In theory it would be better to detect this bug in advance, but this
approach is much simpler, and therefore safer to backport.
This closes tor issue 28973.
Linked connections aren't woken up by libevent due to I/O but rather
artificially so we can, by chunks, empty the spooled object(s).
Commit 5719dfb48f (in 0.3.4.1-alpha) made it
that the schedule_active_linked_connections_event would be only called once at
startup but this is wrong because then we would never go through again the
active linked connections.
Fortunately, everytime a new linked connection is created, the event is
activated and thus we would go through the active list again. On a busy relay,
this issue is mitigated by that but on a slower relays or bridge, a connection
could get stuck for a while until a new directory information request would
show up.
Fixes#28717, #28912
The old implementation did some funky out-of-order lexing, and
tended to parse every port twice if the %d-%d pattern didn't match.
Closes ticket 28853.
I believe we originally added this for "just in case" safety, but it
isn't actually needed -- we never copy uninitialized stack here.
What's more, this one memset is showing up on our startup profiles,
so we ought to remove it.
Closes ticket 28852.
The point of this function is to make sure that the ed25519-based
implementation of curve25519_basepoint() actually works when we
start tor, and use the regular fallback implementation if it
doesn't. But it accounts for 9% of our startup time in the case
when we have directory information, and I think it's safe to make
the test shorter. After all, it has yet to find any actual bugs in
curved25519_scalarmult_basepoint_donna() on any platforms.
Closes ticket 28838.
Add the bootstrap tag name to the log messages, so people
troubleshooting connection problems can look up a symbol instead of a
number. Closes ticket 28731.
Merge Phoul's two lists into teor's list.
Replace the 150 fallbacks originally introduced in Tor 0.3.3.1-alpha in
January 2018 (of which ~115 were still functional), with a list of
157 fallbacks (92 new, 65 existing, 85 removed) generated in
December 2018.
Closes ticket 24803.
Replace the 150 fallbacks originally introduced in Tor 0.3.3.1-alpha in
January 2018 (of which ~115 were still functional), with a list of
148 fallbacks (89 new, 59 existing, 91 removed) generated in
December 2018.
Closes ticket 24803.
If a relay matches at least one fingerprint, IPv4 address, or IPv6
address in the fallback whitelist, it can become a fallback. This
reduces the work required to keep the list up to date.
Closes ticket 28768.
Tor clients on 0.3.5.6-rc? and later will use a consensus that will become
valid up to 24 hours in the future.
Clients on 0.3.5.5-alpha? and earlier won't accept future consensuses.
Update the fallback expiry tolerance to match tor's checks.
Part of 28768, follow-up on 28591.
Tor clients will use a consensus that expired up to 24 hours ago.
Clients on 0.3.5.5-alpha? and earlier won't select guards from an expired
consensus, but they can still bootstrap if they have existing guards.
Update the fallback expiry tolerance to match tor's checks.
Part of 28768, follow-up on 24661.
When retrying all SOCKS connection because new directory information just
arrived, do not BUG() if a connection in state AP_CONN_STATE_RENDDESC_WAIT is
found to have a usable descriptor.
There is a rare case when this can happen as detailed in #28669 so the right
thing to do is put that connection back in circuit wait state so the
descriptor can be retried.
Fixes#28669
Signed-off-by: David Goulet <dgoulet@torproject.org>
Removing a ".auth" file revokes a client access to the service but the
rendezvous circuit is not closed service side because the service simply
doesn't know which circuit is for which client.
This commit notes in the man page that to fully revoke a client access to the
service, the tor process should be restarted.
Closes#28275
Signed-off-by: David Goulet <dgoulet@torproject.org>
This patch explicitly specifies the path to our OpenSSL dependency and
disables the installation of an external OpenSSL version and instead
uses the OpenSSL version available from the MinGW environments.
See: https://bugs.torproject.org/28574
Specifically, if the consensus is older than the (estimted or
measured) release date for this version of tor, we assume that the
required versions may have changed in between that consensus and
this release.
Implements ticket 27735 and proposal 297.
This patch has routers use the same canonicalization logic as
authorities when encoding their family lists. Additionally, they
now warn if any router in their list is given by nickname, since
that's error-prone.
This patch also adds some long-overdue tests for family formatting.
To succesful compile tor-print-ed-signing-cert.exe on Windows we
sometimes need to include the @TOR_LIB_GDI@ library.
See: https://bugs.torproject.org/28485
By adding a file to the ClientOnionAuthDir and sending a HUP signal, tor would
load the new file and use it. However, that doesn't work with the Sandbox
since post initilization, nothing can be changed.
Document in the manpage that limitation within the Sandbox description.
Closes#28128
Signed-off-by: David Goulet <dgoulet@torproject.org>
Correctly identify Windows 8.1, Windows 10, and Windows Server 2008
and later from their NT versions.
On recent Windows versions, the GetVersionEx() function may report
an earlier Windows version than the running OS. To avoid user
confusion, add "[or later]" to Tor's version string on affected
versions of Windows.
Remove Windows versions that were never supported by the
GetVersionEx() function.
Stop duplicating the latest Windows version in get_uname().
Fixes bug 28096; bugfix on 0.2.2.34; reported by Keifer Bly.
In conn_close_if_marked(), we can decide to keep a connection open that still
has data to flush on the wire if it is being rate limited on the write side.
However, in this process, we were also looking at the read() side which can
still have token in its bucket and thus not stop the reading. This lead to a
BUG() introduced in 0.3.4.1-alpha that was expecting the read side to be
closed due to the rate limit but which only applies on the write side.
This commit removes any bandwidth check on the read side and simply stop the
read side on the connection regardless of the bucket state. If we keep the
connection open to flush it out before close, we should not read anything.
Fixes#27750
Signed-off-by: David Goulet <dgoulet@torproject.org>
Apparently some freebsd compilers can't tell that 'c' will never
be used uninitialized.
Fixes bug 28413; bugfix on 0.2.9.3-alpha when we added support for
longer AES keys to this function.
We don't use this syscall, but openssl apparently does.
(This syscall puts a socket into a half-closed state. Don't worry:
It doesn't shut down the system or anything.)
Fixes bug 28183; bugfix on 0.2.5.1-alpha where the sandbox was
introduced.
Our tests showed that this function is responsible for a huge number
of our malloc/free() calls. It's a prime candidate for being
memoized.
Closes ticket 27225.
Resume refusing to start with relative file paths and RunAsDaemon
set (regression from the fix for bug 22731).
Fixes bug 28298; bugfix on 0.3.3.1-alpha.
If tor terminates due to SIGNAL HALT before test_rebind.py calls
tor_process.terminate(), an OSError 3 (no such process) is thrown.
Fixes part of bug 27968 on 0.3.5.1-alpha.
Remember, you can't check to see if there are N bytes left in a
buffer by doing (buf + N < end), since the buf + N computation might
take you off the end of the buffer and result in undefined behavior.
Fixes 28202; bugfix on 0.2.0.3-alpha.
It is not enough to look at protover for v3 rendezvous support but also we
need to make sure that the curve25519 onion key is present or in other words
that the descriptor has been fetched and does contain it.
Fixes#27797.
Signed-off-by: David Goulet <dgoulet@torproject.org>
Treat backtrace test failures as expected on NetBSD, OpenBSD, and
macOS/Darwin, until we solve bug 17808.
(FreeBSD failures have been treated as expected since 18204 in 0.2.8.)
Fixes bug 27948; bugfix on 0.2.5.2-alpha.
Before this commit, we would create the descriptor signing key certificate
when first building the descriptor.
In some extreme cases, it lead to the expiry of the certificate which triggers
a BUG() when encoding the descriptor before uploading.
Ticket #27838 details a possible scenario in which this can happen. It is an
edge case where tor losts internet connectivity, notices it and closes all
circuits. When it came back up, the HS subsystem noticed that it had no
introduction circuits, created them and tried to upload the descriptor.
However, in the meantime, if tor did lack a live consensus because it is
currently seeking to download one, we would consider that we don't need to
rotate the descriptors leading to using the expired signing key certificate.
That being said, this commit does a bit more to make this process cleaner.
There are a series of things that we need to "refresh" before uploading a
descriptor: signing key cert, intro points and revision counter.
A refresh function is added to deal with all mutable descriptor fields. It in
turn simplified a bit the code surrounding the creation of the plaintext data.
We keep creating the cert when building the descriptor in order to accomodate
the unit tests. However, it is replaced every single time the descriptor is
uploaded.
Fixes#27838
Signed-off-by: David Goulet <dgoulet@torproject.org>
When storing a descriptor in the client cache, if we are about to replace an
existing descriptor, make sure to close every introduction circuits of the old
descriptor so we don't have leftovers lying around.
Ticket 27471 describes a situation where tor is sending an INTRODUCE1 cell on
an introduction circuit for which it doesn't have a matching intro point
object (taken from the descriptor).
The main theory is that, after a new descriptor showed up, the introduction
points changed which led to selecting an introduction circuit not used by the
service anymore thus for which we are unable to find the corresponding
introduction point within the descriptor we just fetched.
Closes#27471.
Signed-off-by: David Goulet <dgoulet@torproject.org>
It won't be used if there are no authorized client configured. We do that so
we can easily support the addition of a client with a HUP signal which allow
us to avoid more complex code path to generate that cookie if we have at least
one client auth and we had none before.
Fixes#27995
Signed-off-by: David Goulet <dgoulet@torproject.org>
Both client and service had their own code for this. Consolidate into one
place so we avoid duplication.
Closes#27549
Signed-off-by: David Goulet <dgoulet@torproject.org>
Occasionally, key pinning doesn't catch a relay that shares an ed25519
ID with another relay. Log the identity fingerprints and the shared
ed25519 ID when this happens, instead of making a BUG() warning.
Fixes bug 27800; bugfix on 0.3.2.1-alpha.
Commit 488e2b00bf introduced an issue, most
likely introduced by a bad copy paste, that made us stop reading on the
connection if our write bandwidth limit was reached.
The problem is that because "read_blocked_on_bw" was never set, the connection
was never reenabled for reading.
This is most likely the cause of #27813 where bytes were accumulating in the
kernel TCP bufers because tor was not doing reads. Only relays with
RelayBandwidthRate would suffer from this but affecting all relays connecting
to them. And using that tor option is recommended and best practice so many
many relays have it enabled.
Fixes#28089.
It turns out that if _only_ the ControlPort is set and nothing else, tor would
simply not bootstrap and thus not start properly. Commit 67a41b6306
removed that requirement for tor to be considered a "client".
Unfortunately, this made the mainloop enable basically nothing if only the
ControlPort is set in the torrc.
This commit now makes it that we also consider the ControlPort when deciding
if we are a Client or not. It does not revert 67a41b6306 meaning
options_any_client_port_set() stays the same, not looking at the control port.
Fixes#27849.
Signed-off-by: David Goulet <dgoulet@torproject.org>
OpenSolaris apparently doesn't have timeradd(), so we added a
replacement, but we weren't including it here after the big
refactoring in 0.3.5.1-alpha.
Fixes bug 27963; bugfix on 0.3.5.1-alpha.
Various places in our code try to activate these events or check
their status, so we should make sure they're initialized as early as
possible. Fixes bug 27861; bugfix on 0.3.5.1-alpha.
When freeing a configuration object from confparse.c in
dump_config(), we need to call the appropriate higher-level free
function (like or_options_free()) and not just config_free().
This only happens with options (since they're the one where
options_validate allocates extra stuff) and only when running
--dump-config with something other than minimal (since
OPTIONS_DUMP_MINIMAL doesn't hit this code).
Fixes bug 27893; bugfix on 0.3.2.1-alpha.
If a tor client gets a descriptor that it can't decrypt, chances are that the
onion requires client authorization.
If a tor client is configured with client authorization for an onion but
decryption fails, it means that the configured keys aren't working anymore.
In both cases, we'll log notice the former and log warn the latter and the
rest of the decryption errors are now at info level.
Two logs statement have been removed because it was redundant and printing the
fetched descriptor in the logs when 80% of it is encrypted wat not helping.
Fixes#27550
Signed-off-by: David Goulet <dgoulet@torproject.org>
When sending the INTRODUCE1 cell, we acquire the needed data for the cell but
if the RP node_t has invalid data, we'll fail the send and completely kill the
SOCKS connection.
Instead, close the rendezvous circuit and return a transient error meaning
that Tor can recover by selecting a new rendezvous point. We'll also do the
same when we are unable to encode the INTRODUCE1 cell for which at that point,
we'll simply take another shot at a new rendezvous point.
Fixes#27774
Signed-off-by: David Goulet <dgoulet@torproject.org>
The result of CString::into_raw() is not safe to free
with free() except under finicky and fragile circumstances
that we definitely don't meet right now.
This was missed in be583a34a3.
When we close a socket via tor_tls_free(), we previously had no way
for our socket accounting logic to learn about it. This meant that
the socket accounting code would think we had run out of sockets,
and freak out.
Fixes bug 27795; bugfix on 0.3.5.1-alpha.
Client side, when a descriptor is finally fetched and stored in the cache, we
then go over all pending SOCKS request for that descriptor. If it turns out
that the intro points are unusable, we close the first SOCKS request but not
the others for the same .onion.
This commit makes it that we'll close all SOCKS requests so we don't let
hanging the other ones.
It also fixes another bug which is having a SOCKS connection in RENDDESC_WAIT
state but with a descriptor in the cache. At some point, tor will expire the
intro failure cache which will make that descriptor usable again. When
retrying all SOCKS connection (retry_all_socks_conn_waiting_for_desc()), we
won't end up in the code path where we have already the descriptor for a
pending request causing a BUG().
Bottom line is that we should never have pending requests (waiting for a
descriptor) with that descriptor in the cache (even if unusable).
Fixees #27410.
Signed-off-by: David Goulet <dgoulet@torproject.org>
It is not enough to look at protover for v3 rendezvous support but also we
need to make sure that the curve25519 onion key is present or in other words
that the descriptor has been fetched and does contain it.
Fixes#27797.
Signed-off-by: David Goulet <dgoulet@torproject.org>
Instead, have it call a mockable function. We don't want
crypto_strongest_rand() to be mockable, since doing so creates a
type error when we call it from ed25519-donna, which we do not build
in a test mode.
Fixes bug 27728; bugfix on 0.3.5.1-alpha
Use the Travis Homebrew addon to install packages on macOS. The package
list is the same, but the Homebrew addon does not do a `brew update` by
default.
This makes builds faster, at the cost of using slightly older packages.
Implements ticket 27738.
This argument was added to match an older idea for the C api, but we
decided not to do it that way in C.
Fixes bug 27741; bugfix on 0.3.3.6 / TROVE-2018-005 fix.
Since the default cache directory is the same as the default data
directory, we don't want the default CacheDirectoryGroupReadable
value (0) to override an explicitly set "DataDirectoryGroupReadable
1".
To fix this, I'm making CacheDirectoryGroupReadable into an
autobool, and having the default (auto) value mean "Use the value of
DataDirectoryGroupReadable if the directories are the same, and 0
otherwise."
Fixes bug 26913; bugfix on 0.3.3.1-alpha when the CacheDirectory
option was introduced.
In hs_config.c, we do validate the permission of the hidden service directory
but we do not try to create it. So, in the event that the directory doesn't
exists, we end up in the loading key code path which checks for the
permission and possibly creates the directory. On failure, don't BUG() since
there is a perfectly valid use case for that function to fail.
Fixes#27335
Signed-off-by: David Goulet <dgoulet@torproject.org>
On GCC and Clang, there's a feature to warn you about bad
conditionals like "if (a = b)", which should be "if (a == b)".
However, they don't warn you if there are extra parentheses around
"a = b".
Unfortunately, the tor_assert() macro and all of its kin have been
passing their inputs through stuff like PREDICT_UNLIKELY(expr) or
PREDICT_UNLIKELY(!(expr)), both of which expand to stuff with more
parentheses around "expr", thus suppressing these warnings.
To fix this, this patch introduces new macros that do not wrap
expr. They're only used when GCC or Clang is enabled (both define
__GNUC__), since they require GCC's "({statement expression})"
syntax extension. They're only used when we're building the
unit-test variant of the object files, since they suppress the
branch-prediction hints.
I've confirmed that tor_assert(), tor_assert_nonfatal(),
tor_assert_nonfatal_once(), BUG(), and IF_BUG_ONCE() all now give
compiler warnings when their argument is an assignment expression.
Fixes bug 27709.
Bugfix on 0.0.6, where we first introduced the "tor_assert()" macro.
.retain() would allocating a Vec of billions of integers and check them
one at a time to separate the supported versions from the unsupported.
This leads to a memory DoS.
Closes ticket 27206. Bugfix on e6625113c9.
Before 0.3.3.1-alpha, we would exit() in this case immediately. But
now that we leave tor_main() more conventionally, we need to make
sure we restore things so as not to cause a double free.
Fixes bug 27708; bugfix on 0.3.3.1-alpha.
All node_get_all_orports() does is allocate and return a smartlist
with at most two tor_addr_port_t members that match ORPort's of
node configuration. This is harmful for memory efficiency, as it
allocates the same stuff every time it is called. However,
node_is_a_configured_bridge() does not need to call it, as it
already has all the information to check if there is configured
bridge for a given node.
The new code is arranged in a way that hopefully makes each succeeding
linear search through bridge_list less likely.
It's possible for a unit test to report success via its pipe, but to
fail as it tries to clean up and exit. Notably, this happens on a
leak sanitizer failure.
Fixes bug 27658; bugfix on 0.2.2.4-alpha when tinytest was
introduced.
Existing cached directory information can cause misleadingly high
bootstrap percentages. To improve user experience, defer reporting of
directory information progress until at least one connection has
succeeded to a relay or bridge.
Closes ticket 27169.
There are three reasons we use a cached_dir_t to hold a consensus:
1. to serve that consensus to a client
2. to apply a consensus diff to an existing consensus
3. to send the consensus to a controller.
But case 1 is dircache-only. Case 2 and case 3 both fall back to
networkstatus_read_cached_consensus(). So there's no reason for us
to store this as a client. Avoiding this saves about 23% of our RAM
usage, according to our experiments last month.
This is, semantically, a partial revert of e5c608e535.
Fixes bug 27247; bugfix on 0.3.0.1-alpha.
As of August 29, 2018, Appveyor images come with gcc 8.2.0 by
default. 64-bit Windows executables compiled with gcc 8.2.0 and
tor's --enable-gcc-hardening crash.
Fixes bug 27460; bugfix on 0.3.4.1-alpha.
It is not supported, and always fails. Some compilers warn about the
function pointer cast on 64-bit Windows.
Fixes bug 27461; bugfix on 0.2.2.23-alpha.
gcc 8 warns that extend_info_t.nickname might be truncated by strncpy().
But it doesn't know that nickname can either contain a hex id, or a
nicknames. hex ids are only used for general and HSDir circuits.
Fixes bug 27463; bugfix on 0.1.1.2-alpha.
