- tor-commits - lists.torproject.org

[arm/release] Release notes for version 1.4.5
by atagar＠torproject.org 29 Apr '12

29 Apr '12

commit bd14fdedd4e847ac71698f759a67f2d556dfca79 Author: Damian Johnson <atagar(a)torproject.org> Date: Sat Apr 28 20:49:04 2012 -0700 Release notes for version 1.4.5 --- ChangeLog | 23 +++++++++++++++++++++++ 1 files changed, 23 insertions(+), 0 deletions(-) diff --git a/ChangeLog b/ChangeLog index dd057b1..96270b8 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,5 +1,28 @@ CHANGE LOG +4/28/12 - version 1.4.5 +Software isn't perfect and arm is no exception. This is a bugfix release that corrects most issues that users have reported over the last several months. This does not include any new features, but includes changes that are important for continued interoperability with tor. + + * fix: unrecognized authentication methods (like 'SAFECOOKIE') would make arm crash (caught by E) + * fix: providing users a log message when DisableDebuggerAttachment breaks us + * fix: crashing issue when parsing tor log entries from leap years (fix by Sebastian, https://trac.torproject.org/projects/tor/ticket/5265) + * fix: major terminal glitches related to the import of the readline module (caught by Stephan Seitz) + * fix: adding OSX (and probably BSD) support for pwd lookups (thanks to Sebastian, https://trac.torproject.org/projects/tor/ticket/4236) + * fix: unnecessary validation warnings for DirReqStatistics with new tor versions (caught by Sebastian, https://trac.torproject.org/projects/tor/ticket/4237) + * fix: better validation for path component of circuit-staus output (https://trac.torproject.org/projects/tor/ticket/5267) + * fix: checking that the auth cookie is 32 bytes before reading (caught by rransom, https://trac.torproject.org/projects/tor/ticket/4305) + * fix: tor process detection on Mac/FreeBSD/OpenBSD broken due to different variant of ps (caught thanks to a test system from Jordi Espasa Clofent) + * fix: using only lsof for OpenBSD connection queries (caught thanks to a test system from Jordi Espasa Clofent) + * fix: torrc validation was case sensitive (caught by koolfy, thanks to help from Runa, https://trac.torproject.org/projects/tor/ticket/4601) + * fix: torrc validation warned of special deb Logging default (caught by koolfy, thanks to help from Sebastian and Runa, https://trac.torproject.org/projects/tor/ticket/4602) + * fix: snapshot handler had buggy path handling (caught by Jeff Bonner, http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=646080) + * fix: specifying python version requirement in rpm dependencies (caught by unSpawn, thanks to help from Juan Orti Alcaine) + * fix: proc utils didn't account for big-endian architectures (caught by swalker, https://trac.torproject.org/projects/tor/ticket/4777) + * fix: 'enter' on the connection page was mislabeled (patch by twilde, https://trac.torproject.org/projects/tor/ticket/4621) + * fix: reproduced an ACS failure, noting it in the README and adding a config option to partly work around it (thanks to help from rransom) + * fix: circuit listing would fail to be updated if connection information was unavailable + * fix: errors in checking the auth cookie size prevented alternate authentication methods from being attempted + 9/25/11 - version 1.4.4 (881c67a) Besides the normal bug fixes and minor features, this release introduces the control port interpretor. This is a new page or standalone prompt that gives raw control port access with several usability improvements (tab completion, history, irc style help, interpretor commands, etc).

1 0

[arm/release] Giving users a notice when DisableDebuggerAttachment is set
by atagar＠torproject.org 29 Apr '12

29 Apr '12

commit a386d65d1e70719bb476b6a7c68d8f9f3e59b254 Author: Damian Johnson <atagar(a)torproject.org> Date: Fri Apr 27 09:32:19 2012 -0700 Giving users a notice when DisableDebuggerAttachment is set Ugg this feature is a pita. Giving the user a notice that it'll break the system utilities that arm relies on, telling them how to disable it and where to go for more information. --- src/cli/controller.py | 38 +++++++++++++++++++++++--------------- 1 files changed, 23 insertions(+), 15 deletions(-) diff --git a/src/cli/controller.py b/src/cli/controller.py index c46911c..c1be694 100644 --- a/src/cli/controller.py +++ b/src/cli/controller.py @@ -663,23 +663,31 @@ def startTorMonitor(startTime): torTools.REQ_EVENTS["BW"] = "bandwidth graph won't function" if not CONFIG["startup.blindModeEnabled"]: - torTools.REQ_EVENTS["CIRC"] = "may cause issues in identifying client connections" + # The DisableDebuggerAttachment will prevent our connection panel from really + # functioning. It'll have circuits, but little else. If this is the case then + # notify the user and tell them what they can do to fix it. - # Configures connection resoultions. This is paused/unpaused according to - # if Tor's connected or not. - conn.addStatusListener(connResetListener) - - if torPid: - # use the tor pid to help narrow connection results - torCmdName = sysTools.getProcessName(torPid, "tor") - connections.getResolver(torCmdName, torPid, "tor") + if conn.getOption("DisableDebuggerAttachment") == "1": + log.log(log.NOTICE, "Tor is preventing system utilities like netstat and lsof from working. This means that arm can't provide you with connection information. You can change this by adding 'DisableDebuggerAttachment 0' to your torrc and restarting tor. For more information see...\nhttps://trac.torproject.org/3313") + connections.getResolver("tor").setPaused(True) else: - # constructs singleton resolver and, if tor isn't connected, initizes - # it to be paused - connections.getResolver("tor").setPaused(not conn.isAlive()) - - # hack to display a better (arm specific) notice if all resolvers fail - connections.RESOLVER_FINAL_FAILURE_MSG = "We were unable to use any of your system's resolvers to get tor's connections. This is fine, but means that the connections page will be empty. This is usually permissions related so if you would like to fix this then run arm with the same user as tor (ie, \"sudo -u <tor user> arm\")." + torTools.REQ_EVENTS["CIRC"] = "may cause issues in identifying client connections" + + # Configures connection resoultions. This is paused/unpaused according to + # if Tor's connected or not. + conn.addStatusListener(connResetListener) + + if torPid: + # use the tor pid to help narrow connection results + torCmdName = sysTools.getProcessName(torPid, "tor") + connections.getResolver(torCmdName, torPid, "tor") + else: + # constructs singleton resolver and, if tor isn't connected, initizes + # it to be paused + connections.getResolver("tor").setPaused(not conn.isAlive()) + + # hack to display a better (arm specific) notice if all resolvers fail + connections.RESOLVER_FINAL_FAILURE_MSG = "We were unable to use any of your system's resolvers to get tor's connections. This is fine, but means that the connections page will be empty. This is usually permissions related so if you would like to fix this then run arm with the same user as tor (ie, \"sudo -u <tor user> arm\")." # provides a notice about any event types tor supports but arm doesn't missingEventTypes = cli.logPanel.getMissingEventTypes()

1 0

[arm/release] Circuits wouldn't be updated when connection resolution failed
by atagar＠torproject.org 29 Apr '12

29 Apr '12

commit 5e2bc2229eafc81cc26c46529ed812add14a66f2 Author: Damian Johnson <atagar(a)torproject.org> Date: Thu Apr 26 19:31:24 2012 -0700 Circuits wouldn't be updated when connection resolution failed The connecion panel provides information from two sources: connections from system utilities and circuits from tor. We had a check that prevented the connection panel from being updated when the former didn't have any new information, which in turn broke the panel entirely when connection resolvers either failed or were disabled. The 'is new connecion results available' check was actually pretty pointless. The update method gets called every five seconds and the connection resolvers operate at a similar rate (maybe every ten seconds? it depends on resolver performance) so it didn't actually buy us much. Removed the check. --- src/cli/connections/connPanel.py | 193 +++++++++++++++++++------------------- 1 files changed, 96 insertions(+), 97 deletions(-) diff --git a/src/cli/connections/connPanel.py b/src/cli/connections/connPanel.py index d108252..cf7823a 100644 --- a/src/cli/connections/connPanel.py +++ b/src/cli/connections/connPanel.py @@ -419,105 +419,104 @@ class ConnectionPanel(panel.Panel, threading.Thread): connResolver = connections.getResolver("tor") currentResolutionCount = connResolver.getResolutionCount() - if self._lastResourceFetch != currentResolutionCount: - self.valsLock.acquire() - - newEntries = [] # the new results we'll display - - # Fetches new connections and client circuits... - # newConnections [(local ip, local port, foreign ip, foreign port)...] - # newCircuits {circuitID => (status, purpose, path)...} - - newConnections = connResolver.getConnections() - newCircuits = {} - - for circuitID, status, purpose, path in torTools.getConn().getCircuits(): - # Skips established single-hop circuits (these are for directory - # fetches, not client circuits) - if not (status == "BUILT" and len(path) == 1): - newCircuits[circuitID] = (status, purpose, path) - - # Populates newEntries with any of our old entries that still exist. - # This is both for performance and to keep from resetting the uptime - # attributes. Note that CircEntries are a ConnectionEntry subclass so - # we need to check for them first. - - for oldEntry in self._entries: - if isinstance(oldEntry, circEntry.CircEntry): - newEntry = newCircuits.get(oldEntry.circuitID) - - if newEntry: - oldEntry.update(newEntry[0], newEntry[2]) - newEntries.append(oldEntry) - del newCircuits[oldEntry.circuitID] - elif isinstance(oldEntry, connEntry.ConnectionEntry): - connLine = oldEntry.getLines()[0] - connAttr = (connLine.local.getIpAddr(), connLine.local.getPort(), - connLine.foreign.getIpAddr(), connLine.foreign.getPort()) - - if connAttr in newConnections: - newEntries.append(oldEntry) - newConnections.remove(connAttr) - - # Reset any display attributes for the entries we're keeping - for entry in newEntries: entry.resetDisplay() - - # Adds any new connection and circuit entries. - for lIp, lPort, fIp, fPort in newConnections: - newConnEntry = connEntry.ConnectionEntry(lIp, lPort, fIp, fPort) - newConnLine = newConnEntry.getLines()[0] + self.valsLock.acquire() + + newEntries = [] # the new results we'll display + + # Fetches new connections and client circuits... + # newConnections [(local ip, local port, foreign ip, foreign port)...] + # newCircuits {circuitID => (status, purpose, path)...} + + newConnections = connResolver.getConnections() + newCircuits = {} + + for circuitID, status, purpose, path in torTools.getConn().getCircuits(): + # Skips established single-hop circuits (these are for directory + # fetches, not client circuits) + if not (status == "BUILT" and len(path) == 1): + newCircuits[circuitID] = (status, purpose, path) + + # Populates newEntries with any of our old entries that still exist. + # This is both for performance and to keep from resetting the uptime + # attributes. Note that CircEntries are a ConnectionEntry subclass so + # we need to check for them first. + + for oldEntry in self._entries: + if isinstance(oldEntry, circEntry.CircEntry): + newEntry = newCircuits.get(oldEntry.circuitID) - if newConnLine.getType() != connEntry.Category.CIRCUIT: - newEntries.append(newConnEntry) - - # updates exit port and client locale usage information - if newConnLine.isPrivate(): - if newConnLine.getType() == connEntry.Category.INBOUND: - # client connection, update locale information - clientLocale = newConnLine.foreign.getLocale() - - if clientLocale: - self._clientLocaleUsage[clientLocale] = self._clientLocaleUsage.get(clientLocale, 0) + 1 - elif newConnLine.getType() == connEntry.Category.EXIT: - exitPort = newConnLine.foreign.getPort() - self._exitPortUsage[exitPort] = self._exitPortUsage.get(exitPort, 0) + 1 - - for circuitID in newCircuits: - status, purpose, path = newCircuits[circuitID] - newEntries.append(circEntry.CircEntry(circuitID, status, purpose, path)) - - # Counts the relays in each of the categories. This also flushes the - # type cache for all of the connections (in case its changed since last - # fetched). - - categoryTypes = connEntry.Category.values() - typeCounts = dict((type, 0) for type in categoryTypes) - for entry in newEntries: - if isinstance(entry, connEntry.ConnectionEntry): - typeCounts[entry.getLines()[0].getType()] += 1 - elif isinstance(entry, circEntry.CircEntry): - typeCounts[connEntry.Category.CIRCUIT] += 1 - - # makes labels for all the categories with connections (ie, - # "21 outbound", "1 control", etc) - countLabels = [] - - for category in categoryTypes: - if typeCounts[category] > 0: - countLabels.append("%i %s" % (typeCounts[category], category.lower())) - - if countLabels: self._title = "Connections (%s):" % ", ".join(countLabels) - else: self._title = "Connections:" - - self._entries = newEntries - - self._entryLines = [] - for entry in self._entries: - self._entryLines += entry.getLines() + if newEntry: + oldEntry.update(newEntry[0], newEntry[2]) + newEntries.append(oldEntry) + del newCircuits[oldEntry.circuitID] + elif isinstance(oldEntry, connEntry.ConnectionEntry): + connLine = oldEntry.getLines()[0] + connAttr = (connLine.local.getIpAddr(), connLine.local.getPort(), + connLine.foreign.getIpAddr(), connLine.foreign.getPort()) + + if connAttr in newConnections: + newEntries.append(oldEntry) + newConnections.remove(connAttr) + + # Reset any display attributes for the entries we're keeping + for entry in newEntries: entry.resetDisplay() + + # Adds any new connection and circuit entries. + for lIp, lPort, fIp, fPort in newConnections: + newConnEntry = connEntry.ConnectionEntry(lIp, lPort, fIp, fPort) + newConnLine = newConnEntry.getLines()[0] - self.setSortOrder() - self._lastResourceFetch = currentResolutionCount - self.valsLock.release() + if newConnLine.getType() != connEntry.Category.CIRCUIT: + newEntries.append(newConnEntry) + + # updates exit port and client locale usage information + if newConnLine.isPrivate(): + if newConnLine.getType() == connEntry.Category.INBOUND: + # client connection, update locale information + clientLocale = newConnLine.foreign.getLocale() + + if clientLocale: + self._clientLocaleUsage[clientLocale] = self._clientLocaleUsage.get(clientLocale, 0) + 1 + elif newConnLine.getType() == connEntry.Category.EXIT: + exitPort = newConnLine.foreign.getPort() + self._exitPortUsage[exitPort] = self._exitPortUsage.get(exitPort, 0) + 1 + + for circuitID in newCircuits: + status, purpose, path = newCircuits[circuitID] + newEntries.append(circEntry.CircEntry(circuitID, status, purpose, path)) + + # Counts the relays in each of the categories. This also flushes the + # type cache for all of the connections (in case its changed since last + # fetched). + + categoryTypes = connEntry.Category.values() + typeCounts = dict((type, 0) for type in categoryTypes) + for entry in newEntries: + if isinstance(entry, connEntry.ConnectionEntry): + typeCounts[entry.getLines()[0].getType()] += 1 + elif isinstance(entry, circEntry.CircEntry): + typeCounts[connEntry.Category.CIRCUIT] += 1 + + # makes labels for all the categories with connections (ie, + # "21 outbound", "1 control", etc) + countLabels = [] + + for category in categoryTypes: + if typeCounts[category] > 0: + countLabels.append("%i %s" % (typeCounts[category], category.lower())) + + if countLabels: self._title = "Connections (%s):" % ", ".join(countLabels) + else: self._title = "Connections:" + + self._entries = newEntries + + self._entryLines = [] + for entry in self._entries: + self._entryLines += entry.getLines() + + self.setSortOrder() + self._lastResourceFetch = currentResolutionCount + self.valsLock.release() def _resolveApps(self, flagQuery = True): """

1 0

[arm/release] Arm release 1.4.5
by atagar＠torproject.org 29 Apr '12

29 Apr '12

commit e249dc8f5c4c282326324161cf8421f947cf660e Merge: edcde43 bd14fde Author: Damian Johnson <atagar(a)torproject.org> Date: Sat Apr 28 20:52:33 2012 -0700 Arm release 1.4.5 ChangeLog | 32 +- README | 15 +- armrc.sample | 6 +- src/cli/connections/connPanel.py | 195 ++++---- src/cli/controller.py | 49 ++- src/cli/headerPanel.py | 10 + src/cli/logPanel.py | 23 +- src/resources/tor-arm.desktop | 12 + src/resources/tor-arm.svg | 1074 ++++++++++++++++++++++++++++++++++++++ src/starter.py | 45 ++- src/util/connections.py | 2 +- src/util/procTools.py | 10 +- src/util/sysTools.py | 18 + src/util/torConfig.py | 36 ++- src/util/torInterpretor.py | 8 +- src/util/torTools.py | 39 ++- src/util/uiTools.py | 17 +- src/version.py | 4 +- 18 files changed, 1451 insertions(+), 144 deletions(-) diff --cc src/version.py index 722cde0,0745fb9..38d4708 --- a/src/version.py +++ b/src/version.py @@@ -2,6 -2,6 +2,6 @@@ Provides arm's version and release date. """ - VERSION = '1.4.4.1' - LAST_MODIFIED = "September 29, 2011" -VERSION = '1.4.5_dev' -LAST_MODIFIED = "September 25, 2011" ++VERSION = '1.4.5.0' ++LAST_MODIFIED = "April 28, 2012"

1 0

[arm/release] fix: clearing contents to force curses to refresh
by atagar＠torproject.org 29 Apr '12

29 Apr '12

commit 2b3a9d7f2fc921ccd01dc43b641e98cb056ca799 Author: Damian Johnson <atagar(a)torproject.org> Date: Sun Oct 2 19:56:41 2011 -0700 fix: clearing contents to force curses to refresh Evidently it isn't enough for panels to clear their contents, so also requesting a clear of the uppermost stdscr. Fix suggested by np... https://trac.torproject.org/projects/tor/ticket/2830#comment:9 --- src/cli/controller.py | 5 +++++ 1 files changed, 5 insertions(+), 0 deletions(-) diff --git a/src/cli/controller.py b/src/cli/controller.py index 762f955..aebc6f3 100644 --- a/src/cli/controller.py +++ b/src/cli/controller.py @@ -341,6 +341,11 @@ class Controller: panelImpl.setTop(occupiedContent) occupiedContent += panelImpl.getHeight() + # apparently curses may cache display contents unless we explicitely + # request a redraw here... + # https://trac.torproject.org/projects/tor/ticket/2830#comment:9 + if force: self._screen.clear() + for panelImpl in displayPanels: panelImpl.redraw(force)

1 0

[arm/release] Adding hotfix sig to its changelog entries
by atagar＠torproject.org 29 Apr '12

29 Apr '12

commit 095c7d330b2f068dd68877b69fefca10b1bbab5d Author: Damian Johnson <atagar(a)torproject.org> Date: Thu Sep 29 09:14:28 2011 -0700 Adding hotfix sig to its changelog entries --- ChangeLog | 8 ++++---- 1 files changed, 4 insertions(+), 4 deletions(-) diff --git a/ChangeLog b/ChangeLog index 287ab57..fccf083 100644 --- a/ChangeLog +++ b/ChangeLog @@ -37,10 +37,10 @@ Besides the normal bug fixes and minor features, this release introduces the con * fix: listener wasn't registering GUARD events * fix: dropping nonexistent ORCONN attributes * fix: clarifying the missing torrc message - * fix (9/29/11): dropping gtk/cagraph requirements (caught by monochromec, vinc3nt, and Bill White) - * fix (9/29/11): crash from esc in interpretor panel prompt (caught by Sebastian, https://trac.torproject.org/projects/tor/ticket/4098) - * fix (9/29/11): deduplicating 'Circuit build timeout' and 'looks like we don't need to wait so long' messages (patch by StrangeCharm, https://trac.torproject.org/projects/tor/ticket/4096) - * fix (9/29/11): forcing redraw on ctrl+L reguardless of need (caught by np, https://trac.torproject.org/projects/tor/ticket/2830) + * fix (9/29/11, edcde43): dropping gtk/cagraph requirements (caught by monochromec, vinc3nt, and Bill White) + * fix (9/29/11, edcde43): crash from esc in interpretor panel prompt (caught by Sebastian, https://trac.torproject.org/projects/tor/ticket/4098) + * fix (9/29/11, edcde43): deduplicating 'Circuit build timeout' and 'looks like we don't need to wait so long' messages (patch by StrangeCharm, https://trac.torproject.org/projects/tor/ticket/4096) + * fix (9/29/11, edcde43): forcing redraw on ctrl+L reguardless of need (caught by np, https://trac.torproject.org/projects/tor/ticket/2830) 7/16/11 - version 1.4.3 (6cf4836) This completes the codebase refactoring project that's been a year in the works and provides numerous performance and usability improvements. Most notably this includes a setup wizard for new relays, menuing interface, gui prototype, substantial performance improvements, and support for Mac OSX.

1 0

[arm/release] fix: fixing tor process detection for openbsd
by atagar＠torproject.org 29 Apr '12

29 Apr '12

commit 84cd622e515c97d0c9b0d9779204155978926e9d Author: Damian Johnson <atagar(a)torproject.org> Date: Fri Oct 7 09:18:14 2011 -0700 fix: fixing tor process detection for openbsd OpenBSD uses a different variant of ps causing the tor process detection (among many other things) to fail. Caught thanks to a test system from Jordi Clofent. --- src/util/torTools.py | 13 +++++++++++++ 1 files changed, 13 insertions(+), 0 deletions(-) diff --git a/src/util/torTools.py b/src/util/torTools.py index 8e2d6d5..34c9adc 100644 --- a/src/util/torTools.py +++ b/src/util/torTools.py @@ -352,6 +352,19 @@ def isTorRunning(): # suggestions welcome for making this more reliable commandResults = sysTools.call("ps -A co command") + + if not commandResults: + # OpenBSD uses a weird (and largely broken from the looks of it) version of + # ps. It lacks the -A argument and according to the man page -j, -l, and -u + # all do something similar but they fail. + # + # ucomm is defined in the man page as 'Alias: comm. Name to be used for + # accounting.' The alias part is a lie (it works, but with an error + # message), though this seems to do what we want and prints the bare + # command. + + commandResults = sysTools.call("ps -o ucomm=") + if commandResults: for cmd in commandResults: if cmd.strip() == "tor": return True

1 0

[arm/master] Release notes for version 1.4.5
by atagar＠torproject.org 29 Apr '12

29 Apr '12

commit bd14fdedd4e847ac71698f759a67f2d556dfca79 Author: Damian Johnson <atagar(a)torproject.org> Date: Sat Apr 28 20:49:04 2012 -0700 Release notes for version 1.4.5 --- ChangeLog | 23 +++++++++++++++++++++++ 1 files changed, 23 insertions(+), 0 deletions(-) diff --git a/ChangeLog b/ChangeLog index dd057b1..96270b8 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,5 +1,28 @@ CHANGE LOG +4/28/12 - version 1.4.5 +Software isn't perfect and arm is no exception. This is a bugfix release that corrects most issues that users have reported over the last several months. This does not include any new features, but includes changes that are important for continued interoperability with tor. + + * fix: unrecognized authentication methods (like 'SAFECOOKIE') would make arm crash (caught by E) + * fix: providing users a log message when DisableDebuggerAttachment breaks us + * fix: crashing issue when parsing tor log entries from leap years (fix by Sebastian, https://trac.torproject.org/projects/tor/ticket/5265) + * fix: major terminal glitches related to the import of the readline module (caught by Stephan Seitz) + * fix: adding OSX (and probably BSD) support for pwd lookups (thanks to Sebastian, https://trac.torproject.org/projects/tor/ticket/4236) + * fix: unnecessary validation warnings for DirReqStatistics with new tor versions (caught by Sebastian, https://trac.torproject.org/projects/tor/ticket/4237) + * fix: better validation for path component of circuit-staus output (https://trac.torproject.org/projects/tor/ticket/5267) + * fix: checking that the auth cookie is 32 bytes before reading (caught by rransom, https://trac.torproject.org/projects/tor/ticket/4305) + * fix: tor process detection on Mac/FreeBSD/OpenBSD broken due to different variant of ps (caught thanks to a test system from Jordi Espasa Clofent) + * fix: using only lsof for OpenBSD connection queries (caught thanks to a test system from Jordi Espasa Clofent) + * fix: torrc validation was case sensitive (caught by koolfy, thanks to help from Runa, https://trac.torproject.org/projects/tor/ticket/4601) + * fix: torrc validation warned of special deb Logging default (caught by koolfy, thanks to help from Sebastian and Runa, https://trac.torproject.org/projects/tor/ticket/4602) + * fix: snapshot handler had buggy path handling (caught by Jeff Bonner, http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=646080) + * fix: specifying python version requirement in rpm dependencies (caught by unSpawn, thanks to help from Juan Orti Alcaine) + * fix: proc utils didn't account for big-endian architectures (caught by swalker, https://trac.torproject.org/projects/tor/ticket/4777) + * fix: 'enter' on the connection page was mislabeled (patch by twilde, https://trac.torproject.org/projects/tor/ticket/4621) + * fix: reproduced an ACS failure, noting it in the README and adding a config option to partly work around it (thanks to help from rransom) + * fix: circuit listing would fail to be updated if connection information was unavailable + * fix: errors in checking the auth cookie size prevented alternate authentication methods from being attempted + 9/25/11 - version 1.4.4 (881c67a) Besides the normal bug fixes and minor features, this release introduces the control port interpretor. This is a new page or standalone prompt that gives raw control port access with several usability improvements (tab completion, history, irc style help, interpretor commands, etc).

1 0

[arm/release] Improving tor detection for BSD platforms
by atagar＠torproject.org 29 Apr '12

29 Apr '12

commit b1fce355c3661324f477e2698a08214035d09c10 Author: Damian Johnson <atagar(a)torproject.org> Date: Tue Oct 11 08:47:05 2011 -0700 Improving tor detection for BSD platforms Linux and BSD platforms (mac, openbsd, and freebsd) have distinct argument sets for their ps resolvers. Taking into account for this for tor detection, guessing the family based on uname results then trying both if we guess wrong. Adding the '-a' argument to get results for all users thanks to Jordi Espasa Clofent. Tested this ps command on Mac and OpenBSD, and just guessing that FreeBSD will behave the same. --- src/util/torTools.py | 31 +++++++++++++++++++------------ 1 files changed, 19 insertions(+), 12 deletions(-) diff --git a/src/util/torTools.py b/src/util/torTools.py index 34c9adc..a4ab70c 100644 --- a/src/util/torTools.py +++ b/src/util/torTools.py @@ -350,20 +350,27 @@ def isTorRunning(): then this returns False. """ - # suggestions welcome for making this more reliable - commandResults = sysTools.call("ps -A co command") - + # Linux and the BSD families have different variants of ps. Guess based on + # os.uname() results which to try first, then fall back to the other. + # + # Linux + # -A - Select all processes. Identical to -e. + # -co command - Shows just the base command. + # + # Mac / BSD + # -a - Display information about other users' processes as well as + # your own. + # -o ucomm= - Shows just the ucomm attribute ("name to be used for + # accounting") + + primaryResolver, secondaryResolver = "ps -A co command", "ps -ao ucomm=" + + if os.uname()[0] in ("Darwin", "FreeBSD", "OpenBSD"): + primaryResolver, secondaryResolver = secondaryResolver, primaryResolver + + commandResults = sysTools.call(primaryResolver) if not commandResults: - # OpenBSD uses a weird (and largely broken from the looks of it) version of - # ps. It lacks the -A argument and according to the man page -j, -l, and -u - # all do something similar but they fail. - # - # ucomm is defined in the man page as 'Alias: comm. Name to be used for - # accounting.' The alias part is a lie (it works, but with an error - # message), though this seems to do what we want and prints the bare - # command. - - commandResults = sysTools.call("ps -o ucomm=") + commandResults = sysTools.call(secondaryResolver) if commandResults: for cmd in commandResults:

1 0

[arm/release] Adding OSX support for the getPwd utility
by atagar＠torproject.org 29 Apr '12

29 Apr '12

commit 92ac1930143e51689396563877087d1070a246bd Author: Damian Johnson <atagar(a)torproject.org> Date: Sat Oct 15 14:03:32 2011 -0700 Adding OSX support for the getPwd utility BSD platforms lack either pwdx or proc contents with this information. However, this is available via lsof. Using this for BSD platforms. This was tested by running sysTools.getPwd() on OSX - hopefully it should work on Free/OpenBSD too. Fix is thanks to Sebastian. https://trac.torproject.org/projects/tor/ticket/4236 --- src/util/sysTools.py | 17 +++++++++++++++++ 1 files changed, 17 insertions(+), 0 deletions(-) diff --git a/src/util/sysTools.py b/src/util/sysTools.py index 98361e5..4bb0667 100644 --- a/src/util/sysTools.py +++ b/src/util/sysTools.py @@ -174,6 +174,23 @@ def getPwd(pid): PWD_CACHE[pid] = pwd return pwd except IOError: pass # fall back to pwdx + elif os.uname()[0] in ("Darwin", "FreeBSD", "OpenBSD"): + # BSD neither useres the above proc info nor does it have pwdx. Use lsof to + # determine this instead: + # https://trac.torproject.org/projects/tor/ticket/4236 + # + # ~$ lsof -a -p 75717 -d cwd -Fn + # p75717 + # n/Users/atagar/tor/src/or + + try: + results = call("lsof -a -p %s -d cwd -Fn" % pid) + + if results and len(results) == 2 and results[1].startswith("n/"): + pwd = results[1][1:].strip() + PWD_CACHE[pid] = pwd + return pwd + except IOError, exc: pass try: # pwdx results are of the form:

1 0