Merge remote-tracking branch 'origin/v0.10'

darwin: don't select(&exceptfds) in fallback path

The exceptfds set is for polling OOB data, not errors.

Fixes joyent/node#5155.

unix, windows: remove ngx-queue.h

Avoids an extra #include in public headers and stops the ngx_queue_*
types and macros from leaking into user code.

stream: run try_select only for pipes and ttys

Its not necesary for TCP and other streams, since fd is always working
with kqueue there.

unix: handle POLLERR and POLLHUP in uv__stream_io

Fixes a busy loop when the file descriptor emits POLLHUP but not POLLIN
or POLLOUT, e.g. when polling the read end of a pipe and the wri

unix: handle POLLERR and POLLHUP in uv__stream_io

Fixes a busy loop when the file descriptor emits POLLHUP but not POLLIN
or POLLOUT, e.g. when polling the read end of a pipe and the write end
is closed.

Fixes joyent/node#4923.

show more ...

darwin: fix spurious uv_write2() segfault

We abuse uv_write2() to send over UDP handles to child processes.
Don't call uv__stream_fd() on those handles, it's a macro that on OS X
eva

darwin: fix spurious uv_write2() segfault

We abuse uv_write2() to send over UDP handles to child processes.
Don't call uv__stream_fd() on those handles, it's a macro that on OS X
evaluates to a function that operates on a uv_stream_t with a couple of
OS X specific fields. On other Unices it does (handle)->io_watcher.fd,
which works but only by accident.

Fixes joyent/node#4870.

show more ...

darwin: fix read from uninitialized struct kevent

In the OS X fallback code, don't read from the struct when the kevent()
syscall times out.

unix: make stream.c internal functions static

unix: use uv__set_artificial_error in uv_write2

* Use uv__set_artificial_error(), slightly more efficient than
uv__set_sys_error().

* Return UV_EINVAL instead of UV_EOPNOTSUPP

unix: use uv__set_artificial_error in uv_write2

* Use uv__set_artificial_error(), slightly more efficient than
uv__set_sys_error().

* Return UV_EINVAL instead of UV_EOPNOTSUPP.

* Fix up style.

show more ...

unix: reduce line count, return uv__set_sys_error

unix: fail early on bad send_handle in uv_write2

Return UV_EBADF if send_handle doesn't have a valid fd.

stream: use kevent() information before accept()

Limit number of syscalls by using backlog length information provided by
kevent().

unix: support sending uv_udp_t handles over ipc pipes

unix: make stream.c more DRY

unix: make assertion message more descriptive

unix: make uv__read_start_common() static

stream: fix infinite select() polling on osx

Interruption FD buffer should be emptied after write, otherwise select()
loop will poll indefinitely.

unix: remove stale closing check in uv__write()

Introduced in 0db3274f but no longer necessary; uv__write() no longer
runs when the handle has been closed. Write callbacks receive a spec

unix: remove stale closing check in uv__write()

Introduced in 0db3274f but no longer necessary; uv__write() no longer
runs when the handle has been closed. Write callbacks receive a special
status code that informs them that the handle has been closed.

show more ...

unix: remove uv_write_queue_head()

unix: remove bogus uv__write() comment

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: use select() for specific fds on OS X

kqueue(2) on osx doesn't work (emits EINVAL error) with specific fds
(i.e. /dev/tty, /dev/null, etc). When given such descriptors - start

unix: use select() for specific fds on OS X

kqueue(2) on osx doesn't work (emits EINVAL error) with specific fds
(i.e. /dev/tty, /dev/null, etc). When given such descriptors - start
select(2) watcher thread that will emit io events.

show more ...

unix: only set SO_REUSEADDR on tcp listen sockets

Avoid the extra syscall, it's a no-op for non-listening sockets.

At least, it should be - it remains to be investigated if a FreeBS

unix: only set SO_REUSEADDR on tcp listen sockets

Avoid the extra syscall, it's a no-op for non-listening sockets.

At least, it should be - it remains to be investigated if a FreeBSD kernel bug
affects ephemeral port allocation inside connect(). See [1] for details.

[1] http://www.freebsd.org/cgi/query-pr.cgi?pr=174087

show more ...

unix: rename UV__IO_* constants

unix: remove dependency on libev