Commit 27756b081c (revocation: Check that nonce in OCSP response matches)
introduced strict nonce validation to prevent replay attacks with OCSP
responses having a longer lifetime. However, many commercial CAs (such as
Digicert) do not support nonces in responses, as they reuse once-issued OCSP
responses for the OCSP lifetime. This can be problematic for replay attack
scenarios, but is nothing we can fix at our end.
With the mentioned commit, such OCSP responses get completely unusable,
requiring the fallback to CRL based revocation. CRLs don't provide any
replay protection either, so there is nothing gained security-wise, but may
require a download of several megabytes CRL data.
To make use of replay protection where available, but fix OCSP verification
where it is not, do nonce verification only if the response actually contains
a nonce. To be safe against replay attacks, one has to fix the OCSP responder
or use a different CA, but this is not something we can enforce.
Fixes#3557.
The x509 plugin accepted CRL signers since forever, to be precise, since
dffb176f2b ("CRLSign keyUsage or CA basicConstraint are sufficient
for CRL validation")).
References #3529.
A new global option enables sending this vendor ID to prevent Cisco
devices from narrowing the initiator's local traffic selector to the
requested virtual IP, so e.g. 0.0.0.0/0 can be used instead.
This has been tested with a "tunnel mode ipsec ipv4" Cisco template but
should also work for GRE encapsulation.
Closesstrongswan/strongswan#180.
While `pos` was moved to the end, `len` was not adjusted (i.e. set to 0)
so later calls could write beyond the buffer. However, the last port
written might have been incomplete, so instead we just reset the string.
If it takes a while to start one of the threads, another thread might already
have passed the usleep() call previously used and re-enabled cancelability
so that the loop that checked for it would never terminate.
This way we only have one reference for each CA certificate, whether it
is loaded in an authority section, a connection or via load-certs() command.
It also avoids enumerating CA certificates multiple times if they are
loaded in different ways.
With the previous approach, CA certificates that were not re-loaded via
load-cert() (e.g. from tokens or via absolute paths) would not be available
anymore after the clear-creds() command was used. This avoids this
issue, but can cause duplicate CA certificates to get stored and enumerated,
so there might be a scaling factor.
This reduces the clustering problem (primary clustering) but is not
completely free of it (secondary clustering) it still reduces the maximum
and average probing lengths.
With the previous approach we'd require at least an additional pointer
per item to store them in a list (15-18% increase in the overhead per
item). Instead we switch from handling collisions with overflow lists to
an open addressing scheme and store the actual table as variable-sized
indices pointing into an array of all inserted items in their original
order.
This can reduce the memory overhead even compared to the previous
implementation (especially for smaller tables), but because the array for
items is preallocated whenever the table is resized, it can be worse for
certain numbers of items. However, avoiding all the allocations required
by the previous design is actually a big advantage.
Depending on the usage pattern, the performance can improve quite a bit (in
particular when inserting many items). The raw lookup performance is a bit
slower as probing lengths increase with open addressing, but there are some
caching benefits due to the compact storage. So for general usage the
performance should be better. For instance, one test I did was counting the
occurrences of words in a list of 1'000'000 randomly selected words from a
dictionary of ~58'000 words (i.e. using a counter stored under each word as
key). The new implementation was ~8% faster on average while requiring
10% less memory.
Since we can't remove items from the array (would change the indices of all
items that follow it) we just mark them as removed and remove them once the
hash table is resized/rehashed (the cells in the hash table for these may
be reused). Due to this the latter may also happen if the number of stored
items does not increase e.g. after a series of remove/put operations (each
insertion requires storage in the array, no matter if items were removed).
So if the capacity is exhausted, the table is resized/rehashed (after lots
of removals the size may even be reduced) and all items marked as removed
are simply skipped.
Compared to the previous implementation the load factor/capacity is
lowered to reduce chances of collisions and to avoid primary clustering to
some degree. However, the latter in particular, but the open addressing
scheme in general, make this implementation completely unsuited for the
get_match() functionality (purposefully hashing to the same value and,
therefore, increasing the probing length and clustering). And keeping the
keys optionally sorted would complicate the code significantly. So we just
keep the existing hashlist_t implementation without adding code to maintain
the overall insertion order (we could add that feature optionally later, but
with the mentioned overhead for one or two pointers).
The maximum size is currently not changed. With the new implementation
this translates to a hard limit for the maximum number of items that can be
held in the table (=CAPACITY(MAX_SIZE)). Since this equals 715'827'882
items with the current settings, this shouldn't be a problem in practice,
the table alone would require 20 GiB in memory for that many items. The
hashlist_t implementation doesn't have that limitation due to the overflow
lists (it can store beyond it's capacity) but it itself would require over
29 GiB of memory to hold that many items.
The main intention here is that we can change the hashtable_t
implementation without being impeded by the special requirements imposed
by get_match() and sorting the keys/items in buckets.
This can improve negative lookups, but is mostly intended to be used
with get_match() so keys/items can be matched/enumerated in a specific
order. It's like storing sorted linked lists under a shared key but
with less memory overhead.
They are not marked as temporary addresses so make sure we always return
them whether temporary addresses are preferred as source addresses or not
as we need to enumerate them when searching for addresses in traffic selectors
to install routes.
Fixes: 9f12b8a61c ("kernel-netlink: Enumerate temporary IPv6 addresses according to config")
We don't track CHILD_SA down events anymore and rely on NM's initial timeout
to let the user know if the connection failed initially. So we also don't
have to explicitly differentiate between initial connection failures and
later ones like we do an Android. Also, with the default retransmission
settings, there will only be one keying try as NM's timeout is lower than
the combined retransmission timeout of 165s.
There is no visual indicator while the connection is reestablished later.
Fixes#3300.
If a MOBIKE task is deferred, the retransmission counter is reset to 0
when reinitiating. So if there were retransmits before, this alert would
not be triggered if a response is received now without retransmits.
We usually have a local IP already via ike_sa_t::resolve_hosts() before
build_i() is called but if that's not the case, it's more likely we have
one after we processed the first response (it might also have changed).
There is a potential chance we still don't have one if the socket API
doesn't provide us with the destination address of received messages,
but that seems not very likely nowadays.
We need the PSK/identity already when deriving the keys in process_i().
Fixes: 1665a4e050 ("ikev1: Use actual local identity as initiator or aggressive mode responder")
The native parseInetAddressBytes() method called by that method is not
available when running the tests.
Not very pretty and there are some warnings because PowerMock does
reflection in some illegal way but it fixes the unit tests and does
not require any new dependencies like Apache Commons or Guava just to
parse IP addresses without DNS lookup.
Fixes: 2ef473be15 ("android: Use helper to parse IP addresses where appropriate")
Fixes#3443.
This allows users to ignore whether the app is on the device's power
whitelist without a warning. The flag is currently not set
automatically if the user denies the request.
This is necessary so we can actually schedule events accurately in Doze
mode. Otherwise, we'd only get woken in intervals of several minutes (up to
15 according to the docs) after about an hour.
This uses AlarmManager to schedule events in a way that ensures the app
is woken up (requires whitelisting when in Doze mode to be woken up at
the exact time, otherwise there are delays of up to 15 minutes).
Retransmission jobs for old requests for which we already received a
response previously left the impression that messages were sent more
recently than was actually the case.
task_manager_t always defined INVALID_STATE as possible return value if
no retransmit was sent, this just was never actually returned.
I guess we could further differentiate between actual invalid states
(e.g. if we already received the response) and when we don't send a
retransmit for other reasons e.g. because the IKE_SA became stale.
Previously, if the two utility functions were called while the VPN
connection was established (i.e. charon was initialized) the logger for
libstrongswan would get reset to the initial log handler. So certain
log messages would not get logged to the log file after the TUN device
was created (one of the helpers is used to convert IPs there).
A new NAT mapping might be created even if the IP stays the same. Due to
the DPD fallback with NAT keep-alives this might only be necessary in
corner cases, if at all.
This is useful on Android where the app might not be able to send
keep-alives if the device is asleep for a while. If the NAT mapping
has been deleted in the mean time, the NAT-D payloads allow detecting
this and connectivity can be restored by doing a MOBIKE update or
recreating the SA if the peer already deleted it because the client
wasn't reachable.
This allows measuring the delay between events more accurately if a
device is often suspended.
While CLOCK_BOOTTIME would be preferable, Android's bionic C library
does not support it for condvars.
On some systems it might be preferable to use e.g. CLOCK_BOOTTIME
instead of CLOCK_MONOTONIC, which is also not affected by time
adjustments but includes times when the system was suspended.
XML resources are apparently not supported there. Moving the icon to
the mipmap folders should fix that. Aliases are defined for the icons on
Android < 8.0.
Evidently, onClick() may be called either before onStartListening() or
after onStopListening() has been called, which causes a crash when
trying to load a VpnProfile via mDataSource.
This partially reverts 3716af079e ("android: Avoid crash related to
TileService on Huawei devices").
This change allows to customize the previously hard-coded remote traffic
selectors.
This does not actually write the newly added "remote-ts" configuration option
into NetworkManager's configuration file, but will use an existing value.
Exposing the config setting in the GUI could be done later if this is a
desired change.
Use case: remote firewall appliance wrongly accepts the `0.0.0.0/0` TS but
does not actually route external traffic, leaving the user with a partially
working internet connection.
Closesstrongswan/strongswan#173.
If we fail connecting to the host we got redirected to, we should restart
with the original host where we might get redirected to a different host.
We must not reset this when retrying due to INVALID_KE_PAYLOAD or COOKIE
notifies. Since we keep the initiator SPI in those cases, we use that
flag as indicator.
Since we don't store the original remote_host value, we can't restore
that. So there is a potential conflict with MIPv6.
Closesstrongswan/strongswan#171.
The height of the dialog increased due to the recently added additional
fields for certificate selection and identities. On some screens the
fields to configure custom proposals were not visible anymore.
Together with less spacing on the top level GtkBox this change reduces
the height by about 80 pixels.
Fixes#3448.
The path '/usr/share/appdata' is deprecated as is the .appdata.xml
extension, files should be in installed in '/usr/share/metainfo' with
a .metainfo.xml extension.
According to the docs, the metainfo path should be well supported even
by older distros like Ubuntu 16.04.
Reference: 2.1.2. Filesystem locations
https://www.freedesktop.org/software/appstream/docs/chap-Metadata.html
The need_secrets() method is called before connect() (where we clear the
previous secrets too), so e.g. a password-protected private could be
decrypted with the cached password from earlier but if the password was not
stored with the connection, it would later fail as no password was requested
from the user that could be passed to connect().
References #3428.
On newer desktops the auth dialog is called with --external-ui-mode and
it seems that the password flag has to be set, otherwise the password is
not stored temporarily in the profile and passed to charon-nm (not sure
how this works exactly as need_secrets() is called multiple times even
after the password was already entered, only before doing so the last
time is the password available in that callback, but only if the flag
was set). This now also allows storing the password for the private key
with the profile.
Fixes#3428.
Due to the exponential backoff a high number of retransmits only
makes sense if retransmit_limit is set. However, even with that there
was a problem.
We first calculated the timeout for the next retransmit and only then
compared that to the configured limit. Depending on the configured
base and timeout the calculation overflowed the range of uint32_t after
a relatively low number of retransmits (with the default values after 23)
causing the timeout to first get lower (on a high level) before constantly
resulting in 0 (with the default settings after 60 retransmits).
Since that's obviously lower than any configured limit, all remaining
retransmits were then sent without any delay, causing a lot of concurrent
messages if the number of retransmits was high.
This change determines the maximum number of retransmits until an
overflow occurs based on the configuration and defaults to UINT32_MAX
if that value is exceeded. Note that since the timeout is in milliseconds
UINT32_MAX equals nearly 50 days.
The calculation in task_manager_total_retransmit_timeout() uses a double
variable and the result is in seconds so the maximum number would be higher
there (with the default settings 1205). However, we want its result to
be based on the actual IKE retransmission behavior.
Some peers apparently don't send the notify and still expect to
authenticate with EAP-only authentication. This option allows forcing
the configured use of EAP-only authentication in that scenario.
If such a task was active while reestablishing it will get queued on the
new IKE_SA. If the DH group is already set, the DH groups won't be
stripped from the proposals and a KE payload will be sent, which is invalid
during IKE_AUTH. We don't want to reset the group if the task is part of a
child-rekey task.
Otherwise, the output is buffered when e.g. piping the output to another
command (or file). And it avoids having to call fflush() in the
interactive mode.
Fixes#3404.
The file is usually opened/created by root, however, if user/group IDs
are configured and the configuration is reloaded, the file will be reopened
as configured user. Like with UNIX sockets we only attempt to change
the user if we have CAP_CHOWN allowing a start as regular user.
We don't have chown() on Windows, so check for it.
If cards/libraries don't support signature mechanisms with hashing, we fall
back to do it ourselves in software and pass the PKCS#1 digestInfo ASN.1
structure to sign via CKM_RSA_PKCS mechanism.
Closesstrongswan/strongswan#168.
OpenSSL currently doesn't support squeezing bytes out of an XOF multiple
times. Unfortunately, EVP_DigestFinalXOF() completely resets the context
and later calls not simply fail, they cause a null-pointer dereference in
libcrypto. This fixes the crash at the cost of repeating initializing
the whole state and allocating too much data for subsequent calls.
There is an open issue and PR that might add a function that allows
squeezing more data from an XOF in a future version of OpenSSL.
strtol(3) accepts values in the range of [LONG_MIN;LONG_MAX]. Based
on the architecture (32 or 64 bits), these values expand to either
0x8000000000000000/0x7fffffffffffffff for 64-bit builds, or
0x80000000/0x7fffffff for 32-bit builds.
The behavior when retrieving non-default values for charon.spi_min or
charon.spi_max, for example, depends on the architecture of the target
platform. While 0xC000001/0xCFFFFFFE work fine on a 64-bit build, on a
32-bit build, due to the use of strtol(3), an ERANGE causes get_int()
to return the default values.
By using strtoul(3) the default is only returned if the input value
exceeds 32 or 64 bits, based on the platform. Negative values are still
parsed correctly.
Signed-off-by: Thomas Egerer <thomas.egerer@secunet.com>
Enables us to ignore any future kernel features for routes unless
we actually need to consider them for the source IP routes.
Also enables us to actually really skip IPsec processing for those networks
(because even the routes don't touch those packets). It's more what
users expect.
Co-authored-by: Tobias Brunner <tobias@strongswan.org>
This is mainly an issue on FreeBSD where the current kernel still only
allows the daemon to use reqids < IPSEC_MANUAL_REQID_MAX (0x3fff = 16383).
Fixes#2315.
Charon refuses to make use of algorithms IDs from the private space
for unknown peer implementations [1]. If you chose to ignore and violate
that section of the RFC since you *know* your peers *must* support those
private IDs, there's no way to disable that behavior.
With this commit a strongswan.conf option is introduced which allows to
deliberately ignore parts of section 3.12 from the standard.
[1] http://tools.ietf.org/html/rfc7296#section-3.12
Signed-off-by: Thomas Egerer <thomas.egerer@secunet.com>
Doing this for the self-signed check also (i.e. if this and issuer are
the same) is particularly useful if the issuer uses a different key type.
Otherwise, we'd try to verify the signature with an incompatible key
that would result in a log message.
Fixes#3357.
Previously, we simply used the lifetimes of the first
proposal/transform, which is not correct if the initiator uses different
lifetimes in its proposals/transforms.
If an IKE_SA is terminated while a task is active, the delete task is
simply queued (unless the deletion is forced). If the active task times
out before any optional timeout associated with the termination hits, the
IKE_SA previously was reestablished without considering the termination
request.
Fixes#3335.
Also makes it configurable via configure script. Depending on `$datadir` is
not ideal as package maintainers might set that to a custom value. Depending
on `$datarootdir` might have been better, the default if pkg-config fails is
now based on that.
References #3339.
The code is structured similar to that in the Android client, but two-round
authentication (cert+EAP) is not supported as that might require multiple
secrets ("password" is currently the only secret field used for every
method) and other details are currently missing too (like configurable
client identities).
There are a ton of libsoup/GLib-related "leaks" that we can't whitelist
and with leak detective active there is a delay that interestingly doesn't
happen with soup_session_sync_new(), so tests failed with a timeout (actually
they hung due to the lock in the fetcher manager).
On Travis, the curl plugin is used for the tests, so that's not an issue
there (and without LD the tests complete quickly and successfully).
As noted in 8ea13bbc5c newer compilers might optimize out the
assignment leading to invalid values in the keyUsage extension (as the
length was still set, the extension was encoded, just not with the
intended values).
Fixes#3249.
While it's unlikely that so many (large) items are allocated, this is
technically more correct. The result previously could overflow an
unsigned int (the conversion to size_t happened afterwards).
This clearly never was correct, but didn't cause problems so far.
However, GCC 10 will default to `-fno-common` instead of
`-fcommon` (https://gcc.gnu.org/PR85678), so compilation there fails
with something like:
```
libtool: link: gcc ... -o .libs/swanctl ...
ld: commands/load_authorities.o:strongswan/src/swanctl/./swanctl.h:33:
multiple definition of `swanctl_dir'; commands/load_all.o:strongswan/src/swanctl/./swanctl.h:33: first defined here
```
Fixes: 501bd53a6c ("swanctl: Make credential directories relative to swanctl.conf")
Closesstrongswan/strongswan#163.
This is the recommended location and import config as it allows running the
tests against installed versions of the package. And while the test file
itself is automatically included in the source distribution this way, the
__init__.py file is not, so we still have to update MANIFEST.in.
While the TPM expects and returns the data in big-endian, the SAPI
implementation converts it to native-endianness. As stated in the
SAPI specification (section 3.2):
8. All SAPI data SHALL be in native-endian format. This means that
the SAPI implementation will do any endian conversion required for
both inputs and outputs.
So to use the exponent in a chunk we have to convert it to big-endian again.
Fixes: 7533cedb9a ("libtpmtss: Read RSA public key exponent instead of assuming its value")
RFC 7296, section 2.21.3:
If a peer parsing a request notices that it is badly formatted (after
it has passed the message authentication code checks and window
checks) and it returns an INVALID_SYNTAX notification, then this
error notification is considered fatal in both peers, meaning that
the IKE SA is deleted without needing an explicit Delete payload.
RFC 7296, section 2.21.3:
If a peer parsing a request notices that it is badly formatted (after
it has passed the message authentication code checks and window
checks) and it returns an INVALID_SYNTAX notification, then this
error notification is considered fatal in both peers, meaning that
the IKE SA is deleted without needing an explicit Delete payload.
This happened when installing a duplicate bypass policy for a locally
connected subnet. The destructor and the kernel-net part already
handle this correctly.
Without this we only would learn that the algorithm isn't actually
available (e.g. due to FIPS mode) when set_key() is called later, so there
isn't any automatic fallback to other implementations.
Fixes#3284.
Enforcing CA based constraints previously required the CA certificate file
to be locally installed. This is problematic from a maintencance perspective
when having many intermediate CAs, and is actually redundant if the client
sends its intermediate cert in the request.
The alternative was to use Distinguished Name matching in the subject
identity to indirectly check for the issuing CA by some RDN field, such as OU.
However, this requires trust in the intermediate CA to issue only certificates
with legitime subject identities.
This new approach checks for an intermediate CA by comparing the issuing
identity. This does not require trust in the intermediate, as long as
a path len constraint prevents that intermediate to issue further
intermediate certificates.
This avoids having to register certificates with authority/ca backends
beforehand, which is tricky for intermediate CA certificates loaded
themselves via authority/ca sections. On the other hand, the form of
these URLs can't be determined by config backends anymore (not an issue
for the two current implementations, no idea if custom implementations
ever made use of that possibility). If that became necessary, we could
perhaps pass the certificate to the CDP enumerator or add a new method
to the credential_set_t interface.
Don't define structs for macOS as we don't need them (that's true for
most of the others too, though) and at least one is defined inside an extra
ifdef.
If strings are missing (e.g. because the last value of a range changed
unknowingly or adding a string was simply forgotten) compilation will
now fail.
This could be problematic if the upper limit is out of our control (e.g.
from a system header like pfkeyv2.h), in which case patches might be
required on certain platforms (enforcing at least, and not exactly, the
required number of strings might also be an option to compile against
older versions of such a header - for internal enums it's obviously
better to enforce an exact match, though).
If a CHILD_SA is terminated, the updown event is triggered after the
CHILD_SA is set to state CHILD_DELETED, so no usage stats or detail
information like SPIs were reported. However, when an IKEv2 SA is
terminated, the updown event for its children is triggered without
changing the state first, that is, they usually remain in state
INSTALLED and detailed data was reported in the event. IKEv1
CHILD_SAs are always terminated individually, i.e. with state
change and no extra data so far.
With this change usage stats are also returned for individually deleted
CHILD_SAs as long as the SA has not yet expired.
Fixes#3198.
Many of the messages sent by the kernel, including confirmations to our
requests, are sent as broadcasts to all PF_KEY sockets. So if an
external tool is used to manage SAs/policies (e.g. unrelated to IPsec)
the receive buffer might be filled, resulting in errors like these:
error sending to PF_KEY socket: No buffer space available
To avoid this, just clear the buffer before sending any message.
Fixes#3225.
This avoids having to call strip_dh() in child_cfg_t::get_proposals().
It also inverts the ALLOW_PRIVATE flag (i.e. makes it SKIP_PRIVATE) so
nothing has to be supplied to clone complete proposals.
During proposal selection with ike/child_cfgs a couple of boolean
variables can be set (e.g. private, prefer_self, strip_dh). To simplify
the addition of new parameters, these functions now use a set of flags
instead of indiviual boolean values.
Signed-off-by: Thomas Egerer <thomas.egerer@secunet.com>
Our own implementation ignores NULL values, however, explicit_bzero()
can't handle that, as indicated by the `__nonnull ((1))` attribute in the
function's signature in string.h, and causes a segmentation fault. This
was noticed in one of the unit tests for NewHope. Since we usually use
memwipe() via chunk_clear(), which already ignores NULL pointers, this
is not that much of an issue in practice.
Fixes: 149d1bbb05 ("memory: Use explicit_bzero() as memwipe() if available")
The behavior is undefined if this happens (RFC 7296, section 2.13).
Instead of switching to the non-counter mode, or letting the counter
wrap, this makes it clear that the usage was not as intended.
Martin Willi