tinc
5 weeks agoFix segfault when failing to read random numbers. 1.1
Guus Sliepen [Mon, 21 Sep 2020 21:22:18 +0000 (23:22 +0200)]
Fix segfault when failing to read random numbers.

Because the result of read() was incorrectly stored in an unsigned
variable, an error reading from the random number generator device would
result in an infinite loop that would start writing out of bounds and
eventually corrupt the stack.

7 months agofix macos build
leptonyu [Sun, 15 Mar 2020 01:39:42 +0000 (01:39 +0000)]
fix macos build

9 months agoInclude stddef.h if available.
Guus Sliepen [Mon, 20 Jan 2020 20:12:17 +0000 (21:12 +0100)]
Include stddef.h if available.

This is necessary to compile fd_device.c.

9 months agofd_device: allow fd to be passed through a unix socket
pacien [Mon, 20 Jan 2020 12:58:13 +0000 (13:58 +0100)]
fd_device: allow fd to be passed through a unix socket

New restrictions on the Android OS forbid direct leaking of file descriptors.
This patch allows the tinc daemon to have an fd and the associated
permissions transferred to it through a Unix domain socket.

15 months agoTry harder to connect to unreachable nodes.
Guus Sliepen [Wed, 17 Jul 2019 23:49:48 +0000 (01:49 +0200)]
Try harder to connect to unreachable nodes.

15 months agoDon't keep an address cache in an outgoing_t.
Guus Sliepen [Wed, 17 Jul 2019 23:33:38 +0000 (01:33 +0200)]
Don't keep an address cache in an outgoing_t.

Address caches are associated with nodes, so just use the address cache
in the node_t struct. Also ensure we always have opened an address cache in
setup_outgoing_connection().

Thanks to admincheg for finding this issue.

15 months agoDisable AutoConnect in the ns-ping test.
Guus Sliepen [Wed, 17 Jul 2019 22:31:54 +0000 (00:31 +0200)]
Disable AutoConnect in the ns-ping test.

15 months agoAvoid void pointer arithmetic.
Guus Sliepen [Wed, 17 Jul 2019 22:31:11 +0000 (00:31 +0200)]
Avoid void pointer arithmetic.

18 months agoFix strict aliasing violation in inet_checksum()
Maciej S. Szmigiero [Tue, 16 Apr 2019 13:00:50 +0000 (15:00 +0200)]
Fix strict aliasing violation in inet_checksum()

inet_checksum() accesses packet data as an array of uint16_t, but the
packet data can be for example of "anonymous struct pseudo" type from
route_ipv6_unreachable().
This type isn't a compatible type with uint16_t so a strict aliasing
violation occurs and causes the checksum to be computed incorrectly.

Fix this by using the memcpy() idiom to read the packet data as an array of
uint16_t in inet_checksum() (this should be understood by compilers and
optimized accordingly, so no actual copy occurs).

18 months agoRevert "Work around a GCC bug that causes inet_checksum() to give wrong results."
Maciej S. Szmigiero [Tue, 16 Apr 2019 12:19:48 +0000 (14:19 +0200)]
Revert "Work around a GCC bug that causes inet_checksum() to give wrong results."

This reverts commit 7c73cb3ace6659df58ec2382b8d47bb521dad886.

19 months agofix: use EVP_DecryptUpdate while decrypting
Andreas Rammhold [Thu, 28 Feb 2019 19:38:14 +0000 (20:38 +0100)]
fix: use EVP_DecryptUpdate while decrypting

With OpenSSL versions 1.0.2r & 1.1.1b there were changes in regards to
how OpenSSL treats misuse of Encrypt/Decrypt EVP methods in the opposite
case. E.g. using the encrypt methods in a decrypt context. OpenSSL now
returns an error in these situations. [1]
Since tinc used the EVP_EncryptUpdate function in the cipher_decrypt
function the new sanity check was triggered causing tinc to be unusable
with said OpenSSL versions.

[1] https://github.com/openssl/openssl/pull/7852

22 months agoPrevent large amounts of UDP probes being sent consecutively.
Guus Sliepen [Tue, 18 Dec 2018 16:44:08 +0000 (17:44 +0100)]
Prevent large amounts of UDP probes being sent consecutively.

We cannot reset udp_ping_sent to zero when we receive a valid reply to
an UDP probe, because that would cause a new one to be sent immediately
in try_udp(). Instead, add a bit to node_status_t to keep track of whether we
have a UDP probe that's waiting for a reply.

Thanks to Ronny Nilsson for spotting the source of the problem.

22 months agoFix the scripts test.
Guus Sliepen [Fri, 30 Nov 2018 13:47:52 +0000 (14:47 +0100)]
Fix the scripts test.

This was broken because of the changes in commit 44f3023.

22 months agoDouble-quote node names in dump graph output.
Guus Sliepen [Fri, 30 Nov 2018 13:41:55 +0000 (14:41 +0100)]
Double-quote node names in dump graph output.

This is needed for all nodes with a name starting with a digit,
otherwise the ID would be interpreted as a numeral.

Based on the patch from Quentin Rameau for tinc 1.0.

22 months agoGenerate tinc-up.bat on windows
Fabian Maurer [Mon, 19 Nov 2018 21:11:10 +0000 (22:11 +0100)]
Generate tinc-up.bat on windows

23 months agoFix interface spelling
Fabian Maurer [Mon, 19 Nov 2018 21:05:13 +0000 (22:05 +0100)]
Fix interface spelling

2 years agoHandle DOS line endings in invitation files.
Guus Sliepen [Mon, 22 Oct 2018 18:34:19 +0000 (20:34 +0200)]
Handle DOS line endings in invitation files.

2 years agoAttempt to make the test suite work with Windows executables.
Guus Sliepen [Mon, 22 Oct 2018 18:31:37 +0000 (20:31 +0200)]
Attempt to make the test suite work with Windows executables.

The test suite still assumes a POSIX shell to run the tests, but now handles
the case when the executables themselves are (cross-)compiled for Windows,
with a .exe extension. Also, DOS line endings must be converted to UNIX
line endings in some cases.

Some tests now pass on Linux+Wine, but others do not, mainly due to Wine
not handling services very well.

2 years agoPrevent sptps_test from sending overly large UDP packets.
Guus Sliepen [Mon, 22 Oct 2018 15:20:30 +0000 (17:20 +0200)]
Prevent sptps_test from sending overly large UDP packets.

This fixes the test suite on macOS, which has a lo0 interface with an MTU
of 16384, which is smaller than the maximum packet size we could send.

2 years agoMake sure the stop command works on Windows if tincd is running in the foreground.
Guus Sliepen [Mon, 22 Oct 2018 15:01:08 +0000 (17:01 +0200)]
Make sure the stop command works on Windows if tincd is running in the foreground.

2 years agoDrop support for Cygwin.
Guus Sliepen [Mon, 22 Oct 2018 14:53:12 +0000 (16:53 +0200)]
Drop support for Cygwin.

Tinc can be built as a native Windows binary using MinGW, and should support
all of the features that the Cygwin version did.

2 years agoFix compiling test binaries on Windows.
Guus Sliepen [Sun, 21 Oct 2018 14:24:10 +0000 (16:24 +0200)]
Fix compiling test binaries on Windows.

2 years agoFix warnings when compiling for Windows.
Guus Sliepen [Sun, 21 Oct 2018 14:23:45 +0000 (16:23 +0200)]
Fix warnings when compiling for Windows.

2 years agoSkip the legacy protocol test if that protocol is disabled.
Guus Sliepen [Thu, 18 Oct 2018 19:41:52 +0000 (21:41 +0200)]
Skip the legacy protocol test if that protocol is disabled.

2 years agoAllow "tinc --force join" to accept all variables sent in an invitaiton.
Guus Sliepen [Thu, 18 Oct 2018 15:19:47 +0000 (17:19 +0200)]
Allow "tinc --force join" to accept all variables sent in an invitaiton.

2 years agoMake more variables safe for use in invitations.
Guus Sliepen [Thu, 18 Oct 2018 15:17:20 +0000 (17:17 +0200)]
Make more variables safe for use in invitations.

When writing one's own invitation files, more variables are now accepted
by the invitee. The goal is to allow anything that doesn't interfere
with the existing network configuration of the invitee and that doesn't
cause any unexpected behaviour, such as starting running commands.

2 years agoFix segfault when dest->mtu is 0.
Werner Schreiber [Wed, 10 Oct 2018 17:16:59 +0000 (19:16 +0200)]
Fix segfault when dest->mtu is 0.

2 years agoFix building with --disable-legacy-protocol.
Guus Sliepen [Thu, 18 Oct 2018 14:42:18 +0000 (16:42 +0200)]
Fix building with --disable-legacy-protocol.

2 years agoReformat the code using astyle.
Guus Sliepen [Thu, 18 Oct 2018 14:15:19 +0000 (16:15 +0200)]
Reformat the code using astyle.

2 years agoCheck all Address statements when making outgoing connections.
Guus Sliepen [Thu, 18 Oct 2018 13:59:24 +0000 (15:59 +0200)]
Check all Address statements when making outgoing connections.

The logic in get_recent_address() caused tinc to only consider the first
Address statement that could be resolved to one or more IP addresses.

2 years agoUse the onlink flag when adding routes on Linux.
Guus Sliepen [Thu, 18 Oct 2018 14:10:32 +0000 (16:10 +0200)]
Use the onlink flag when adding routes on Linux.

As reported by iczero, adding gateway routes on Linux can fail if the
gateway address is not reachable yet, either because the interface is
down, or if the gateway address is not inside any route that already is
added or is being added to that interface. By adding the onlink flag,
iproute2 will just add the route without questions.

2 years agoFix tinc-up generation on windows
iczero [Wed, 17 Oct 2018 05:58:52 +0000 (22:58 -0700)]
Fix tinc-up generation on windows

- use `%INTERFACE%` instead of `$INTERFACE` on windows
- correct typo in `netsh interface` (was `netsh inetface`)
- remove `static` when setting ipv6 address

2 years agoFix manpage mdoc syntax
Shengjing Zhu [Wed, 10 Oct 2018 07:00:47 +0000 (15:00 +0800)]
Fix manpage mdoc syntax

2 years agoInstall the bash completion file when running make install.
Guus Sliepen [Mon, 8 Oct 2018 20:12:23 +0000 (22:12 +0200)]
Install the bash completion file when running make install.

Closes #209 on GitHub.

2 years agoFix the compiler attribute test to work with Clang.
Guus Sliepen [Mon, 8 Oct 2018 19:27:08 +0000 (21:27 +0200)]
Fix the compiler attribute test to work with Clang.

Clang doesn't like the __nonnull__ attribute being applied to functions
that don't take pointer arguments, and this causes errors when running
the configure script.

2 years agoReleasing 1.1pre17. release-1.1pre17
Guus Sliepen [Mon, 8 Oct 2018 09:00:01 +0000 (11:00 +0200)]
Releasing 1.1pre17.

2 years agoUpdate THANKS.
Guus Sliepen [Sun, 7 Oct 2018 16:05:50 +0000 (18:05 +0200)]
Update THANKS.

2 years agoOpenBSD has a proper tap device.
Rafael Sadowski [Mon, 1 Oct 2018 13:14:24 +0000 (15:14 +0200)]
OpenBSD has a proper tap device.

2 years agoUpdate README and links to required libraries.
Guus Sliepen [Sun, 7 Oct 2018 11:41:23 +0000 (13:41 +0200)]
Update README and links to required libraries.

2 years agoDon't check for NULL-pointers before calling free().
Guus Sliepen [Sun, 7 Oct 2018 11:32:25 +0000 (13:32 +0200)]
Don't check for NULL-pointers before calling free().

2 years agoFix spelling errors.
Guus Sliepen [Sun, 7 Oct 2018 11:05:49 +0000 (13:05 +0200)]
Fix spelling errors.

Found by codespell.

2 years agoFix all warnings when compiling with -Wall -W -pedantic.
Guus Sliepen [Sat, 6 Oct 2018 21:31:05 +0000 (23:31 +0200)]
Fix all warnings when compiling with -Wall -W -pedantic.

2 years agoFix warnings from the Clang static analyzer.
Guus Sliepen [Sat, 6 Oct 2018 16:18:45 +0000 (18:18 +0200)]
Fix warnings from the Clang static analyzer.

2 years agoFix compiler warnings.
Guus Sliepen [Sat, 6 Oct 2018 15:51:41 +0000 (17:51 +0200)]
Fix compiler warnings.

2 years agoAdd a test for backwards compatibility with the legacy protocol.
Guus Sliepen [Sun, 9 Sep 2018 20:13:43 +0000 (22:13 +0200)]
Add a test for backwards compatibility with the legacy protocol.

2 years agoPrevent oracle attacks in the legacy protocol (CVE-2018-16737, CVE-2018-16738)
Guus Sliepen [Sun, 9 Sep 2018 16:19:15 +0000 (18:19 +0200)]
Prevent oracle attacks in the legacy protocol (CVE-2018-16737, CVE-2018-16738)

The legacy authentication protocol allows an oracle attack that could
potentially be exploited. This commit contains several mitigations:

- Connections are no longer closed immediately on error, but put in
  a "tarpit".
- The authentication protocol now requires a valid CHAL_REPLY from the
  initiator of a connection before sending a CHAL_REPLY of its own.
- Reduce the amount of connections per second accepted.
- Null ciphers or digests are no longer allowed in METAKEYs.
- Connections that claim to have the same name as the local node are
  rejected.

Just to be on the safe side:

- The new protocol now requires a valid SIG from the initiator of a
  connection before sending a SIG of its own.

2 years agoEnable AutoConnect by default.
Guus Sliepen [Sun, 24 Jun 2018 14:19:10 +0000 (16:19 +0200)]
Enable AutoConnect by default.

2 years agoRemove address cache debug messages printed to stderr.
Guus Sliepen [Sat, 23 Jun 2018 20:32:19 +0000 (22:32 +0200)]
Remove address cache debug messages printed to stderr.

2 years agoAvoid treating compressed MTU probes as having a negative length.
Guus Sliepen [Sat, 23 Jun 2018 20:26:12 +0000 (22:26 +0200)]
Avoid treating compressed MTU probes as having a negative length.

This was not harmful, but caused negative values being logged.

2 years agoPrint UDP RTT on its own line.
Guus Sliepen [Wed, 13 Jun 2018 20:41:02 +0000 (22:41 +0200)]
Print UDP RTT on its own line.

2 years agoMerge remote-tracking branch 'volth/release-1.1pre16-rtt' into 1.1
Guus Sliepen [Wed, 13 Jun 2018 20:23:27 +0000 (22:23 +0200)]
Merge remote-tracking branch 'volth/release-1.1pre16-rtt' into 1.1

Also, reformat the code and fix a compiler warning.

2 years agoAdd missing item and attribution to NEWS.
Guus Sliepen [Tue, 12 Jun 2018 18:50:58 +0000 (20:50 +0200)]
Add missing item and attribution to NEWS.

2 years agoRemove the ping test.
Guus Sliepen [Tue, 12 Jun 2018 18:50:37 +0000 (20:50 +0200)]
Remove the ping test.

This was kind of a hack. The ns-ping test is a much better test, although
it only works on Linux.

2 years agominor
volth [Wed, 13 Jun 2018 18:10:47 +0000 (18:10 +0000)]
minor

2 years agoexpose traffic stats to 'tinc info ___' and 'tinc dump nodes'
volth [Wed, 13 Jun 2018 14:12:02 +0000 (14:12 +0000)]
expose traffic stats to 'tinc info ___' and 'tinc dump nodes'

2 years agokeep track of round trip times of UDP pings
volth [Tue, 12 Jun 2018 21:17:30 +0000 (21:17 +0000)]
keep track of round trip times of UDP pings

2 years agoReleasing 1.1pre16. release-1.1pre16
Guus Sliepen [Tue, 12 Jun 2018 18:01:43 +0000 (20:01 +0200)]
Releasing 1.1pre16.

2 years agoRemove the wxPython GUI.
Guus Sliepen [Tue, 12 Jun 2018 17:47:02 +0000 (19:47 +0200)]
Remove the wxPython GUI.

This GUI is missing a lot of functionality, and won't be part of an 1.1.0
release. Therefore, it's being removed, and might be brought back after
1.1.0.

2 years agoFix compiling when support for UML sockets is enabled.
Oliver Freyermuth [Wed, 4 Apr 2018 20:01:52 +0000 (22:01 +0200)]
Fix compiling when support for UML sockets is enabled.

2 years agoReformat all code using astyle.
Guus Sliepen [Mon, 2 Apr 2018 14:49:06 +0000 (16:49 +0200)]
Reformat all code using astyle.

2 years agoAdd the ability to set a firewall mark on sockets.
Guus Sliepen [Mon, 2 Apr 2018 14:33:14 +0000 (16:33 +0200)]
Add the ability to set a firewall mark on sockets.

The FWMark option is added, when set it will use setsockopt(SOL_SOCKET, SO_MARK)
to set the given value as the mark on all sockets created by tinc.

Thanks to Olivier Tirat for submitting a similar patch in the past.

2 years agoFix invitation tests if port 655 is available.
Guus Sliepen [Mon, 2 Apr 2018 10:29:46 +0000 (12:29 +0200)]
Fix invitation tests if port 655 is available.

Running the "del Port" command when no Port is set returns an error. But
it is not necessary anyway since the later "set Port" command will
overwrite it.

2 years agoPrevent an infinite loop in get_recent_address().
Guus Sliepen [Fri, 30 Mar 2018 09:50:40 +0000 (11:50 +0200)]
Prevent an infinite loop in get_recent_address().

When a node is offline, but we still have edges to it that have the same
address as we already have in our address cache, an infinite loop would
happen in get_recent_address(), because we forgot to advance the pointer
in the list of known addresses, and kept looking at the same one over
and over.

Thanks to Sven-Haegar Koch for spotting the bug and providing
diagnostics.

2 years agoProperly implement tinc.texi's dependency on tincinclude.texi.
Guus Sliepen [Fri, 30 Mar 2018 09:38:22 +0000 (11:38 +0200)]
Properly implement tinc.texi's dependency on tincinclude.texi.

With this, make distcheck succeeds even without the info-in-builddir
option to AM_INIT_AUTOMAKE.

2 years agoWarn if we cannot reload the tincd when creating an invitation.
Guus Sliepen [Tue, 27 Mar 2018 20:54:15 +0000 (22:54 +0200)]
Warn if we cannot reload the tincd when creating an invitation.

2 years agoFix handling partial SPTPS messages in sptps_test.
Mike Sullivan [Mon, 26 Mar 2018 19:19:04 +0000 (14:19 -0500)]
Fix handling partial SPTPS messages in sptps_test.

2 years agoEnsure we call CloseServiceHandle() in case of errors.
Guus Sliepen [Tue, 6 Mar 2018 18:31:17 +0000 (19:31 +0100)]
Ensure we call CloseServiceHandle() in case of errors.

2 years agoReformat all code using astyle.
Guus Sliepen [Tue, 6 Mar 2018 18:24:14 +0000 (19:24 +0100)]
Reformat all code using astyle.

2 years agofix service removal.
Gjergji [Tue, 6 Mar 2018 10:10:41 +0000 (11:10 +0100)]
fix service removal.
Windows service was not removed until tincctl exits

2 years agoFix compatibility with LibreSSL and OpenSSL < 1.1.
Guus Sliepen [Wed, 28 Feb 2018 20:34:48 +0000 (21:34 +0100)]
Fix compatibility with LibreSSL and OpenSSL < 1.1.

Closes #184 on GitHub.

2 years agoTry to process all pending events after select().
Guus Sliepen [Wed, 28 Feb 2018 20:28:16 +0000 (21:28 +0100)]
Try to process all pending events after select().

If we break out of the loop every time at the first filedescriptor that
is read/writeable, we risk starving the other filedescriptors.

2 years agoCall WSAWaitForMultipleEvents() in a loop until we have checked all events.
Todd C. Miller [Tue, 27 Feb 2018 21:20:46 +0000 (14:20 -0700)]
Call WSAWaitForMultipleEvents() in a loop until we have checked all events.
WSAWaitForMultipleEvents() only returns the index of the first event that is read.  We need to call WSAWaitForMultipleEvents() repeatedly to check if other events are also ready.  Otherwise, a single busy event (such as the TAP device) can starve the other events.

2 years agoWork around a GCC bug that causes inet_checksum() to give wrong results.
Guus Sliepen [Tue, 27 Feb 2018 20:08:57 +0000 (21:08 +0100)]
Work around a GCC bug that causes inet_checksum() to give wrong results.

Valgrind reports the following bug:

==24877== Conditional jump or move depends on uninitialised value(s)
==24877==    at 0x12283E: inet_checksum (route.c:80)
==24877==    by 0x12283E: route_ipv6_unreachable (route.c:315)
==24877==    by 0x1236AC: route_ipv6 (route.c:751)
==24877==    by 0x1236AC: route (route.c:1160)
==24877==    by 0x113DE0: receive_tcppacket (net_packet.c:493)
==24877==    by 0x1119D4: receive_meta (meta.c:315)
==24877==    by 0x113288: handle_meta_connection_data (net.c:287)
==24877==    by 0x11A091: handle_meta_io (net_socket.c:491)
==24877==    by 0x10FB0C: event_loop (event.c:370)
==24877==    by 0x11362E: main_loop (net.c:489)
==24877==    by 0x10CACA: main (tincd.c:551)

Clearing the variable pseudo in route_ipv6_unreachable removes this error,
but the resulting checksum is still bad. If one instead adds a dummy
write that depends on checksum, the error goes away and the checksum is
correct.

2 years agoRevert "Unconditionally remove timeouts from the queue before calling the callback."
Guus Sliepen [Tue, 27 Feb 2018 18:11:38 +0000 (19:11 +0100)]
Revert "Unconditionally remove timeouts from the queue before calling the callback."

This reverts commit e8a60109fc91a42420ec626b63956771675f89b0.

2 years agoUnconditionally remove timeouts from the queue before calling the callback.
Guus Sliepen [Mon, 26 Feb 2018 21:19:43 +0000 (22:19 +0100)]
Unconditionally remove timeouts from the queue before calling the callback.

We are going to unlink the timeout from the splay tree anyway, so do it
unconditionally before the callback, instead of waiting until after the
callback to check whether or not to remove it based on its expiration
time.

2 years agoIn device_handle_read() we need to reset the read event on error or
Todd C. Miller [Thu, 22 Feb 2018 21:27:37 +0000 (14:27 -0700)]
In device_handle_read() we need to reset the read event on error or
it will keep firing.  This is easy to reproduce by suspending the
machine while tinc is running.

2 years agoUpdate the documentation of the control protocol.
Guus Sliepen [Wed, 21 Feb 2018 19:34:42 +0000 (20:34 +0100)]
Update the documentation of the control protocol.

2 years agoFix heap corruption on Windows exposed by the use-after free fix.
Todd C. Miller [Wed, 21 Feb 2018 03:18:38 +0000 (20:18 -0700)]
Fix heap corruption on Windows exposed by the use-after free fix.
reset_address_cache() could call free_known_addresses() on a struct
addrinfo * that was returned by getaddrinfo().  It seems safest to just
make a copy of the addresses returned by getaddrinfo() so we can always
use free_known_addresses() instead of trying to determine whether or
not we need to use freeaddrinfo().

2 years agoDocument the control protocol.
Guus Sliepen [Mon, 19 Feb 2018 19:41:21 +0000 (20:41 +0100)]
Document the control protocol.

2 years agoReduce memory allocations due to HMAC() and EVP_MD_*().
Guus Sliepen [Sun, 18 Feb 2018 15:51:06 +0000 (16:51 +0100)]
Reduce memory allocations due to HMAC() and EVP_MD_*().

HMAC() allocates a temporary buffer on the heap each time it is called.
Similarly, we called EVP_MD_CTX_create() every time we wanted to
calculate a hash. Use HMAC_CTX and EVP_MD_CTX variables to store the
state so no (re)allocations are necessary. HMAC() was called for every
legacy packet sent and received.

This issue was found thanks to heaptrack.

2 years agoReduce memory allocations due to zlib's uncompress().
Guus Sliepen [Sun, 18 Feb 2018 14:38:12 +0000 (15:38 +0100)]
Reduce memory allocations due to zlib's uncompress().

Everytime uncompress() is called, zlib allocates some buffer on the heap
and frees it again. When compression is enabled, this is the biggest source
of memory allocations in tinc. Instead of using this function, use
inflate(), which can store its state in a z_stream variable, which avoids
(re)allocating memory for every packet received.

This issue was found thanks to heaptrack.

2 years agoAdd code coverage testing support.
Guus Sliepen [Sun, 18 Feb 2018 14:33:36 +0000 (15:33 +0100)]
Add code coverage testing support.

Allows configure to be run with the --enable-code-coverage flag, allowing
one to run make check-code-coverage, which runs the test suite and produces
a code coverage report.

2 years agoReformat all code using astyle.
Guus Sliepen [Sat, 17 Feb 2018 13:39:29 +0000 (14:39 +0100)]
Reformat all code using astyle.

2 years agoFix a use-after-free bug in get_recent_address() and two related issues.
Todd C. Miller [Fri, 16 Feb 2018 21:17:39 +0000 (14:17 -0700)]
Fix a use-after-free bug in get_recent_address() and two related issues.
1) The sockaddr_t * returned may be part of memory freed by the call to
   freeaddrinfo().
2) The sockaddr_t * returned from a recently seen address not in the
   cache was cast from struct addrinfo *ai, not the struct sockaddr *
   inside of it.
3) In do_outgoing_connection(), when filling in the address in the
   connection_t, there is a buffer overflow (read, not write) if
   the sa returned by get_recent_address() didn't come from the
   cache of recently seen addresses.  That is, it was really a
   struct sockaddr * and not a sockaddr_t *.  This last was
   found by building tinc with address sanitizer.

2 years agoIn device_issue_read() there is no need to reset Offset and OffsetHigh
Todd C. Miller [Wed, 31 Jan 2018 21:55:20 +0000 (14:55 -0700)]
In device_issue_read() there is no need to reset Offset and OffsetHigh
to 0; they are only used for seekable files (not sockets).

Reset the write event before the call to WriteFile().  This is
consistent with how the read event is reset before ReadFile().

Clear device_write_packet.len() if WriteFile() fails with an error
other than ERROR_IO_PENDING; otherwise write_packet() will call
GetOverlappedResult() the next time it is run even though there is
no write in progress.

2 years agoWSAEVENT is a pointer, so we cannot simply return the different of two
Todd C. Miller [Tue, 23 Jan 2018 22:57:58 +0000 (15:57 -0700)]
WSAEVENT is a pointer, so we cannot simply return the different of two
events in io_compare(), which returns an int.  This can return the wrong
result for 64-bit executables.

2 years agoAdd some missing freeaddrinfo() calls to avoid leaking memory.
Todd C. Miller [Mon, 22 Jan 2018 17:27:16 +0000 (10:27 -0700)]
Add some missing freeaddrinfo() calls to avoid leaking memory.

2 years agoFix calling freeaddrinfo() on the wrong pointer.
Guus Sliepen [Mon, 22 Jan 2018 17:05:09 +0000 (18:05 +0100)]
Fix calling freeaddrinfo() on the wrong pointer.

Thanks to Todd C. Miller for finding this issue.

2 years agoMove ResetEvent() call before ReadFile().
Etienne Dechamps [Wed, 17 Jan 2018 19:37:53 +0000 (19:37 +0000)]
Move ResetEvent() call before ReadFile().

Commit 313a752 changed the Windows device code such that ResetEvent() is
called on the read OVERLAPPED structure before GetOverlappedResult(), as
opposed to before ReadFile(). In [1] Guus pointed out that this doesn't
make a ton of sense, and I agree with him; it must have been an
oversight on my part when I wrote this code.

Surprisingly, none of this makes any difference in my testing, at least
with the standard TAP 9.0.0.9 driver. Nevertheless, this code is
probably wrong and fixing it will make me sleep better at night.

[1]: https://www.tinc-vpn.org/pipermail/tinc/2018-January/005091.html

2 years agoFix "use of GNU empty initializer extension" warning.
Etienne Dechamps [Sun, 7 Jan 2018 14:48:08 +0000 (14:48 +0000)]
Fix "use of GNU empty initializer extension" warning.

2 years agoFix "void function should not return void expression" warning.
Etienne Dechamps [Sun, 7 Jan 2018 14:44:12 +0000 (14:44 +0000)]
Fix "void function should not return void expression" warning.

2 years agoFix AC_CHECK_DECLS usage in openssl.m4.
Etienne Dechamps [Sun, 7 Jan 2018 14:26:00 +0000 (14:26 +0000)]
Fix AC_CHECK_DECLS usage in openssl.m4.

See:

  https://www.gnu.org/software/autoconf/manual/autoconf-2.62/html_node/Generic-Declarations.html

"For each of the symbols (comma-separated list)"

When building with aggressive warning settings the current code results
in the following configure test code being generated:

  #ifndef OpenSSL_add_all_algorithms EVP_aes_256_cfb
  #ifdef __cplusplus
    (void) OpenSSL_add_all_algorithms EVP_aes_256_cfb;
  #else
    (void) OpenSSL_add_all_algorithms EVP_aes_256_cfb;
  #endif
  #endif

Which is obviously wrong and makes the configure check fail.

2 years agoAdd a cache of recently seen addresses.
Guus Sliepen [Fri, 5 Jan 2018 21:49:30 +0000 (22:49 +0100)]
Add a cache of recently seen addresses.

This maintains a cache file for each host we have communicated with, either
via TCP or UDP. The cache is used when trying to make outgoing connections,
and is updated whenever a successful TCP or UDP connection is established.
Up to 8 addresses are stored in the cache.

Currently, the cache is stored in /etc/tinc/NETNAME/cache. The directory
has to be manually created to opt in to this feature for now.

2 years agoFix all spelling errors found by codespell.
Guus Sliepen [Sat, 6 Jan 2018 19:46:22 +0000 (20:46 +0100)]
Fix all spelling errors found by codespell.

2 years agoDocument how to enable tinc at boot time using systemd.
Guus Sliepen [Sat, 6 Jan 2018 19:34:37 +0000 (20:34 +0100)]
Document how to enable tinc at boot time using systemd.

2 years agoDon't include generated files into the tarball.
Guus Sliepen [Sat, 6 Jan 2018 15:59:21 +0000 (16:59 +0100)]
Don't include generated files into the tarball.

2 years agoUpdate .gitignore.
Guus Sliepen [Sat, 6 Jan 2018 15:21:19 +0000 (16:21 +0100)]
Update .gitignore.

2 years agoEnsure the sptps-basic test doesn't fail during make distcheck.
Guus Sliepen [Sat, 6 Jan 2018 15:20:25 +0000 (16:20 +0100)]
Ensure the sptps-basic test doesn't fail during make distcheck.

2 years agoSet default systemd unit path to ${libdir}/systemd/system.
Guus Sliepen [Sat, 6 Jan 2018 15:20:03 +0000 (16:20 +0100)]
Set default systemd unit path to ${libdir}/systemd/system.

This installs systemd unit files into /usr/local, just like the binaries.
The systemd documentation claims to read this directory as well.

2 years agoRemove hardcoded paths from systemd unit files.
Guus Sliepen [Sat, 6 Jan 2018 15:17:35 +0000 (16:17 +0100)]
Remove hardcoded paths from systemd unit files.

Closes #160 on GitHub.