| Commit message (Collapse) | Author | Age | |
|---|---|---|---|
| * | Dead code removal (stuff for removed sync transport apis.) | Garrett D'Amore | 2017-07-07 |
| | | |||
| * | TCP asynchronous working now. | Garrett D'Amore | 2017-07-07 |
| | | | | | | | | | | | | | | | | | 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. | ||
| * | Fix lost event during poll. | Garrett D'Amore | 2017-07-06 |
| | | |||
| * | Fixes for async resolver, plus a test suite for it. | Garrett D'Amore | 2017-07-06 |
| | | |||
| * | Initial swag at asynchronous name resolution. | Garrett D'Amore | 2017-07-06 |
| | | |||
| * | Make ipc work 100% async. | Garrett D'Amore | 2017-07-05 |
| | | | | | | | | | | | | | | | | | | | The connect & accept logic for IPC is now fully asynchronous. This will serve as a straight-forward template for TCP. Note that the upper logic still uses a thread to run this "synchronously", but that will be able to be removed once the last transport (TCP) is made fully async. The unified ipcsock is also now separated, and we anticipate being able to remove the posix_sock.c logic shortly. Separating out the endpoint logic from the pipe logic helps makes things clearer, and may faciliate a day where endpoints have multiple addresses (for example with a connect() endpoint that uses a round-robin DNS list and tries to run the entire list in parallel, stopping with the first connection made.) The platform header got a little cleanup while we were here. | ||
| * | Move TCP negotiation out of sync accept/connect. | Garrett D'Amore | 2017-07-05 |
| | | |||
| * | Move IPC negotiation out of connect/accept. | Garrett D'Amore | 2017-07-05 |
| | | | | | | | This prevents a slow partner from blocking new connections from being established on the server. Before this a single partner could cause the server to block waiting to complete the negotiation. | ||
| * | Return void on send & recv for transports. | Garrett D'Amore | 2017-07-05 |
| | | |||
| * | Cleanup some unused things. | Garrett D'Amore | 2017-07-05 |
| | | |||
| * | Remove obsolete poll.c. | Garrett D'Amore | 2017-07-05 |
| | | |||
| * | epdesc functionality. | Garrett D'Amore | 2017-07-05 |
| | | |||
| * | Separate out poller/pollq from basic socket operations. | Garrett D'Amore | 2017-07-04 |
| | | |||
| * | Improved routines for list management. | Garrett D'Amore | 2017-07-04 |
| | | |||
| * | All pipes are nonblocking. | Garrett D'Amore | 2017-07-04 |
| | | |||
| * | Rename aio functions. Enhanced epdesc_finish. | Garrett D'Amore | 2017-07-04 |
| | | |||
| * | IPC & TCP negotiation done using aio. Remove old sync send/recv. | Garrett D'Amore | 2017-07-03 |
| | | |||
| * | inproc transport uses aio for connect/accept. | Garrett D'Amore | 2017-07-03 |
| | | |||
| * | Rename legacy connect/accept as _sync for now. | Garrett D'Amore | 2017-07-03 |
| | | |||
| * | Remove the unused infinite timeout versions of msgq. | Garrett D'Amore | 2017-07-03 |
| | | |||
| * | Delete unused msgq_notify. (We use aios instead.) | Garrett D'Amore | 2017-07-03 |
| | | |||
| * | Use common aio cancellation. | Garrett D'Amore | 2017-07-02 |
| | | |||
| * | More aio name cleanups. | Garrett D'Amore | 2017-07-02 |
| | | |||
| * | Remove the extra _aio_ part of pipe send and recv functions. | Garrett D'Amore | 2017-07-02 |
| | | |||
| * | pipe_send and recv can be void. | Garrett D'Amore | 2017-07-02 |
| | | |||
| * | Transports allocate their pipe structures during connect & accept. | Garrett D'Amore | 2017-07-02 |
| | | |||
| * | Properly initialize eps during listen. | Garrett D'Amore | 2017-06-30 |
| | | |||
| * | pipedesc_read/write -> pipedesc_send/recv. | Garrett D'Amore | 2017-06-30 |
| | | |||
| * | More progress on POSIX async connect stuff. | Garrett D'Amore | 2017-06-30 |
| | | | | | | | | | | Note that we're going to refactor this again, for both TCP and IPC, to actually push the endpoint abstraction further down instead of using a combined "socket" abstraction. This may help solve other problems, such as parallel outgoing connections. Nonetheless, most of the work to make POSIX sockets fully async is now done. | ||
| * | IPC fixes: correct handling of path removal, and path absence. | Garrett D'Amore | 2017-06-30 |
| | | |||
| * | Fixes for IPC: don't try to disable Nagle, and use SUN_LEN properly. | Garrett D'Amore | 2017-06-29 |
| | | |||
| * | More plumbing for async connections. | Garrett D'Amore | 2017-06-29 |
| | | |||
| * | Begin work on async connect/accept for POSIX. Not referenced yet. | Garrett D'Amore | 2017-06-29 |
| | | |||
| * | Use common POSIX socket handling for IPC. | Garrett D'Amore | 2017-06-29 |
| | | |||
| * | Avoid Solaris / illumos namespace collision. | Garrett D'Amore | 2017-06-29 |
| | | |||
| * | Use common socket handling on POSIX (tcp done, ipc pending.) | Garrett D'Amore | 2017-06-29 |
| | | |||
| * | Adjust npds count properly. | Garrett D'Amore | 2017-06-29 |
| | | |||
| * | Pass cancel of IPC and TCP all the way down to POSIX pipedescs. | Garrett D'Amore | 2017-06-29 |
| | | |||
| * | We don't need pipe_hold or pipe_rele anymore. | Garrett D'Amore | 2017-06-28 |
| | | |||
| * | Refactor stop again, closing numerous races (thanks valgrind!) | Garrett D'Amore | 2017-06-28 |
| | | |||
| * | Clean up the pipe system (no leaks). | Garrett D'Amore | 2017-06-28 |
| | | |||
| * | Cancellation plumbing for message queues. | Garrett D'Amore | 2017-06-27 |
| | | |||
| * | Put errors go on the putq. | Garrett D'Amore | 2017-06-27 |
| | | |||
| * | Convert to POSIX polled I/O for async; start of cancelable aio. | Garrett D'Amore | 2017-06-27 |
| | | | | | | | | | | | | | | | | | | | 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. | ||
| * | More race condition fixes. | Garrett D'Amore | 2017-06-26 |
| | | |||
| * | Set the expiration under lock (data race). | Garrett D'Amore | 2017-06-26 |
| | | |||
| * | Fix uninitialized use (thanks valgrind.) | Garrett D'Amore | 2017-06-26 |
| | | |||
| * | Adding mutex lock during fini to prevent false positive data race. | Garrett D'Amore | 2017-06-25 |
| | | |||
| * | Simplify timer locks, hopefully resolving potential deadlock. | Garrett D'Amore | 2017-06-25 |
| | | |||
| * | Fix a few race condition panics. | Garrett D'Amore | 2017-06-25 |
| | | |||
 |
index : nng | |
| A mirror of https://github.com/nanomsg/nng |
| aboutsummaryrefslogtreecommitdiff |