A failed region lookup was inserted into the thread-local cache as
nullptr, making the failure permanent for the lifetime of the cache
generation - retrying never healed because the fast path would return
nullptr without calling GetRegion again. Skip the cache insert on
failure so subsequent calls retry the slow path.
Add two public entry points needed by the ALICE use case where shmem
messages are allocated via a transport but never sent — their metadata
is instead serialised into Arrow tables and delivered over a separate
channel, allowing consumer devices to resolve the payload pointer
without taking ownership.
shmem::Message::GetMeta() returns the MetaHeader of the message,
mirroring the existing positional-init pattern already used in Socket.h.
shmem::GetDataAddressFromHandle(TransportFactory&, const MetaHeader&)
is a free function declared in Common.h and defined in Manager.cxx.
Keeping it out of the TransportFactory class body means callers only
need to include Common.h (available transitively via Message.h) and do
not drag in Socket.h or zmq.h. The implementation handles both managed
segments and unmanaged regions, and throws SharedMemoryError with a
typed message on a bad segment or region id. TransportFactory also
gains a same-named member for callers that already have the concrete
type. Lifetime of the returned pointer is the caller's responsibility;
the cache device is expected to hold the messages alive.
A SideChannel test covers the GetMeta/GetDataAddressFromHandle
round-trip for both standard and expanded-metadata configurations.
Boost 1.88 replaced Boost.Process with v2, breaking the v1 API.
Boost 1.89 restores v1 compatibility via <boost/process/v1.hpp>.
- Fail configuration if Boost 1.88 is detected
- Define FAIRMQ_BOOST_PROCESS_V1_HEADER for Boost >= 1.89
- Use conditional includes to select v1.hpp or process.hpp
- Add namespace aliases (bp, bp_this) for portable API access
Avoid disseminating every compile unit including Message.h with TransportNames and
TransportTypes and the associated unordered_map helper methods (e.g.
murmur_hash).
The shm metadata msg will be right-padded to the given size. This
tunable may be used to saturate the kernel msg buffers more quickly with
the effect that the ZeroMQ message queue size - on which the FairMQ
shmem transport relies upon - behaves more accurately for very small
queue sizes.
This introduces a change for the meta msg format in the multipart case:
old: | MetaHeader 1 | ... | MetaHeader n |
new: | n | MetaHeader 1 | ... | MetaHeader n | padded to fMetadataMsgSize |
where `n` is a `size_t` and contains the number of following meta headers.
Previously, this number was infered from the msg buffer size itself which is
no longer possible due to the potential padding.
Implements #432
