Libuv loaded and unloaded the Core Services and Application Services for
every call to uv_set_process_title().
Change that to load them on the first call to uv_set_process_title() and
delay unloading until libuv is unloaded.
Speeds up process_title_threadsafe by about 10x on my system. It should
fail less often (hopefully not at all) on the CI now.
PR-URL: https://github.com/libuv/libuv/pull/2480
Reviewed-By: Saúl Ibarra Corretgé <s@saghul.net>
Add an API for obtaining cryptographically strong random data from the
system PRNG.
Co-authored-by: Saúl Ibarra Corretgé <s@saghul.net>
Refs: https://github.com/libuv/libuv/pull/1055
PR-URL: https://github.com/libuv/libuv/pull/2347
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Santiago Gimeno <santiago.gimeno@gmail.com>
Platforms that support O_CLOEXEC have supported it for a long time
and don't need feature detection. Libuv can just assume it works as
advertised.
The feature detection was broken for old Linux kernels because they
don't report EINVAL for unknown open() flags, they simply open the
file without setting the FD_CLOEXEC flag.
Libuv could fix that by checking afterwards with fcntl() if the flag
was actually set but it doesn't support kernels that old in the first
place so let's simplify the logic.
Fixes: https://github.com/libuv/libuv/issues/2450
PR-URL: https://github.com/libuv/libuv/pull/2454
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Santiago Gimeno <santiago.gimeno@gmail.com>
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
The preprocessor was used to select a special implementation when
building on AIX using XL C++ (strangely, not XL C). This code
implemented `cmpxchgi()` by directly reading the old value and then
calling `__compare_and_swap()`, an intrinsic that does not provide
any sort of memory barrier guarantees. The return value was not
used, and the value read prior to the `__compare_and_swap()` call
was returned. There is no way that this code could provide the
required semantics of the function and it causes observable data
races and strange library failures in production under load.
XL C/C++ for AIX has provided support for the GCC intrinsic used in
the GCC/Clang cases since version 12.1 of the compiler. This
version of the compiler is old enough that it doesn't warrant a
version check. (The compiler was released 8-Jun-2012, maintenance
ended 9-Jul-2019, and service ends 30-Apr-2020.)
This change fixes all observed atomic issues and unifies XL C/C++
with GCC/Clang.
Relevant XL C/C++ for AIX V12.1 documentation links:
- [__compare_and_swap](https://www.ibm.com/support/knowledgecenter/en/SSGH3R_12.1.0/com.ibm.xlcpp121.aix.doc/compiler_ref/bif_compare_and_swap_compare_and_swaplp.html)
- [__sync_val_compare_and_swap](https://www.ibm.com/support/knowledgecenter/en/SSGH3R_12.1.0/com.ibm.xlcpp121.aix.doc/compiler_ref/bif_gcc_atomic_val_comp_swap.html)
PR-URL: https://github.com/libuv/libuv/pull/2455
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
PR-URL: https://github.com/libuv/libuv/pull/2447
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Michael Dawson <michael_dawson@ca.ibm.com>
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
`_NSGetEnviron()` is not available on that platform.
PR-URL: https://github.com/libuv/libuv/pull/2449
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Signed-off-by: MaYuming <maym@appexnetworks.com>
Use the correct socket options: `MCAST_JOIN_SOURCE_GROUP` and
`MCAST_LEAVE_SOURCE_GROUP`.
Set mreq.gsr_interface = 0 if iface_addr = NULL.
PR-URL: https://github.com/libuv/libuv/pull/2202
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
All the supported platforms support specific source multicast.
PR-URL: https://github.com/libuv/libuv/pull/2202
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
It resets a TCP connection by sending a RST packet. Due to some platform
inconsistencies, mixing of `uv_shutdown` and `uv_tcp_close_reset` calls
is not allowed.
Fixes: https://github.com/libuv/libuv/issues/1991
PR-URL: https://github.com/libuv/libuv/pull/2425
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Make sure the `uv_cpu_info_t` array is fully freed in case of error.
Also, make it more robust by setting `count` to `0` and `cpu_info` to
`NULL` in case of error, so a subsequent call to `uv_free_cpu_info()`
does not segfault.
Refs: https://github.com/libuv/libuv/issues/2068
PR-URL: https://github.com/libuv/libuv/pull/2432
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
This commit fixes a warning about uv__fs_preadv() being
unused on BSD.
PR-URL: https://github.com/libuv/libuv/pull/2367
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
The UV_HANDLE_READING stream flag should be removed before
stopping the IO watcher and calling the callback. If the
callback calls uv_read_start(), the IO watcher will be started
then after the callback the UV_HANDLE_READING flag is removed.
This will result in epoll constantly reporting POLLIN events
while no one handles it, causing 100% CPU usage.
PR-URL: https://github.com/libuv/libuv/pull/2382
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Newer computers may have some cores offline because of the
Meltdown/Spectre patches from a while ago, making uv_cpu_info
potentially misleading and leading to people attempting to use more
threads than they actually have available. This makes it so only the
cores that are online and doing any work are counted.
PR-URL: https://github.com/libuv/libuv/pull/2326
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Santiago Gimeno <santiago.gimeno@gmail.com>
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
This reverts commit 00c6b1649d.
It's been reported (and I can confirm) that this change breaks
`process.title = 'foo'` in Node.js.
Since libuv just calls out to Core Services and Application Services,
and since those frameworks are really just black boxes that you can't
look inside, it's impossible to debug what exactly goes wrong. Revert
it is then.
Fixes: https://github.com/nodejs/node/issues/28945
PR-URL: https://github.com/libuv/libuv/pull/2405
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
uClibc does not have header "gnu/libc-version.h"
PR-URL: https://github.com/libuv/libuv/pull/2329
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
PR-URL: https://github.com/libuv/libuv/pull/2318
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
Reviewed-By: Сковорода Никита Андреевич <chalkerx@gmail.com>
Reviewed-By: Vladimír Čunát <v@cunat.cz>
The fallback for systems that lack preadv() only filled the first
buffer.
This commit rectifies that to fill all (or at least as many as possible)
buffers.
Fixes: https://github.com/libuv/libuv/issues/2332
PR-URL: https://github.com/libuv/libuv/pull/2338
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Jameson Nash <vtjnash@gmail.com>
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Make uv_cwd() do what the documentation says it did when the destination
buffer is too small: report UV_ENOBUFS and set the `size` in/out param
to the size of the path including the trailing nul byte.
Fixes: https://github.com/libuv/libuv/issues/2333
PR-URL: https://github.com/libuv/libuv/pull/2335
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
F_BARRIERFSYNC isn't defined when building on that platform.
PR-URL: https://github.com/libuv/libuv/pull/2334
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
So far, for some (?) errors, `uv_try_write()` returns `EAGAIN`
regardless of the actual error, so `ECONNRESET` and `EPIPE` errors
can be swallowed here.
This commit changes `uv_try_write()` so that it prefers to return
the actual error it has seen.
PR-URL: https://github.com/libuv/libuv/pull/2321
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Santiago Gimeno <santiago.gimeno@gmail.com>
uv_thread_create() inspects min(PTHREAD_STACK_MIN, RLIMIT_STACK) to
find out the minimum allowed stack size. After spelunking through
kernel and libc code, I came to the conclusion that allowing such
small stacks does not mix well with signals.
Musl's PTHREAD_STACK_MIN is 2 KB but signal handlers on many
architectures need at least 1 KB to store the signal context,
making it almost impossible to do anything on the thread without
signal delivery overflowing the stack.
Therefore, increase the lower bound to 8 KB. That corresponds to
PTHREAD_STACK_MIN + MINSIGSTKSZ on arm64, which has the largest
MINSIGSTKSZ of the architectures that musl supports.
PR-URL: https://github.com/libuv/libuv/pull/2310
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
This commit add support for Haiku, an open-source operating system
inspired by BeOS.
PR-URL: https://github.com/libuv/libuv/pull/2301
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
The link symbol for close$NOCANCEL is slightly different for 32 bit
builds.
PR-URL: https://github.com/libuv/libuv/pull/2309
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Fall back to F_BARRIERFSYNC if F_FULLFSYNC is not supported by the file
system, only fall back to fsync() if both fcntls fail.
F_BARRIERFSYNC should be at least as safe as fsync() because it's fsync
coupled with a barrier.
PR-URL: https://github.com/libuv/libuv/pull/2317
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Use `"%" PRIu64` and uint64_t for sscanf-ing large values, not `"%lu"`
and unsigned long. The latter can overflow when libuv is a 32 bits
application running on a 64 bits machine.
Fixes: https://github.com/libuv/libuv/issues/2297
PR-URL: https://github.com/libuv/libuv/pull/2305
Reviewed-By: Santiago Gimeno <santiago.gimeno@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Same as the previous commit but this time for Linux.
Specifically, the glibc close() wrapper is a cancellation point;
the thread is unwound when close() is called when the thread is
in the "cancel" state.
That's according to spec and therefore arguably less severe than
MacOS's "EINTR without actually closing" behavior but we can avoid
the file descriptor leak altogether by sidestepping glibc and
making the system call directly.
Musl libc is unaffected, its close() wrapper doesn't check
the cancel state.
PR-URL: https://github.com/libuv/libuv/pull/2291
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
The close() system call tests for thread cancellation first. That has
the unfortunate side effect of aborting the sytem call with EINTR
without actually closing the file descriptor when the thread is in
the "cancel" state. Work around that by calling close$NOCANCEL().
This might well qualify as an academic bug because approximately no one
uses thread cancellation but let's aim for correctness anyway.
PR-URL: https://github.com/libuv/libuv/pull/2291
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Newer macOS systems (10.7+) use fsevents() for the fs_event backend,
but older macOS systems don't have full functionality for this in
the macOS fsevents() API. Fall back to kqueue for these systems.
PR-URL: https://github.com/libuv/libuv/pull/2290
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Reviewed-By: Santiago Gimeno <santiago.gimeno@gmail.com>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
When building libuv with clang scan-build, an "Assigned value is
garbage or undefined" warning is emitted unless the return value
of open() is verified to be >= 0. This commit fixes the warning.
PR-URL: https://github.com/libuv/libuv/pull/2140
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
