Deleting a writer with unacknowledged data present in its WHC causes it
to linger for a configurable duration. Once it is lingering, there are
two routes to actually deleting the writer: because the samples get
acknowledged, or because the linger duration elapses.
When these two happen roughly concurrently, there was a possibility of
both succeeding in looking up the writer by its GUID, in which case one
of them then asserts on removing it from the entity index (if assertions
are enabled, if not, things are worse).
This fixes that by ensuring only one of the two actually does something,
as was always the intent.
Signed-off-by: Erik Boasson <eb@ilities.com>
As opposed to NOT_ALLOWED_BY_SECURITY. There is a meaningful
difference between something being disallowed and something being
impossible.
Co-Authored-By: Kyle Fazzari <github@status.e4ward.com>
Signed-off-by: Erik Boasson <eb@ilities.com>
Currently:
* DDS_HAS_SECURITY for DDS Security support
* DDS_HAS_LIFESPAN for lifespan QoS support
* DDS_HAS_DEADLINE_MISSED for "deadline missed" event support
These are defined to 1 if support for the feature is included in the
build and left undefined if it isn't.
Signed-off-by: Erik Boasson <eb@ilities.com>
When built without support for DDS Security, any attempt to create a
participant QoS settings in the security name space (those prefixed by
"dds.sec.") must fail.
Signed-off-by: Erik Boasson <eb@ilities.com>
* read/take failed to restore the null pointer in the first entry of the
sample pointer array it gets passed, in the case no "loan" had been
allocated yet and it returned an empty set. The consequence is that
on a subsequence read it will reuse the address without marking at as
in use, so that a *second* read using with a null pointer in that
first entry will overwrite the first result. (Introduced by
d16264fd82.)
* return_loan failed to free all memory if its argument wasn't actually
a loan. There are many good arguments why the read/take/return_loan
interface is messed up, but in the context of the existing interface
this is a perfectly reasonable case: there is at most one "loan" for
each reader, but one can keep calling read/take and return_loan as if
there's an infinite number of "loans". It's just that the first gets
cached and the others don't.
Signed-off-by: Erik Boasson <eb@ilities.com>
Fixed a bug in the subject compare function for identity subjects, that
could cause using the incorrect permission grant in case multiple grants
are provided in the permissions configuration of the access control plugin.
Signed-off-by: Dennis Potman <dennis.potman@adlinktech.com>
The moving around and cleaning up of network code broke the IPv6
multicast support by memcpy'ing a sockaddr_in6 instead of an in6_addr in
a multicast join record.
Signed-off-by: Erik Boasson <eb@ilities.com>
* Remove duplicated code in authentication plugin
Signed-off-by: Dennis Potman <dennis.potman@adlinktech.com>
* Fix build warnings
Signed-off-by: Dennis Potman <dennis.potman@adlinktech.com>
* Fix memory leak and call create_validate_asymmetrical_signature directly from create_validate_signature_impl
Signed-off-by: Dennis Potman <dennis.potman@adlinktech.com>
* Fix refcount issue (assert in openssl) for identity cert in hs remote info
Signed-off-by: Dennis Potman <dennis.potman@adlinktech.com>
* Refactoring of validate_handshake_token function
Co-authored-by: Erik Boasson <eb@ilities.com>
Signed-off-by: Dennis Potman <dennis.potman@adlinktech.com>
The participant listener creates a pong writer, setting a publication
matched listener on it. That listener can be invoked immediately and as
it queries the subscriptions reader, it must not be enabled before the
latter reader has been created.
Signed-off-by: Erik Boasson <eb@ilities.com>
* Move wctime, mtime, etime types to ddsrt
* Add ddsrt_time_wallclock
* Change ddsrt_time_monontic, elapsed to use mtime, etime types
* Remove now, now_mt, now_et
* Rename X_to_sec_usec to ddsrt_X_to_sec_usec
* add_duration_to_X to ddsrt_X_add_duration (to be in line with the
existing ddsrt_time_add_duration)
* elimination of ddsrt/timeconv.h, it added more in the way of
complications than it did in making things more elegant
* rename of q_time.[ch] to ddsi_time.[ch]: that now only deals with DDSI
timestamps and durations on the wire
Signed-off-by: Erik Boasson <eb@ilities.com>
* Fix code formatting, fix for memory leak in validate_handshake_reply_token and
make error handling and return values more consistent with the other two
plugins.
Signed-off-by: Dennis Potman <dennis.potman@adlinktech.com>
* Processed review comments: fixed memory leaks and more consistent error handling and function returns
Signed-off-by: Dennis Potman <dennis.potman@adlinktech.com>
* Fix trusted ca dir max exceeded
Signed-off-by: Dennis Potman <dennis.potman@adlinktech.com>
The test gates access-control plugin invocation and with the inverted
condition all remote readers/writers requiring access control are
blocked eiter because of the permissions handle, or because a NIL handle
is passed to the access control plugin.
Signed-off-by: Erik Boasson <eb@ilities.com>
* access-control check_remote_datareader has "relay_only" as an out
parameter, so should pass in an address instead of "false";
* value of "relay_only" returned by check_remote_datareader must be
passed to crypto register_matched_remote_datareader
Signed-off-by: Erik Boasson <eb@ilities.com>
Security plugins are built but not installed. Add target to CMakeLists.txt
for three security plugins.
Signed-off-by: Sid Faber <sid.faber@canonical.com>
wsock32.lib is only needed for the legacy version of Winsock and is not needed with Winsock2 (the current version).
This appears to be a root cause of the multicast issue on Win10 and may allow us to reverse #404
Signed-off-by: Dan Rose <dan@digilabs.io>
This is a workaround for interoperability issues, ultimately driven by a
Windows quirk that makes multicast delivery within a machine utterly
unreliable if the transmitting socket is bound to 0.0.0.0 (despite all
sockets having multicast interfaces set correctly) when there are also
sockets transmitting to the same multicast group that have been bound to
non-0.0.0.0. (Note: there may be other factors at play, but this is
what it looks like after experimentation.)
At least Fast-RTPS in some versions binds the socket it uses for
transmitting multicasts to non-0.0.0.0, so interoperability with
Fast-RTPS on Windows requires us to bind the socket we use for
transmitting multicasts (which was the same as the one we use for
receiving unicast data) also to non-0.0.0.0 or our multicasts get
dropped often.
This would work fine if other implementations honoured the set of
advertised addresses. However, at least Fast-RTPS and Connext (in some
versions) fail to do this and happily substitute 127.0.0.1 for the
advertised IP address. If we bind to, e.g., 192.168.1.1, then suddenly
those packets won't arrive anymore, breaking interoperability.
The only work around is to use a separate socket for sending.
Signed-off-by: Erik Boasson <eb@ilities.com>
* IPv6 extensions to patterns
* use full GUID prefix for Cyclone
* pattern fixes to deal with small changes in the formatting of QoS
* suppressinof local built-in topic publications
* asymmetrical disconnect detection improvements (better chance of
detecting it, plus better suppression of spurious notifications)
Signed-off-by: Erik Boasson <eb@ilities.com>
Isolate domains by using unique tags when running tests that are using security handshake.
And updated coding style and some minor refactoring in fsm and timed_cb tests.
Signed-off-by: Dennis Potman <dennis.potman@adlinktech.com>
