unix: use uv__hrtime() internally

This commit renames the various uv_hrtime() implementations to uv__hrtime().

Libuv uses the high-res timer internally in performance-critical code

unix: use uv__hrtime() internally

This commit renames the various uv_hrtime() implementations to uv__hrtime().

Libuv uses the high-res timer internally in performance-critical code paths.
Calling the non-public version avoids going through the PLT when libuv is
compiled as a shared object.

The exported uv_hrtime() now has a single definition in src/unix/core.c that
calls uv__hrtime().

A future optimization is to lift the uv__hrtime() declarations into header
files so they can be inlined at the call sites. Then again, linking with -flto
should accomplish the same thing.

show more ...

unix: don't run loop when all handles are unref'd

Implemented uv_run2

Allows for running the event loop in 3 modes:
* default: loop runs until the refcount drops to zero
* once: poll for events only once and block until one is h

Implemented uv_run2

Allows for running the event loop in 3 modes:
* default: loop runs until the refcount drops to zero
* once: poll for events only once and block until one is handled
* nowait: poll for events only once but don't block if there are
no pending events

show more ...

unix: fix event loop stall

Fix a rather obscure bug where the event loop stalls when an I/O watcher is
stopped while an artificial event, generated with uv__io_feed(), is pending.

unix: change uv_backend_timeout() prototype

* change return value to signed int
* constify loop argument

unix: add uv_backend_fd() and uv_backend_timeout()

This can be used in conjuction with uv_run_once() to poll in one thread and run
the event loop's event callbacks in another.

U

unix: add uv_backend_fd() and uv_backend_timeout()

This can be used in conjuction with uv_run_once() to poll in one thread and run
the event loop's event callbacks in another.

Useful for embedding libuv's event loop in another event loop.

show more ...

darwin, freebsd: set SO_NOSIGPIPE on sockets

Suppresses delivery of SIGPIPE signals when the remote end of the socket has
gone away.

unix: rename UV__IO_* constants

unix: remove dependency on libev

unix: support signal handlers outside the main loop

unix: make it possible to delay close callbacks

unix: move getaddrinfo code to getaddrinfo.c

unix: remove __read_mostly macro

It creates trouble with llvm-gcc on OS X.

darwin, freebsd: use ioctl(FIOCLEX) and ioctl(FIONBIO)

Set the non-blocking and close-on-exec flags with ioctl() instead of fcntl(),
it's about 10-25% faster.

Stick with fcntl()

darwin, freebsd: use ioctl(FIOCLEX) and ioctl(FIONBIO)

Set the non-blocking and close-on-exec flags with ioctl() instead of fcntl(),
it's about 10-25% faster.

Stick with fcntl() on Solaris. ioctl(FIONBIO) works but is twice as slow as
fcntl(O_NONBLOCK). ioctl(FIOCLEX) doesn't raise an error but doesn't actually
work either.

show more ...

linux: tag no_accept4 __read_mostly

unix, windows: preliminary signal handler support

* a no-op on Windows for now
* only supports the main loop on UNIX (again, for now)

unix: add relaxed accept() setting

Mitigates unfair scheduling in multi-process setups that share a single listen
socket across multiple processes.

Merge branch 'v0.8'

unix: fix 'zero handles, one request' busy loop

Fixes #484.

Merge branch 'v0.8'

linux: don't use accept4() syscall after ENOSYS

Repeatedly calling the syscall when it's not supported has a small but
measurable performance impact.

Besides, it's a silly thing

linux: don't use accept4() syscall after ENOSYS

Repeatedly calling the syscall when it's not supported has a small but
measurable performance impact.

Besides, it's a silly thing to do.

show more ...

linux: set close-on-exec flag with ioctl(FIOCLEX)

ioctl(FIOCLEX) is 25% faster than fcntl(F_SETFD) on a stock 2.6.32 kernel.

unix: malloc only once in uv_getaddrinfo()

unix: clarify that uv_getaddrinfo_t is a req

unix, windows: add stat() based file watcher

Monitors a file path for changes. Supersedes ev_stat.