This commit adds support for the liveliness QoS for the liveliness
kinds automatic and manual-by-participant. It also implements the
lease_duration from this QoS, which was ignored until now. In the
api the function dds_assert_liveliness is added to assert liveliness
on a participant, which can be used when using liveliness kind
manual-by-participant.
Liveliness kind manual-by-topic is not yet supported, this will be
added in a later commit.
* Proxy participants now have 2 fibheaps to keep leases: one for leases
of pwrs with automatic liveliness and one for leases of the pwrs with
manual-by-participant liveliness (both protected by the proxypp lock).
The minl_auto and minl_man members represent the shortest lease from
these fibheaps and these leases are renewed when receiving data.
Replacing the minl_ leases is now done by replacing the lease object
(atomic ptr) with delayed deletion of the old lease using the gc.
* Proxy writers are set not-alive when the lease expired, and reset to
alive then data is received. When data is received by a pwr, the other
pwrs in the proxypp might also be set alive. I think the specification
is not clear at this point, and for now I have not implemented this
* I refactored out the counter for man-by-pp proxy writers and improved
locking when updating the min-leases on the proxy participant, so I
think this fixes the race conditions.
Some additional tests are required, e.g. to test the not-alive->alive
transition for pwrs. I will add these in short term, as well as the
implementation of the manual-by-topic liveliness kind.
Signed-off-by: Dennis Potman <dennis.potman@adlinktech.com>
Sign extension causes some character values to be interpreted as special
tokens, potentially causing the parser to hang, pass a null pointer as
element name to the "elem_open" callback (which would crash the
configuration processing).
Signed-off-by: Erik Boasson <eb@ilities.com>
When scanning buckets while looking for a specific element, it is faster
to only inspect those buckets for which the corresponding bit in
"hopinfo" is set. Sadly, simple scanning is faster than the far more
elegant perfect hash based on a De Bruijn sequence.
Add a simple test/performance measurement.
Signed-off-by: Erik Boasson <eb@ilities.com>
* Moved bswap functions to ddsrt
Moved the byte swapping functions from ddsi to ddsrt so that
these can be re-used in e.g. the security plugins and tests.
Signed-off-by: Dennis Potman <dennis.potman@adlinktech.com>
* Moved decarations for bswap functions to ddsrt
Signed-off-by: Dennis Potman <dennis.potman@adlinktech.com>
CMAKE_PROJECT_NAME refers to the top-level project name, not the most recent project. So any CMake project that pulls this in as a dependency was in for a nasty surprise.
https://cmake.org/cmake/help/latest/variable/CMAKE_PROJECT_NAME.html
Signed-off-by: Dan Rose <dan@digilabs.io>
The test only works on Linux and macOS because of the system
dependencies in catching an abort. The logging code is platform
independent so testing only on these platforms still gives a good sanity
check.
Signed-off-by: Erik Boasson <eb@ilities.com>
* macOS < 10.13: skip log tests for want of fmemopen
There is a dependency on fmemopen in the tests for the logging
functions, which macOS introduced only in version 10.13. There should
be no other obstacle to building/running/testing on older versions of
macOS. This commit does the pragmatic thing: remove the dependency and
turns the tests into no-ops in this one case.
Signed-off-by: Erik Boasson <eb@ilities.com>
* Fix macOS clock_gettime_nsec_np availability test
Signed-off-by: Erik Boasson <eb@ilities.com>
* Fix type of num reliable readers (int to int32_t)
* Conversion codes in debug monitor printf formats
* Dead code elimination
* Skipping a test case where SIZE_MAX is assumed > INT32_MAX if
assumption is false on target platform
* Error handling in os_sockWaitsetNew
* Stick to unsigned in fragment size calculations
This check is actually guarded by valid_DataFrag and was safe for
datagrams up to 2GB, but the unintended and implicit conversion to is
still best eliminated.
* A "server" connection never has an invalid socket in TCP wrapper
* Handle error return from gethostname in SPDP write (CID 248183)
* Handle extended retcodes in dds_strretcode
CID 248131, introduced by 19aec98b8a
* Remove dead code in ddsrt logging test (CID 248195)
* Validate command-line argument in process test (CID 248117)
* Allow for extremely delayed store in test
Test is constructed to have the events trigger only at the appropriate
times, but it does assume that the store to cb_called becomes visible
prior to the listener callback. I'm pretty sure that will always be
the case in practice, but I'm also pretty sure there is no formal
guarantee without a memory barrier, which mutex_unlock provides.
CID 248088, 248136, 248177, 253590, 253591, 253593
* Check unsetenv return value in test (CID 248099)
Signed-off-by: Erik Boasson <eb@ilities.com>
Initialisation of ddsrt layer uses a hand-rolled CAS/sleep device not
unlike a spin lock. This so initialisation doesn't depend on, e.g.,
ddsrt_once.
Checking or changing thread states between "awake" and "asleep" can end
up in ddsrt_init if the thread is unknown at the time of the call.
Once really only ends up in those cases when the library is initialised
already, in which case no sleeping occurs.
In any case, the sleep is just a friendly yielding of the CPU. Coverity
will still see the loop, just not the sleep.
Signed-off-by: Erik Boasson <eb@ilities.com>
This commit adds two entity types: a "domain", which is the parent of
participants and which is instantiated for each domain that has at least
one participant in it; and "cyclonedds", which is a representation of
the (initialized) Cyclone DDS library in the process and that is the
parent of all domain entities. The handle of the latter is a
compile-constant, DDS_CYCLONEDDS_HANDLE.
This changes the return value from dds_get_parent when executed on a
participant: it now returns the handle of the entity representing the
domain the participant is attached to. Two participants in the same
domain self-evidently return the same domain entity.
This allows deleting all participants in a domain by calling dds_delete
on the domain entity, or tearing down everything and deinitializing the
library by calling dds_delete on the top-level entity.
Signed-off-by: Erik Boasson <eb@ilities.com>
* Fix compilation errors for RMW build on Windows
This commit fixes the compilation errors that occur when building
the ROS2 rmw_cyclonedds_cpp module on Windows with the msvc C++
compiler. The error are fixed by adding explicit casts when calling
operations on atomics and a different syntax is used for compound
literals (the c99 syntax is not supported in msvc in C++ mode).
In additionally some warnings related to emtpy array usage
in structs are suppressed for msvc.
Signed-off-by: Dennis Potman <dennis.potman@adlinktech.com>
* Restructured macros for vendor id cast and casts for atomic operations
To improve readability I've restructured the macros that are used
for vendor id casts on msvc in c++ mode and macros for type-casting
arguments in operations on atomics (as suggested in the review of
my previous commit)
Signed-off-by: Dennis Potman <dennis.potman@adlinktech.com>
* Moved atomic function suffix to macro
Moved the suffix for (msvc) 64 bit atomic functions to the
DDSRT_ATOMIC_OP64 macro
Signed-off-by: Dennis Potman <dennis.potman@adlinktech.com>
It currently uses process id and timestamp, but on a low-resolution
clock that might result in two subsequent initialisations in the same
process yielding the same seed, and that wasn't the intent.
Signed-off-by: Erik Boasson <eb@ilities.com>
Change the structure of the configuration file (in a backwards
compatible manner) to allow specifying configurations for multiple
domains in a file. (Listing multiple files in CYCLONEDDS_URI was
already supported.) A configuration specifies an id, with a default of
any, configurations for an incompatible id are ignored.
If the application specifies an id other than DDS_DOMAIN_DEFAULT in the
call to create_participant, then only configuration specifications for
Domain elements with that id or with id "any" will be used. If the
application does specify DDS_DOMAIN_DEFAULT, then the id will be taken
from the first Domain element that specifies an id. If none do, the
domain id defaults to 0. Each applicable domain specification is taken
as a separate source and may override settings made previously.
All settings moved from the top-level CycloneDDS element to the
CycloneDDS/Domain element. The CycloneDDS/Domain/Id element moved to
become the "id" attribute of CycloneDDS/Domain. The old locations still
work, with appropriate deprecation warnings.
Signed-off-by: Erik Boasson <eb@ilities.com>
* constness in ternary expressions
* removal of OS_MAX_INTEGER
* inclusion of dds/ddsrt/attributes.h everywhere DDS_EXPORT inline
occurs
* _POSIX_PTHREAD_SEMANTICS in ddsperf
Signed-off-by: Erik Boasson <eb@ilities.com>
This commit moves all but a handful of the global variables into the
domain object, in particular including the DDSI configuration, globals
and all transport internal state.
The goal of this commit is not to produce the nicest code possible, but
to get a working version that can support multiple simultaneous domains.
Various choices are driven by this desire and it is expected that some
of the changes will have to be undone. (E.g., passing the DDSI globals
into address set operations and locator printing because there is no
other way to figure out what transport to use for a given locator;
storing the transport pointer inside the locator would solve that.)
Signed-off-by: Erik Boasson <eb@ilities.com>
The main change is a longer sleep in the main thread while waiting for
the select call to timeout. Still not perfect, but more practical than
checking the stacktrace of the thread calling ddsrt_select() to see
whether it has spent too much time in the call.
Signed-off-by: Erik Boasson <eb@ilities.com>
Some Linux versions sometimes return this (undocumented) error,
presumably because of firewalling. Better to ignore it.
Signed-off-by: Erik Boasson <eb@ilities.com>
* use multicast only for participant discovery if using a WiFi network
* default to using unicast for retransmits
Signed-off-by: Erik Boasson <eb@ilities.com>
It is an excellent platform for catching bugs: big-endian, slow enough
that a context switch in the middle of an operation becomes a regular
occurrence, and all that on a SMP box. Or: I just wanted to see if it
would work.
Signed-off-by: Erik Boasson <eb@ilities.com>
The two do essentially the same think, and ddsrt_strtok_r was only used
in one place. (Triggered by Solaris 2.6 not providing strtok_r.)
Signed-off-by: Erik Boasson <eb@ilities.com>
On 32-bit machines uintmax_t is likely to be slower than uintptr_t, and
for that reason, using an uintmax_t for a thread id seems highly
unlikely. For the current platforms, uintptr_t works.
Signed-off-by: Erik Boasson <eb@ilities.com>
Signed-off-by: Thijs Sassen <thijs.sassen@adlinktech.com>
Adjusted the close methode not to expand by the lwip close macro and added a check for DDSI_INCLUDE_SSM to match the correct pid table size.
Signed-off-by: Thijs Sassen <thijs.sassen@adlinktech.com>
Rather than allocate a HH_HOP_RANGE large array of buckets, allocate
just 1 if the initial size is 1, then jump to HH_HOP_RANGE as soon as a
second element is added to the table. There are quite a few cases where
hash tables are created where there never be more than 1 (or even 0)
elements in the table (e.g., a writer without readers, a reader for a
keyless topic).
Signed-off-by: Erik Boasson <eb@ilities.com>
Thread sanitizer warns about reads and writes of variables that are
meant to be read without holding a lock:
* Global "keep_going" is now a ddsrt_atomic_uint32_t
* Thread "vtime" is now a ddsrt_atomic_uint32_t
Previously the code relied on the assumption that a 32-bit int would be
treated as atomic, now that is all wrapped in ddsrt_atomic_{ld,st}32.
These being inline functions doing exactly the same thing, there is no
functional change, but it does allow annotating the loads and stores for
via function attributes on the ddsrt_atomic_{ld,st}X.
The concurrent hashtable implementation is replaced by a locked version
of the non-concurrent implementation if thread sanitizer is used. This
changes eliminates the scores of problems signalled by thread sanitizer
in the GUID-to-entity translation and the key-to-instance id lookups.
Other than that, this replaces a flag used in a waitset test case to be
a ddsrt_atomic_uint32_t.
Signed-off-by: Erik Boasson <eb@ilities.com>
