arm64 doesn't have a epoll_wait() system call but a logic error stopped
libuv from falling back to epoll_pwait().
This bug was introduced in commit 67bb2b5 ("linux: fix epoll_pwait()
regression with < 2.6.19") which sadly exchanged one regression for
another.
This is a backport of 1d8332f (v1.x)
PR-URL: https://github.com/libuv/libuv/pull/308
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Linux before kernel 2.6.19 does not support epoll_pwait(). Due to a
logic error in commit 2daf944 ("unix: add flag for blocking SIGPROF
during poll"), the fallback path for ENOSYS was not taken.
This commit also adds epoll_pwait() emulation using pthread_sigmask().
The block/unblock operations are not atomic but that is fine for our
particular use case, to wit, sleep through SIGPROF signals.
This is a back-port of commit 67bb2b5 from the v1.x branch.
Original-PR-URL: https://github.com/libuv/libuv/pull/162
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
PR-URL: https://github.com/libuv/libuv/pull/165
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Revisit the fix from commit b705b53. The problem with using sigset_t
and _NSIG is that the size of sigset_t and the value of _NSIG depend
on what headers libuv picks up first, <signal.h> or <asm/signal.h>.
With the former, sizeof(sigset_t) = 128; with the latter, it's 8.
Simply sidestep the issue by calculating the signal mask as a 64 bits
integer, without using sigset_t or _NSIG.
This is a partial cherry-pick of commit 751ac48 from the v1.x branch.
Original PR-URL: https://github.com/libuv/libuv/pull/83
PR-URL: https://github.com/libuv/libuv/pull/84
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
Add a per-event loop flag for blocking SIGPROF signals when polling for
events.
The motivation for this addition is to reduce the number of wakeups and
subsequent clock_gettime() system calls when using a sampling profiler.
On Linux, this switches from epoll_wait() to epoll_pwait() when enabled.
Other platforms bracket the poll syscall with pthread_sigmask() calls.
Refs strongloop/strong-agent#3 and strongloop-internal/scrum-cs#37.
PR-URL: https://github.com/libuv/libuv/pull/15
Reviewed-By: Saúl Ibarra Corretgé <saghul@gmail.com>
It seems that epoll_wait is implemented in glibc in terms of epoll_pwait and
new architectures (like arm64) do not implement the epoll_wait syscall at all.
So if epoll_wait errors with ENOSYS, just call epoll_pwait.
This is a backport of 861de3d71d for v0.10
branch.
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
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>
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
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.
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.
RSS is a reflection of the number of pages that a process has mapped.
glibc implements fopen() in terms of mmap() which means that trying
to read the number of mapped pages changes it. Switch to open().
Make changes to the process title visible to tools like `ps`.
The argv clobber technique is reasonably portable across Unices;
the common code has been moved into src/unix/proctitle.c and is used
on Linux and OS X. Other platforms will probably follow in the future.
Don't add the io watcher to the watcher queue if the requested change
is effectively a no-op, that is, when the event mask doesn't change.
The exception here is sunos because the event ports backend requires
that watched file descriptors are re-added on every turn of the event
loop.
This commit is a micro-optimization, it does not change the event
loop's observable behavior in any way.
Don't check the return value of epoll_ctl(EPOLL_CTL_DEL). When the
file descriptor is closed, we're potentially racing with another
thread and that means the errno is not a reliable indicator of
the actual error.
The other event mechanisms (kqueue, event ports) are not affected
because:
* kqueue returns either EBADF or ENOENT. Both are handled by libuv.
* event ports rearms all file descriptors on each call to port_getn().
Fixesjoyent/node#4558.
