Description

This PR is a revision of #3046, in which the data race between WriterProxy::stop and perform_initial_acknack was addressed by forcing the thread invoking the former to wait for the completion of the latter. However, a deadlock may occur as the reader waiting for the event to end does so with Endpoint::mp_mutex locked, which is also locked when sending an acknack at perform_initial_acknack.

Two different solutions are suggested:

• Introducing a new mutex in StatefulReader which is taken before reader's, and released after modifying the writer proxy's state and before waiting for the executing event to finish.
• Releasing reader's mutex before stopping a WriterProxy, and locking it again after doing so. This solution implies that StatefulReader::matched_writer_add and StatefulReader::matched_writer_remove may no longer be executed "atomically", a relevant change that should be carefully analyzed.

In addition, this PR also addresses the potential (rarely or never seen in our tests) datarace that may occur between WriterProxy::stop and perform_heartbeat_response, as StatefulReader::send_acknack reads attributes from WriterProxy which are modified when being stopped.

Contributor Checklist

• Commit messages follow the project guidelines.
• The code follows the style guidelines of this project.
• Tests that thoroughly check the new feature have been added/Regression tests checking the bug and its fix have been added.
• Any new/modified methods have been properly documented using Doxygen.
• Fast DDS test suite has been run locally.
• Changes are ABI compatible.
• Changes are API compatible.
• Documentation builds and tests pass locally.
• New feature has been added to the versions.md file (if applicable).
• New feature has been documented/Current behavior is correctly described in the documentation.

Reviewer Checklist

• Check contributor checklist is correct.
• Check CI results: changes do not issue any warning.
• Check CI results: failing tests are unrelated with the changes.