, it was zero. But stop that anyway. - Correct "your server is reachable" log entries to indicate that it was self-testing that told us so. Changes in version 0.1.1.6-alpha - 2005-09-09 o Fixes on 0.1.1.5-alpha: - We broke fascistfirewall in 0.1.1.5-alpha. Oops. - Fix segfault in unit tests in 0.1.1.5-alpha. Oops. - Fix bug with tor_memmem finding a match at the end of the string. - Make unit tests run without segfaulting. - Resolve some solaris x86 compile warnings. - Handle duplicate lines in approved-routers files without warning. - Fix bug where as soon as a server refused any requests due to his exit policy (e.g. when we ask for localhost and he tells us that's 127.0.0.1 and he won't do it), we decided he wasn't obeying his exit policy using him for any exits. - Only do openssl hardware accelerator stuff if openssl version is at least 0.9.7. o New controller features/fixes: - Add a "RESETCONF" command so you can set config options like AllowUnverifiedNodes and LongLivedPorts to "". Also, if you give a config option in the torrc with no value, then it clears it entirely (rather than setting it to its default). - Add a "GETINFO config-file" to tell us where torrc is. - Avoid sending blank lines when GETINFO replies should be empty. - Add a QUIT command for the controller (for using it manually). - Fix a bug in SAVECONF that was adding default dirservers and other redundant entries to the torrc file. o Start on the new directory design: - Generate, publish, cache, serve new network-status format. - Publish individual descriptors (by fingerprint, by "all", and by "tell me yours"). - Publish client and server recommended versions separately. - Allow tor_gzip_uncompress() to handle multiple concatenated compressed strings. Serve compressed groups of router descriptors. The compression logic here could be more memory-efficient. - Distinguish v1 authorities (all currently trusted directories) from v2 authorities (all trusted directories). - Change DirServers config line to note which dirs are v1 authorities. - Add configuration option "V1AuthoritativeDirectory 1" which moria1, moria2, and tor26 should set. - Remove option when getting directory cache to see whether they support running-routers; they all do now. Replace it with one to see whether caches support v2 stuff. o New features: - Dirservers now do their own external reachability testing of each Tor server, and only list them as running if they've been found to be reachable. We also send back warnings to the server's logs if it uploads a descriptor that we already believe is unreachable. - Implement exit enclaves: if we know an IP address for the destination, and there's a running Tor server at that address which allows exit to the destination, then extend the circuit to that exit first. This provides end-to-end encryption and end-to-end authentication. Also, if the user wants a .exit address or enclave, use 4 hops rather than 3, and cannibalize a general circ for it if you can. - Permit transitioning from ORPort=0 to ORPort!=0, and back, from the controller. Also, rotate dns and cpu workers if the controller changes options that will affect them; and initialize the dns worker cache tree whether or not we start out as a server. - Only upload a new server descriptor when options change, 18 hours have passed, uptime is reset, or bandwidth changes a lot. - Check [X-]Forwarded-For headers in HTTP requests when generating log messages. This lets people run dirservers (and caches) behind Apache but still know which IP addresses are causing warnings. o Config option changes: - Replace (Fascist)Firewall* config options with a new ReachableAddresses option that understands address policies. For example, "ReachableAddresses *:80,*:443" - Get rid of IgnoreVersion undocumented config option, and make us only warn, never exit, when we're running an obsolete version. - Make MonthlyAccountingStart config option truly obsolete now. o Fixes on 0.1.0.x: - Reject ports 465 and 587 in the default exit policy, since people have started using them for spam too. - It turns out we couldn't bootstrap a network since we added reachability detection in 0.1.0.1-rc. Good thing the Tor network has never gone down. Add an AssumeReachable config option to let servers and dirservers bootstrap. When we're trying to build a high-uptime or high-bandwidth circuit but there aren't enough suitable servers, try being less picky rather than simply failing. - Our logic to decide if the OR we connected to was the right guy was brittle and maybe open to a mitm for unverified routers. - We weren't cannibalizing circuits correctly for CIRCUIT_PURPOSE_C_ESTABLISH_REND and CIRCUIT_PURPOSE_S_ESTABLISH_INTRO, so we were being forced to build those from scratch. This should make hidden services faster. - Predict required circuits better, with an eye toward making hidden services faster on the service end. - Retry streams if the exit node sends back a 'misc' failure. This should result in fewer random failures. Also, after failing from resolve failed or misc, reset the num failures, so we give it a fair shake next time we try. - Clean up the rendezvous warn log msgs, and downgrade some to info. - Reduce severity on logs about dns worker spawning and culling. - When we're shutting down and we do something like try to post a server descriptor or rendezvous descriptor, don't complain that we seem to be unreachable. Of course we are, we're shutting down. - Add TTLs to RESOLVED, CONNECTED, and END_REASON_EXITPOLICY cells. We don't use them yet, but maybe one day our DNS resolver will be able to discover them. - Make ContactInfo mandatory for authoritative directory servers. - Require server descriptors to list IPv4 addresses -- hostnames are no longer allowed. This also fixes some potential security problems with people providing hostnames as their address and then preferentially resolving them to partition users. - Change log line for unreachability to explicitly suggest /etc/hosts as the culprit. Also make it clearer what IP address and ports we're testing for reachability. - Put quotes around user-supplied strings when logging so users are more likely to realize if they add bad characters (like quotes) to the torrc. - Let auth dir servers start without specifying an Address config option. - Make unit tests (and other invocations that aren't the real Tor) run without launching listeners, creating subdirectories, and so on. Changes in version 0.1.1.5-alpha - 2005-08-08 o Bugfixes included in 0.1.0.14. o Bugfixes on 0.1.0.x: - If you write "HiddenServicePort 6667 127.0.0.1 6668" in your torrc rather than "HiddenServicePort 6667 127.0.0.1:6668", it would silently using ignore the 6668. Changes in version 0.1.0.14 - 2005-08-08 o Bugfixes on 0.1.0.x: - Fix the other half of the bug with crypto handshakes (CVE-2005-2643). - Fix an assert trigger if you send a 'signal term' via the controller when it's listening for 'event info' messages. Changes in version 0.1.1.4-alpha - 2005-08-04 o Bugfixes included in 0.1.0.13. o Features: - Improve tor_gettimeofday() granularity on windows. - Make clients regenerate their keys when their IP address changes. - Implement some more GETINFO goodness: expose helper nodes, config options, getinfo keys. Changes in version 0.1.0.13 - 2005-08-04 o Bugfixes on 0.1.0.x: - Fix a critical bug in the security of our crypto handshakes. - Fix a size_t underflow in smartlist_join_strings2() that made it do bad things when you hand it an empty smartlist. - Fix Windows installer to ship Tor license (thanks to Aphex for pointing out this oversight) and put a link to the doc directory in the start menu. - Explicitly set no-unaligned-access for sparc: it turns out the new gcc's let you compile broken code, but that doesn't make it not-broken. Changes in version 0.1.1.3-alpha - 2005-07-23 o Bugfixes on 0.1.1.2-alpha: - Fix a bug in handling the controller's "post descriptor" function. - Fix several bugs in handling the controller's "extend circuit" function. - Fix a bug in handling the controller's "stream status" event. - Fix an assert failure if we have a controller listening for circuit events and we go offline. - Re-allow hidden service descriptors to publish 0 intro points. - Fix a crash when generating your hidden service descriptor if you don't have enough intro points already. o New features on 0.1.1.2-alpha: - New controller function "getinfo accounting", to ask how many bytes we've used in this time period. - Experimental support for helper nodes: a lot of the risk from a small static adversary comes because users pick new random nodes every time they rebuild a circuit. Now users will try to stick to the same small set of entry nodes if they can. Not enabled by default yet. o Bugfixes on 0.1.0.12: - If you're an auth dir server, always publish your dirport, even if you haven't yet found yourself to be reachable. - Fix a size_t underflow in smartlist_join_strings2() that made it do bad things when you hand it an empty smartlist. Changes in version 0.1.0.12 - 2005-07-18 o New directory servers: - tor26 has changed IP address. o Bugfixes on 0.1.0.x: - Fix a possible double-free in tor_gzip_uncompress(). - When --disable-threads is set, do not search for or link against pthreads libraries. - Don't trigger an assert if an authoritative directory server claims its dirport is 0. - Fix bug with removing Tor as an NT service: some people were getting "The service did not return an error." Thanks to Matt Edman for the fix. Changes in version 0.1.1.2-alpha - 2005-07-15 o New directory servers: - tor26 has changed IP address. o Bugfixes on 0.1.0.x, crashes/leaks: - Port the servers-not-obeying-their-exit-policies fix from 0.1.0.11. - Fix an fd leak in start_daemon(). - On Windows, you can't always reopen a port right after you've closed it. So change retry_listeners() to only close and re-open ports that have changed. - Fix a possible double-free in tor_gzip_uncompress(). o Bugfixes on 0.1.0.x, usability: - When tor_socketpair() fails in Windows, give a reasonable Windows-style errno back. - Let people type "tor --install" as well as "tor -install" when they want to make it an NT service. - NT service patch from Matt Edman to improve error messages. - When the controller asks for a config option with an abbreviated name, give the full name in our response. - Correct the man page entry on TrackHostExitsExpire. - Looks like we were never delivering deflated (i.e. compressed) running-routers lists, even when asked. Oops. - When --disable-threads is set, do not search for or link against pthreads libraries. o Bugfixes on 0.1.1.x: - Fix a seg fault with autodetecting which controller version is being used. o Features: - New hidden service descriptor format: put a version in it, and let people specify introduction/rendezvous points that aren't in "the directory" (which is subjective anyway). - Allow the DEBUG controller event to work again. Mark certain log entries as "don't tell this to controllers", so we avoid cycles. Changes in version 0.1.0.11 - 2005-06-30 o Bugfixes on 0.1.0.x: - Fix major security bug: servers were disregarding their exit policies if clients behaved unexpectedly. - Make OS X init script check for missing argument, so we don't confuse users who invoke it incorrectly. - Fix a seg fault in "tor --hash-password foo". - The MAPADDRESS control command was broken. Changes in version 0.1.1.1-alpha - 2005-06-29 o Bugfixes: - Make OS X init script check for missing argument, so we don't confuse users who invoke it incorrectly. - Fix a seg fault in "tor --hash-password foo". - Fix a possible way to DoS dirservers. - When we complain that your exit policy implicitly allows local or private address spaces, name them explicitly so operators can fix it. - Make the log message less scary when all the dirservers are temporarily unreachable. - We were printing the number of idle dns workers incorrectly when culling them. o Features: - Revised controller protocol (version 1) that uses ascii rather than binary. Add supporting libraries in python and java so you can use the controller from your applications without caring how our protocol works. - Spiffy new support for crypto hardware accelerators. Can somebody test this? Changes in version 0.0.9.10 - 2005-06-16 o Bugfixes on 0.0.9.x (backported from 0.1.0.10): - Refuse relay cells that claim to have a length larger than the maximum allowed. This prevents a potential attack that could read arbitrary memory (e.g. keys) from an exit server's process (CVE-2005-2050). Changes in version 0.1.0.10 - 2005-06-14 o Allow a few EINVALs from libevent before dying. Warn on kqueue with libevent before 1.1a. Changes in version 0.1.0.9-rc - 2005-06-09 o Bugfixes: - Reset buf->highwater every time buf_shrink() is called, not just on a successful shrink. This was causing significant memory bloat. - Fix buffer overflow when checking hashed passwords. - Security fix: if seeding the RNG on Win32 fails, quit. - Allow seeding the RNG on Win32 even when you're not running as Administrator. - Disable threading on Solaris too. Something is wonky with it, cpuworkers, and reentrant libs. - Reenable the part of the code that tries to flush as soon as an OR outbuf has a full TLS record available. Perhaps this will make OR outbufs not grow as huge except in rare cases, thus saving lots of CPU time plus memory. - Reject malformed .onion addresses rather then passing them on as normal web requests. - Adapt patch from Adam Langley: fix possible memory leak in tor_lookup_hostname(). - Initialize libevent later in the startup process, so the logs are already established by the time we start logging libevent warns. - Use correct errno on win32 if libevent fails. - Check and warn about known-bad/slow libevent versions. - Pay more attention to the ClientOnly config option. - Have torctl.in/tor.sh.in check for location of su binary (needed on FreeBSD) - Correct/add man page entries for LongLivedPorts, ExitPolicy, KeepalivePeriod, ClientOnly, NoPublish, HttpProxy, HttpsProxy, HttpProxyAuthenticator - Stop warning about sigpipes in the logs. We're going to pretend that getting these occassionally is normal and fine. - Resolve OS X installer bugs: stop claiming to be 0.0.9.2 in certain installer screens; and don't put stuff into StartupItems unless the user asks you to. - Require servers that use the default dirservers to have public IP addresses. We have too many servers that are configured with private IPs and their admins never notice the log entries complaining that their descriptors are being rejected. - Add OSX uninstall instructions. An actual uninstall script will come later. Changes in version 0.1.0.8-rc - 2005-05-23 o Bugfixes: - It turns out that kqueue on OS X 10.3.9 was causing kernel panics. Disable kqueue on all OS X Tors. - Fix RPM: remove duplicate line accidentally added to the rpm spec file. - Disable threads on openbsd too, since its gethostaddr is not reentrant either. - Tolerate libevent 0.8 since it still works, even though it's ancient. - Enable building on Red Hat 9.0 again. - Allow the middle hop of the testing circuit to be running any version, now that most of them have the bugfix to let them connect to unknown servers. This will allow reachability testing to work even when 0.0.9.7-0.0.9.9 become obsolete. - Handle relay cells with rh.length too large. This prevents a potential attack that could read arbitrary memory (maybe even keys) from the exit server's process. - We screwed up the dirport reachability testing when we don't yet have a cached version of the directory. Hopefully now fixed. - Clean up router_load_single_router() (used by the controller), so it doesn't seg fault on error. - Fix a minor memory leak when somebody establishes an introduction point at your Tor server. - If a socks connection ends because read fails, don't warn that you're not sending a socks reply back. o Features: - Add HttpProxyAuthenticator config option too, that works like the HttpsProxyAuthenticator config option. - Encode hashed controller passwords in hex instead of base64, to make it easier to write controllers. Changes in version 0.1.0.7-rc - 2005-05-17 o Bugfixes: - Fix a bug in the OS X package installer that prevented it from installing on Tiger. - Fix a script bug in the OS X package installer that made it complain during installation. - Find libevent even if it's hiding in /usr/local/ and your CFLAGS and LDFLAGS don't tell you to look there. - Be able to link with libevent as a shared library (the default after 1.0d), even if it's hiding in /usr/local/lib and even if you haven't added /usr/local/lib to your /etc/ld.so.conf, assuming you're running gcc. Otherwise fail and give a useful error message. - Fix a bug in the RPM packager: set home directory for _tor to something more reasonable when first installing. - Free a minor amount of memory that is still reachable on exit. Changes in version 0.1.0.6-rc - 2005-05-14 o Bugfixes: - Implement --disable-threads configure option. Disable threads on netbsd by default, because it appears to have no reentrant resolver functions. - Apple's OS X 10.4.0 ships with a broken kqueue. The new libevent release (1.1) detects and disables kqueue if it's broken. - Append default exit policy before checking for implicit internal addresses. Now we don't log a bunch of complaints on startup when using the default exit policy. - Some people were putting "Address " in their torrc, and they had a buggy resolver that resolved " " to 0.0.0.0. Oops. - If DataDir is ~/.tor, and that expands to /.tor, then default to LOCALSTATEDIR/tor instead. - Fix fragmented-message bug in TorControl.py. - Resolve a minor bug which would prevent unreachable dirports from getting suppressed in the published descriptor. - When the controller gave us a new descriptor, we weren't resolving it immediately, so Tor would think its address was 0.0.0.0 until we fetched a new directory. - Fix an uppercase/lowercase case error in suppressing a bogus libevent warning on some Linuxes. o Features: - Begin scrubbing sensitive strings from logs by default. Turn off the config option SafeLogging if you need to do debugging. - Switch to a new buffer management algorithm, which tries to avoid reallocing and copying quite as much. In first tests it looks like it uses *more* memory on average, but less cpu. - First cut at support for "create-fast" cells. Clients can use these when extending to their first hop, since the TLS already provides forward secrecy and authentication. Not enabled on clients yet. - When dirservers refuse a router descriptor, we now log its contactinfo, platform, and the poster's IP address. - Call tor_free_all instead of connections_free_all after forking, to save memory on systems that need to fork. - Whine at you if you're a server and you don't set your contactinfo. - Implement --verify-config command-line option to check if your torrc is valid without actually launching Tor. - Rewrite address "serifos.exit" to "localhost.serifos.exit" rather than just rejecting it. Changes in version 0.1.0.5-rc - 2005-04-27 o Bugfixes: - Stop trying to print a null pointer if an OR conn fails because we didn't like its cert. o Features: - Switch our internal buffers implementation to use a ring buffer, to hopefully improve performance for fast servers a lot. - Add HttpsProxyAuthenticator support (basic auth only), based on patch from Adam Langley. - Bump the default BandwidthRate from 1 MB to 2 MB, to accommodate the fast servers that have been joining lately. - Give hidden service accesses extra time on the first attempt, since 60 seconds is often only barely enough. This might improve robustness more. - Improve performance for dirservers: stop re-parsing the whole directory every time you regenerate it. - Add more debugging info to help us find the weird dns freebsd pthreads bug; cleaner debug messages to help track future issues. Changes in version 0.0.9.9 - 2005-04-23 o Bugfixes on 0.0.9.x: - If unofficial Tor clients connect and send weird TLS certs, our Tor server triggers an assert. This release contains a minimal backport from the broader fix that we put into 0.1.0.4-rc. Changes in version 0.1.0.4-rc - 2005-04-23 o Bugfixes: - If unofficial Tor clients connect and send weird TLS certs, our Tor server triggers an assert. Stop asserting, and start handling TLS errors better in other situations too. - When the controller asks us to tell it about all the debug-level logs, it turns out we were generating debug-level logs while telling it about them, which turns into a bad loop. Now keep track of whether you're sending a debug log to the controller, and don't log when you are. - Fix the "postdescriptor" feature of the controller interface: on non-complete success, only say "done" once. o Features: - Clients are now willing to load balance over up to 2mB, not 1mB, of advertised bandwidth capacity. - Add a NoPublish config option, so you can be a server (e.g. for testing running Tor servers in other Tor networks) without publishing your descriptor to the primary dirservers. Changes in version 0.1.0.3-rc - 2005-04-08 o Improvements on 0.1.0.2-rc: - Client now retries when streams end early for 'hibernating' or 'resource limit' reasons, rather than failing them. - More automated handling for dirserver operators: - Automatically approve nodes running 0.1.0.2-rc or later, now that the the reachability detection stuff is working. - Now we allow two unverified servers with the same nickname but different keys. But if a nickname is verified, only that nickname+key are allowed. - If you're an authdirserver connecting to an address:port, and it's not the OR you were expecting, forget about that descriptor. If he *was* the one you were expecting, then forget about all other descriptors for that address:port. - Allow servers to publish descriptors from 12 hours in the future. Corollary: only whine about clock skew from the dirserver if he's a trusted dirserver (since now even verified servers could have quite wrong clocks). - Adjust maximum skew and age for rendezvous descriptors: let skew be 48 hours rather than 90 minutes. - Efficiency improvements: - Keep a big splay tree of (circid,orconn)->circuit mappings to make it much faster to look up a circuit for each relay cell. - Remove most calls to assert_all_pending_dns_resolves_ok(), since they're eating our cpu on exit nodes. - Stop wasting time doing a case insensitive comparison for every dns name every time we do any lookup. Canonicalize the names to lowercase and be done with it. - Start sending 'truncated' cells back rather than destroy cells, if the circuit closes in front of you. This means we won't have to abandon partially built circuits. - Only warn once per nickname from add_nickname_list_to_smartlist per failure, so an entrynode or exitnode choice that's down won't yell so much. - Put a note in the torrc about abuse potential with the default exit policy. - Revise control spec and implementation to allow all log messages to be sent to controller with their severities intact (suggested by Matt Edman). Update TorControl to handle new log event types. - Provide better explanation messages when controller's POSTDESCRIPTOR fails. - Stop putting nodename in the Platform string in server descriptors. It doesn't actually help, and it is confusing/upsetting some people. o Bugfixes on 0.1.0.2-rc: - We were printing the host mask wrong in exit policies in server descriptors. This isn't a critical bug though, since we were still obeying the exit policy internally. - Fix Tor when compiled with libevent but without pthreads: move connection_unregister() from _connection_free() to connection_free(). - Fix an assert trigger (already fixed in 0.0.9.x): when we have the rare mysterious case of accepting a conn on 0.0.0.0:0, then when we look through the connection array, we'll find any of the cpu/dnsworkers. This is no good. o Bugfixes on 0.0.9.8: - Fix possible bug on threading platforms (e.g. win32) which was leaking a file descriptor whenever a cpuworker or dnsworker died. - When using preferred entry or exit nodes, ignore whether the circuit wants uptime or capacity. They asked for the nodes, they get the nodes. - chdir() to your datadirectory at the *end* of the daemonize process, not the beginning. This was a problem because the first time you run tor, if your datadir isn't there, and you have runasdaemon set to 1, it will try to chdir to it before it tries to create it. Oops. - Handle changed router status correctly when dirserver reloads fingerprint file. We used to be dropping all unverified descriptors right then. The bug was hidden because we would immediately fetch a directory from another dirserver, which would include the descriptors we just dropped. - When we're connecting to an OR and he's got a different nickname/key than we were expecting, only complain loudly if we're an OP or a dirserver. Complaining loudly to the OR admins just confuses them. - Tie MAX_DIR_SIZE to MAX_BUF_SIZE, so now directory sizes won't get artificially capped at 500kB. Changes in version 0.0.9.8 - 2005-04-07 o Bugfixes on 0.0.9.x: - We have a bug that I haven't found yet. Sometimes, very rarely, cpuworkers get stuck in the 'busy' state, even though the cpuworker thinks of itself as idle. This meant that no new circuits ever got established. Here's a workaround to kill any cpuworker that's been busy for more than 100 seconds. Changes in version 0.1.0.2-rc - 2005-04-01 o Bugfixes on 0.1.0.1-rc: - Fixes on reachability detection: - Don't check for reachability while hibernating. - If ORPort is reachable but DirPort isn't, still publish the descriptor, but zero out DirPort until it's found reachable. - When building testing circs for ORPort testing, use only high-bandwidth nodes, so fewer circuits fail. - Complain about unreachable ORPort separately from unreachable DirPort, so the user knows what's going on. - Make sure we only conclude ORPort reachability if we didn't initiate the conn. Otherwise we could falsely conclude that we're reachable just because we connected to the guy earlier and he used that same pipe to extend to us. - Authdirservers shouldn't do ORPort reachability detection, since they're in clique mode, so it will be rare to find a server not already connected to them. - When building testing circuits, always pick middle hops running Tor 0.0.9.7, so we avoid the "can't extend to unknown routers" bug. (This is a kludge; it will go away when 0.0.9.x becomes obsolete.) - When we decide we're reachable, actually publish our descriptor right then. - Fix bug in redirectstream in the controller. - Fix the state descriptor strings so logs don't claim edge streams are in a different state than they actually are. - Use recent libevent features when possible (this only really affects win32 and osx right now, because the new libevent with these features hasn't been released yet). Add code to suppress spurious libevent log msgs. - Prevent possible segfault in connection_close_unattached_ap(). - Fix newlines on torrc in win32. - Improve error msgs when tor-resolve fails. o Improvements on 0.0.9.x: - New experimental script tor/contrib/ExerciseServer.py (needs more work) that uses the controller interface to build circuits and fetch pages over them. This will help us bootstrap servers that have lots of capacity but haven't noticed it yet. - New experimental script tor/contrib/PathDemo.py (needs more work) that uses the controller interface to let you choose whole paths via addresses like "...path" - When we've connected to an OR and handshaked but didn't like the result, we were closing the conn without sending destroy cells back for pending circuits. Now send those destroys. Changes in version 0.0.9.7 - 2005-04-01 o Bugfixes on 0.0.9.x: - Fix another race crash bug (thanks to Glenn Fink for reporting). - Compare identity to identity, not to nickname, when extending to a router not already in the directory. This was preventing us from extending to unknown routers. Oops. - Make sure to create OS X Tor user in <500 range, so we aren't creating actual system users. - Note where connection-that-hasn't-sent-end was marked, and fix a few really loud instances of this harmless bug (it's fixed more in 0.1.0.x). Changes in version 0.1.0.1-rc - 2005-03-28 o New features: - Add reachability testing. Your Tor server will automatically try to see if its ORPort and DirPort are reachable from the outside, and it won't upload its descriptor until it decides they are. - Handle unavailable hidden services better. Handle slow or busy hidden services better. - Add support for CONNECTing through https proxies, with "HttpsProxy" config option. - New exit policy: accept most low-numbered ports, rather than rejecting most low-numbered ports. - More Tor controller support (still experimental). See http://tor.eff.org/doc/control-spec.txt for all the new features, including signals to emulate unix signals from any platform; redirectstream; extendcircuit; mapaddress; getinfo; postdescriptor; closestream; closecircuit; etc. - Make nt services work and start on startup on win32 (based on patch by Matt Edman). - Add a new AddressMap config directive to rewrite incoming socks addresses. This lets you, for example, declare an implicit required exit node for certain sites. - Add a new TrackHostExits config directive to trigger addressmaps for certain incoming socks addresses -- for sites that break when your exit keeps changing (based on patch by Mike Perry). - Redo the client-side dns cache so it's just an addressmap too. - Notice when our IP changes, and reset stats/uptime/reachability. - When an application is using socks5, give him the whole variety of potential socks5 responses (connect refused, host unreachable, etc), rather than just "success" or "failure". - A more sane version numbering system. See http://tor.eff.org/cvs/tor/doc/version-spec.txt for details. - New contributed script "exitlist": a simple python script to parse directories and find Tor nodes that exit to listed addresses/ports. - New contributed script "privoxy-tor-toggle" to toggle whether Privoxy uses Tor. Seems to be configured for Debian by default. - Report HTTP reasons to client when getting a response from directory servers -- so you can actually know what went wrong. - New config option MaxAdvertisedBandwidth which lets you advertise a low bandwidthrate (to not attract as many circuits) while still allowing a higher bandwidthrate in reality. o Robustness/stability fixes: - Make Tor use Niels Provos's libevent instead of its current poll-but-sometimes-select mess. This will let us use faster async cores (like epoll, kpoll, and /dev/poll), and hopefully work better on Windows too. - pthread support now too. This was forced because when we forked, we ended up wasting a lot of duplicate ram over time. Also switch to foo_r versions of some library calls to allow reentry and threadsafeness. - Better handling for heterogeneous / unreliable nodes: - Annotate circuits w/ whether they aim to contain high uptime nodes and/or high capacity nodes. When building circuits, choose appropriate nodes. - This means that every single node in an intro rend circuit, not just the last one, will have a minimum uptime. - New config option LongLivedPorts to indicate application streams that will want high uptime circuits. - Servers reset uptime when a dir fetch entirely fails. This hopefully reflects stability of the server's network connectivity. - If somebody starts his tor server in Jan 2004 and then fixes his clock, don't make his published uptime be a year. - Reset published uptime when you wake up from hibernation. - Introduce a notion of 'internal' circs, which are chosen without regard to the exit policy of the last hop. Intro and rendezvous circs must be internal circs, to avoid leaking information. Resolve and connect streams can use internal circs if they want. - New circuit pooling algorithm: make sure to have enough circs around to satisfy any predicted ports, and also make sure to have 2 internal circs around if we've required internal circs lately (and with high uptime if we've seen that lately too). - Split NewCircuitPeriod option into NewCircuitPeriod (30 secs), which describes how often we retry making new circuits if current ones are dirty, and MaxCircuitDirtiness (10 mins), which describes how long we're willing to make use of an already-dirty circuit. - Cannibalize GENERAL circs to be C_REND, C_INTRO, S_INTRO, and S_REND circ as necessary, if there are any completed ones lying around when we try to launch one. - Make hidden services try to establish a rendezvous for 30 seconds, rather than for n (where n=3) attempts to build a circuit. - Change SHUTDOWN_WAIT_LENGTH from a fixed 30 secs to a config option "ShutdownWaitLength". - Try to be more zealous about calling connection_edge_end when things go bad with edge conns in connection.c. - Revise tor-spec to add more/better stream end reasons. - Revise all calls to connection_edge_end to avoid sending "misc", and to take errno into account where possible. o Bug fixes: - Fix a race condition that can trigger an assert, when we have a pending create cell and an OR connection fails right then. - Fix several double-mark-for-close bugs, e.g. where we were finding a conn for a cell even if that conn is already marked for close. - Make sequence of log messages when starting on win32 with no config file more reasonable. - When choosing an exit node for a new non-internal circ, don't take into account whether it'll be useful for any pending x.onion addresses -- it won't. - Turn addr_policy_compare from a tristate to a quadstate; this should help address our "Ah, you allow 1.2.3.4:80. You are a good choice for google.com" problem. - Make "platform" string in descriptor more accurate for Win32 servers, so it's not just "unknown platform". - Fix an edge case in parsing config options (thanks weasel). If they say "--" on the commandline, it's not an option. - Reject odd-looking addresses at the client (e.g. addresses that contain a colon), rather than having the server drop them because they're malformed. - tor-resolve requests were ignoring .exit if there was a working circuit they could use instead. - REUSEADDR on normal platforms means you can rebind to the port right after somebody else has let it go. But REUSEADDR on win32 means to let you bind to the port _even when somebody else already has it bound_! So, don't do that on Win32. - Change version parsing logic: a version is "obsolete" if it is not recommended and (1) there is a newer recommended version in the same series, or (2) there are no recommended versions in the same series, but there are some recommended versions in a newer series. A version is "new" if it is newer than any recommended version in the same series. - Stop most cases of hanging up on a socks connection without sending the socks reject. o Helpful fixes: - Require BandwidthRate to be at least 20kB/s for servers. - When a dirserver causes you to give a warn, mention which dirserver it was. - New config option DirAllowPrivateAddresses for authdirservers. Now by default they refuse router descriptors that have non-IP or private-IP addresses. - Stop publishing socksport in the directory, since it's not actually meant to be public. For compatibility, publish a 0 there for now. - Change DirFetchPeriod/StatusFetchPeriod to have a special "Be smart" value, that is low for servers and high for clients. - If our clock jumps forward by 100 seconds or more, assume something has gone wrong with our network and abandon all not-yet-used circs. - Warn when exit policy implicitly allows local addresses. - If we get an incredibly skewed timestamp from a dirserver mirror that isn't a verified OR, don't warn -- it's probably him that's wrong. - Since we ship our own Privoxy on OS X, tweak it so it doesn't write cookies to disk and doesn't log each web request to disk. (Thanks to Brett Carrington for pointing this out.) - When a client asks us for a dir mirror and we don't have one, launch an attempt to get a fresh one. - If we're hibernating and we get a SIGINT, exit immediately. - Add --with-dmalloc ./configure option, to track memory leaks. - And try to free all memory on closing, so we can detect what we're leaking. - Cache local dns resolves correctly even when they're .exit addresses. - Give a better warning when some other server advertises an ORPort that is actually an apache running ssl. - Add "opt hibernating 1" to server descriptor to make it clearer whether the server is hibernating. Changes in version 0.0.9.6 - 2005-03-24 o Bugfixes on 0.0.9.x (crashes and asserts): - Add new end stream reasons to maintainance branch. Fix bug where reason (8) could trigger an assert. Prevent bug from recurring. - Apparently win32 stat wants paths to not end with a slash. - Fix assert triggers in assert_cpath_layer_ok(), where we were blowing away the circuit that conn->cpath_layer points to, then checking to see if the circ is well-formed. Backport check to make sure we dont use the cpath on a closed connection. - Prevent circuit_resume_edge_reading_helper() from trying to package inbufs for marked-for-close streams. - Don't crash on hup if your options->address has become unresolvable. - Some systems (like OS X) sometimes accept() a connection and tell you the remote host is 0.0.0.0:0. If this happens, due to some other mis-features, we get confused; so refuse the conn for now. o Bugfixes on 0.0.9.x (other): - Fix harmless but scary "Unrecognized content encoding" warn message. - Add new stream error reason: TORPROTOCOL reason means "you are not speaking a version of Tor I understand; say bye-bye to your stream." - Be willing to cache directories from up to ROUTER_MAX_AGE seconds into the future, now that we are more tolerant of skew. This resolves a bug where a Tor server would refuse to cache a directory because all the directories it gets are too far in the future; yet the Tor server never logs any complaints about clock skew. - Mac packaging magic: make man pages useable, and do not overwrite existing torrc files. - Make OS X log happily to /var/log/tor/tor.log Changes in version 0.0.9.5 - 2005-02-22 o Bugfixes on 0.0.9.x: - Fix an assert race at exit nodes when resolve requests fail. - Stop picking unverified dir mirrors--it only leads to misery. - Patch from Matt Edman to make NT services work better. Service support is still not compiled into the executable by default. - Patch from Dmitri Bely so the Tor service runs better under the win32 SYSTEM account. - Make tor-resolve actually work (?) on Win32. - Fix a sign bug when getrlimit claims to have 4+ billion file descriptors available. - Stop refusing to start when bandwidthburst == bandwidthrate. - When create cells have been on the onion queue more than five seconds, just send back a destroy and take them off the list. Changes in version 0.0.9.4 - 2005-02-03 o Bugfixes on 0.0.9: - Fix an assert bug that took down most of our servers: when a server claims to have 1 GB of bandwidthburst, don't freak out. - Don't crash as badly if we have spawned the max allowed number of dnsworkers, or we're out of file descriptors. - Block more file-sharing ports in the default exit policy. - MaxConn is now automatically set to the hard limit of max file descriptors we're allowed (ulimit -n), minus a few for logs, etc. - Give a clearer message when servers need to raise their ulimit -n when they start running out of file descriptors. - SGI Compatibility patches from Jan Schaumann. - Tolerate a corrupt cached directory better. - When a dirserver hasn't approved your server, list which one. - Go into soft hibernation after 95% of the bandwidth is used, not 99%. This is especially important for daily hibernators who have a small accounting max. Hopefully it will result in fewer cut connections when the hard hibernation starts. - Load-balance better when using servers that claim more than 800kB/s of capacity. - Make NT services work (experimental, only used if compiled in). Changes in version 0.0.9.3 - 2005-01-21 o Bugfixes on 0.0.9: - Backport the cpu use fixes from main branch, so busy servers won't need as much processor time. - Work better when we go offline and then come back, or when we run Tor at boot before the network is up. We do this by optimistically trying to fetch a new directory whenever an application request comes in and we think we're offline -- the human is hopefully a good measure of when the network is back. - Backport some minimal hidserv bugfixes: keep rend circuits open as long as you keep using them; actually publish hidserv descriptors shortly after they change, rather than waiting 20-40 minutes. - Enable Mac startup script by default. - Fix duplicate dns_cancel_pending_resolve reported by Giorgos Pallas. - When you update AllowUnverifiedNodes or FirewallPorts via the controller's setconf feature, we were always appending, never resetting. - When you update HiddenServiceDir via setconf, it was screwing up the order of reading the lines, making it fail. - Do not rewrite a cached directory back to the cache; otherwise we will think it is recent and not fetch a newer one on startup. - Workaround for webservers that lie about Content-Encoding: Tor now tries to autodetect compressed directories and compression itself. This lets us Proxypass dir fetches through apache. Changes in version 0.0.9.2 - 2005-01-04 o Bugfixes on 0.0.9 (crashes and asserts): - Fix an assert on startup when the disk is full and you're logging to a file. - If you do socks4 with an IP of 0.0.0.x but *don't* provide a socks4a style address, then we'd crash. - Fix an assert trigger when the running-routers string we get from a dirserver is broken. - Make worker threads start and run on win32. Now win32 servers may work better. - Bandaid (not actually fix, but now it doesn't crash) an assert where the dns worker dies mysteriously and the main Tor process doesn't remember anything about the address it was resolving. o Bugfixes on 0.0.9 (Win32): - Workaround for brain-damaged __FILE__ handling on MSVC: keep Nick's name out of the warning/assert messages. - Fix a superficial "unhandled error on read" bug on win32. - The win32 installer no longer requires a click-through for our license, since our Free Software license grants rights but does not take any away. - Win32: When connecting to a dirserver fails, try another one immediately. (This was already working for non-win32 Tors.) - Stop trying to parse $HOME on win32 when hunting for default DataDirectory. - Make tor-resolve.c work on win32 by calling network_init(). o Bugfixes on 0.0.9 (other): - Make 0.0.9.x build on Solaris again. - Due to a fencepost error, we were blowing away the \n when reporting confvalue items in the controller. So asking for multiple config values at once couldn't work. - When listing circuits that are pending on an opening OR connection, if we're an OR we were listing circuits that *end* at us as being pending on every listener, dns/cpu worker, etc. Stop that. - Dirservers were failing to create 'running-routers' or 'directory' strings if we had more than some threshold of routers. Fix them so they can handle any number of routers. - Fix a superficial "Duplicate mark for close" bug. - Stop checking for clock skew for OR connections, even for servers. - Fix a fencepost error that was chopping off the last letter of any nickname that is the maximum allowed nickname length. - Update URLs in log messages so they point to the new website. - Fix a potential problem in mangling server private keys while writing to disk (not triggered yet, as far as we know). - Include the licenses for other free software we include in Tor, now that we're shipping binary distributions more regularly. Changes in version 0.0.9.1 - 2004-12-15 o Bugfixes on 0.0.9: - Make hibernation actually work. - Make HashedControlPassword config option work. - When we're reporting event circuit status to a controller, don't use the stream status code. Changes in version 0.0.9 - 2004-12-12 o Cleanups: - Clean up manpage and torrc.sample file. - Clean up severities and text of log warnings. o Mistakes: - Make servers trigger an assert when they enter hibernation. Changes in version 0.0.9rc7 - 2004-12-08 o Bugfixes on 0.0.9rc: - Fix a stack-trashing crash when an exit node begins hibernating. - Avoid looking at unallocated memory while considering which ports we need to build circuits to cover. - Stop a sigpipe: when an 'end' cell races with eof from the app, we shouldn't hold-open-until-flush if the eof arrived first. - Fix a bug with init_cookie_authentication() in the controller. - When recommending new-format log lines, if the upper bound is LOG_ERR, leave it implicit. o Bugfixes on 0.0.8.1: - Fix a whole slew of memory leaks. - Fix isspace() and friends so they still make Solaris happy but also so they don't trigger asserts on win32. - Fix parse_iso_time on platforms without strptime (eg win32). - win32: tolerate extra "readable" events better. - win32: when being multithreaded, leave parent fdarray open. - Make unit tests work on win32. Changes in version 0.0.9rc6 - 2004-12-06 o Bugfixes on 0.0.9pre: - Clean up some more integer underflow opportunities (not exploitable we think). - While hibernating, hup should not regrow our listeners. - Send an end to the streams we close when we hibernate, rather than just chopping them off. - React to eof immediately on non-open edge connections. o Bugfixes on 0.0.8.1: - Calculate timeout for waiting for a connected cell from the time we sent the begin cell, not from the time the stream started. If it took a long time to establish the circuit, we would time out right after sending the begin cell. - Fix router_compare_addr_to_addr_policy: it was not treating a port of * as always matching, so we were picking reject *:* nodes as exit nodes too. Oops. o Features: - New circuit building strategy: keep a list of ports that we've used in the past 6 hours, and always try to have 2 circuits open or on the way that will handle each such port. Seed us with port 80 so web users won't complain that Tor is "slow to start up". - Make kill -USR1 dump more useful stats about circuits. - When warning about retrying or giving up, print the address, so the user knows which one it's talking about. - If you haven't used a clean circuit in an hour, throw it away, just to be on the safe side. (This means after 6 hours a totally unused Tor client will have no circuits open.) Changes in version 0.0.9rc5 - 2004-12-01 o Bugfixes on 0.0.8.1: - Disallow NDEBUG. We don't ever want anybody to turn off debug. - Let resolve conns retry/expire also, rather than sticking around forever. - If we are using select, make sure we stay within FD_SETSIZE. o Bugfixes on 0.0.9pre: - Fix integer underflow in tor_vsnprintf() that may be exploitable, but doesn't seem to be currently; thanks to Ilja van Sprundel for finding it. - If anybody set DirFetchPostPeriod, give them StatusFetchPeriod instead. Impose minima and maxima for all *Period options; impose even tighter maxima for fetching if we are a caching dirserver. Clip rather than rejecting. - Fetch cached running-routers from servers that serve it (that is, authdirservers and servers running 0.0.9rc5-cvs or later.) o Features: - Accept *:706 (silc) in default exit policy. - Implement new versioning format for post 0.1. - Support "foo.nickname.exit" addresses, to let Alice request the address "foo" as viewed by exit node "nickname". Based on a patch by Geoff Goodell. - Make tor --version --version dump the cvs Id of every file. Changes in version 0.0.9rc4 - 2004-11-28 o Bugfixes on 0.0.8.1: - Make windows sockets actually non-blocking (oops), and handle win32 socket errors better. o Bugfixes on 0.0.9rc1: - Actually catch the -USR2 signal. Changes in version 0.0.9rc3 - 2004-11-25 o Bugfixes on 0.0.8.1: - Flush the log file descriptor after we print "Tor opening log file", so we don't see those messages days later. o Bugfixes on 0.0.9rc1: - Make tor-resolve work again. - Avoid infinite loop in tor-resolve if tor hangs up on it. - Fix an assert trigger for clients/servers handling resolves. Changes in version 0.0.9rc2 - 2004-11-24 o Bugfixes on 0.0.9rc1: - I broke socks5 support while fixing the eof bug. - Allow unitless bandwidths and intervals; they default to bytes and seconds. - New servers don't start out hibernating; they are active until they run out of bytes, so they have a better estimate of how long it takes, and so their operators can know they're working. Changes in version 0.0.9rc1 - 2004-11-23 o Bugfixes on 0.0.8.1: - Finally fix a bug that's been plaguing us for a year: With high load, circuit package window was reaching 0. Whenever we got a circuit-level sendme, we were reading a lot on each socket, but only writing out a bit. So we would eventually reach eof. This would be noticed and acted on even when there were still bytes sitting in the inbuf. - When poll() is interrupted, we shouldn't believe the revents values. o Bugfixes on 0.0.9pre6: - Fix hibernate bug that caused pre6 to be broken. - Don't keep rephist info for routers that haven't had activity for 24 hours. (This matters now that clients have keys, since we track them too.) - Never call close_temp_logs while validating log options. - Fix backslash-escaping on tor.sh.in and torctl.in. o Features: - Implement weekly/monthly/daily accounting: now you specify your hibernation properties by AccountingMax N bytes|KB|MB|GB|TB AccountingStart day|week|month [day] HH:MM Defaults to "month 1 0:00". - Let bandwidth and interval config options be specified as 5 bytes, kb, kilobytes, etc; and as seconds, minutes, hours, days, weeks. - kill -USR2 now moves all logs to loglevel debug (kill -HUP to get back to normal.) - If your requested entry or exit node has advertised bandwidth 0, pick it anyway. - Be more greedy about filling up relay cells -- we try reading again once we've processed the stuff we read, in case enough has arrived to fill the last cell completely. - Apply NT service patch from Osamu Fujino. Still needs more work. Changes in version 0.0.9pre6 - 2004-11-15 o Bugfixes on 0.0.8.1: - Fix assert failure on malformed socks4a requests. - Use identity comparison, not nickname comparison, to choose which half of circuit-ID-space each side gets to use. This is needed because sometimes we think of a router as a nickname, and sometimes as a hex ID, and we can't predict what the other side will do. - Catch and ignore SIGXFSZ signals when log files exceed 2GB; our write() call will fail and we handle it there. - Add a FAST_SMARTLIST define to optionally inline smartlist_get and smartlist_len, which are two major profiling offenders. o Bugfixes on 0.0.9pre5: - Fix a bug in read_all that was corrupting config files on windows. - When we're raising the max number of open file descriptors to 'unlimited', don't log that we just raised it to '-1'. - Include event code with events, as required by control-spec.txt. - Don't give a fingerprint when clients do --list-fingerprint: it's misleading, because it will never be the same again. - Stop using strlcpy in tor_strndup, since it was slowing us down a lot. - Remove warn on startup about missing cached-directory file. - Make kill -USR1 work again. - Hibernate if we start tor during the "wait for wakeup-time" phase of an accounting interval. Log our hibernation plans better. - Authoritative dirservers now also cache their directory, so they have it on start-up. o Features: - Fetch running-routers; cache running-routers; compress running-routers; serve compressed running-routers.z - Add NSI installer script contributed by J Doe. - Commit VC6 and VC7 workspace/project files. - Commit a tor.spec for making RPM files, with help from jbash. - Add contrib/torctl.in contributed by Glenn Fink. - Implement the control-spec's SAVECONF command, to write your configuration to torrc. - Get cookie authentication for the controller closer to working. - Include control-spec.txt in the tarball. - When set_conf changes our server descriptor, upload a new copy. But don't upload it too often if there are frequent changes. - Document authentication config in man page, and document signals we catch. - Clean up confusing parts of man page and torrc.sample. - Make expand_filename handle ~ and ~username. - Use autoconf to enable largefile support where necessary. Use ftello where available, since ftell can fail at 2GB. - Distinguish between TOR_TLS_CLOSE and TOR_TLS_ERROR, so we can log more informatively. - Give a slightly more useful output for "tor -h". - Refuse application socks connections to port 0. - Check clock skew for verified servers, but allow unverified servers and clients to have any clock skew. - Break DirFetchPostPeriod into: - DirFetchPeriod for fetching full directory, - StatusFetchPeriod for fetching running-routers, - DirPostPeriod for posting server descriptor, - RendPostPeriod for posting hidden service descriptors. - Make sure the hidden service descriptors are at a random offset from each other, to hinder linkability. Changes in version 0.0.9pre5 - 2004-11-09 o Bugfixes on 0.0.9pre4: - Fix a seg fault in unit tests (doesn't affect main program). - Fix an assert bug where a hidden service provider would fail if the first hop of his rendezvous circuit was down. - Hidden service operators now correctly handle version 1 style INTRODUCE1 cells (nobody generates them still, so not a critical bug). - If do_hup fails, actually notice. - Handle more errnos from accept() without closing the listener. Some OpenBSD machines were closing their listeners because they ran out of file descriptors. - Send resolve cells to exit routers that are running a new enough version of the resolve code to work right. - Better handling of winsock includes on non-MSV win32 compilers. - Some people had wrapped their tor client/server in a script that would restart it whenever it died. This did not play well with our "shut down if your version is obsolete" code. Now people don't fetch a new directory if their local cached version is recent enough. - Make our autogen.sh work on ksh as well as bash. o Major Features: - Hibernation: New config option "AccountingMaxKB" lets you set how many KBytes per month you want to allow your server to consume. Rather than spreading those bytes out evenly over the month, we instead hibernate for some of the month and pop up at a deterministic time, work until the bytes are consumed, then hibernate again. Config option "MonthlyAccountingStart" lets you specify which day of the month your billing cycle starts on. - Control interface: a separate program can now talk to your client/server over a socket, and get/set config options, receive notifications of circuits and streams starting/finishing/dying, bandwidth used, etc. The next step is to get some GUIs working. Let us know if you want to help out. See doc/control-spec.txt . - Ship a contrib/tor-control.py as an example script to interact with the control port. - "tor --hash-password zzyxz" will output a salted password for use in authenticating to the control interface. - New log format in config: "Log minsev[-maxsev] stdout|stderr|syslog" or "Log minsev[-maxsev] file /var/foo" o Minor Features: - DirPolicy config option, to let people reject incoming addresses from their dirserver. - "tor --list-fingerprint" will list your identity key fingerprint and then exit. - Add "pass" target for RedirectExit, to make it easier to break out of a sequence of RedirectExit rules. - Clients now generate a TLS cert too, in preparation for having them act more like real nodes. - Ship src/win32/ in the tarball, so people can use it to build. - Make old win32 fall back to CWD if SHGetSpecialFolderLocation is broken. - New "router-status" line in directory, to better bind each verified nickname to its identity key. - Deprecate unofficial config option abbreviations, and abbreviations not on the command line. - Add a pure-C tor-resolve implementation. - Use getrlimit and friends to ensure we can reach MaxConn (currently 1024) file descriptors. o Code security improvements, inspired by Ilja: - Replace sprintf with snprintf. (I think they were all safe, but hey.) - Replace strcpy/strncpy with strlcpy in more places. - Avoid strcat; use snprintf or strlcat instead. - snprintf wrapper with consistent (though not C99) overflow behavior. Changes in version 0.0.9pre4 - 2004-10-17 o Bugfixes on 0.0.9pre3: - If the server doesn't specify an exit policy, use the real default exit policy, not reject *:*. - Ignore fascistfirewall when uploading/downloading hidden service descriptors, since we go through Tor for those; and when using an HttpProxy, since we assume it can reach them all. - When looking for an authoritative dirserver, use only the ones configured at boot. Don't bother looking in the directory. - The rest of the fix for get_default_conf_file() on older win32. - Make 'Routerfile' config option obsolete. o Features: - New 'MyFamily nick1,...' config option for a server to specify other servers that shouldn't be used in the same circuit with it. Only believed if nick1 also specifies us. - New 'NodeFamily nick1,nick2,...' config option for a client to specify nodes that it doesn't want to use in the same circuit. - New 'Redirectexit pattern address:port' config option for a server to redirect exit connections, e.g. to a local squid. Changes in version 0.0.9pre3 - 2004-10-13 o Bugfixes on 0.0.8.1: - Better torrc example lines for dirbindaddress and orbindaddress. - Improved bounds checking on parsed ints (e.g. config options and the ones we find in directories.) - Better handling of size_t vs int, so we're more robust on 64 bit platforms. - Fix the rest of the bug where a newly started OR would appear as unverified even after we've added his fingerprint and hupped the dirserver. - Fix a bug from 0.0.7: when read() failed on a stream, we would close it without sending back an end. So 'connection refused' would simply be ignored and the user would get no response. o Bugfixes on 0.0.9pre2: - Serving the cached-on-disk directory to people is bad. We now provide no directory until we've fetched a fresh one. - Workaround for bug on windows where cached-directories get crlf corruption. - Make get_default_conf_file() work on older windows too. - If we write a *:* exit policy line in the descriptor, don't write any more exit policy lines. o Features: - Use only 0.0.9pre1 and later servers for resolve cells. - Make the dirservers file obsolete. - Include a dir-signing-key token in directories to tell the parsing entity which key is being used to sign. - Remove the built-in bulky default dirservers string. - New config option "Dirserver %s:%d [fingerprint]", which can be repeated as many times as needed. If no dirservers specified, default to moria1,moria2,tor26. - Make moria2 advertise a dirport of 80, so people behind firewalls will be able to get a directory. - Http proxy support - Dirservers translate requests for http://%s:%d/x to /x - You can specify "HttpProxy %s[:%d]" and all dir fetches will be routed through this host. - Clients ask for /tor/x rather than /x for new enough dirservers. This way we can one day coexist peacefully with apache. - Clients specify a "Host: %s%d" http header, to be compatible with more proxies, and so running squid on an exit node can work. Changes in version 0.0.8.1 - 2004-10-13 o Bugfixes: - Fix a seg fault that can be triggered remotely for Tor clients/servers with an open dirport. - Fix a rare assert trigger, where routerinfos for entries in our cpath would expire while we're building the path. - Fix a bug in OutboundBindAddress so it (hopefully) works. - Fix a rare seg fault for people running hidden services on intermittent connections. - Fix a bug in parsing opt keywords with objects. - Fix a stale pointer assert bug when a stream detaches and reattaches. - Fix a string format vulnerability (probably not exploitable) in reporting stats locally. - Fix an assert trigger: sometimes launching circuits can fail immediately, e.g. because too many circuits have failed recently. - Fix a compile warning on 64 bit platforms. Changes in version 0.0.9pre2 - 2004-10-03 o Bugfixes: - Make fetching a cached directory work for 64-bit platforms too. - Make zlib.h a required header, not an optional header. Changes in version 0.0.9pre1 - 2004-10-01 o Bugfixes: - Stop using separate defaults for no-config-file and empty-config-file. Now you have to explicitly turn off SocksPort, if you don't want it open. - Fix a bug in OutboundBindAddress so it (hopefully) works. - Improve man page to mention more of the 0.0.8 features. - Fix a rare seg fault for people running hidden services on intermittent connections. - Change our file IO stuff (especially wrt OpenSSL) so win32 is happier. - Fix more dns related bugs: send back resolve_failed and end cells more reliably when the resolve fails, rather than closing the circuit and then trying to send the cell. Also attach dummy resolve connections to a circuit *before* calling dns_resolve(), to fix a bug where cached answers would never be sent in RESOLVED cells. - When we run out of disk space, or other log writing error, don't crash. Just stop logging to that log and continue. - We were starting to daemonize before we opened our logs, so if there were any problems opening logs, we would complain to stderr, which wouldn't work, and then mysteriously exit. - Fix a rare bug where sometimes a verified OR would connect to us before he'd uploaded his descriptor, which would cause us to assign conn->nickname as though he's unverified. Now we look through the fingerprint list to see if he's there. - Fix a rare assert trigger, where routerinfos for entries in our cpath would expire while we're building the path. o Features: - Clients can ask dirservers for /dir.z to get a compressed version of the directory. Only works for servers running 0.0.9, of course. - Make clients cache directories and use them to seed their router lists at startup. This means clients have a datadir again. - Configuration infrastructure support for warning on obsolete options. - Respond to content-encoding headers by trying to uncompress as appropriate. - Reply with a deflated directory when a client asks for "dir.z". We could use allow-encodings instead, but allow-encodings isn't specified in HTTP 1.0. - Raise the max dns workers from 50 to 100. - Discourage people from setting their dirfetchpostperiod more often than once per minute. - Protect dirservers from overzealous descriptor uploading -- wait 10 seconds after directory gets dirty, before regenerating. Changes in version 0.0.8 - 2004-08-25 o Port it to SunOS 5.9 / Athena Changes in version 0.0.8rc2 - 2004-08-20 o Make it compile on cygwin again. o When picking unverified routers, skip those with low uptime and/or low bandwidth, depending on what properties you care about. Changes in version 0.0.8rc1 - 2004-08-18 o Changes from 0.0.7.3: - Bugfixes: - Fix assert triggers: if the other side returns an address 0.0.0.0, don't put it into the client dns cache. - If a begin failed due to exit policy, but we believe the IP address should have been allowed, switch that router to exitpolicy reject *:* until we get our next directory. - Features: - Clients choose nodes proportional to advertised bandwidth. - Avoid using nodes with low uptime as introduction points. - Handle servers with dynamic IP addresses: don't replace options->Address with the resolved one at startup, and detect our address right before we make a routerinfo each time. - 'FascistFirewall' option to pick dirservers and ORs on specific ports; plus 'FirewallPorts' config option to tell FascistFirewall which ports are open. (Defaults to 80,443) - Be more aggressive about trying to make circuits when the network has changed (e.g. when you unsuspend your laptop). - Check for time skew on http headers; report date in response to "GET /". - If the entrynode config line has only one node, don't pick it as an exitnode. - Add strict{entry|exit}nodes config options. If set to 1, then we refuse to build circuits that don't include the specified entry or exit nodes. - OutboundBindAddress config option, to bind to a specific IP address for outgoing connect()s. - End truncated log entries (e.g. directories) with "[truncated]". o Patches to 0.0.8preX: - Bugfixes: - Patches to compile and run on win32 again (maybe)? - Fix crash when looking for ~/.torrc with no $HOME set. - Fix a race bug in the unit tests. - Handle verified/unverified name collisions better when new routerinfo's arrive in a directory. - Sometimes routers were getting entered into the stats before we'd assigned their identity_digest. Oops. - Only pick and establish intro points after we've gotten a directory. - Features: - AllowUnverifiedNodes config option to let circuits choose no-name routers in entry,middle,exit,introduction,rendezvous positions. Allow middle and rendezvous positions by default. - Add a man page for tor-resolve. Changes in version 0.0.7.3 - 2004-08-12 o Stop dnsworkers from triggering an assert failure when you ask them to resolve the host "". Changes in version 0.0.8pre3 - 2004-08-09 o Changes from 0.0.7.2: - Allow multiple ORs with same nickname in routerlist -- now when people give us one identity key for a nickname, then later another, we don't constantly complain until the first expires. - Remember used bandwidth (both in and out), and publish 15-minute snapshots for the past day into our descriptor. - You can now fetch $DIRURL/running-routers to get just the running-routers line, not the whole descriptor list. (But clients don't use this yet.) - When people mistakenly use Tor as an http proxy, point them at the tor-doc.html rather than the INSTALL. - Remove our mostly unused -- and broken -- hex_encode() function. Use base16_encode() instead. (Thanks to Timo Lindfors for pointing out this bug.) - Rotate onion keys every 12 hours, not every 2 hours, so we have fewer problems with people using the wrong key. - Change the default exit policy to reject the default edonkey, kazaa, gnutella ports. - Add replace_file() to util.[ch] to handle win32's rename(). o Changes from 0.0.8preX: - Fix two bugs in saving onion keys to disk when rotating, so hopefully we'll get fewer people using old onion keys. - Fix an assert error that was making SocksPolicy not work. - Be willing to expire routers that have an open dirport -- it's just the authoritative dirservers we want to not forget. - Reject tor-resolve requests for .onion addresses early, so we don't build a whole rendezvous circuit and then fail. - When you're warning a server that he's unverified, don't cry wolf unpredictably. - Fix a race condition: don't try to extend onto a connection that's still handshaking. - For servers in clique mode, require the conn to be open before you'll choose it for your path. - Fix some cosmetic bugs about duplicate mark-for-close, lack of end relay cell, etc. - Measure bandwidth capacity over the last 24 hours, not just 12 - Bugfix: authoritative dirservers were making and signing a new directory for each client, rather than reusing the cached one. Changes in version 0.0.8pre2 - 2004-08-04 o Changes from 0.0.7.2: - Security fixes: - Check directory signature _before_ you decide whether you're you're running an obsolete version and should exit. - Check directory signature _before_ you parse the running-routers list to decide who's running or verified. - Bugfixes and features: - Check return value of fclose while writing to disk, so we don't end up with broken files when servers run out of disk space. - Log a warning if the user uses an unsafe socks variant, so people are more likely to learn about privoxy or socat. - Dirservers now include RFC1123-style dates in the HTTP headers, which one day we will use to better detect clock skew. o Changes from 0.0.8pre1: - Make it compile without warnings again on win32. - Log a warning if you're running an unverified server, to let you know you might want to get it verified. - Only pick a default nickname if you plan to be a server. Changes in version 0.0.8pre1 - 2004-07-23 o Bugfixes: - Made our unit tests compile again on OpenBSD 3.5, and tor itself compile again on OpenBSD on a sparc64. - We were neglecting milliseconds when logging on win32, so everything appeared to happen at the beginning of each second. o Protocol changes: - 'Extend' relay cell payloads now include the digest of the intended next hop's identity key. Now we can verify that we're extending to the right router, and also extend to routers we hadn't heard of before. o Features: - Tor nodes can now act as relays (with an advertised ORPort) without being manually verified by the dirserver operators. - Uploaded descriptors of unverified routers are now accepted by the dirservers, and included in the directory. - Verified routers are listed by nickname in the running-routers list; unverified routers are listed as "$". - We now use hash-of-identity-key in most places rather than nickname or addr:port, for improved security/flexibility. - To avoid Sybil attacks, paths still use only verified servers. But now we have a chance to play around with hybrid approaches. - Nodes track bandwidth usage to estimate capacity (not used yet). - ClientOnly option for nodes that never want to become servers. - Directory caching. - "AuthoritativeDir 1" option for the official dirservers. - Now other nodes (clients and servers) will cache the latest directory they've pulled down. - They can enable their DirPort to serve it to others. - Clients will pull down a directory from any node with an open DirPort, and check the signature/timestamp correctly. - Authoritative dirservers now fetch directories from other authdirservers, to stay better synced. - Running-routers list tells who's down also, along with noting if they're verified (listed by nickname) or unverified (listed by hash-of-key). - Allow dirservers to serve running-router list separately. This isn't used yet. - ORs connect-on-demand to other ORs - If you get an extend cell to an OR you're not connected to, connect, handshake, and forward the create cell. - The authoritative dirservers stay connected to everybody, and everybody stays connected to 0.0.7 servers, but otherwise clients/servers expire unused connections after 5 minutes. - When servers get a sigint, they delay 30 seconds (refusing new connections) then exit. A second sigint causes immediate exit. - File and name management: - Look for .torrc if no CONFDIR "torrc" is found. - If no datadir is defined, then choose, make, and secure ~/.tor as datadir. - If torrc not found, exitpolicy reject *:*. - Expands ~/ in filenames to $HOME/ (but doesn't yet expand ~arma). - If no nickname is defined, derive default from hostname. - Rename secret key files, e.g. identity.key -> secret_id_key, to discourage people from mailing their identity key to tor-ops. - Refuse to build a circuit before the directory has arrived -- it won't work anyway, since you won't know the right onion keys to use. - Try other dirservers immediately if the one you try is down. This should tolerate down dirservers better now. - Parse tor version numbers so we can do an is-newer-than check rather than an is-in-the-list check. - New socks command 'resolve', to let us shim gethostbyname() locally. - A 'tor_resolve' script to access the socks resolve functionality. - A new socks-extensions.txt doc file to describe our interpretation and extensions to the socks protocols. - Add a ContactInfo option, which gets published in descriptor. - Publish OR uptime in descriptor (and thus in directory) too. - Write tor version at the top of each log file - New docs in the tarball: - tor-doc.html. - Document that you should proxy your SSL traffic too. Changes in version 0.0.7.2 - 2004-07-07 o A better fix for the 0.0.0.0 problem, that will hopefully eliminate the remaining related assertion failures. Changes in version 0.0.7.1 - 2004-07-04 o When an address resolves to 0.0.0.0, treat it as a failed resolve, since internally we use 0.0.0.0 to signify "not yet resolved". Changes in version 0.0.7 - 2004-06-07 o Updated the man page to reflect the new features. Changes in version 0.0.7rc2 - 2004-06-06 o Changes from 0.0.7rc1: - Make it build on Win32 again. o Changes from 0.0.6.2: - Rotate dnsworkers and cpuworkers on SIGHUP, so they get new config settings too. Changes in version 0.0.7rc1 - 2004-06-02 o Bugfixes: - On sighup, we were adding another log without removing the first one. So log messages would get duplicated n times for n sighups. - Several cases of using a connection after we'd freed it. The problem was that connections that are pending resolve are in both the pending_resolve tree, and also the circuit's resolving_streams list. When you want to remove one, you must remove it from both. - Fix a double-mark-for-close where an end cell arrived for a resolving stream, and then the resolve failed. - Check directory signatures based on name of signer, not on whom we got the directory from. This will let us cache directories more easily. o Features: - Crank up some of our constants to handle more users. Changes in version 0.0.7pre1 - 2004-06-02 o Fixes for crashes and other obnoxious bugs: - Fix an epipe bug: sometimes when directory connections failed to connect, we would give them a chance to flush before closing them. - When we detached from a circuit because of resolvefailed, we would immediately try the same circuit twice more, and then give up on the resolve thinking we'd tried three different exit nodes. - Limit the number of intro circuits we'll attempt to build for a hidden service per 15-minute period. - Check recommended-software string *early*, before actually parsing the directory. Thus we can detect an obsolete version and exit, even if the new directory format doesn't parse. o Fixes for security bugs: - Remember which nodes are dirservers when you startup, and if a random OR enables his dirport, don't automatically assume he's a trusted dirserver. o Other bugfixes: - Directory connections were asking the wrong poll socket to start writing, and not asking themselves to start writing. - When we detached from a circuit because we sent a begin but didn't get a connected, we would use it again the first time; but after that we would correctly switch to a different one. - Stop warning when the first onion decrypt attempt fails; they will sometimes legitimately fail now that we rotate keys. - Override unaligned-access-ok check when $host_cpu is ia64 or arm. Apparently they allow it but the kernel whines. - Dirservers try to reconnect periodically too, in case connections have failed. - Fix some memory leaks in directory servers. - Allow backslash in Win32 filenames. - Made Tor build complain-free on FreeBSD, hopefully without breaking other BSD builds. We'll see. o Features: - Doxygen markup on all functions and global variables. - Make directory functions update routerlist, not replace it. So now directory disagreements are not so critical a problem. - Remove the upper limit on number of descriptors in a dirserver's directory (not that we were anywhere close). - Allow multiple logfiles at different severity ranges. - Allow *BindAddress to specify ":port" rather than setting *Port separately. Allow multiple instances of each BindAddress config option, so you can bind to multiple interfaces if you want. - Allow multiple exit policy lines, which are processed in order. Now we don't need that huge line with all the commas in it. - Enable accept/reject policies on SOCKS connections, so you can bind to 0.0.0.0 but still control who can use your OP. Changes in version 0.0.6.2 - 2004-05-16 o Our integrity-checking digest was checking only the most recent cell, not the previous cells like we'd thought. Thanks to Stefan Mark for finding the flaw! Changes in version 0.0.6.1 - 2004-05-06 o Fix two bugs in our AES counter-mode implementation (this affected onion-level stream encryption, but not TLS-level). It turns out we were doing something much more akin to a 16-character polyalphabetic cipher. Oops. Thanks to Stefan Mark for finding the flaw! o Retire moria3 as a directory server, and add tor26 as a directory server. Changes in version 0.0.6 - 2004-05-02 [version bump only] Changes in version 0.0.6rc4 - 2004-05-01 o Update the built-in dirservers list to use the new directory format o Fix a rare seg fault: if a node offering a hidden service attempts to build a circuit to Alice's rendezvous point and fails before it reaches the last hop, it retries with a different circuit, but then dies. o Handle windows socket errors correctly. Changes in version 0.0.6rc3 - 2004-04-28 o Don't expire non-general excess circuits (if we had enough circuits open, we were expiring rendezvous circuits -- even when they had a stream attached. oops.) o Fetch randomness from /dev/urandom better (not via fopen/fread) o Better debugging for tls errors o Some versions of openssl have an SSL_pending function that erroneously returns bytes when there is a non-application record pending. o Set Content-Type on the directory and hidserv descriptor. o Remove IVs from cipher code, since AES-ctr has none. o Win32 fixes. Tor now compiles on win32 with no warnings/errors. o We were using an array of length zero in a few places. o win32's gethostbyname can't resolve an IP to an IP. o win32's close can't close a socket. Changes in version 0.0.6rc2 - 2004-04-26 o Fix a bug where we were closing tls connections intermittently. It turns out openssl keeps its errors around -- so if an error happens, and you don't ask about it, and then another openssl operation happens and succeeds, and you ask if there was an error, it tells you about the first error. Fun fun. o Fix a bug that's been lurking since 27 may 03 (!) When passing back a destroy cell, we would use the wrong circ id. 'Mostly harmless', but still worth fixing. o Since we don't support truncateds much, don't bother sending them; just close the circ. o check for so we build on NetBSD again (I hope). o don't crash if a conn that sent a begin has suddenly lost its circuit (this was quite rare). Changes in version 0.0.6rc1 - 2004-04-25 o We now rotate link (tls context) keys and onion keys. o CREATE cells now include oaep padding, so you can tell if you decrypted them correctly. o Add bandwidthburst to server descriptor. o Directories now say which dirserver signed them. o Use a tor_assert macro that logs failed assertions too. Changes in version 0.0.6pre5 - 2004-04-18 o changes from 0.0.6pre4: - make tor build on broken freebsd 5.2 installs - fix a failed assert when you try an intro point, get a nack, and try a second one and it works. - when alice uses a port that the hidden service doesn't accept, it now sends back an end cell (denied by exit policy). otherwise alice would just have to wait to time out. - fix another rare bug: when we had tried all the intro points for a hidden service, we fetched the descriptor again, but we left our introcirc thinking it had already sent an intro, so it kept waiting for a response... - bugfix: when you sleep your hidden-service laptop, as soon as it wakes up it tries to upload a service descriptor, but socketpair fails for some reason (localhost not up yet?). now we simply give up on that upload, and we'll try again later. i'd still like to find the bug though. - if an intro circ waiting for an ack dies before getting one, then count it as a nack - we were reusing stale service descriptors and refetching usable ones. oops. Changes in version 0.0.6pre4 - 2004-04-14 o changes from 0.0.6pre3: - when bob fails to connect to the rendezvous point, and his circ didn't fail because of the rendezvous point itself, then he retries a couple of times - we expire introduction and rendezvous circs more thoroughly (sometimes they were hanging around forever) - we expire unattached rendezvous streams that have been around too long (they were sticking around forever). - fix a measly fencepost error that was crashing everybody with a strict glibc. Changes in version 0.0.6pre3 - 2004-04-14 o changes from 0.0.6pre2: - make hup work again - fix some memory leaks for dirservers - allow more skew in rendezvous descriptor timestamps, to help handle people like blanu who don't know what time it is - normal circs are 3 hops, but some rend/intro circs are 4, if the initiator doesn't get to choose the last hop - send acks for introductions, so alice can know whether to try again - bob publishes intro points more correctly o changes from 0.0.5: - fix an assert trigger that's been plaguing us since the days of 0.0.2prexx (thanks weasel!) - retry stream correctly when we fail to connect because of exit-policy-reject (should try another) or can't-resolve-address (also should try another, because dns on random internet servers is flaky). - when we hup a dirserver and we've *removed* a server from the approved-routers list, now we remove that server from the in-memory directories too Changes in version 0.0.6pre2 - 2004-04-08 o We fixed our base32 implementation. Now it works on all architectures. Changes in version 0.0.6pre1 - 2004-04-08 o Features: - Hidden services and rendezvous points are implemented. Go to http://6sxoyfb3h2nvok2d.onion/ for an index of currently available hidden services. (This only works via a socks4a proxy such as Privoxy, and currently it's quite slow.) Changes in version 0.0.5 - 2004-03-30 [version bump only] Changes in version 0.0.5rc3 - 2004-03-29 o Install torrc as torrc.sample -- we no longer clobber your torrc. (Woo!) o Re-enable recommendedversion checking (we broke it in rc2, oops) o Add in a 'notice' log level for things the operator should hear but that aren't warnings Changes in version 0.0.5rc2 - 2004-03-29 o Hold socks connection open until reply is flushed (if possible) o Make exit nodes resolve IPs to IPs immediately, rather than asking the dns farm to do it. o Fix c99 aliasing warnings in rephist.c o Don't include server descriptors that are older than 24 hours in the directory. o Give socks 'reject' replies their whole 15s to attempt to flush, rather than seeing the 60s timeout and assuming the flush had failed. o Clean automake droppings from the cvs repository Changes in version 0.0.5rc1 - 2004-03-28 o Fix mangled-state bug in directory fetching (was causing sigpipes). o Only build circuits after we've fetched the directory: clients were using only the directory servers before they'd fetched a directory. This also means longer startup time; so it goes. o Fix an assert trigger where an OP would fail to handshake, and we'd expect it to have a nickname. o Work around a tsocks bug: do a socks reject when AP connection dies early, else tsocks goes into an infinite loop. Changes in version 0.0.4 - 2004-03-26 o When connecting to a dirserver or OR and the network is down, we would crash. Changes in version 0.0.3 - 2004-03-26 o Warn and fail if server chose a nickname with illegal characters o Port to Solaris and Sparc: - include missing header fcntl.h - have autoconf find -lsocket -lnsl automatically - deal with hardware word alignment - make uname() work (solaris has a different return convention) - switch from using signal() to sigaction() o Preliminary work on reputation system: - Keep statistics on success/fail of connect attempts; they're published by kill -USR1 currently. - Add a RunTesting option to try to learn link state by creating test circuits, even when SocksPort is off. - Remove unused open circuits when there are too many. Changes in version 0.0.2 - 2004-03-19 - Include strlcpy and strlcat for safer string ops - define INADDR_NONE so we compile (but still not run) on solaris Changes in version 0.0.2pre27 - 2004-03-14 o Bugfixes: - Allow internal tor networks (we were rejecting internal IPs, now we allow them if they're set explicitly). - And fix a few endian issues. Changes in version 0.0.2pre26 - 2004-03-14 o New features: - If a stream times out after 15s without a connected cell, don't try that circuit again: try a new one. - Retry streams at most 4 times. Then give up. - When a dirserver gets a descriptor from an unknown router, it logs its fingerprint (so the dirserver operator can choose to accept it even without mail from the server operator). - Inform unapproved servers when we reject their descriptors. - Make tor build on Windows again. It works as a client, who knows about as a server. - Clearer instructions in the torrc for how to set up a server. - Be more efficient about reading fd's when our global token bucket (used for rate limiting) becomes empty. o Bugfixes: - Stop asserting that computers always go forward in time. It's simply not true. - When we sent a cell (e.g. destroy) and then marked an OR connection expired, we might close it before finishing a flush if the other side isn't reading right then. - Don't allow dirservers to start if they haven't defined RecommendedVersions - We were caching transient dns failures. Oops. - Prevent servers from publishing an internal IP as their address. - Address a strcat vulnerability in circuit.c Changes in version 0.0.2pre25 - 2004-03-04 o New features: - Put the OR's IP in its router descriptor, not its fqdn. That way we'll stop being stalled by gethostbyname for nodes with flaky dns, e.g. poblano. o Bugfixes: - If the user typed in an address that didn't resolve, the server crashed. Changes in version 0.0.2pre24 - 2004-03-03 o Bugfixes: - Fix an assertion failure in dns.c, where we were trying to dequeue a pending dns resolve even if it wasn't pending - Fix a spurious socks5 warning about still trying to write after the connection is finished. - Hold certain marked_for_close connections open until they're finished flushing, rather than losing bytes by closing them too early. - Correctly report the reason for ending a stream - Remove some duplicate calls to connection_mark_for_close - Put switch_id and start_daemon earlier in the boot sequence, so it will actually try to chdir() to options.DataDirectory - Make 'make test' exit(1) if a test fails; fix some unit tests - Make tor fail when you use a config option it doesn't know about, rather than warn and continue. - Make --version work - Bugfixes on the rpm spec file and tor.sh, so it's more up to date Changes in version 0.0.2pre23 - 2004-02-29 o New features: - Print a statement when the first circ is finished, so the user knows it's working. - If a relay cell is unrecognized at the end of the circuit, send back a destroy. (So attacks to mutate cells are more clearly thwarted.) - New config option 'excludenodes' to avoid certain nodes for circuits. - When it daemonizes, it chdir's to the DataDirectory rather than "/", so you can collect coredumps there. o Bugfixes: - Fix a bug in tls flushing where sometimes data got wedged and didn't flush until more data got sent. Hopefully this bug was a big factor in the random delays we were seeing. - Make 'connected' cells include the resolved IP, so the client dns cache actually gets populated. - Disallow changing from ORPort=0 to ORPort>0 on hup. - When we time-out on a stream and detach from the circuit, send an end cell down it first. - Only warn about an unknown router (in exitnodes, entrynodes, excludenodes) after we've fetched a directory. Changes in version 0.0.2pre22 - 2004-02-26 o New features: - Servers publish less revealing uname information in descriptors. - More memory tracking and assertions, to crash more usefully when errors happen. - If the default torrc isn't there, just use some default defaults. Plus provide an internal dirservers file if they don't have one. - When the user tries to use Tor as an http proxy, give them an http 501 failure explaining that we're a socks proxy. - Dump a new router.desc on hup, to help confused people who change their exit policies and then wonder why router.desc doesn't reflect it. - Clean up the generic tor.sh init script that we ship with. o Bugfixes: - If the exit stream is pending on the resolve, and a destroy arrives, then the stream wasn't getting removed from the pending list. I think this was the one causing recent server crashes. - Use a more robust poll on OSX 10.3, since their poll is flaky. - When it couldn't resolve any dirservers, it was useless from then on. Now it reloads the RouterFile (or default dirservers) if it has no dirservers. - Move the 'tor' binary back to /usr/local/bin/ -- it turns out many users don't even *have* a /usr/local/sbin/. Changes in version 0.0.2pre21 - 2004-02-18 o New features: - There's a ChangeLog file that actually reflects the changelog. - There's a 'torify' wrapper script, with an accompanying tor-tsocks.conf, that simplifies the process of using tsocks for tor. It even has a man page. - The tor binary gets installed to sbin rather than bin now. - Retry streams where the connected cell hasn't arrived in 15 seconds - Clean up exit policy handling -- get the default out of the torrc, so we can update it without forcing each server operator to fix his/her torrc. - Allow imaps and pop3s in default exit policy o Bugfixes: - Prevent picking middleman nodes as the last node in the circuit Changes in version 0.0.2pre20 - 2004-01-30 o New features: - We now have a deb package, and it's in debian unstable. Go to it, apt-getters. :) - I've split the TotalBandwidth option into BandwidthRate (how many bytes per second you want to allow, long-term) and BandwidthBurst (how many bytes you will allow at once before the cap kicks in). This better token bucket approach lets you, say, set BandwidthRate to 10KB/s and BandwidthBurst to 10MB, allowing good performance while not exceeding your monthly bandwidth quota. - Push out a tls record's worth of data once you've got it, rather than waiting until you've read everything waiting to be read. This may improve performance by pipelining better. We'll see. - Add an AP_CONN_STATE_CONNECTING state, to allow streams to detach from failed circuits (if they haven't been connected yet) and attach to new ones. - Expire old streams that haven't managed to connect. Some day we'll have them reattach to new circuits instead. o Bugfixes: - Fix several memory leaks that were causing servers to become bloated after a while. - Fix a few very rare assert triggers. A few more remain. - Setuid to User _before_ complaining about running as root. Changes in version 0.0.2pre19 - 2004-01-07 o Bugfixes: - Fix deadlock condition in dns farm. We were telling a child to die by closing the parent's file descriptor to him. But newer children were inheriting the open file descriptor from the parent, and since they weren't closing it, the socket never closed, so the child never read eof, so he never knew to exit. Similarly, dns workers were holding open other sockets, leading to all sorts of chaos. - New cleaner daemon() code for forking and backgrounding. - If you log to a file, it now prints an entry at the top of the logfile so you know it's working. - The onionskin challenge length was 30 bytes longer than necessary. - Started to patch up the spec so it's not quite so out of date. Changes in version 0.0.2pre18 - 2004-01-02 o Bugfixes: - Fix endian issues with the 'integrity' field in the relay header. - Fix a potential bug where connections in state AP_CONN_STATE_CIRCUIT_WAIT might unexpectedly ask to write. Changes in version 0.0.2pre17 - 2003-12-30 o Bugfixes: - Made --debuglogfile (or any second log file, actually) work. - Resolved an edge case in get_unique_circ_id_by_conn where a smart adversary could force us into an infinite loop. o Features: - Each onionskin handshake now includes a hash of the computed key, to prove the server's identity and help perfect forward secrecy. - Changed cell size from 256 to 512 bytes (working toward compatibility with MorphMix). - Changed cell length to 2 bytes, and moved it to the relay header. - Implemented end-to-end integrity checking for the payloads of relay cells. - Separated streamid from 'recognized' (otherwise circuits will get messed up when we try to have streams exit from the middle). We use the integrity-checking to confirm that a cell is addressed to this hop. - Randomize the initial circid and streamid values, so an adversary who breaks into a node can't learn how many circuits or streams have been made so far. Changes in version 0.0.2pre16 - 2003-12-14 o Bugfixes: - Fixed a bug that made HUP trigger an assert - Fixed a bug where a circuit that immediately failed wasn't being counted as a failed circuit in counting retries. o Features: - Now we close the circuit when we get a truncated cell: otherwise we're open to an anonymity attack where a bad node in the path truncates the circuit and then we open streams at him. - Add port ranges to exit policies - Add a conservative default exit policy - Warn if you're running tor as root - on HUP, retry OR connections and close/rebind listeners - options.EntryNodes: try these nodes first when picking the first node - options.ExitNodes: if your best choices happen to include any of your preferred exit nodes, you choose among just those preferred exit nodes. - options.ExcludedNodes: nodes that are never picked in path building Changes in version 0.0.2pre15 - 2003-12-03 o Robustness and bugfixes: - Sometimes clients would cache incorrect DNS resolves, which would really screw things up. - An OP that goes offline would slowly leak all its sockets and stop working. - A wide variety of bugfixes in exit node selection, exit policy handling, and processing pending streams when a new circuit is established. - Pick nodes for a path only from those the directory says are up - Choose randomly from all running dirservers, not always the first one - Increase allowed http header size for directory fetch. - Stop writing to stderr (if we're daemonized it will be closed). - Enable -g always, so cores will be more useful to me. - Switch "-lcrypto -lssl" to "-lssl -lcrypto" for broken distributions. o Documentation: - Wrote a man page. It lists commonly used options. o Configuration: - Change default loglevel to warn. - Make PidFile default to null rather than littering in your CWD. - OnionRouter config option is now obsolete. Instead it just checks ORPort>0. - Moved to a single unified torrc file for both clients and servers. Changes in version 0.0.2pre14 - 2003-11-29 o Robustness and bugfixes: - Force the admin to make the DataDirectory himself - to get ownership/permissions right - so clients no longer make a DataDirectory and then never use it - fix bug where a client who was offline for 45 minutes would never pull down a directory again - fix (or at least hide really well) the dns assert bug that was causing server crashes - warnings and improved robustness wrt clockskew for certs - use the native daemon(3) to daemonize, when available - exit if bind() fails - exit if neither socksport nor orport is defined - include our own tor_timegm (Win32 doesn't have its own) - bugfix for win32 with lots of connections - fix minor bias in PRNG - make dirserver more robust to corrupt cached directory o Documentation: - Wrote the design document (woo) o Circuit building and exit policies: - Circuits no longer try to use nodes that the directory has told them are down. - Exit policies now support bitmasks (18.0.0.0/255.0.0.0) and bitcounts (18.0.0.0/8). - Make AP connections standby for a circuit if no suitable circuit exists, rather than failing - Circuits choose exit node based on addr/port, exit policies, and which AP connections are standing by - Bump min pathlen from 2 to 3 - Relay end cells have a payload to describe why the stream ended. - If the stream failed because of exit policy, try again with a new circuit. - Clients have a dns cache to remember resolved addresses. - Notice more quickly when we have no working circuits o Configuration: - APPort is now called SocksPort - SocksBindAddress, ORBindAddress, DirBindAddress let you configure where to bind - RecommendedVersions is now a config variable rather than hardcoded (for dirservers) - Reloads config on HUP - Usage info on -h or --help - If you set User and Group config vars, it'll setu/gid to them. Changes in version 0.0.2pre13 - 2003-10-19 o General stability: - SSL_write no longer fails when it returns WANTWRITE and the number of bytes in the buf has changed by the next SSL_write call. - Fix segfault fetching directory when network is down - Fix a variety of minor memory leaks - Dirservers reload the fingerprints file on HUP, so I don't have to take down the network when I approve a new router - Default server config file has explicit Address line to specify fqdn o Buffers: - Buffers grow and shrink as needed (Cut process size from 20M to 2M) - Make listener connections not ever alloc bufs o Autoconf improvements: - don't clobber an external CFLAGS in ./configure - Make install now works - create var/lib/tor on make install - autocreate a tor.sh initscript to help distribs - autocreate the torrc and sample-server-torrc with correct paths o Log files and Daemonizing now work: - If --DebugLogFile is specified, log to it at -l debug - If --LogFile is specified, use it instead of commandline - If --RunAsDaemon is set, tor forks and backgrounds on startup