| Commit message (Collapse) | Author | Age |
|---|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This eliminates the two threads per pipe that were being used to provide
basic I/O handling, replacing them with a single global thread for now,
that uses poll and nonblocking I/O. This should lead to great scalability.
The infrastructure is in place to easily expand to multiple polling worker
threads. Some thought needs to be given about how to scale this to engage
multiple CPUs. Horizontal scaling may also shorten the poll() lists easing
C10K problem.
We should look into better solutions than poll() for platforms that have
them (epoll on Linux, kqueue on BSD, and event ports on illumos).
Note that the file descriptors start out in blocking mode for now, but
then are placed into non-blocking mode. This is because the negotiation
phase is not yet callback driven, and so needs to be synchronous.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
| |
As with TCP, we're still using threads under the hood. But this
completes the send/recv logic conversion for POSIX to our AIO framework,
and hence represents a substantial milestone towards full asyncronous
operation.
We still need to do accept/connect operations asynchronously, then making.
Windows overlapped IO work properly. After that, poll/epoll/kqueue, etc.
|
| | |
|
| |
|
|
|
|
|
| |
Transport-level pipe initialization is now sepearate and explicit.
The POSIX send/recv logic still uses threads under the hood, but
makes use of the AIO framework for send/recv. This is a key stepping
stone towards enabling poll() or similar async I/O approaches.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
| |
I seem to be having a very difficult time getting dual-stack sockets
to function properly on Windows. I've sort of abandoned it for now.
I need to think about how to solve this -- it's not clear to me
right now whether dual stack sockets are the right answer or not.
People do expect these to work, but a tcp6:// url might be more
elegant.
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Sleep() on Win32 rounds *down*, leading to truncated timeouts.
What we do is change our sleep routing to start incrementally
sleeping by 1ms until the tick count is reached. This ensures
we don't wake early.
This problem affects condition variables too, which means that some
timeouts may occur up to one clock tick early (15ish ms). This should
not be a problem for most users, who should really only be setting
timeouts in quantities of a second or greater.
|
| |
|
|
|
|
|
|
|
|
| |
This compiles correctly, but doesn't actually deliver events yet.
As part of this, I've made most of the initializables in nng
safe to tear-down if uninitialized (or set to zero e.g. via calloc).
This makes it loads easier to write the teardown on error code, since
I can deinit everything, without worrying about which things have been
initialized and which have not.
|
| | |
|
| | |
|
| |
|
|
|
| |
Test code needs to use the static libraries so that they can get access
to the entire set of symbols, including private ones that are not exported.
|
| |
|
|
|
|
| |
It turns out that I didn't quite understand overlapped I/O. We can and
should always do the GetOverlappedResult(), regardless of how the routine
returns.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Since we use the tick counter to sleep, we should use the same clock
for validation. The problem is that the high performance tick counter
on the CPU may be slightly out of agreement with the windows clock.
Furthermore, the tick counter is probably lots faster to retrieve since
it is already updated, and needn't be recalculated each time.
(We should consider just switching to millisecond clock resolution
internally as well. It turns out that I don't think that timers that
are shorter than 1ms are very useful.)
|
| |
|
|
|
|
|
| |
There are lots of changes here, mostly stuff we did in support of
Windows TCP. However, there are some bugs that were fixed, and we
added some new error codes, and generalized the handling of some failures
during accept. Windows IPC (NamedPipes) is still missing.
|
| |
|
|
|
|
| |
Windows is getting there. Needs a couple of more more hours to enable
everything, especially IPC, and most of the work at this point is probably
some combination of debug and tweaking things like error handling.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
| |
Platforms must seed the pRNGs by offering an nni_plat_seed_prng()
routine. Implementations for POSIX using various options (including
the /dev/urandom device) are supplied.
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
There is an occasional use-after-free bug we need to fix still.
|
| |
|
|
|
| |
This also adds checks in the protocols to verify that pipe peers
are of the proper protocol.
|
| | |
|
| |
|
|
|
|
|
| |
This uncovered a few problems - inproc was not moving the headers
to the body on transmit, and the message chunk allocator had a serious
bug leading to memory corruption. I've also added a message dumper,
which turns out to be incredibly useful during debugging.
|
