This makes libuv more tolerant to the properties of the pipes
that it can use without any issue. This is necessary because Cygwin
(and hence Mintty) opens STDIN without FILE_WRITE_ATTRIBUTES.
This functionality is present in stream and uv_udp_t has a queue
as well so it makes sense for udp to have a send_write_size.
Since udp sends entire messages atomically, the send_queue_count field
lets the user determine how many messages are there left to send.
Debug build failed on mingw32 because CRT assertion disable code was stubbed out.
Replace __declspec(thread) with UV_THREAD_LOCAL which is defined as __thread on GCC.
According to @aktau, the call to `mach_timebase_info` costs 90% of the
total execution time of `uv_hrtime()`. The result of the call is static
on all existing platforms, so there is no need in invoking it multiple
times.
Signed-off-by: Fedor Indutny <fedor@indutny.com>
We add a function called uv_loop_size which returns the size of
the loop. It is an additional function which complements
`uv_handle_size` and` uv_req_size` in order to provide full FFI
support.
Signed-off-by: Fedor Indutny <fedor@indutny.com>
Switches from using ReadConsoleA to ReadConsoleW.
Uses an auxiliary buffer to store the temporary utf-16 string.
Signed-off-by: Peter Atashian <retep998@gmail.com>
Signed-off-by: Saúl Ibarra Corretgé <saghul@gmail.com>
Make uv_barrier_wait() return a boolean value indicating whether this
thread was the "serializer".
From `man pthread_barrier_wait`:
Upon successful completion, the pthread_barrier_wait() function
shall return PTHREAD_BARRIER_SERIAL_THREAD for a single (arbitrary)
thread synchronized at the barrier and zero for each of the other
threads.
Exposing that information from libuv is useful because it can make
cleanup significantly easier:
if (uv_barrier_wait(&barrier) > 0)
uv_barrier_destroy(&barrier);
Signed-off-by: Fedor Indutny <fedor@indutny.com>
Remove should shuffle items in both directions, not just down. It is
required, because `max` node could be not the actual maximum value in
the tree.
fix#1267
Signed-off-by: Fedor Indutny <fedor@indutny.com>
Only these functions will trigger an implicit binding of a UDP handle:
- uv_udp_send
- uv_udp_recv_start
- uv_udp_set_membership
All other functions will return UV_EBADF in case the socket was not
bound.
Note: currently the socket is created and bound at the same time. This
may change in the future.
If the handle was opened using `uv_udp_open` ift's possible that the
kernel doesn't fill in the msg_name field, so return NULL as the address
in that case.
fixes#1252
The PATH-parsing code for windows erroneously contained an infinite
loop when the PATH started with a leading semicolon. Each iteration of
the loop usually bumped over the separator, but if the first character
was a semicolon then it would never skip it, causing the infinite
loop.
Closes#909
Fix https://github.com/joyent/node/issues/7345
Google Chrome is launching native messaging hosts by invoking cmd.exe
with input/output redirected from/to named pipes. The host ends up
with a read-only handle to the stdin pipe. This is causing
SetNamedPipeHandleState to fail.
Reopen one of the pipe descriptors as read/write through the procfs.
Allows us to close the original pipe file descriptors, saving a file
descriptor on kernels that don't support eventfd(2).
The VirtualBox shared file system doesn't seem to be capable of
`readv()` syscall. Since we don't need them in the most of the cases
anyway, it won't hurt to just use `read()` and `write()` when we are
reading just one buffer.
see https://github.com/joyent/node/issues/7343 for details.
Add UV_UDP_REUSEADDR flag instead, which can be passed to uv_udp_bind.
If the udp handle is unbound when uv_udp_set_memberhsip or
uv_udp_set_multicast_interface is called, the handle will be bound with
UV_UDP_REUSEADDR set.
In case reopening the file descriptor fails, fallback to the old
behavior where we do blocking writes, to avoid disrupting other
processes potentially using the file descriptor.
Invalidate file descriptor when closing `uv_fs_event_t` handle. Note
that `uv__io_stop` is just removing `fd` from `loop->watchers` and not
actually invalidating all consequent events in a `kevent()` results.
fixjoyent/node#1101
Reopen the file descriptor when it refers to a tty. This lets us put the
tty in non-blocking mode without affecting other processes that share it
with us.
This brings back commit 31f9fbc, which was reverted in 20bb1bf. The OSX
select trick is working now.
Original patch by @bnoordhuis
If the same file description is open in two different processes, then
closing the file descriptor is not sufficient to deregister it from the
epoll instance (as described in epoll(7)), resulting in spurious events
that cause the event loop to spin repeatedly. So always explicitly
deregister it.
Fixes#1099.
Conflicts:
test/test-spawn.c
Introduce `int uv_pipe_pending_count(uv_pipe_t*)` and
`uv_handle_type uv_pipe_pending_type(uv_pipe_t*)`. They should be
used in IPC pipe's read cb to accept incoming handles:
int count = uv_pipe_pending_count(pipe);
int i;
for (i = 0; i < count; i++) {
uv_handle_type type = uv_pipe_pending_type(pipe);
/* ... */
uv_accept(...);
}
When the supplied buffer is not big enough and UV_ENOBUFS is
returned, hint the user about the required size by setting
the len paramemeter to the required value.
Applies to:
- uv_pipe_getsockname
- uv_fs_event_getpath
- uv_fs_poll_getpath
This improves API consistency with uv_read and uv_write and may
improve efficiency for some uses. Vectored IO is emulated when the
requisite system calls are unavailable.
These functions supersede uv_loop_new and uv_loop_delete.
uv_loop_init initialized a user allocated loop and uv_loop_close
removes all associated resources a loop uses after it has finished
execution.
uv_loop_new and uv_loop_delete are now deprecated.
If the same file description is open in two different processes, then
closing the file descriptor is not sufficient to deregister it from the
epoll instance (as described in epoll(7)), resulting in spurious events
that cause the event loop to spin repeatedly. So always explicitly
deregister it.
Fixes#1099.
Replace the red-black tree with a heap. The most common operation that
libuv performs on timers is looking up the first timer to expire. With
a red-black tree, that operation is O(log n). With a heap, it's O(1).
The event_handle field of unused accept requests was not being
initialized properly. As a result, uv_tcp_endgame would try to close
an invalid handle and this could lead to unexpected behavior.
This fixes node.js test test-cluster-disconnect.js on Windows.
This improves API consistency with uv_read and uv_write and may
improve efficiency for some uses. Vectored IO is emulated when the
requisite system calls are unavailable.
The unix implementation of uv_spawn always starts out with a
uv__handle_init, but the windows implementation sometimes bails out
early before calling uv__handle_init. This means that uv_close on a
failed uv_spawn will always succeed on unix but sometimes fail on
windows.
This commit lifts the initialization of the uv_process_t above all of
the error checking to ensure that uv_close will always work when
uv_spawn returns an error.
The crtdbg.h header was added in c0716b3d, but MinGW does not have this
header present on the system. This commit takes the same approach of
2684f876a and just ignores this header and functionality on MinGW
Every file descriptor opened using libuv should be automatically marked
as CLOEXEC to prevent it from leaking to a child process. Note that
since we are opening fds in a thread pool, there is a possible race
condition between `uv_spawn()` and the `open()` + `uv__cloexec()`. The
rwlock was added to avoid it.
see https://github.com/joyent/node/issues/6905
Changed the order of the member assignments since the thread
may start before the parameters are assigned. This especially
happens when the osx scheduler is very busy.
Changes since version 0.11.17:
* osx: Fix a possible segfault in uv__io_poll (Alex Crichton)
* windows: improved handling of invalid FDs (Alexis Campailla)
* doc: adding ARCHS flag to OS X build command (Nathan Sweet)
* tcp: reveal bind-time errors before listen (Alexis Campailla)
* tcp: uv_tcp_dualstack() (Fedor Indutny)
* linux: relax assumption on /proc/stat parsing (Luca Bruno)
* openbsd: fix obvious bug in uv_cpu_info (Fedor Indutny)
* process: close stdio after dup2'ing it (Fedor Indutny)
* linux: move sscanf() out of the assert() (Trevor Norris)
When building using gyp and BUILDTYPE=Release using clang 3.4 received
this warning:
../src/unix/linux-core.c:640:34: warning: variable 'n' is uninitialized
for (len = sizeof("cpu0"); n /= 10; len++);
^
Initializing n = 0 silences this build warning.
Changes since version 0.10.22:
* linux: relax assumption on /proc/stat parsing (Luca Bruno)
* openbsd: fix obvious bug in uv_cpu_info (Fedor Indutny)
* process: close stdio after dup2'ing it (Fedor Indutny)
Thus allow passing the same file descriptor as the source of multiple
stdios.
Committed with the help and code parts from:
* Sam Roberts <sam@strongloop.com>
fix#1074
CPU entries in /proc/stat are not guaranteed to be monotonically
increasing, asserting on this assumption can break in cases such
as the UltraSparc II machine shown in #1080.
Signed-off-by: Luca Bruno <lucab@debian.org>
Changed uv_tcp_duplicate_socket to reveal any bind-time errors
before calling listen().
This fix is 100% windows specific.
It helps fix Node.js unite test test-cluster-bind-twice on Windows.
If passed and invalid FD, _get_osfhandle() sets an error code
through errno, not _doserrno. Hence we need to use
SET_REQ_WIN32_ERROR insted of SET_REQ_RESULT.
In debug builds, _get_osfhandle() also raises a superfluous
assert. I implemented a wrapper that disables all asserts
around the call to _get_osfhandle().
This fixes node.js unit tests test-fs-read-stream.js and
test-listen-fd-ebadf.js.
Changes since version 0.11.16:
* stream: allow multiple buffers for uv_try_write (Fedor Indutny)
* unix: fix a possible memory leak in uv_fs_readdir (Alex Crichton)
* unix, windows: add uv_loop_alive() function (Sam Roberts)
* windows: avoid assertion failure when pipe server is closed (Bert
Belder)
* osx: Fix a possible segfault in uv__io_poll (Alex Crichton)
* stream: fix uv__stream_osx_select (Fedor Indutny)
uv__stream_osx_interrupt_select was only defined on OS X, but is used
elsewhere on all platforms. The intention was to implement it as a
no-op on other platforms, but the entire definition was inside
"#ifdef __APPLE__", so this didn't actually work.
Fix: move it above the #ifdef.
Relying on `readable`/`writable` when polling will lead to the select
thread spinning in the loop and calling `uv_async_send()`, because
stream may never become not `readable`/`writable`.
We should rely on `uv__io_active()` instead and interruprt select thread
every time it changes.
If multiple handles arrive to the IPC pipe at the same time (happens on
some platforms), libuv will queue them internally, and call `read2_cb`
multiple times with a null-buffer and proper `handle_type`.
In our build infrastructure, I've seen a lot of segfaults recently that
were all only happening on OSX. Upon inspecting the coredumps, it
appearded that all segfaults happened at the same instruction, and upon
translating the assembly back to the source, I found that an array could
be indexed with a -1 index before the index was checked to be not -1.
As concrete evidence, here is the situation that I found caused the
segfault. The instruction in question along with the relevant register
values was:
mov (%r8,%r15,8),%r12
r8 = 0x7fb0ba800000
r15 = 0xffffffffffffffff
r8 + r15 * 8 == 0x7fb0ba7ffff8
It appears that the base of loop->watchers was page aligned, and by
going back one word I guess that the page wasn't mapped, causing our
segfaults.
In our build infrastructure, I've seen a lot of segfaults recently that
were all only happening on OSX. Upon inspecting the coredumps, it
appearded that all segfaults happened at the same instruction, and upon
translating the assembly back to the source, I found that an array could
be indexed with a -1 index before the index was checked to be not -1.
As concrete evidence, here is the situation that I found caused the
segfault. The instruction in question along with the relevant register
values was:
mov (%r8,%r15,8),%r12
r8 = 0x7fb0ba800000
r15 = 0xffffffffffffffff
r8 + r15 * 8 == 0x7fb0ba7ffff8
It appears that the base of loop->watchers was page aligned, and by
going back one word I guess that the page wasn't mapped, causing our
segfaults.
When a pipe server is closed, all pending accept requests and their
associated HANDLEs are closed to force windows to cancel the
ConnectNamedPipe IRP. The returned request has the `pipeHandle` field
set to INVALID_HANDLE_VALUE, which trips an assert in
uv_pipe_process_accept_req. This patch fixes that.
Useful to know when the the event loop is empty, this can't be done with
uv_run() without possibly blocking, or running some events (which might
empty the event loop as a side-effect).
Some scandir implementations allocate the dirent struct even if the
directory is empty, so if `scandir` returns 0 there may still be memory
that needs to get deallocated. I have altered uv__fs_readdir to go to
the "deallocation exit area" when 0 files are found in the directory
and continue to return early on a return value of -1.
I went to add a test for this functionality, but it appears that one
already exists (reading an empty directory), so I imagine that the
valgrind builds must only be happening on linux instead of OSX as well?
I have confirmed manually that a program without this fix will
infinitely leak memory, and with this fix the memory usage stays
constant.
`uv_try_write(stream, buf, size)` acts like `uv_write()`,
but without queueing actual write until UV_POLLOUT (or IOCP completion).
This is useful for doing writes using on-stack `uv_write_t` requests.
fix#1025
We weren't always translating system errors for in uv_spawn.
This is half a fix for node's unit test
test-child-process-cwd.js on win.
The rest will go into node lib.
It also fixes test-child-process-exec-error
Large performance counter frequency values would cause overflows, even
when 64-bit integers were used to do the multiplication with NANOSEC.
Fix this by using floating point math instead.
Fixes#850
First of all, a bit of explanation of what happens there:
1. FSEvents emits absolute paths to changed files or directories
2. We cut off the first part of such paths, which is equal to handle's
real path ('/dir/subdir/subsubdir`, without trailing slash)
3. Then, if we are running in non-recursive mode, we discard paths that
have slashes ('/') as a non-first character in them
Commit 3d2c820 back-ports a patch from the master branch that disables
the use of SO_REUSEPORT on Linux for reasons mentioned in the commit
log.
Unfortunately, the back-port was incomplete; another setsockopt() call
site in src/unix/udp.c was overlooked. This commit rectifies that.
Hat tip to Luca Bruno for helping troubleshoot the issue.
Fix a bug that was introduced in commit 3ee4d3f ("unix, windows:
return error codes directly") and add a regression test for good
measure.
Hat tip to Fedor for pointing out the issue.
Fixes#1007.
Print the error message rather than just the errno. The fact that the
errno is 24 is only informative to people that have their operating
system's error codes memorized.
Set the close-on-exec flag on file descriptors that we've received with
recvmsg() so we don't leak them when calling fork() afterwards.
On Linux, we use the MSG_CMSG_CLOEXEC flag when supported (2.6.23 and
up.)
On older Linux versions and other platforms, we walk the received file
descriptors and set the close-on-exec flag for each fd manually. That
won't entirely avoid race conditions when other threads call fork() or
clone() but at least we're less likely to leak file descriptors now.
When fd is closed and new one (with the same number) is opened inside
kqueue/epoll/port loop's callback - stale events might invoke callbacks
on wrong watchers.
Check if watcher was changed after invocation and invalidate all events
with the same fd.
fix#826
Otherwise `FSEventStreamCreate()` will coalesce events, even if they're
happening in the interval, bigger than supplied `latency`. In other
words, if this flag is not set events will happen in separate callback
only if there was a delay bigger than `latency` between two consecutive
events.
Drop the _CRT_NON_CONFORMING_SWPRINTFS hack and just use _snwprintf().
It's a long and complicated story but the gist of it is that the MS CRT
had a swprintf() function before ISO C did, with a different function
prototype to boot: the ISO C one takes a |size| argument, the MS one
does not.
The function prototype that's exported by mingw and mingw-w64 depends
on the mingw version and the _CRT_NON_CONFORMING_SWPRINTFS define.
If they don't match up, you get the wrong prototype and things will
crash at run-time.
Reduce the phase space by sidestepping the whole issue: drop swprintf()
altogether and use _snwprintf() from now on.
Fixes#990.
Work around an epoll quirk where it sometimes reports just the EPOLLERR
or EPOLLHUP event. In order to force the event loop to move forward,
we merge in the read/write events that the watcher is interested in;
uv__read() and uv__write() will then deal with the error or hangup in
the usual fashion.
Fixes#982.
This is a back-port of commit 24bfef2 from the master branch.
Work around an epoll quirk where it sometimes reports just the EPOLLERR
or EPOLLHUP event. In order to force the event loop to move forward,
we merge in the read/write events that the watcher is interested in;
uv__read() and uv__write() will then deal with the error or hangup in
the usual fashion.
Fixes#982.
Create file paths with CFStringCreateWithFileSystemRepresentation(),
not CFStringCreateWithCString().
Reapplies 3780e12 ("fsevents: support japaneese characters in path")
from the v0.10 branch. Was dropped in the last v0.10 -> master merge
for failing to apply.
Drops commit 3780e12 ("fsevents: support japaneese characters in path")
for being quite inapplicable to the master branch. Will be reworked
and applied in a follow-up commit.
Conflicts:
README.md
build.mk
src/unix/fsevents.c
src/unix/udp.c
If spawning a process fails due to an exec() failure (but it succeeded
in forking), then this should be considered a spawn failure instead of
an asynchronous termination of the process. This allows to check for
common exec() failure conditions such as a bad path quickly instead of
having to rely on keeping track of the async callback.
Additionally, the meaning of the two fields returned in the callback are
now exactly what they advertise to be. The process exit argument is not
one of two values depending on what happened to the child.
Fixes#978.
Rewrite some of the macros in a way that:
a) makes them more likely to trigger compile-time errors if used
inappropriately, and
b) makes C++ compilers happy
On the BSDs, SO_REUSEPORT is pretty much SO_REUSEADDR with some special
casing for IP multicast. When two processes (that don't do multicast)
bind to the same address, only the last one receives traffic. It allows
one to "steal" the bound address from another process. (Both processes
have to enable SO_REUSEPORT though, so it only works in a cooperative
setting.)
On Linux however, it enables port sharing, not stealing - both processes
receive a share of the traffic. This is a desirable trait but pre-3.9
kernels don't support the socket option and a libuv program therefore
behaves differently with older kernels or on another platform.
This is a back-port of commit 9d60f1e from the master branch.
Fixesjoyent/node#6454.
Watchers could be stopped between two `kevent()`/`epoll_wait()` calls
(which may happen in the same loop in `uv__io_poll()`), in such cases
`watcher->events` could be stale and won't be updated to
`watcher->pevents`.
Try to use and rely on `watcher->pevents` instead of blindly expecting
`watcher->events` to be always correct.
On some systems, clock_gettime(CLOCK_MONOTONIC) is only serviced from
the vDSO when the __vdso_clock_gettime() wrapper is confident enough
that the vDSO timestamp is highly accurate. When in doubt, it falls
back to making a traditional SYS_clock_gettime system call with all
the overhead that entails.
While a commendable approach, it's overkill for our purposes because we
don't usually need high precision time. That's why this commit switches
to CLOCK_MONOTONIC_COARSE for low-precision timekeeping, provided said
clock has at least a one millisecond resolution.
This change should eliminate the system call on almost all systems,
including virtualized ones, provided the kernel is >= 2.6.32 and glibc
is new enough to find and parse the vDSO.
Cheking if the loop is alive is covered in the while loop afterwards.
Also, the stop flag will be cleared if necessary, which didn't happen
before this patch.
Mea culpa, the previous commit added another ERROR_FILENAME_EXCED_RANGE
case to the switch statement in uv_translate_sys_error(). This commit
fixes up the build error.
Make it possible to call uv_tty_reset_mode() from inside a signal
handler. The primary motivation is to make it possible to restore
the TTY from inside a SIGINT or SIGTERM signal handler.
Fixes#954.
Before this commit, multiple event loops raced with each other when a
SIGCHLD signal was received. More concretely, it was possible for
event loop A to consume waitpid() events that should have been
delivered to event loop B.
This commit addresses that by doing a linear scan over the list of
child processes. An O(n) scan is not terribly efficient but the
actual performance impact is not measurable in a benchmark that spawns
rounds of several thousands instances of /bin/false. For the time
being, this patch will suffice; we can always revisit it later.
Fixes#887.
Before this commit, uv_uptime() returned the nanoseconds as the
fractional part of the uptime. It's a bit inconsistent with the
output on other platforms because those only return whole seconds
(with the possible exception of Windows.)
This commit changes linux-core.c to only return whole seconds.
There're could be a situation, where one fsevents handle gets created
and another one is destroyed simultaneously. In such cases
`fsevent_need_reschedule` will be set to 1 twice and reset only once,
leaving handle destructor hanging in uv_sem_wait().
This commit reverts the following commits:
983fa68 darwin: fix 10.6 build error in fsevents.c
684e212 fsevents: use shared FSEventStream
ea4cb77 fsevents: FSEvents is most likely not thread-safe
9bae606 darwin: create fsevents thread on demand
Several people have reported stability issues on OS X 10.8 and bus
errors on the 10.9 developer preview.
See also joyent/node#6296 and joyent/node#6251.
The cleanup-after-error code path in uv_spawn() was closing file
descriptors indiscriminately. Only close file descriptors that we
created ourselves, not the ones that are passed in by the user.
Fixesjoyent/node#6297.
