| Commit message (Collapse) | Author | Age |
|---|
| |
|
|
|
| |
This is not quite complete, but it sets the stage for other
protocols (such as zmq or mqtt) to be added to the project.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is a sweeping cleanup of the transport logic around options,
and also harmonizes the names used when setting or getting options.
Additionally, legacy methods are now moved into a separate file and
can be elided via CMake or a preprocessor define.
Fundamentally, the ability to set to transport options via the socket
is deprecated; there are numerous problems with this and my earlier
approaches to deal with this have been somewhat misguided. Further
these approaches will not work with future protocol work that is
planned (were some options need to be negotiated with peers at the
time of connection establishment.)
Documentation has been updated to reflect this. The test suites still
make rather broad use of the older APIs, and will be converted later.
|
| |
|
|
|
| |
Also, addressed a number of Clang-tidy complaints. Potential hangs
in close addressed as well.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
This introduces reference counting on messages to reduce the data copies.
This should have a marked improvement when moving large messages through
the system, or when publishing to many subscribers. For some transports,
when using large messages, the copy time can be the dominant factor.
Note that when a message is actually shared, inproc will still perform
an extra copy in order to ensure that it can modify the headers.
This will unfortunately always be the case with REQ, as the REQ protocol
keeps a copy of the original message so it can retry.
|
| |
|
|
|
|
| |
In addition, the message queue code is rather heavy weight, so this may
make the inproc code a bit leaner. It also makes backpressure conditions
much more reliable since it now is completely unbuffered.
|
| | |
|
| |
|
|
|
|
|
| |
This also eliminates the enforcement of NNG_OPT_RECVMAXSZ for inproc,
which never really made much sense. This helps inproc go faster.
While here, also clean up the entry point for protocols to support
a drain option, since we don't use that anywhere.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
fixes #698 Need TCP stats
fixes #699 Need IPC stats
fixes #701 Need TLS stats
This commit addresses a problem when negotiating using one of the stream
based negotiation APIs -- a slow or misbehaving peer can prevent well
behaved ones from establishing a connection. The fix is a fairly
significant change in how these transports link up, and it does rely
on the fact that the socket only has a single accept() or connect()
pending at a time (on a given endpoint that is).
While here, we have completely revamped the way transport statistics are
done, offering a standard API for collecting these statistics.
Unfortunately, this completely borks the statistics for inproc. As we
are planning to change the way inproc works soon, in order to provide
more control and work on performance fixes for the message queue, we feel
this is an acceptable trade-off. Furthermore, almost nobody uses inproc
for anything, and even fewer people are making use of the statistics
at this time.
|
| |
|
|
|
| |
Some more changes to use nni_type instead of nni_opt_type
are included as well.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This introduces new public APIs for obtaining statistics,
and adds some generic stats for dialers, listeners, pipes, and
sockets. Also added are stats for inproc and pairv1 protocol.
The other protocols and transports will have stats added
incrementally as time goes on.
A simple test program, and man pages are provided for this.
Start by looking at nng_stat(5).
Statistics does have some impact, and they can be disabled by
using the advanced NNG_ENABLE_STATS (setting it to OFF, it's
ON by default) if you need to build a minimized configuration.
|
| | |
|
| |
|
|
|
|
|
|
|
| |
This changes the signature of the aio cancellation routines
to take the argument for cancellation directly, so we do not
need to lookup the argument using the nni_aio_get_prov_data.
We should probably consider eliminating nni_aio_get_prov_data,
and co, and changing the prov_extra to reflect prov_data. Later.
|
| | |
|
| |
|
|
|
|
|
|
|
|
| |
This separates the plumbing for endpoints into distinct
dialer and listeners. Some of the transports could benefit
from further separation, but we've done some rather larger
separation e.g. for the websocket transport.
IPC would be a good one to update later, when we start looking
at exposing a more natural underlying API.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
fixes #538 setopt should have an explicit chkopt routine
fixes #537 Internal TCP API needs better name separation
fixes #524 Option types should be "typed"
This is a rework of the option management code, to make it both clearer
and to prepare for further work to break up endpoints. This reduces
a certain amount of dead or redundant code, and actually saves cycles
when setting options, as some loops were not terminated that should have
been.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* fixes #419 want to nni_aio_stop without blocking
This actually introduces an nni_aio_close() API that causes
nni_aio_begin to return NNG_ECLOSED, while scheduling a callback
on the AIO to do an NNG_ECLOSED as well. This should be called
in non-blocking close() contexts instead of nni_aio_stop(), and
the cases where we call nni_aio_fini() multiple times are updated
updated to add nni_aio_stop() calls on all "interlinked" aios before
finalizing them.
Furthermore, we call nni_aio_close() as soon as practical in the
close path. This closes an annoying race condition where the
callback from a lower subsystem could wind up rescheduling an
operation that we wanted to abort.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
fixes #326 consider nni_taskq_exec_synch()
fixes #410 kqueue implementation could be smarter
fixes #411 epoll_implementation could be smarter
fixes #426 synchronous completion can lead to panic
fixes #421 pipe close race condition/duplicate destroy
This is a major refactoring of two significant parts of the code base,
which are closely interrelated.
First the aio and taskq framework have undergone a number of simplifications,
and improvements. We have ditched a few parts of the internal API (for
example tasks no longer support cancellation) that weren't terribly useful
but added a lot of complexity, and we've made aio_schedule something that
now checks for cancellation or other "premature" completions. The
aio framework now uses the tasks more tightly, so that aio wait can
devolve into just nni_task_wait(). We did have to add a "task_prep()"
step to prevent race conditions.
Second, the entire POSIX poller framework has been simplified, and made
more robust, and more scalable. There were some fairly inherent race
conditions around the shutdown/close code, where we *thought* we were
synchronizing against the other thread, but weren't doing so adequately.
With a cleaner design, we've been able to tighten up the implementation
to remove these race conditions, while substantially reducing the chance
for lock contention, thereby improving scalability. The illumos poller
also got a performance boost by polling for multiple events.
In highly "busy" systems, we expect to see vast reductions in lock
contention, and therefore greater scalability, in addition to overall
improved reliability.
One area where we currently can do better is that there is still only
a single poller thread run. Scaling this out is a task that has to be done
differently for each poller, and carefuly to ensure that close conditions
are safe on all pollers, and that no chance for deadlock/livelock waiting
for pfd finalizers can occur.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This closes a fundamental flaw in the way aio structures were
handled. In paticular, aio expiration could race ahead, and
fire before the aio was properly registered by the provider.
This ultimately led to the possibility of duplicate completions
on the same aio.
The solution involved breaking up nni_aio_start into two functions.
nni_aio_begin (which can be run outside of external locks) simply
validates that nni_aio_fini() has not been called, and clears certain
fields in the aio to make it ready for use by the provider.
nni_aio_schedule does the work to register the aio with the expiration
thread, and should only be called when the aio is actually scheduled
for asynchronous completion. nni_aio_schedule_verify does the same thing,
but returns NNG_ETIMEDOUT if the aio has a zero length timeout.
This change has a small negative performance impact. We have plans to
rectify that by converting nni_aio_begin to use a locklesss flag for
the aio->a_fini bit.
While we were here, we fixed some error paths in the POSIX subsystem,
which would have returned incorrect error codes, and we made some
optmizations in the message queues to reduce conditionals while holding
locks in the hot code path.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
fixes #302 nng_dialer/listener/pipe_getopt_sockaddr desired
This adds plumbing to pass and check the type of options
all the way through.
NNG_ZT_OPT_ORBIT is type UINT64, but you can use the untyped form to
pass two of them if needed.
No typed access for retrieving strings yet. I think this should allocate
a pointer and copy that out, but that's for later.
|
| |
|
|
| |
fixes #290 sockaddr improvements
|
| |
|
|
|
|
|
|
| |
This addresses the use of the pipe special field, and eliminates it.
The message APIs (recvmsg, sendmsg) need to be updated as well still,
but I want to handle that as part of a separate issue.
While here we fixed various compiler warnings, etc.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This introduces enough of the HTTP API to support fully server
applications, including creation of websocket style protocols,
pluggable handlers, and so forth.
We have also introduced scatter/gather I/O (rudimentary) for
aios, and made other enhancements to the AIO framework. The
internals of the AIOs themselves are now fully private, and we
have eliminated the aio->a_addr member, with plans to remove the
pipe and possibly message members as well.
A few other minor issues were found and fixed as well.
The HTTP API includes request, response, and connection objects,
which can be used with both servers and clients. It also defines
the HTTP server and handler objects, which support server applications.
Support for client applications will require a client object to be
exposed, and that should be happening shortly.
None of this is "documented" yet, bug again, we will follow up shortly.
|
| |
|
|
|
|
|
|
|
|
| |
This eliminates a bunch of redundant URL parsing, using the common
URL logic we already have in place.
While here I fixed a problem with the TLS and WSS test suites that
was failing on older Ubuntu -- apparently older versions of mbedTLS
were unhappy if selecting OPTIONAL verification without a validate
certificate chain.
|
| |
|
|
|
| |
fixes #210 Want NNG_OPT_TLS_* options for TLS transport
fixes #212 Eliminate a_endpt member of aio
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This makes all the protocols and transports optional. All
of them except ZeroTier are enabled by default, but you can
now disable them (remove from the build) with cmake options.
The test suite is modified so that tests still run as much
as they can, but skip over things caused by missing functionality
from the library (due to configuration).
Further, the constant definitions and prototypes for functions
that are specific to transports or protocols are moved into
appropriate headers, which should be included directly by
applications wishing to use these.
We have also added and improved documentation -- all of the
transports are documented, and several more man pages for
protocols have been added. (Req/Rep and Surveyor are still
missing.)
|
| |
|
|
|
|
|
| |
We added nng_inproc man page, and a header with an explicit
registration method for it. We also fixed up the markup on
several other pages, and corrected the include statements for
nng_zerotier.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
This makes the APIs use string keys, and largely eliminates the use of
integer option IDs altogether. The underlying registration for options
is also now a bit richer, letting protcols and transports declare the
actual options they use, rather than calling down into each entry point
carte blanche and relying on ENOTSUP.
This code may not be as fast as the integers was, but it is more intuitive,
easier to extend, and is not on any hot code paths. (If you're diddling
options on a hot code path you're doing something wrong.)
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This eliminates all the old #define's or enum values, making all
option IDs now totally dynamic, and providing well-known string
values for well-behaved applications.
We have added tests of some of these options, including lookups, and
so forth. We have also fixed a few problems; including at least
one crasher bug when the timeouts on reconnect were zero.
Protocol specific options are now handled in the protocol. We will
be moving the initialization for a few of those well known entities
to the protocol startup code, following the PAIRv1 pattern, later.
Applications must therefore not depend on the value of the integer IDs,
at least until the application has opened a socket of the appropriate
type.
|
| |
|
|
|
|
|
| |
This eliminates tests for code that we cannot reach, because the
upper layer endpoint code already ensures that we don't get called
if we are closing, that the mode is correct, and that only one
outstanding endpoint operation is in progress on any given endpoint.
|
| |
|
|
|
|
| |
This also includes tests for some of the edge cases surrounding
pluggable transports, such as version mismatch, duplication registration,
and failure to initialize.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If the underlying platform fails (FreeBSD is the only one I'm aware
of that does this!), we use a global lock or condition variable instead.
This means that our lock initializers never ever fail.
Probably we could eliminate most of this for Linux and Darwin, since
on those platforms, mutex and condvar initialization reasonably never
fails. Initial benchmarks show little difference either way -- so we
can revisit (optimize) later.
This removes a lot of otherwise untested code in error cases and so forth,
improving coverage and resilience in the face of allocation failures.
Platforms other than POSIX should follow a similar pattern if they need
this. (VxWorks, I'm thinking of you.) Most sane platforms won't have
an issue here, since normally these initializations do not need to allocate
memory. (Reportedly, even FreeBSD has plans to "fix" this in libthr2.)
While here, some bugs were fixed in initialization & teardown.
The fallback code is properly tested with dedicated test cases.
|
| |
|
|
|
| |
Also enables creating endpoints that are idle (first part of
endpoint options API) and shutting down endpoints.
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
This makes the operations that work on headers start with
nni_msg_header or nng_msg_header. It also renames _trunc to
_chop (same strlen as _trim), and renames prepend to insert.
We add a shorthand for clearing message content, and make
better use of the endian safe 32-bit accessors too.
This also fixes a bug in inserting large headers into messages.
A test suite for message handling is included.
|
| |
|
|
|
|
|
|
|
| |
This passes valgrind 100% clean for both helgrind and deep leak
checks. This represents a complete rethink of how the AIOs work,
and much simpler synchronization; the provider API is a bit simpler
to boot, as a number of failure modes have been simply eliminated.
While here a few other minor bugs were squashed.
|
| |
|
|
|
|
|
|
|
|
|
| |
We have seen leaks of pipes causing test failures (e.g. the Windows
IPC test) due to EADDRINUSE. This was caused by a case where we
failed to pass the pipe up because the AIO had already been canceled,
and we didn't realize that we had oprhaned the pipe. The fix is to
add a return value to nni_aio_finish, and verify that we did finish
properly, or if we did not then we must free the pipe ourself. (The
zero return from nni_aio_finish indicates that it accepts ownership
of resources passed via the aio.)
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
| |
Most of the races around close were probably here - the cancellation was
not getting through on endpoint close, which meant that we could actually
toss endpoints while they were in use.
We need to fix the timeouts stuff -- especially for reconnects etc, but
we are just about ready for this stuff to be reintegrated into master.
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
It turns out that I had to fix a number of subtle asynchronous
handling bugs, but now TCP is fully asynchronous. We need to
change the high-level dial and listen interfaces to be async
as well.
Some of the transport APIs have changed here, and I've elected
to change what we expose to consumers as endpoints into seperate
dialers and listeners. Under the hood they are the same, but
it turns out that its helpful to know the intended use of the
endpoint at initialization time.
Scalability still occasionally hangs on Linux. Investigation
pending.
|
| | |
|
| | |
|
