| Date | Commit message (Collapse) |
|---|
|
There are still some places where on_destroy isn't suitable,
This gets rid of getpid() calls in most of those cases to
reduce syscall costs and cleanup syscall trace output.
|
|
The epoll implementation is the only one which respects the
limit (kevent would, but IO::KQueue does not). In any case,
I'm not a fan of the maxevents=1000 historical default since
it leads to fairness problems with shared non-blocking listeners
across multiple daemon workers.
|
|
This allows us to cut down on imports and reduce code.
This also makes it easier (in the next commit) to provide an option
to disable epoll/kqueue when saving an FD is valued over scalability.
|
|
The awaitpid API turns out to be quite handy for managing
long-lived worker processes. This allows us to ensure all our
uses of signalfd (and kevent emulation) are non-blocking.
|
|
Destruction order is unpredictable at process teardown,
so confessing or warning here is unnecessary, just break
out of the sub since it's for to delete an entry, anyways.
|
|
Another tiny step towards an eventual startup time improvements
by avoiding strict.pm
|
|
|
|
`do {} while(...)' loops in Perl don't support `next', actually :x
This only affects *BSD platforms with IO::KQueue installed.
Fixes: d6674af04cb74a4e "httpd|nntpd: avoid missed signal wakeups"
|
|
Since signalfd is often combined with our event loop, give it a
convenient API and reduce the code duplication required to use it.
EventLoop is replaced with ::event_loop to allow consistent
parameter passing and avoid needlessly passing the package name
on stack.
We also avoid exporting SFD_NONBLOCK since it's the only flag we
support. There's no sense in having the memory overhead of a
constant function when it's in cold code.
|
|
Using "make update-copyrights" after setting GNULIB_PATH in my
config.mak
|
|
Since Perl exposes O_NONBLOCK as a constant, we can safely make
SFD_NONBLOCK a constant, too. This is not the case for
SFD_CLOEXEC, since O_CLOEXEC is not exposed by Perl despite
being used internally in the interpreter.
|
|
Consistently returning the equivalent of pollfd.revents in a
portable manner was never worth the effort for us, as we use the
same ->event_step callback regardless of POLLIN/POLLOUT/POLLHUP.
Being a Perl, @events knows it size and we don't have to return
a maximum index for the caller to iterate on.
We can also avoid redundant integer coercion ("+0") since we
ensure everything is an IV in other places.
Finally, vec() is preferable to ("\0" x $size) for resizing
buffers because it only needs to write the extended portion
and not overwrite the entire buffer.
|
|
Thanks to the GCC compile farm project, we can wire up syscalls
for sparc64 and set system-specific SFD_* constants properly.
I've FINALLY figured out how to use POSIX::SigSet to generate
a usable buffer for the syscall perlfunc. This is required
for endian-neutral behavior and relevant to sparc64, at least.
There's no need for signalfd-related stuff to be constants,
either. signalfd initialization is never a hot path and a stub
subroutine for constants uses several KB of memory in the
interpreter.
We'll drop the needless SEEK_CUR import while we're importing
O_NONBLOCK, too.
|
|
Just like the EPOLL_CTL_ADD emulation path, the EPOLL_CTL_MOD
and EPOLL_CTL_DEL emulation paths can fail if attempting to
install an EVFILT_WRITE for a read-only pipe.
I've only observed this on the EPOLL_CTL_DEL emulation path, but
I suspect it could happen on the EPOLL_CTL_MOD path as well.
Increasing the amount of read-only pipes we rely on with altid
exports via sqlite3 made this old bug more apparent and
reproducible while looping the test suite.
This may be adjusted in the future to deal with write-only
pipes, but we currently don't have any of those watched by
kqueue.
|
|
NNTP TLS and COMPRESS support and cgit spawning from
the WWW interface were implemented last year.
Given the lack of syscall number stability guarantee on the
OpenBSD and FreeBSD, I don't think supporting a pure-Perl kevent
is feasible. Inline::C may still be an option since IO::KQueue
is abandoned, though, as it is for some Linux-only syscalls and
maybe some POSIX ones not covered by POSIX.pm.
|
|
I didn't wait until September to do it, this year!
|
|
Our attempt at using a self-pipe in signal handlers was
ineffective, since pure Perl code execution is deferred
and Perl doesn't use an internal self-pipe/eventfd. In
retrospect, I actually prefer the simplicity of Perl in
this regard...
We can use sigprocmask() from Perl, so we can introduce
signalfd(2) and EVFILT_SIGNAL support on Linux and *BSD-based
systems, respectively. These OS primitives allow us to avoid a
race where Perl checks for signals right before epoll_wait() or
kevent() puts the process to sleep.
The (few) systems nowadays without signalfd(2) or IO::KQueue
will now see wakeups every second to avoid missed signals.
|
|
Oops, IO::KQueue support was broken due to this missing
constant. Add a new ds-kqxs.t test case to ensure we
test the IO::KQueue path if IO::KQueue is available.
|
|
EV_DISPATCH is actually a better match for EPOLLONESHOT
semantics than EV_ONESHOT in that it doesn't require EV_ADD
for every mod operation.
Blindly using EV_ADD everywhere forces the FreeBSD kernel to
do extra allocations up front, so it's best avoided.
|
|
We don't need extra wakeups from the kernel when we know a
listener is already active.
|
|
On Linux systems with epoll support, we don't want to be
clobbering defined subs in the t/ds-poll.t test; so use
OO ->method dispatch instead and require users to explicitly
import subs via EXPORT_OK.
|
|
We don't need to code multiple event loops or have branches in
watch() if we can easily make the IO::KQueue-based interface
look like our lower-level epoll_* API.
|