tor-commits July 2017

tor-commits@lists.torproject.org

14 participants
1743 discussions

[stem/master] Move release checklist into setup.py
by atagar＠torproject.org 03 Jul '17

03 Jul '17

commit e10700e57a3cf5408279b4f3f1c46d0a6790519b Author: Damian Johnson <atagar(a)torproject.org> Date: Mon Jul 3 11:24:47 2017 -0700 Move release checklist into setup.py When I first started stem I made a wiki page, but we don't actually use it for much of anything... https://trac.torproject.org/projects/tor/wiki/doc/stem About the only useful information is the release checklist, and this could reside in our setup.py just as well. --- setup.py | 27 +++++++++++++++++++++++++++ 1 file changed, 27 insertions(+) diff --git a/setup.py b/setup.py index 8f4a6e8..2024826 100644 --- a/setup.py +++ b/setup.py @@ -1,6 +1,33 @@ #!/usr/bin/env python # Copyright 2012-2016, Damian Johnson and The Tor Project # See LICENSE for licensing information +# +# Release Checklist +# ================= +# +# * Recache latest information (cache_manual.py and cache_fallback_directories.py) +# +# * Tag the release +# |- Bump stem's version (in stem/__init__.py and docs/index.rst). +# |- git commit -a -m "Stem release 1.0.0" +# |- git tag -u 9ABBEEC6 -m "stem release 1.0.0" 1.0.0 d0bb81a +# +- git push --tags +# +# * Dry-run release on https://pypi.python.org/pypi/stem/ +# |- python setup.py sdist +# |- gpg --detach-sig --armor dist/stem-dry-run-1.0.0.tar.gz +# |- twine upload dist/* +# +- Check that https://pypi.python.org/pypi/stem-dry-run/ looks correct, comparing it to https://pypi.python.org/pypi/stem/ +# +# * Final release +# |- rm dist/* +# |- Change the setup.py's DRY_RUN flag to false. +# |- python setup.py sdist +# |- gpg --detach-sig --armor dist/stem-1.0.0.tar.gz +# +- twine upload dist/* +# +# * Contact package maintainers +# * Announce the release (example: https://blog.torproject.org/blog/stem-release-11) import distutils.core import os

1 0

[stem/master] Replace wiki link with jenkins
by atagar＠torproject.org 03 Jul '17

03 Jul '17

commit fe561fb3ddfe2de3dae71ecfc01b0c10c9f3d98b Author: Damian Johnson <atagar(a)torproject.org> Date: Mon Jul 3 11:27:20 2017 -0700 Replace wiki link with jenkins Gonna drop our wiki page since we don't actually use it. Last thing there is just a link to jenkins. --- docs/_templates/layout.html | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/_templates/layout.html b/docs/_templates/layout.html index e3c8d2c..c0b443c 100644 --- a/docs/_templates/layout.html +++ b/docs/_templates/layout.html @@ -52,7 +52,7 @@ <li><a href="{{ pathto('faq') }}">FAQ</a></li> <li><a href="{{ pathto('change_log') }}">Change Log</a></li> <li><a href="https://trac.torproject.org/projects/tor/wiki/doc/stem/bugs">Bug Tracker</a></li> - <li><a href="https://trac.torproject.org/projects/tor/wiki/doc/stem">Wiki</a></li> + <li><a href="https://jenkins.torproject.org/job/stem-tor-ci/">Jenkins</a></li> <li><a href="{{ pathto('download') }}">Download</a></li> </ul> </li>

1 0

[stem/master] Drop workaround for relative cookie paths
by atagar＠torproject.org 03 Jul '17

03 Jul '17

commit 32a37dcd211a8292c757c9da7c628939bf37555a Author: Damian Johnson <atagar(a)torproject.org> Date: Mon Jul 3 10:57:03 2017 -0700 Drop workaround for relative cookie paths Five years ago we worked around a tor bug where PROTOCOLINFO responses could have relative cookie paths... https://trac.torproject.org/projects/tor/ticket/4881 https://trac.torproject.org/projects/tor/ticket/1101 This has easily been fixed long enough that we can drop this workaround. Yay, less code! --- stem/connection.py | 60 -------------------------------------- test/unit/response/protocolinfo.py | 35 ---------------------- 2 files changed, 95 deletions(-) diff --git a/stem/connection.py b/stem/connection.py index db3f218..cce00bf 100644 --- a/stem/connection.py +++ b/stem/connection.py @@ -985,11 +985,6 @@ def get_protocolinfo(controller): the tor process running on it. If the socket is already closed then it is first reconnected. - According to the control spec the cookie_file is an absolute path. However, - this often is not the case (especially for the Tor Browser Bundle). If the - path is relative then we'll make an attempt (which may not work) to correct - this (:trac:`1101`). - This can authenticate to either a :class:`~stem.control.BaseController` or :class:`~stem.socket.ControlSocket`. @@ -1021,27 +1016,6 @@ def get_protocolinfo(controller): raise stem.SocketError(exc) stem.response.convert('PROTOCOLINFO', protocolinfo_response) - - # attempt to expand relative cookie paths - - if protocolinfo_response.cookie_path: - _expand_cookie_path(protocolinfo_response, stem.util.system.pid_by_name, 'tor') - - # attempt to expand relative cookie paths via the control port or socket file - - if isinstance(controller, stem.socket.ControlSocket): - control_socket = controller - else: - control_socket = controller.get_socket() - - if isinstance(control_socket, stem.socket.ControlPort): - if control_socket.get_address() == '127.0.0.1': - pid_method = stem.util.system.pid_by_port - _expand_cookie_path(protocolinfo_response, pid_method, control_socket.get_port()) - elif isinstance(control_socket, stem.socket.ControlSocketFile): - pid_method = stem.util.system.pid_by_open_file - _expand_cookie_path(protocolinfo_response, pid_method, control_socket.get_socket_path()) - return protocolinfo_response @@ -1122,40 +1096,6 @@ def _read_cookie(cookie_path, is_safecookie): raise UnreadableCookieFile(exc_msg, cookie_path, is_safecookie) -def _expand_cookie_path(protocolinfo_response, pid_resolver, pid_resolution_arg): - """ - Attempts to expand a relative cookie path with the given pid resolver. This - leaves the cookie_path alone if it's already absolute, **None**, or the - system calls fail. - """ - - cookie_path = protocolinfo_response.cookie_path - if cookie_path and not os.path.isabs(cookie_path): - try: - tor_pid = pid_resolver(pid_resolution_arg) - - if not tor_pid: - raise IOError('pid lookup failed') - - tor_cwd = stem.util.system.cwd(tor_pid) - - if not tor_cwd: - raise IOError('cwd lookup failed') - - cookie_path = stem.util.system.expand_path(cookie_path, tor_cwd) - except IOError as exc: - resolver_labels = { - stem.util.system.pid_by_name: ' by name', - stem.util.system.pid_by_port: ' by port', - stem.util.system.pid_by_open_file: ' by socket file', - } - - pid_resolver_label = resolver_labels.get(pid_resolver, '') - log.debug('unable to expand relative tor cookie path%s: %s' % (pid_resolver_label, exc)) - - protocolinfo_response.cookie_path = cookie_path - - class AuthenticationFailure(Exception): """ Base error for authentication failures. diff --git a/test/unit/response/protocolinfo.py b/test/unit/response/protocolinfo.py index 9b2c9b1..ab6dd0e 100644 --- a/test/unit/response/protocolinfo.py +++ b/test/unit/response/protocolinfo.py @@ -2,7 +2,6 @@ Unit tests for the stem.response.protocolinfo.ProtocolInfoResponse class. """ -import os import unittest import stem.response @@ -156,37 +155,3 @@ class TestProtocolInfoResponse(unittest.TestCase): control_message = ControlMessage.from_str(UNICODE_COOKIE_PATH, 'PROTOCOLINFO', normalize = True) self.assertEqual(EXPECTED_UNICODE_PATH, control_message.cookie_path) - - @patch('stem.util.proc.is_available', Mock(return_value = False)) - @patch('stem.util.system.is_available', Mock(return_value = True)) - def test_relative_cookie(self): - """ - Checks an authentication cookie with a relative path where expansion both - succeeds and fails. - """ - - # we need to mock both pid and cwd lookups since the general cookie - # expanion works by... - # - resolving the pid of the "tor" process - # - using that to get tor's cwd - - def call_function(command, default): - if command == stem.util.system.GET_PID_BY_NAME_PGREP % 'tor': - return ['10'] - elif command == stem.util.system.GET_CWD_PWDX % 10: - return ['10: /tmp/foo'] - - with patch('stem.util.system.call') as call_mock: - call_mock.side_effect = call_function - - control_message = ControlMessage.from_str(RELATIVE_COOKIE_PATH, 'PROTOCOLINFO', normalize = True) - stem.connection._expand_cookie_path(control_message, stem.util.system.pid_by_name, 'tor') - - self.assertEqual(os.path.join('/tmp/foo', 'tor-browser_en-US', 'Data', 'control_auth_cookie'), control_message.cookie_path) - - # exercise cookie expansion where both calls fail (should work, just - # leaving the path unexpanded) - - with patch('stem.util.system.call', Mock(return_value = None)): - control_message = ControlMessage.from_str(RELATIVE_COOKIE_PATH, 'PROTOCOLINFO', normalize = True) - self.assertEqual('./tor-browser_en-US/Data/control_auth_cookie', control_message.cookie_path)

1 0

[stem/master] Python3 fix for fallback_directories.cfg
by atagar＠torproject.org 03 Jul '17

03 Jul '17

commit c945bc6538eafe56104a283dd8fe83922415e716 Author: Damian Johnson <atagar(a)torproject.org> Date: Mon Jul 3 11:06:50 2017 -0700 Python3 fix for fallback_directories.cfg Drop our use of cmp so this works with python3... * /home/atagar/Desktop/stem/cache_fallback_directories.py line 57 - undefined name 'cmp' | for directory in sorted(latest_fallback_directories.values(), lambda x, y: cmp(x.fingerprint, y.fingerprint)): --- cache_fallback_directories.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/cache_fallback_directories.py b/cache_fallback_directories.py index 4d5665b..0e61cf2 100755 --- a/cache_fallback_directories.py +++ b/cache_fallback_directories.py @@ -54,7 +54,7 @@ if __name__ == '__main__': conf.set('tor_commit', fallback_dir_commit) conf.set('stem_commit', stem_commit) - for directory in sorted(latest_fallback_directories.values(), lambda x, y: cmp(x.fingerprint, y.fingerprint)): + for directory in sorted(latest_fallback_directories.values(), key = lambda x: x.fingerprint): fingerprint = directory.fingerprint conf.set('%s.address' % fingerprint, directory.address) conf.set('%s.or_port' % fingerprint, str(directory.or_port))

1 0

[stem/master] Suppress unused prereq import notices for python3
by atagar＠torproject.org 03 Jul '17

03 Jul '17

commit 8b64ed0dc1e962b3a05529838d6927e14f5381a8 Author: Damian Johnson <atagar(a)torproject.org> Date: Mon Jul 3 11:14:29 2017 -0700 Suppress unused prereq import notices for python3 We already have suppressions that do the trick for python2 which complain about just the method name in 'from' imports... 'signing' imported but unused However, with python3 pyflakes complains about the whole name... 'nacl.signing' imported but unused This cleans up the following noise... * /home/atagar/Desktop/stem/stem/prereq.py line 99 - 'cryptography.utils.int_to_bytes' imported but unused | from cryptography.utils import int_from_bytes, int_to_bytes line 99 - 'cryptography.utils.int_from_bytes' imported but unused | from cryptography.utils import int_from_bytes, int_to_bytes line 100 - 'cryptography.hazmat.backends.default_backend' imported but unused | from cryptography.hazmat.backends import default_backend line 101 - 'cryptography.hazmat.primitives.serialization.load_der_public_key' imported but unused | from cryptography.hazmat.primitives.serialization import load_der_public_key line 102 - 'cryptography.hazmat.primitives.ciphers.modes' imported but unused | from cryptography.hazmat.primitives.ciphers import Cipher, algorithms, modes line 102 - 'cryptography.hazmat.primitives.ciphers.Cipher' imported but unused | from cryptography.hazmat.primitives.ciphers import Cipher, algorithms, modes line 102 - 'cryptography.hazmat.primitives.ciphers.algorithms' imported but unused | from cryptography.hazmat.primitives.ciphers import Cipher, algorithms, modes line 164 - 'nacl.encoding' imported but unused | from nacl import encoding line 165 - 'nacl.signing' imported but unused | from nacl import signing --- test/settings.cfg | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/test/settings.cfg b/test/settings.cfg index 963ba59..4d532ff 100644 --- a/test/settings.cfg +++ b/test/settings.cfg @@ -162,6 +162,15 @@ pyflakes.ignore stem/prereq.py => 'unittest.mock' imported but unused pyflakes.ignore stem/prereq.py => 'long_to_bytes' imported but unused pyflakes.ignore stem/prereq.py => 'encoding' imported but unused pyflakes.ignore stem/prereq.py => 'signing' imported but unused +pyflakes.ignore stem/prereq.py => 'cryptography.utils.int_to_bytes' imported but unused +pyflakes.ignore stem/prereq.py => 'cryptography.utils.int_from_bytes' imported but unused +pyflakes.ignore stem/prereq.py => 'cryptography.hazmat.backends.default_backend' imported but unused +pyflakes.ignore stem/prereq.py => 'cryptography.hazmat.primitives.serialization.load_der_public_key' +pyflakes.ignore stem/prereq.py => 'cryptography.hazmat.primitives.ciphers.modes' imported but unused +pyflakes.ignore stem/prereq.py => 'cryptography.hazmat.primitives.ciphers.Cipher' imported but unused +pyflakes.ignore stem/prereq.py => 'cryptography.hazmat.primitives.ciphers.algorithms' +pyflakes.ignore stem/prereq.py => 'nacl.encoding' imported but unused +pyflakes.ignore stem/prereq.py => 'nacl.signing' imported but unused pyflakes.ignore stem/interpreter/__init__.py => undefined name 'raw_input' pyflakes.ignore stem/util/conf.py => undefined name 'unicode' pyflakes.ignore stem/util/test_tools.py => 'pyflakes' imported but unused

1 0

[sandboxed-tor-browser/master] Bug 20773: Stop mounting /proc in the Tor Browser container.
by yawning＠torproject.org 03 Jul '17

03 Jul '17

commit 95857360ec7f84cf9f0a01855c15881c89919133 Author: Yawning Angel <yawning(a)schwanenlied.me> Date: Mon Jul 3 17:27:35 2017 +0000 Bug 20773: Stop mounting /proc in the Tor Browser container. Wasn't so bad to figure out, though upstream still should fall back gracefully, since the only reason why it crashes horribly is because pthread_attr_getstack() returns a stack size of 0 for the default thread when /proc is missing. In our case we can safely fall back to querying RLIMIT_STACK, because the sandbox containerization explicity sets it to something large, but relatively sane. --- ChangeLog | 1 + README.md | 12 ++-- .../internal/sandbox/application.go | 1 + src/tbb_stub/tbb_stub.c | 76 ++++++++++++++++++++++ 4 files changed, 83 insertions(+), 7 deletions(-) diff --git a/ChangeLog b/ChangeLog index b42022a..51f693a 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,5 +1,6 @@ Changes in version 0.0.9 - UNRELEASED: * Bug 22712: Suppress ATK Bridge initialization which will never work. + * Bug 20773: Stop mounting /proc in the Tor Browser container. * Fix the build being broken on Debian Jessie due to #22648. * Remove the undocumented command line options that enable unsafe behavior. diff --git a/README.md b/README.md index dc4fc18..6cad572 100644 --- a/README.md +++ b/README.md @@ -41,18 +41,16 @@ Things that the sandbox breaks: Places where the sandbox could be better: - * More about the host system is exposed than neccecary, primarily because - Firefox crashes without `/proc`. + * The updater container still mounts `/proc`. + * PulseAudio is likely unsafe without a protocol filter like X11. + * X11 is still X11, and despite mitigations is likely still unsafe. Upstream Bugs: - * Tor Browser should run without a `/proc` filesystem. - (https://bugs.torproject.org/20283) + * Tor Browser should run without a `/proc` filesystem, worked around in + the worst possible way. (https://bugs.torproject.org/20283) * OpenGL software rendering is broken on certain Linux systems. (https://bugs.torproject.org/20866) - * On certain systems, Firefox crashes with an X11 error due to the lack of - MIT-SHM support. This appears to be a race condition, and still occurs - despite mitigations. (https://bugzilla.mozilla.org/show_bug.cgi?id=1271100) Notes: diff --git a/src/cmd/sandboxed-tor-browser/internal/sandbox/application.go b/src/cmd/sandboxed-tor-browser/internal/sandbox/application.go index d3162dd..a160b67 100644 --- a/src/cmd/sandboxed-tor-browser/internal/sandbox/application.go +++ b/src/cmd/sandboxed-tor-browser/internal/sandbox/application.go @@ -69,6 +69,7 @@ func RunTorBrowser(cfg *config.Config, manif *config.Manifest, tor *tor.Tor) (pr h.stderr = logger h.seccompFn = installTorBrowserSeccompProfile h.fakeDbus = true + h.mountProc = false // Gtk+ and PulseAudio. hasAdwaita := h.appendGtk2Theme() diff --git a/src/tbb_stub/tbb_stub.c b/src/tbb_stub/tbb_stub.c index e7ca733..84a14e8 100644 --- a/src/tbb_stub/tbb_stub.c +++ b/src/tbb_stub/tbb_stub.c @@ -37,6 +37,9 @@ #define _GNU_SOURCE /* Fuck *BSD and Macintoys. */ +#include <sys/types.h> +#include <sys/syscall.h> +#include <sys/resource.h> #include <sys/socket.h> #include <sys/un.h> #include <arpa/inet.h> @@ -47,11 +50,13 @@ #include <stdbool.h> #include <stdio.h> #include <stdlib.h> +#include <unistd.h> static pthread_once_t stub_init_once = PTHREAD_ONCE_INIT; static int (*real_connect)(int, const struct sockaddr *, socklen_t) = NULL; static int (*real_socket)(int, int, int) = NULL; static void *(*real_dlopen)(const char *, int) = NULL; +static int (*real_pthread_attr_getstack)(const pthread_attr_t *, void **, size_t *); static struct sockaddr_un socks_addr; static struct sockaddr_un control_addr; @@ -228,6 +233,73 @@ pa_start_child_for_read(const char *name, const char *argv1, pid_t *pid) return -1; } +/* Firefox will crash if pthread_attr_getstack doesn't return a sensible stack + * size, which will happen if /proc is missing, since glibc grovels through + * /proc/self/maps to determine this information for the default thread. + * + * See: glibc/nptl/pthread_getattr_np.c + */ +int +pthread_attr_getstack(const pthread_attr_t *attr, void **stackaddr, size_t *stacksize) +{ + int ret; + + ret = real_pthread_attr_getstack(attr, stackaddr, stacksize); + if (ret != 0) { + fprintf(stderr, "WARN: pthread_attr_getstack(%p, %p, %p) = %d\n", attr, stackaddr, stacksize, ret); + return ret; + } + +#if 0 + fprintf(stderr, "tbb_stub: pthread_attr_getstack(%p, %p, %p) = %d\n", attr, stackaddr, stacksize, ret); + fprintf(stderr, "tbb_stub: stackaddr: %p\n", stackaddr); + fprintf(stderr, "tbb_stub: stacksize: %ld\n", *stacksize); +#endif + + /* If we got a sensible value for the stack size, then return. */ + if (*stacksize != 0) { + return ret; + } else { + /* Otherwise, we should be the initial thread (pid == tid). */ + pid_t tid = syscall(__NR_gettid); + pid_t pid = getpid(); + + if (tid != pid) { + fprintf(stderr, "ERROR: Got a 0 stack size when pid = %d != tid = %d\n", pid, tid); + abort(); + } + } + + /* First try pthread_attr_getstacksize(), which works on glibc 2.25. */ + ret = pthread_attr_getstacksize(attr, stacksize); + if (ret != 0 || *stacksize == 0) { + /* Fall back to getrlimit(). */ + struct rlimit rl; + + ret = getrlimit(RLIMIT_STACK, &rl); + if (ret != 0) { + fprintf(stderr, "ERROR: Failed to query rlimit: %d", ret); + abort(); + } + + /* So, the main reason why glibc digs through proc is so that it can + * return the current committed stack size, and not the upper bound on + * the stack size. But without /proc, there's no good way to get this + * information. + * + * This is probably ok, pthread_attr_getstacksize() without proc mounted + * appears to do the same thing. + */ + *stacksize = rl.rlim_cur; + } + +#if 0 + fprintf(stderr, "tbb_stub: Fallback stacksize: %ld\n", *stacksize); +#endif + + return ret; +} + /* Initialize the stub. */ static void stub_init(void) @@ -257,6 +329,10 @@ stub_init(void) fprintf(stderr, "ERROR: Failed to find `socket()` symbol: %s\n", dlerror()); goto out; } + if ((real_pthread_attr_getstack = dlsym(RTLD_NEXT, "pthread_attr_getstack")) == NULL) { + fprintf(stderr, "ERROR: Failed to find `pthread_attr_getstack()` symbol: %s\n", dlerror()); + goto out; + } /* Initialize the SOCKS target address. */ socks_addr.sun_family = AF_LOCAL;

1 0

[tor/master] Update fuzzing_include_am to include updates from include.am
by nickm＠torproject.org 03 Jul '17

03 Jul '17

commit d0111264678704d0c2394a4ddd28621d3ef23d97 Author: Nick Mathewson <nickm(a)torproject.org> Date: Mon Jul 3 13:24:57 2017 -0400 Update fuzzing_include_am to include updates from include.am --- scripts/codegen/fuzzing_include_am.py | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/scripts/codegen/fuzzing_include_am.py b/scripts/codegen/fuzzing_include_am.py index 91e9e5f..6e45c21 100755 --- a/scripts/codegen/fuzzing_include_am.py +++ b/scripts/codegen/fuzzing_include_am.py @@ -1,4 +1,4 @@ - +#!/usr/bin/python FUZZERS = """ consensus @@ -33,7 +33,10 @@ FUZZING_LIBS = \ @TOR_ZLIB_LIBS@ @TOR_LIB_MATH@ \ @TOR_LIBEVENT_LIBS@ \ @TOR_OPENSSL_LIBS@ @TOR_LIB_WS32@ @TOR_LIB_GDI@ @CURVE25519_LIBS@ \ - @TOR_SYSTEMD_LIBS@ + @TOR_SYSTEMD_LIBS@ \ + @TOR_LZMA_LIBS@ \ + @TOR_ZSTD_LIBS@ \ + $(rust_ldadd) oss-fuzz-prereqs: \ src/or/libtor-testing.a \ @@ -48,7 +51,7 @@ oss-fuzz-prereqs: \ noinst_HEADERS += \ src/test/fuzz/fuzzing.h -LIBFUZZER = /home/nickm/build/libfuzz/libFuzzer.a +LIBFUZZER = -lFuzzer LIBFUZZER_CPPFLAGS = $(FUZZING_CPPFLAGS) -DLLVM_FUZZ LIBFUZZER_CFLAGS = $(FUZZING_CFLAGS) LIBFUZZER_LDFLAG = $(FUZZING_LDFLAG)

1 0

[tor/maint-0.3.1] Update fuzzing_include_am to include updates from include.am
by nickm＠torproject.org 03 Jul '17

03 Jul '17

1 0

[tor/release-0.3.1] Merge branch 'maint-0.3.1' into release-0.3.1
by nickm＠torproject.org 03 Jul '17

03 Jul '17

commit ba45eec74363cd77553a7e3edb0f5bb355241806 Merge: 26c9afc d011126 Author: Nick Mathewson <nickm(a)torproject.org> Date: Mon Jul 3 13:26:01 2017 -0400 Merge branch 'maint-0.3.1' into release-0.3.1 scripts/codegen/fuzzing_include_am.py | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-)

1 0

[tor/master] Merge branch 'maint-0.3.1'
by nickm＠torproject.org 03 Jul '17

03 Jul '17

commit 2778032f4a0b65429c5279db003fd5b354115d1b Merge: 0fe7c42 d011126 Author: Nick Mathewson <nickm(a)torproject.org> Date: Mon Jul 3 13:26:38 2017 -0400 Merge branch 'maint-0.3.1' scripts/codegen/fuzzing_include_am.py | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-)

1 0

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

tor-commits July 2017