Date | Commit message (Collapse) |
|
This is cleaner in most cases and may allow Perl to reuse memory
from unused fields.
We can do this now that we no longer support Perl 5.8; since
Danga::Socket was written with struct-like pseudo-hash support
in mind, and Perl 5.9+ dropped support for pseudo-hashes over
a decade ago.
|
|
We don't need to keep track of that field since we always
know what events we're interested in when using one-shot
wakeups.
|
|
It may be reinstated at a later time if there's interest; but I
want to be able to use one-shot notifications for certain events
while retaining level-triggered notifications others.
OTOH, I intend to fully support kqueue; via IO::KQueue for now,
but via syscall() eventually to take advantage of the syscall
reduction kevent(2) can provide over (current) epoll APIs.
|
|
There was much duplicate logic between watch_read and
watch_write. Share that logic, and give us room to enable
edge-triggered or one-shot notifications in the future.
|
|
Make the rest of our IO::KQueue-using code less verbose and
closer to the C equivalent.
|
|
We can avoid the EPOLL_CTL_ADD && EPOLL_CTL_MOD sequence with
a single EPOLL_CTL_ADD.
|
|
EPOLL* constants already match their POLL* counterparts and
there's no way Linux can ever diverge or change the values
of those constants. So we'll favor the EPOLL* ones since we
use EPOLLEXCLUSIVE, already.
For weird stuff like kqueue, we'd need to keep maintaining
the mapping, anyways.
|
|
Data which can't fit into a generously-sized socket buffer,
has no business being stored in heap.
|
|
No sense in having similar Linux-specific functionality in
both our NNTP.pm and HTTP.pm
|
|
We call ->flush_write directly, now; so we can eliminate a
needless check.
|
|
There's no point in passing a mask of interesting events
when removing an item from the epoll watch set.
|
|
We don't need write buffering unless we encounter slow clients
requesting large responses. So don't waste a hash slot or
(empty) arrayref for it.
|
|
Get rid of the confusing $need_queue variable and all
the associated documentation for it. Instead, make it
obvious that we're either skipping the write buffer or
flushing the write buffer by splitting the sub in two.
|
|
Since Perl 5.10+, "fields" makes a restricted hash; not a
compile-time-defined array (struct) with fixed offsets as
it did in Perl <= 5.8.
Thus in-use fields cost memory, and since the write buffer
offset is rarely needed; stop relying on it.
|
|
Wrong place to be wrapping this method.
|
|
We rely on immediate timers often, so we can avoid the overhead
of an extra subroutine call to retrieve the monotonic time (and
a sometimes-system call on some platforms).
|
|
All of our internal timing code should use monotonic clocks
for consistency against system clock adjustments.
This can be shared by our Daemon and NNTP packages.
|
|
Over a decade of using Danga::Socket and I never found the
built-in debug functionality useful.
|
|
Merely checking the presence of the {sock} field is
enough, and having multiple sources of truth increases
confusion and the likelyhood of bugs.
|
|
Having separate read/write callbacks in every class is too
confusing to my easily-confused mind. Instead, give every class
an "event_step" callback which is easier to wrap my head around.
This will make future code to support IO::Socket::SSL-wrapped
sockets easier-to-digest, since SSL_write() can require waiting
on POLLIN events, and SSL_read() can require waiting on POLLOUT
events.
|
|
If we got something to write, then write it. Otherwise, try
reading; and continue dealing with errors which normally occur
along the way.
Trying to read requests while we need to buffer in luserspace
is suicidal from a memory management standpoint.
The only adjustment needed for existing callers is EvCleanup;
where we need to ensure we're always calling the dummy
EvCleanup::event_write callback to accomplish nothing.
|
|
In my experience, both are worthless as any normal read/write
call path will be wanting to check errors and deal with them
appropriately; so we can just call event_read, for now.
Eventually, there'll probably be only one callback for dealing
with all in/out/err/hup events to simplify logic, especially w.r.t
TLS socket negotiation.
|
|
Since we stop using it in NNTP, we don't need it at all.
|
|
We won't be needing it, not even for TLS support.
|
|
Storing the file descriptor was redundant as we can quickly call
fileno($self->{sock}) and not have to store an extra hash table
entry. Multiple sources of truth leads to confusion, confusion
leads to bugs.
|
|
ECONNRESET and EPIPE are common on a big Internet filled with
unreliable connections, and there's nothing our code can do
about it.
So no point in wasting code to log them and there are plenty of
tracing tools to choose from if such diagnostics are needed.
|
|
Stop importing unused constants, and favor integer comparisons
of `$!' over `$!{EFOO}' hash lookups. Integer comparisons are
slightly faster, even:
Benchmark: timing 30 iterations of cmp_eq, cmp_ne, hash_hit, hash_miss...
cmp_eq: 1 wallclock secs ( 1.61 usr + 0.00 sys = 1.61 CPU) @ 18.63/s (n=30)
cmp_ne: 2 wallclock secs ( 1.57 usr + 0.00 sys = 1.57 CPU) @ 19.11/s (n=30)
hash_hit: 4 wallclock secs ( 3.85 usr + 0.00 sys = 3.85 CPU) @ 7.79/s (n=30)
hash_miss: 4 wallclock secs ( 3.74 usr + 0.00 sys = 3.74 CPU) @ 8.02/s (n=30)
#!/usr/bin/perl -w
use Benchmark qw(:all);
use Errno qw(EAGAIN EINTR);
my ($r, $w);
pipe($r, $w) or die 'pipe';
require IO::Handle;
$r->blocking(0);
my $buf;
my $n = 30000;
timethese(30, {
hash_hit => sub {
sysread($r, $buf, 1);
for (0..$n) {
next if $!{EAGAIN};
die 'FAIL';
}
}
,
'cmp_eq' => sub {
sysread($r, $buf, 1);
for (0..$n) {
next if $! == EAGAIN;
die 'FAIL';
}
},
hash_miss => sub {
sysread($r, $buf, 1);
for (0..$n) {
die 'FAIL' if $!{EINTR};
}
},
'cmp_ne' => sub {
sysread($r, $buf, 1);
for (0..$n) {
die 'FAIL' if $! == EINTR;
}
},
});
|
|
Keeping track of write_buf_size was redundant and pointless when
we can simply check the number of elements in the buffer array.
Multiple sources of truth leads to confusion; confusion leads to
bugs.
Finally, rename the prefixes to 'wbuf' to ensure we loudly
(instead of silently) break any external dependencies being
ported over from Danga::Socket, as further changes are pending.
|
|
We don't need and won't be needing per-socket PostLoopCallbacks.
|
|
We never enable write watches ourselves for HTTP and NNTP,
and only enable the write watch with EvCleanup because it's
an "always on" watch.
|
|
This was never used in Danga::Socket 1.61, either.
|
|
I've used Danga::Socket for well over a decade in various
projects at this point and have never seen the need for it.
If such a bug ever happens; the process should fall over so
it gets fixed ASAP.
|
|
This is not used by perlbal for OpenSSL support, either;
and it does not appear to be the right layer for doing
write translations anyways (IO::Socket::SSL uses `tie').
|
|
Sometimes I get bored with the email part of this project and
need a distraction :P
|
|
ToClose and HaveEpoll are of no use to us and I see no
future use for them, either.
|
|
Even though we currently don't use it repeatedly, ->Reset
should close() kqueue FDs and not cause the process to run
out of descriptors.
Add a close-on-exec test while we're at it.
|
|
We should not be leaking these FDs to git(1) processes,
in case git has a bug that causes it to access the wrong FD.
|
|
Enabling deprecation warnings didn't seem to have any
noticeable effects with "perl -w -c", so whatever reason
Danga had for it is long irrelevant.
|
|
Unlike Danga::Socket, we do not support TCP_CORK, either
|
|
It was only relevant to Danga::Socket.
|
|
It's easy enough to wrap FDs in classes that can use
all of the functionality of the event loop, not just
the read-only interface AddOtherFds provided.
|
|
"make syntax" is clean, now
|
|
Any operations on an fd after POSIX::close() are invalid, so
epoll_ctl will fail. Worse off, in a multi-threaded Perl, the
fd may be reused by another thread and EPOLL_CTL_DEL can hit the
wrong file description as a result.
cf. https://rt.cpan.org/Ticket/Display.html?id=129487
|
|
No longer used since we removed the *_ip_string fields
|
|
IO::Poll::_poll returns -1, which is "true" to Perl.
cf. https://rt.cpan.org/Ticket/Display.html?id=129484
|
|
IO::Kqueue seems unmaintained, so workaround a long-standing
bug where it falls over on signals:
https://rt.cpan.org/Ticket/Display.html?id=116615
|
|
There's other ways to profile and we don't need to add runtime
branches to do this.
|
|
More will likely be dropped in the future, but drop the obvious
ones we aren't using, for now; especially since some of them are
set at ->new time and unavoidable.
This saves 579 bytes per-client on my 64-bit Debian stable
system as measured by Devel::Size::total_size from
PublicInbox::HTTP::event_read. This adds up in C10K or C100K
situations.
Things we drop are:
* corked - MSG_MORE requires fewer syscalls
* read_push_back - tried to use it, ate CPU with slow clients
* IP/port fields - accept() already returns what we care about
|
|
Since our listen sockets are non-blocking and we may run
multiple httpd|nntpd processes; we need a way to avoid
thundering herds when there are multiple httpd|nntpd worker
processes.
EPOLLEXCLUSIVE was added just for that in Linux 4.5
|
|
These modules are unmaintained upstream at the moment, but I'll
be able to help with the intended maintainer once/if CPAN
ownership is transferred. OTOH, we've been waiting for that
transfer for several years, now...
Changes I intend to make:
* EPOLLEXCLUSIVE for Linux
* remove unused fields wasting memory
* kqueue bugfixes e.g. https://rt.cpan.org/Ticket/Display.html?id=116615
* accept4 support
And some lower priority experiments:
* switch to EV_ONESHOT / EPOLLONESHOT (incompatible changes)
* nginx-style buffering to tmpfile instead of string array
* sendfile off tmpfile buffers
* io_uring maybe?
|