Arne Schwabe [Mon, 27 Jul 2015 15:33:11 +0000 (17:33 +0200)]
Fix commit e473b7c if an inline file happens to have a line break exactly at buffer limit
The check does only for strlen(line) space and buf_printf will only use at
most space -1 and not print the final character ('\n') in this corner.
Since a missing \n only breaks certificates at the start and end marker,
missing line breaks otherwise do not trigger this error.
reintroduce md5_digest wrapper struct to fix gcc warnings
I was wrong to assume that adding the const qualifier to the pointer-to-
fixed-size-array contruction used in options_hash_changed_or_zero() was
allowed. GCC actually warns about this, but I was using clang and clang
seems to be fine with the contruction. To make GCC happy too, reintroduce
the md5_digest wrapped struct, and use that when passing around the digest.
This reverts the "struct md5_digest" parts of 827de237860813d2859a, but
keeps the rest.
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1437910039-30101-1-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9949 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Commits da9b292733e929a2900dc32d37f0424c3d588366 and 315f6fbc7f657a7f1127628bd714f468709d5185 broke the use case where we are
asking password from the management client. The password is always asked
before daemonization. With this fix we avoid this and ask it via management
interface in the same spot as before the mentioned commits. Tested on Linux.
v2: This patch was first submitted by Christian Pellegrin (from Google),
and reworked by Steffan Karger (from the OpenVPN team) to also work for
setups with --management-query-passwords but without --auth-user-pass.
Signed-off-by: Steffan Karger <steffan@karger.me> Signed-off-by: Christian Pellegrin <chripell@google.com> Tested-by: Christian Pellegrin <chripell@google.com> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <55A6C46C.5080601@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9927 Signed-off-by: Gert Doering <gert@greenie.muc.de>
James Geboski [Tue, 8 Jan 2013 22:52:57 +0000 (17:52 -0500)]
Fix --askpass not allowing for password input via stdin
This resolves --askpass treating stdin as a file during the file access
check. In turn, this leads to openvpn failing to start if this option is
set to stdin.
By default, --askpass reads the certificate's password from stdin rather
than a file. Without passing the CHKACC_ACPTSTDIN flag to
check_file_access(), stdin is marked as being a nonexistent file.
Trac #248
Signed-off-by: James Geboski <jgeboski@gmail.com> Acked-by: Steffan Karger <steffan.karger@fox-it.com>
Message-Id: <55A41225.2020705@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9918 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Gert Doering [Mon, 13 Jul 2015 19:10:07 +0000 (21:10 +0200)]
Produce a meaningful error message if --daemon gets in the way of asking for passwords.
With the --daemon / SSL init reordering in da9b292733, we fail if we
daemonize first and then try to ask for a private key passphrase (or,
for that matter, username+password if --auth-nocache is set) - but
no meaningful error message was printed, instead depending on operating
system and library versions, either we looped around "ssl init failed"
or died with an unspecified "fatal error".
So: check if get_user_pass_cr() is called in a context that needs
"from_stdin", but both stdin and stderr are not connected to a tty
device (which getpass() needs). In that case, print a meaningful
error message pointing to --askpass, and die.
fix regression: query password before becoming daemon
The init sequence was changed to daemonize before the crypto init to fix
issues on FreeBSD some commits ago. This introduced a regression where
we would no longer query for passwords before daemonizing, as described
in trac #574 and #576.
This commit restores the correct order, and adds a bit of const
correctness since we're touching this now code anyway.
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1436477759-5884-1-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9901 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Signed-off-by: Steffan Karger <steffan@karger.me> Tested-by: Fish Wang <fish.thss@gmail.com> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <CAA1AbxKrPXnc1+PPH4dQV1OksY_0t3ewqRnqVrjhZL+i2Pbkmw@mail.gmail.com>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9859 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Steffan Karger [Tue, 30 Jun 2015 19:44:56 +0000 (21:44 +0200)]
Increase control channel packet size for faster handshakes
Instead of limiting the control channel TCP/UDP packet payload size at
'100 bytes + real control channel overhead' (~140 bytes ethernet payload),
increase the max TCP/UDP payload size to '1250 bytes - calculated overhead'
(~1210 bytes ethernet payload).
Note that this patch does *not* yield an optimal solution, but it is a
simple and rather safe change that will improve connection setup times
significantly.
v2: use the mininum value of --link-mtu and 1250 to give the user a way to
reduce control packet size if really needed.
trac #545
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1435693496-10931-1-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9841 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Holger Kummert [Thu, 25 Jun 2015 16:01:20 +0000 (18:01 +0200)]
Del ipv6 addr on close of linux tun interface
When a linux tun interface is closed (e.g. on disconnect)
an optional ipv6 addr that was previously set is deleted now.
Without this patch a later reconnect could fail with
'Linux ip -6 addr add failed: external program exited with error status: 2'
and openvpn would exit.
This is mainly relevant for persistant tun devices (staying around
after openvpn exits) but can also happen at reconnect.
If addresses are *supposed* to stay around on openvpn exit, run
openvpn with --ifconfig-noexec and configure IPv4/IPv6 addresses
manually before openvpn starts (or using an --up script).
Trac #141
Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1435248080-12670-1-git-send-email-Holger.Kummert@Sophos.com>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9810
Throw an error if an option has extra parameters; previously they
were silently ignored (see also trac #557)
This feature was discussed on the openvpn-devel mailing list
(http://thread.gmane.org/gmane.network.openvpn.devel/9599).
The (modified) message "Unrecognized option or missing or extra
parameter(s)" is used except for a few options:
* The --help option: An extra parameter for --help generates a
specific error message after showing the syntax message. This
is done to help a user who tries "--help tls-cipher" or similar,
hoping to get more information about the "tls-cipher" option.
* The --dhcp-option option: It has its own similar message, into
which " or extra" has been inserted.
* Ten options such as --up that accept a command (instead of a
path) already detect extra parameters and generate specific
error messages that mention double-quoting commands which
contain embedded spaces.
Acked-by: Arne Schwabe <arne@rfc2549.org>
Message-Id: <CAEsd45RvRfmqetu-EoFjSdyBk5F9X1K5muPGu-+_+30TMNOU_A@mail.gmail.com>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9783
URL: https://community.openvpn.net/openvpn/ticket/557
Steffan Karger [Thu, 18 Jun 2015 22:08:45 +0000 (00:08 +0200)]
write pid file immediately after daemonizing
Since we split daemonizing from changing directory in commit da9b292
(f025de005d719201a69ad0313d545a1ddd244752 in release/2.3), we can
now simply write the pid file immediately after daemonizing.
This not only fixes the bug reported in trac #563, but also further
simplifies the code.
trac #563
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1434665325-3225-1-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9793 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Gert Doering [Sun, 26 Apr 2015 18:03:58 +0000 (20:03 +0200)]
Fix FreeBSD ifconfig for topology subnet tunnels.
For "topology subnet", we only pretend to have a subnet and keep
using the tun if in point-to-point mode - but for that to fully
work, the "remote" address needs to be different from the "local"
address. So just arbitrarily construct one from the on-link
subnet - base+1, if "that is not us", base+2, otherwise.
Fix trac #481
See also: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=194745
Signed-off-by: Gert Doering <gert@greenie.muc.de> Tested-By: Anton Sayetsky <vsasjason@gmail.com> Lazy-ACK-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1430071438-31675-1-git-send-email-gert@greenie.muc.de>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9606
Gert Doering [Sun, 31 May 2015 13:59:09 +0000 (15:59 +0200)]
Move res_init() call to inner openvpn_getaddrinfo() loop
A non-working nameserver in /etc/resolv.conf could lead to endless
loops inside openvpn_getaddrinfo(), because many systems will only
pick up changes to resolv.conf if res_init() is called again.
To reproduce, run openvpn with --resolv-retry infinite (2.3) or
--resolv-retry "a high number" (master) on a BSD system. Linux
glibc seems to stat() resolv.conf on calls to getaddrinfo() and
pick up changes automatically.
Trac #523
Signed-off-by: Gert Doering <gert@greenie.muc.de> Acked-by: Arne Schwabe <arne@rfc2549.org>
Message-Id: <1433080749-6892-1-git-send-email-gert@greenie.muc.de>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9763
Guy Yur [Mon, 1 Jun 2015 19:51:13 +0000 (21:51 +0200)]
Fix --redirect-private in --dev tap mode.
When specifying redirect-private option and not specifying route-gateway
or ifconfig options, OpenVPN fails to add the route to the remote host
with the following message:
NOTE: unable to redirect default gateway -- VPN gateway parameter
(--route-gateway or --ifconfig) is missing
In redirect_default_route_to_vpn() the check for remote endpoint happens
even though it is not used by redirect-private - make check conditional
on RG_REROUTE_GW (= --redirect-gateway).
Trac #261
Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <20150531120327.GE382@greenie.muc.de>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9761 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Gert Doering [Mon, 1 Jun 2015 19:04:47 +0000 (21:04 +0200)]
Use configure.ac hack to apply serial_test AM option only if supported.
Inspired by libguestfs' configure.ac hack - test automake version,
and if 1.12 or newer, use m4 magic to pass "serial_tests" option
to AM_INIT_AUTOMAKE().
Gert Doering [Sun, 31 May 2015 20:41:58 +0000 (22:41 +0200)]
On signal reception, return EAI_SYSTEM from openvpn_getaddrinfo().
A signal (except SIGUSR1) received while waiting for getaddrinfo() is
considered fatal, so openvpn_getaddrinfo() is destroying the returned
information with freeaddrinfo(), but still signalled "success" (0)
to the caller - so if the caller accessed *res before checking
*signal_received, it would access just-free()ed memory, which on some
platforms still worked and on others caused a crash.
Also, ensure that *ai is also NULLed in the caller now.
Trac #276
Signed-off-by: Gert Doering <gert@greenie.muc.de> Acked-by: Arne Schwabe <arne@rfc2549.org>
Message-Id: <1433104918-9523-1-git-send-email-gert@greenie.muc.de>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9764
Gert Doering [Thu, 28 May 2015 09:09:50 +0000 (11:09 +0200)]
Revert "Enforce "serial-tests" behaviour for tests/Makefile"
This reverts commit fc03ca9d13e35c40bdf1c3c676db2adf48c60223,
because it breaks "autoreconf -vif" on autoconf versions older
than 1.12 - like, CentOS 6, Debian 7, ...
Gert Doering [Wed, 27 May 2015 18:31:38 +0000 (20:31 +0200)]
Enforce "serial-tests" behaviour for tests/Makefile
Our "make check" testsuite creates quite a bit of output which is
intended to help pinpointing the exact reason for failure - hidden by
default by automake 1.12 and up, which default to "parallel-tests" which
has no benefit for us. So, just set the automake option to revert to
the old behaviour.
See also: https://www.gnu.org/software/automake/manual/html_node/Serial-Test-Harness.html
Trac #427
Signed-off-by: Gert Doering <gert@greenie.muc.de>
Message-ID: <20150524193011.GK382@greenie.muc.de>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9738 Acked-by: David Sommerseth <openvpn.list@topphemmelig.net>
But keep the chdir to / at the place where deamon() was before, to preserve
the current behaviour wrt relative paths in the config.
This should fix the issue reported in trac #480, without changing the
behaviour visible to the end user.
Note that by moving the daemon() call to an earlier stage of the init
process, we no longer have to call platform_mlockall() again, or do a
pkcs11_forkFixup().
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1430144937-4149-1-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9609 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Gert Doering [Sun, 24 May 2015 13:02:34 +0000 (15:02 +0200)]
Disallow usage of --server-poll-timeout in --secret key mode.
The internal machinery wants TLS for this to work, so just add this
to the (long) list of options not allowed unless either --tls-client
or --tls-server is active. For added sanity, add an ASSERT() call
to the place where this combination caused a NULL ptr reference, and
document the restriction.
Steffan Karger [Sun, 24 May 2015 09:45:40 +0000 (11:45 +0200)]
Clarify --capath option in manpage
Prevent confusion as described in trac #422 by better explaining the
behaviour of --capath, and providing pointers to relevant openssl man
pages.
Attached are patches for the master and release/2.3 branches. The only
difference is that in the master patch, a line referencing the
requirement for OpenSSL 0.9.7 is removed, since master already requires
OpenSSL >= 0.9.8.
Gert Doering [Sun, 24 May 2015 07:56:12 +0000 (09:56 +0200)]
Correct note about DNS randomization in openvpn.8
Commit 4880739c17b502d00a removed DNS randomization, and the dual-stack
patches for 2.4 completely changed the getaddrinfo() result handling again,
but neither fact ever made it into the man page.
Trac #411
Signed-off-by: Gert Doering <gert@greenie.muc.de> Acked-by: Arne Schwabe <arne@rfc2549.org>
Message-Id: <1432454172-1318-1-git-send-email-gert@greenie.muc.de>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9730
Gert Doering [Mon, 27 Apr 2015 19:27:21 +0000 (21:27 +0200)]
assume res_init() is always there.
Previously, the code tried to find res_init(), and on some systems
got it wrong in configure, silently not-using res_init(), leading
to unexpected failures to re-init the resolver.
We know that all supported OSes (except Windows) have res_init(), so
change the call to "#ifndef WIN32", and adjust configure.ac to just
find the library to link (if any). With that, failures to find
res_init() are no longer "hidden" but clearly visible at link time.
AC_SEARCH_LIBS() bits inspired by CUPS' cups_network.m4 (GPLv2)
Fix (part of) trac #523
Signed-off-by: Gert Doering <gert@greenie.muc.de> Lazy-ACK-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1430162841-5840-1-git-send-email-gert@greenie.muc.de>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9614
Steffan Karger [Sat, 23 May 2015 13:02:25 +0000 (15:02 +0200)]
Re-read auth-user-pass file on (re)connect if required
Fixes trac #225 ('--auth-user-pass FILE' and '--auth-nocache' problem).
This patch is based on the changes suggested by ye_olde_iron in the trac
ticket. Also added a note to the manpage to inform people to use
absolute paths when combining --auth-user-pass file and --auth-nocache.
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1432386145-15045-1-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9717 Signed-off-by: Gert Doering <gert@greenie.muc.de>
"make check" self-test was broken after commit db950be85d37 due to
do_ifconfig_setenv() not checking whether tt->did_ifconfig_setup
was set (which isn't, for "dev null" type setups)
Signed-off-by: Gert Doering <gert@greenie.muc.de> Acked-by: Steffan Karger <steffan@karger.me>
Message-Id: <1432404098-29401-1-git-send-email-gert@greenie.muc.de>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9718 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Steffan Karger [Sun, 3 May 2015 15:07:11 +0000 (17:07 +0200)]
cleanup: remove md5 helper functions
The MD5 wrapper functions were used in just a few places, which imho is
not worth the extra code. Instead of using these wrappers, just use
the generic md_ctx_*() functions directly.
The md5sum() function was only used for logging information that was not
useful to a user; first the full options string would be printed, and
later just the hash. That hash is less informative than the full
string, so why print it at all?
Finally, also removed save_pulled_options_digest(). The two times it
was called, it executed either one of the possible branches in the
function, where one of these needed a comment to explain what passing
NULL as newdigest is supposed to do...
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1430665631-4022-1-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9642 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Felix Janda [Sat, 16 May 2015 13:59:50 +0000 (15:59 +0200)]
Use OPENVPN_ETH_P_* so that <netinet/if_ether.h> is unecessary
Fixes compilation error on linux with musl libc because of
conflicting ethhdr declarations in <netinet/if_ether.h> and
<linux/if_ether.h> which is included from <linux/if_tun.h>
Signed-off-by: Felix Janda <felix.janda@posteo.de> Acked-by: Steffan Karger <steffan.karger@fox-it.com>
Message-Id: <20150516134604.GA2302@euler>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9690 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Samuli Seppänen [Mon, 11 May 2015 10:19:14 +0000 (13:19 +0300)]
Properly escape dashes on the man-page
On UTF-8 systems groff interprets unescaped dashes as hyphens and escaped
dashes
as minus signs. Unescaped dashes can cause problems when searching for or
copying and pasting options. This patch ensures that dashes in command-line
options are escaped and that everything else is left unescaped. This patch
is
for the Git "master" branch.
Trac: 512 Signed-off-by: Samuli Seppänen <samuli@openvpn.net> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1431339554-20553-1-git-send-email-samuli@openvpn.net>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9674 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Steffan Karger [Mon, 4 May 2015 19:06:38 +0000 (21:06 +0200)]
polarssl: disable 1/n-1 record splitting
Disable record splitting (for now). OpenVPN assumes records are sent
unfragmented, which is no longer a valid assumption when record splitting
is enabled (which polarssl/mbedtls did in 1.3.10, see trac #524).
Changing the code to deal with record splitting will require intrusive
changes that need thorough review and testing. Since OpenVPN is not
susceptible to BEAST (the data transmitted over the control channel is
very hard to influence for a remote attacker), we can just disable record
splitting as a quick fix. This gives us the time to develop a proper
solution in the mean time, and test that thoroughly.
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1430766398-17209-1-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9646 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Steffan Karger [Tue, 5 May 2015 15:47:37 +0000 (17:47 +0200)]
Improve --tls-cipher and --show-tls man page description
As reported in trac tickets #304, #358 and #359 (and possibly more), the
usage and interpretation of --tls-cipher (and --show-tls) is tricky. This
patch extends the man page to explain those a bit better and point out
that --tls-cipher is an expert feature (i.e. easy to get wrong). Also add
a notice to the --show-tls output, referring to the man page explanation.
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Arne Schwabe <arne@rfc2549.org>
Message-Id: <1430840857-6123-1-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9651 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Steffan Karger [Sun, 3 May 2015 12:06:01 +0000 (14:06 +0200)]
polarssl: remove code duplication in key_state_write_plaintext{, _const}()
Both functions had almost exactly the same code. Instead of the code
duplication, have key_state_write_plaintext() call
key_state_write_plaintext_const() to do the actual work. This is just
a bit of cleanup, it should not change any behaviour.
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1430654761-26563-2-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9640 Signed-off-by: Gert Doering <gert@greenie.muc.de>
As described in trac #484, the current inline file size limit of 10000
bytes is becoming an issue for some users. Since RSA keys and signature
sizes are increasing, we need to adjust our limits. As #484 reports,
10000 can be too small for PKCS#12 files with 4K RSA keys. Instead of
postponing this issue by increasing the static limit, dynamically increase
the buffer size while reading. This keeps the memory usage limited but does
allow for larger inlined files.
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1430122342-11742-1-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9607 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Gert Doering [Tue, 28 Apr 2015 10:20:19 +0000 (12:20 +0200)]
explain effect of --topology subnet on --ifconfig
The fact that the second parameter of --ifconfig is no longer
a "remote address" but a "netmask" when using --dev tun and
--topology subnet was not documented clearly enough.
Gert Doering [Tue, 28 Apr 2015 11:04:23 +0000 (13:04 +0200)]
Print helpful error message on --mktun/--rmtun if not available.
OpenVPN only supports --mktun/--rmtun to create/destroy persistant
tunnels on Linux. On BSD OSes, "ifconfig tun0 create" can do the
same job, so we do not actually need to support it - but the previous
error message ("unknown option") wasn't helpful. So always accept
the option now, and on non-supported systems, direct user to manpage.
Trac #85
Signed-off-by: Gert Doering <gert@greenie.muc.de> Acked-by: David Sommerseth <dazo@users.sourceforge.net>
Message-Id: <1430219063-12291-1-git-send-email-gert@greenie.muc.de>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9617
Gert Doering [Tue, 21 Apr 2015 20:49:12 +0000 (22:49 +0200)]
Fix leftover 'if (false) ;' statements
Commit a4b8f653ee5be9c2292c removed the #ifdefs for ENABLE_HTTP_PROXY and
ENABLE_SOCKS, thus making this "if (false) ; else if (...)" construct
superfluous. Spotted by David Sommerseth.
Signed-off-by: Gert Doering <gert@greenie.muc.de> Acked-by: Arne Schwabe <arne@rfc2549.org>
Message-Id: <1429649352-21034-1-git-send-email-gert@greenie.muc.de>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9603
Steffan Karger [Thu, 26 Mar 2015 00:01:29 +0000 (01:01 +0100)]
Remove unneeded parameter 'first_time' from possibly_become_daemon()
The static helper function possibly_become_daemon() is called only once,
by do_init_first_time(), which checks 'first_time' to be true before
calling possibly_become_daemon(). This makes the parameter useless.
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1427328089-886-1-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9555 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Vasily Kulikov [Wed, 25 Feb 2015 16:07:18 +0000 (19:07 +0300)]
Mac OS X Keychain management client
This patch adds support for using certificates stored in the Mac OSX
Keychain to authenticate with the OpenVPN server. This works with
certificates stored on the computer as well as certificates on hardware
tokens that support Apple's tokend interface. The patch is based on
the Windows Crypto API certificate functionality that currently exists
in OpenVPN.
This patch version implements management client which handles RSA-SIGN
command for RSA offloading. Also it handles new 'NEED-CERTIFICATE'
request to pass a certificate from the keychain to OpenVPN.
OpenVPN itself gets new 'NEED-CERTIFICATE" command which is called when
--management-external-cert is used. It is implemented as a multiline
command very similar to an existing 'RSA-SIGN' command.
v4:
- added '--management-external-cert' argument
- keychain-mcd now parses NEED-CERTIFICATE argument if 'auto' is passed
as cmdline's identity template
- fixed typo in help output option name
- added '--management-external-cert' info in openvpn(8) manpage
- added 'certificate' command documentation into doc/management-notes.txt
v3:
- used new 'NEED-CERTIFICATE' command for certificate data request
instead of 'NEED-OK'
- improved option checking
- improved invalid certificate selection string handling
- added man page for keychain-mcd
- handle INFO, FATAL commands from openvpn and show them to user
* ACK from Arne Schwabe for OpenVPN part
* ACK from James based on Arne's testing
v2 (http://sourceforge.net/p/openvpn/mailman/message/33225603/):
- used management interface to communicate with OpenVPN process
v1 (http://sourceforge.net/p/openvpn/mailman/message/33125844/):
- used RSA_METHOD to extend openvpn itself
Signed-off-by: Vasily Kulikov <segoon@openwall.com>
-- Acked-by: Arne Schwabe <arne@rfc2549.org>
Message-Id: <20150225160718.GA6306@cachalot>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9486 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Gert Doering [Sat, 27 Dec 2014 20:59:43 +0000 (21:59 +0100)]
Remove count_netmask_bits(), convert users to use netmask_to_netbits2()
The previous Linux/iproute2 code converted binary netmasks to string
representation (print_in_addr_t()), just to immediately scanf() it back
to binary to count bits. netmask_to_netbits2() directly works on the
in_addr_t.
Lev Stipakov [Thu, 5 Mar 2015 20:33:56 +0000 (22:33 +0200)]
Fix mssfix default value in connection_list context
Due to this bug, mssfix hasn't been assigned to fragment value
and used default value (1450) instead. As a consequence, TCP packets
get fragmented, which causes performance penalty.
This fix assigns mssfix value to fragment value for connection_entry
inside connection_list instead of connection_entry inside options struct
(which does not work for connection_list case).
Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1425587636-23338-1-git-send-email-lstipakov@gmail.com>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9507
Steffan Karger [Thu, 5 Mar 2015 21:37:31 +0000 (22:37 +0100)]
Allow for CN/username of 64 characters (fixes off-by-one)
This is an alternative patch to fix the issue reported in trac #515 by
Jorge Peixoto. Instead of increasing the TLS_USERNAME_LEN define, do +1 at
the relevant places in the code.
Also see Jorge's original patch and the discussion on the maillinglist:
http://thread.gmane.org/gmane.network.openvpn.devel/9438
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <54F8CC9B.9040104@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9508 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Steffan Karger [Thu, 5 Mar 2015 21:58:35 +0000 (22:58 +0100)]
Get rid of old OpenSSL workarounds.
We now only support OpenSSL 0.9.8+, so we don't have to work around the bug
in 0.9.6b anymore. Also, OBJ_txt2nid() now takes a const char * (instead
of a char *), so we no langer have to cast away const.
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1425592716-14243-1-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9512 Signed-off-by: Gert Doering <gert@greenie.muc.de>
CBC mode is the only mode that OpenVPN supports that needs padding. So,
only include the worst case padding size in the frame size calculation when
using CBC mode.
While doing so, rewrite crypto_adjust_frame_parameters() to be better
readable, and provide debug output (for high debug levels).
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1406667144-17674-1-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/8952 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Gert Doering [Mon, 29 Dec 2014 17:48:45 +0000 (18:48 +0100)]
Print remote IPv4 address on a dual-stack v6 socket in IPv4 format
Previously, the code would print IPv4-mapped format ("::ffff:1.2.3.4"),
which is technically correct but adds no extra value, and is confusingly
different from the output if using a v4 socket. Print "1.2.3.4" instead,
whatever socket type is used.
Affects client IP address in log file, status output (mroute_addr_print_ex),
and environment (setenv_sockaddr).
The fact that a dual-stack socket was used is still visible in the initial
peer connect message in the log:
'... Peer Connection Initiated with [AF_INET6]::ffff:1.1.3.4:53779'
Signed-off-by: Gert Doering <gert@greenie.muc.de> Acked-by: Arne Schwabe <arne@rfc2549.org>
Message-Id: <1419875325-96015-1-git-send-email-gert@greenie.muc.de>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9363
Gert Doering [Sun, 8 Feb 2015 10:18:45 +0000 (11:18 +0100)]
New approach to handle peer-id related changes to link-mtu.
Instead of statically increasing link-mtu by +3, keep the old value for
OCC compatibility with old servers/clients, and only increase link-mtu
if peer-id option is enabled (right now: is pushed by server).
If link-mtu has been set in the config, keep configured value, and log
warning (because the extra overhead has to decrease tun-mtu).
Reserve extra +3 bytes in frame->extra_link.
v2: use frame->extra_link, not frame->extra_buffer (receive path on server)
introduce frame_add_to_link_mtu() to manipulate frame->link_mtu value
rework comments to make more clear what is happening
Steffan Karger [Sun, 15 Feb 2015 14:24:26 +0000 (15:24 +0100)]
Disable SSL compression
As reported in trac #502, SSL compression can cause problems in some corner
cases. OpenVPN does not need SSL compression, since the control channel is
low bandwidth. This does not influence the data channel compressen (i.e.
--comp or --comp-lzo).
Even though this has not yet been relevant for OpenVPN (since an attacker
can not easily control contents of control channel messages), SSL
compression has been used in the CRIME and BREACH attacks on TLS. TLS 1.3
will probably even remove support for compression all together, for
exactly this reason.
Since we don't need it, and SSL compression causes issues, let's just
disable it in OpenSSL builds. PolarSSL has no run-time flag to disable
compression, but is by default compiled without compression.
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Arne Schwabe <arne@rfc2549.org> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1424010266-5910-1-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9453 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Lev Stipakov [Fri, 6 Feb 2015 12:38:00 +0000 (14:38 +0200)]
Fix NULL dereferencing
In certain cases buf.len can be -1, which causes BPTR to return NULL and
NULL pointer dereferencing on the next line.
As a fix, process only packets with non-zero length. Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1423226280-9580-1-git-send-email-lstipakov@gmail.com>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9444
Lev Stipakov [Wed, 7 Jan 2015 19:26:38 +0000 (21:26 +0200)]
Disallow lameduck's float to an address taken by another client
Existing check didn't take into account the case when floated client is
lame duck (CN for lame duck is NULL), which allowed lame duck to float
to an address taken by another client.
As a fix we use cert_hash_compare function which, besides fixing
mentioned case, also allows lame duck to float to an address already
taken by the same client. Acked-by: Steffan Karger <steffan.karger@fox-it.com>
Message-Id: <1420658798-29943-1-git-send-email-lstipakov@gmail.com>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9386
Steffan Karger [Thu, 1 Jan 2015 19:46:09 +0000 (20:46 +0100)]
Add option to disable Diffie Hellman key exchange by setting '--dh none'
As requested on the mailing list and in trac ticket #410, add an option to
disable 'traditional' Diffie Hellman key exchange. People want to be able
to create ecdh-only configurations.
This patch also disables RSA key exchange by default for OpenSSL builds, to
prevent that people who set "--dh none" but have an OpenSSL version that
doesn't support ECDH end up with a less secure connection. Note that users
that specify their own --tls-cipher override these defaults and thus can
still use whatever OpenSSL supports (and might thus end up with less secure
connections).
PolarSSL does not allow to easily disable RSA key exchange during runtime,
but its default compile options do not include RSA key exchange based
cipher suites.
Finally update the manpage to reflect the new behaviour, and while touching
it change the text to motivate users towards a more secure configuration.
v2 - disable RSA key exchange by default
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Arne Schwabe <arne@rfc2549.org>
Message-Id: <1420141569-11773-1-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9376 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Remove the --disable-ssl configure option and accompanying ENABLE_SSL
defines in the master/2.4 branch, to reduce the code and testing
complexity a bit.
This does not remove to runtime option to run without SSL, just the compile
time option to not include any SSL-related code.
During the community meeting in November 2014 there were no objections
amongst he developers present. Also, this has been announced on the -users
and -devel mailing lists two weeks ago, without any response whatsoever.
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <54A4248A.1090501@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9371 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Steffan Karger [Sat, 25 Oct 2014 18:49:26 +0000 (20:49 +0200)]
openssl: add more descriptive message for 'no shared cipher' error
Overzealous users using the --tls-cipher option, or users with actual
incompatible crypto libaries often waste quite some time debugging the
'no shared cipher' error from openssl. See e.g. trac ticket #359:
https://community.openvpn.net/openvpn/ticket/359
This change adds a more clear, verb 1 error message reporting the problem
directly to the user, instead of just printing the openssl error.
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Arne Schwabe <arne@rfc2549.org>
Message-Id: <544EB12E.40200@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9209 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Steffan Karger [Sat, 25 Oct 2014 20:35:22 +0000 (22:35 +0200)]
openssl: add crypto_msg(), to easily log openssl errors
This works towards removing OpenSSL-specific error printing code from
error.c. The crypto_msg() functions provide convenience wrappers, specific
to OpenSSL. Instead of passing the magical 'M_SSLERR' flag to msg(), a
developer now just calls crypto_msg() to get OpenSSL errors dumped to log.
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1414269324-14102-5-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9199 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Default gateway can't be determined on illumos/Solaris platforms
The logic which is used on BSD is preserved. The only distinction
is that illumos doesn't provide sa_len field in sockaddr structures. Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1413315802-62359-5-git-send-email-alp@rsu.ru>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9129
Steffan Karger [Sun, 28 Dec 2014 10:25:13 +0000 (11:25 +0100)]
Set tls-version-max to 1.1 if cryptoapicert is used
OpenVPN's current cryptoapicert implementation does not support TLS 1.2
(and newer). Fixing this requires a rewrite of our cryptoapi code to use
Microsofts' "Cryptography API: Next Generation", and several hacks to work
around that API. As long as we don't fix that, make openvpn automatically
cap the TLS version to 1.1 when using cryptoapi (and tell the user we're
doing so). This enables the user to use cryptoapi + TLS version
negotiation (upto TLS 1.1) without having to change his configuration.
This patch has been tested on Windows 8.1 for both the master and
release/2.3 branches.
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1419762313-31233-1-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9361 Signed-off-by: Gert Doering <gert@greenie.muc.de>
David Woodhouse [Thu, 11 Dec 2014 13:03:35 +0000 (13:03 +0000)]
pkcs11: Load p11-kit-proxy.so module by default
If the user specifies --pkcs11-id or --pkcs-id-management but neglects
to explicitly provide a --pkcs11-provider argument, and if the system
has p11-kit installed, then load the p11-kit proxy module so that the
system-configured tokens are available.
Trac: 490 Signed-off-by: David Woodhouse <David.Woodhouse@intel.com> Acked-by: Steffan Karger <steffan.karger@fox-it.com>
Message-Id: <1418303015.31745.78.camel@infradead.org>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9342 Signed-off-by: Gert Doering <gert@greenie.muc.de>
(cherry picked from commit 6f1d3cf062d5c33cbad4d521d157d43d53ffc7d1)
The coding style was somewhat chaotic. Cleaning it up using the astyle
tool. The style parameters are coherent to what was agreed upon at the
Munich Hackathon 2014 [1].
astyle --style=allman --indent=spaces=4 -c
Also included a "Local variables" section which some editors may pick
up automatically.
Signed-off-by: David Sommerseth <davids@redhat.com> Acked-by: Steffan Karger <steffan.karger@fox-it.com>
Message-Id: <1418078751-3614-1-git-send-email-openvpn.list@topphemmelig.net>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9331 Signed-off-by: Gert Doering <gert@greenie.muc.de>
sockets: Remove the limitation of --tcp-nodelay to be server-only
The assert(0) happening if trying to use --tcp-nodelay in a client
config is really not helpful at all. When this assert(0) was removed,
another warning appeared that this could only be used in server
configs. That itself is also quite silly, as clients can choose to
use --socket-flags TCP_NODELAY in the client config instead. This
behaviour does not help the user in any way.
This patch removes the server-only restriction and rather provides
a more helpful warning when using --tcp-nodelay on the client side.
Trac: 489 Signed-off-by: David Sommerseth <davids@redhat.com> Acked-by: Arne Schwabe <arne@rfc2549.org>
Message-Id: 1418118764-17846-1-git-send-email-openvpn.list@topphemmelig.net
URL: http://article.gmane.org/gmane.network.openvpn.devel/9334
Lev Stipakov [Mon, 8 Dec 2014 16:48:45 +0000 (18:48 +0200)]
Prevent memory drain for long lasting floating sessions
For every float event we generate prefix, which allocates 256 + 64
bytes. That memory is reclaimed when client disconnects, so long lasting
and constantly floating sessions drain memory.
As a fix use preallocated buffer inside multi_instance for storing
multi_prefix.
Signed-off-by: Lev Stipakov <lstipakov@gmail.com> Acked-by: Steffan Karger <steffan.karger@fox-it.com>
Message-Id: 1418057325-13265-1-git-send-email-lstipakov@gmail.com
URL: http://article.gmane.org/gmane.network.openvpn.devel/9321 Signed-off-by: David Sommerseth <davids@redhat.com>
Lev Stipakov [Mon, 8 Dec 2014 17:06:02 +0000 (19:06 +0200)]
Add the peer-id to the output of the status command
This adds peer-id to the status output which might help analyze floating
logs. This will change the output of status in the same way commit 662ce6acc065bddf6490b3494725b8b3987b7def did.
Signed-off-by: Lev Stipakov <lstipakov@gmail.com> Acked-by: Arne Schwabe <arne@rfc2549.org>
Message-Id: 1418058362-13480-1-git-send-email-lstipakov@gmail.com
URL: http://article.gmane.org/gmane.network.openvpn.devel/9322 Signed-off-by: David Sommerseth <davids@redhat.com>
Steffan Karger [Sun, 7 Dec 2014 18:48:15 +0000 (19:48 +0100)]
Update doxygen (a bit)
This is not a full update, but just updates some data channel-related docs
I came across. Other pages probably need a bit of attention too.
Stuff that was changed:
* Explain data channel crypto format in crypto.h
* Add P_DATA_V1 and P_DATA_V2 packet format spec
* Remove '2.1' from title
* Update some OpenSSL-specific text
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1417978095-19427-1-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9318 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Arne Schwabe [Sat, 6 Dec 2014 13:15:03 +0000 (14:15 +0100)]
Remove possibility of using --tls-auth with non OpenVPN Static key files
In older version OpenVPN would hash a --tls-auth file
if it does not conform to the expected format Acked-by: Steffan Karger <steffan.karger@fox-it.com>
Message-Id: <1417871704-30273-1-git-send-email-arne@rfc2549.org>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9306
David Sommerseth [Sun, 16 Nov 2014 14:31:02 +0000 (15:31 +0100)]
down-root plugin: Replaced system() calls with execve()
The system() call is prone to shell expansions and provides far more
environments variables to the executable run than what is usually
preferred. By moving over to exevce() shell expansions are far more
difficult to achieve and only the OpenVPN provided environment
variables are available.
This is a response to the patch submitted to openvpn-devel ML:
http://article.gmane.org/gmane.network.openvpn.devel/7919
v2 - Pulling it up again, fixing a few whitespace and spelling issues
Signed-off-by: David Sommerseth <davids@redhat.com> Acked-by: Steffan Karger <steffan.karger@fox-it.com>
Message-Id: <1416148262-20978-1-git-send-email-openvpn.list@topphemmelig.net>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9238 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Steffan Karger [Tue, 2 Dec 2014 20:42:00 +0000 (21:42 +0100)]
Really fix '--cipher none' regression
... by not incorrectly hinting to the compiler the function argument of
cipher_kt_mode_{cbc,ofb_cfb}() is nonnull, since that no longer is the
case.
Verified the fix on Debian Wheezy, one of the platforms the reporter in
trac #473 mentions with a compiler that would optimize out the required
checks.
Also add a testcase for --cipher none to t_lpback, to prevent further
regressions.
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1417552920-31770-1-git-send-email-steffan@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9300 Signed-off-by: Gert Doering <gert@greenie.muc.de>
David Sommerseth [Mon, 24 Nov 2014 18:09:38 +0000 (19:09 +0100)]
autotools: Fix wrong ./configure help screen default values
enable_crypto_ofb_cfb is "yes" by default, so the --help screen
should show --disable-ofb-cfb and not --enable-ofb-cfb.
enable_small and enable_password_save are both "no" by default, so
the --help screen should state "default: no". Now it says "yes" as
default, but is really disabled in the reality.
Signed-off-by: David Sommerseth <davids@redhat.com> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1416852578-7581-1-git-send-email-openvpn.list@topphemmelig.net>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9278 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Steffan Karger [Thu, 20 Nov 2014 12:43:05 +0000 (13:43 +0100)]
Drop too-short control channel packets instead of asserting out.
This fixes a denial-of-service vulnerability where an authenticated client
could stop the server by triggering a server-side ASSERT().
OpenVPN would previously ASSERT() that control channel packets have a
payload of at least 4 bytes. An authenticated client could trigger this
assert by sending a too-short control channel packet to the server.
Thanks to Dragana Damjanovic for reporting the issue.
This bug has been assigned CVE-2014-8104.
Signed-off-by: Steffan Karger <steffan.karger@fox-it.com> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1CED409804E2164C8104F9E623B08B9018803B0FE7@FOXDFT02.FOX.local> Signed-off-by: Gert Doering <gert@greenie.muc.de>
Lev Stipakov [Sun, 23 Nov 2014 15:17:11 +0000 (17:17 +0200)]
Peer-id patch v7
Added new packet format P_DATA_V2, which includes peer-id. If server
supports, client sends all data packets in the new format. When data
packet arrives, server identifies peer by peer-id. If peer's ip/port has
changed, server assumes that client has floated, verifies HMAC and
updates ip/port in internal structs.
Changes in v7:
A few nitpicks.
Changes in v6:
Fixed: Make sure float won't happen if hmac check failed (regression).
Fixed: Access outside of bounds of array, which has caused memory
corruption and crash.
Various review fixes.
Changes in v5:
Protection agains replay attack by commiting float changes only after
existing packet processing flow has completed.
If peer floats to an address which is already taken by another active
session, drop float packet, otherwise disconnect existing session.
Changes in v4:
Handles correctly float to an address which is used by another peer.
This also has fixed crash on assert in multi_client_disconnect.
Changes in v3:
Bugfix: If float happens after TLS renegotiation and there are no
data packets between reneg and float, server will not recognize floated
client. Acked-by: Steffan Karger <steffan.karger@fox-it.com> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1416755831-21250-1-git-send-email-lstipakov@gmail.com>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9270
David Sommerseth [Thu, 13 Nov 2014 14:43:37 +0000 (15:43 +0100)]
systemd: Reworked the systemd unit file to handle server and client configs better
Systemd can delay starting a service if the network isn't fully available
yet. This feature is useful in client configurations, where OpenVPN will
not be started before the client can reach the Internet. It is the network
service manager which tells systemd if the system is "online" or not.
For server configurations, the OpenVPN should be able to be started,
regardless if the system is "online" or not. This is also the old
behaviour of most of the old init.d script and the last systemd unit file.
This patch splits the previous systemd unit file into to two files. One
which is aimed at clients (openvpn-client@.service) and one for server
configurations (openvpn-server@.service). These files will also pick
the configurations from different sub-directories. The unit file for
openvpn-client@ will use /etc/openvpn/client and the server unit file
will use /etc/openvpn/server. This also ensures that config files
are not started in the wrong manner.
The arguments given to the openvpn binary have also shifted order,
to ensure that some of them cannot be overridden by the config file,
such as --daemon and --writepid. For server configurations a
--status file is also added with the status format set to 2. This
can be overridden by the configuration file.
Signed-off-by: David Sommerseth <davids@redhat.com> Acked-by: Gert Doering <gert@greenie.muc.de>
Message-Id: <1415889817-28049-1-git-send-email-openvpn.list@topphemmelig.net>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9222 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Heiko Hund [Sat, 15 Nov 2014 16:25:07 +0000 (17:25 +0100)]
Fix compilation on Windows
The local member in struct link_socket_address went away a while ago.
Fixing the code to compile again under Windows. While there, also
fix the code to deal with struct link_socket_info.af == AF_UNSPEC
Steffan Karger [Wed, 22 Oct 2014 22:14:29 +0000 (00:14 +0200)]
Modernize sample keys and sample configs
I kept most of the certificate properties equal to the old
certs, since some people's test scripts might rely on them (and
it does not require any creativity from my part).
Changes:
* Add script to generate fresh test/sample keys
(but keep sample keys in git for simple testing)
* Switch from 1024 to 4096 bits RSA CA
* Switch from 1024 to 2048 bits client/server RSA keys
* Switch from 1024 to 2048 bits Diffie-Hellman parameters
* Generate EC client and server cert, but sign with RSA CA
(lets us test EC <-> RSA interoperability)
* Remove 3DES cipher from 'sample' config
* Add 'remote-cert-tls server' to client config
* Update config files to deprecate nsCertType in favour of the
keyUsage and extendedKeyUsage extensions.
* Make naming more consistent
Steffan Karger [Mon, 25 Aug 2014 22:03:23 +0000 (00:03 +0200)]
Add --tls-version-max
Because using TLS 1.2 breaks certain setups, a user might want to enforce
a maximum TLS version to use. This patch adds that option.
This patch removes a number of #ifdefs from ssl_polarssl.c, because the
polarssl versions we currently support (polar 1.2 for openvpn 2.3, and
polar 1.3 for openvpn-master) have all versions unconditionally enabled.
Signed-off-by: Steffan Karger <steffan.karger@fox-it.com> Acked-by: Arne Schwabe <arne@rfc2549.org>
Message-Id: <544EC052.3080809@fox-it.com>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9210 Signed-off-by: Gert Doering <gert@greenie.muc.de>
Steffan Karger [Sat, 8 Nov 2014 10:15:08 +0000 (11:15 +0100)]
Fix assertion error when using --cipher none
Some commits ago, the cipher mode checks were cleaned up to
remove code duplication (and fix the issue in #471), but broke
'--cipher none' (reported in #473). This commit fixes that.
Signed-off-by: Steffan Karger <steffan@karger.me> Acked-by: Arne Schwabe <arne@rfc2549.org>
Message-Id: <545DED2C.5070002@karger.me>
URL: http://article.gmane.org/gmane.network.openvpn.devel/9217 Signed-off-by: Gert Doering <gert@greenie.muc.de>
projects / thirdparty / openvpn.git / log
