user/dev discussion of public-inbox itself
 help / color / Atom feed
From: Eric Wong <e@80x24.org>
To: meta@public-inbox.org
Subject: [PATCH 33/57] nntp: simplify re-arming/requeue logic
Date: Mon, 24 Jun 2019 02:52:34 +0000
Message-ID: <20190624025258.25592-34-e@80x24.org> (raw)
In-Reply-To: <20190624025258.25592-1-e@80x24.org>

We can be smarter about requeuing clients to run and avoid
excessive epoll_ctl calls since we can trust event_step to do
the right thing depending on the state of the client.
---
 lib/PublicInbox/NNTP.pm | 33 ++++++++++-----------------------
 1 file changed, 10 insertions(+), 23 deletions(-)

diff --git a/lib/PublicInbox/NNTP.pm b/lib/PublicInbox/NNTP.pm
index 42fbb255..468a22f5 100644
--- a/lib/PublicInbox/NNTP.pm
+++ b/lib/PublicInbox/NNTP.pm
@@ -58,6 +58,11 @@ sub next_tick () {
 	}
 }
 
+sub requeue ($) {
+	push @$nextq, $_[0];
+	$nextt ||= PublicInbox::EvCleanup::asap(*next_tick);
+}
+
 sub update_idle_time ($) {
 	my ($self) = @_;
 	my $sock = $self->{sock} or return;
@@ -633,7 +638,7 @@ sub long_response ($$) {
 			}
 			if ($self->{sock}) {
 				update_idle_time($self);
-				check_read($self);
+				requeue($self);
 			} else {
 				out($self, " deferred[$fd] aborted - %0.6f",
 				           now() - $t0);
@@ -642,14 +647,12 @@ sub long_response ($$) {
 			# no recursion, schedule another call ASAP
 			# but only after all pending writes are done
 			update_idle_time($self);
-
-			push @$nextq, $self;
-			$nextt ||= PublicInbox::EvCleanup::asap(*next_tick);
+			requeue($self);
 		} else { # all done!
 			delete $self->{long_res};
-			check_read($self);
 			res($self, '.');
 			out($self, " deferred[$fd] done - %0.6f", now() - $t0);
+			requeue($self);
 		}
 	};
 	$self->{long_res}->(); # kick off!
@@ -930,6 +933,7 @@ sub out ($$;@) {
 	printf { $self->{nntpd}->{out} } $fmt."\n", @args;
 }
 
+# callback used by PublicInbox::DS for any (e)poll (in/out/hup/err)
 sub event_step {
 	my ($self) = @_;
 
@@ -965,24 +969,7 @@ sub event_step {
 
 	# maybe there's more pipelined data, or we'll have
 	# to register it for socket-readiness notifications
-	check_read($self) unless ($self->{long_res} || $self->{wbuf});
-}
-
-sub check_read {
-	my ($self) = @_;
-	if (index($self->{rbuf}, "\n") >= 0) {
-		# Force another read if there is a pipelined request.
-		# We don't know if the socket has anything for us to read,
-		# and we must double-check again by the time the timer fires
-		# in case we really did dispatch a read event and started
-		# another long response.
-		push @$nextq, $self;
-		$nextt ||= PublicInbox::EvCleanup::asap(*next_tick);
-	} else {
-		# no pipelined requests available, let the kernel know
-		# to wake us up if there's more
-		$self->watch_in1; # PublicInbox::DS::watch_in1
-	}
+	requeue($self) unless ($self->{long_res} || $self->{wbuf});
 }
 
 sub not_idle_long ($$) {
-- 
EW


  parent reply index

Thread overview: 61+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-06-24  2:52 [PATCH 00/57] ds: shrink, TLS support, buffer writes to FS Eric Wong
2019-06-24  2:52 ` [PATCH 01/57] ds: get rid of {closed} field Eric Wong
2019-06-24  2:52 ` [PATCH 02/57] ds: get rid of more unused debug instance methods Eric Wong
2019-06-24  2:52 ` [PATCH 03/57] ds: use and export monotonic now() Eric Wong
2019-06-24  2:52 ` [PATCH 04/57] AddTimer: avoid clock_gettime for the '0' case Eric Wong
2019-06-24  2:52 ` [PATCH 05/57] ds: get rid of on_incomplete_write wrapper Eric Wong
2019-06-24  2:52 ` [PATCH 06/57] ds: lazy initialize wbuf_off Eric Wong
2019-06-24  2:52 ` [PATCH 07/57] ds: split out from ->flush_write and ->write Eric Wong
2019-06-24  2:52 ` [PATCH 08/57] ds: lazy-initialize wbuf Eric Wong
2019-06-24  2:52 ` [PATCH 09/57] ds: don't pass `events' arg to EPOLL_CTL_DEL Eric Wong
2019-06-24  2:52 ` [PATCH 10/57] ds: remove support for DS->write(undef) Eric Wong
2019-06-24  2:52 ` [PATCH 11/57] http: favor DS->write(strref) when reasonable Eric Wong
2019-06-24  2:52 ` [PATCH 12/57] ds: share send(..., MSG_MORE) logic Eric Wong
2019-06-24  2:52 ` [PATCH 13/57] ds: switch write buffering to use a tempfile Eric Wong
2019-06-24  2:52 ` [PATCH 14/57] ds: get rid of redundant and unnecessary POLL* constants Eric Wong
2019-06-24  2:52 ` [PATCH 15/57] syscall: get rid of unused EPOLL* constants Eric Wong
2019-06-24  2:52 ` [PATCH 16/57] syscall: get rid of unnecessary uname local vars Eric Wong
2019-06-24  2:52 ` [PATCH 17/57] ds: set event flags directly at initialization Eric Wong
2019-06-24  2:52 ` [PATCH 18/57] ds: import IO::KQueue namespace Eric Wong
2019-06-24  2:52 ` [PATCH 19/57] ds: share watch_chg between watch_read/watch_write Eric Wong
2019-06-24  2:52 ` [PATCH 20/57] ds: remove IO::Poll support (for now) Eric Wong
2019-06-24  2:52 ` [PATCH 21/57] ds: get rid of event_watch field Eric Wong
2019-06-24  2:52 ` [PATCH 22/57] httpd/async: remove EINTR check Eric Wong
2019-06-24  2:52 ` [PATCH 23/57] spawn: remove `Blocking' flag handling Eric Wong
2019-06-24  2:52 ` [PATCH 24/57] qspawn: describe where `$rpipe' come from Eric Wong
2019-06-24  2:52 ` [PATCH 25/57] http|nntp: favor "$! == EFOO" over $!{EFOO} checks Eric Wong
2019-06-24  2:52 ` [PATCH 26/57] ds: favor `delete' over assigning fields to `undef' Eric Wong
2019-06-24  2:52 ` [PATCH 27/57] http: don't pass extra args to PublicInbox::DS::close Eric Wong
2019-06-24  2:52 ` [PATCH 28/57] ds: pass $self to code references Eric Wong
2019-06-24  2:52 ` [PATCH 29/57] evcleanup: replace _run_asap with `event_step' callback Eric Wong
2019-06-24  2:52 ` [PATCH 30/57] ds: remove pointless exit calls Eric Wong
2019-06-24  2:52 ` [PATCH 31/57] http|nntp: be explicit about bytes::length on rbuf Eric Wong
2019-06-24  2:52 ` [PATCH 32/57] ds: hoist out do_read from NNTP and HTTP Eric Wong
2019-06-24  2:52 ` Eric Wong [this message]
2019-06-24  2:52 ` [PATCH 34/57] allow use of PerlIO layers for filesystem writes Eric Wong
2019-06-24  2:52 ` [PATCH 35/57] ds: deal better with FS-related errors IO buffers Eric Wong
2019-06-24  2:52 ` [PATCH 36/57] nntp: wait for writability before sending greeting Eric Wong
2019-06-24  2:52 ` [PATCH 37/57] nntp: NNTPS and NNTP+STARTTLS working Eric Wong
2019-06-24  2:52 ` [PATCH 38/57] certs/create-certs.perl: fix cert validity on 32-bit Eric Wong
2019-06-24  2:52 ` [PATCH 39/57] daemon: map inherited sockets to well-known schemes Eric Wong
2019-06-24  2:52 ` [PATCH 40/57] ds|nntp: use CORE::close on socket Eric Wong
2019-06-24  2:52 ` [PATCH 41/57] nntp: call SSL_shutdown in normal cases Eric Wong
2019-06-24  2:52 ` [PATCH 42/57] t/nntpd-tls: slow client connection test Eric Wong
2019-06-24  2:52 ` [PATCH 43/57] daemon: use SSL_MODE_RELEASE_BUFFERS Eric Wong
2019-06-24  2:52 ` [PATCH 44/57] ds: allow ->write callbacks to syswrite directly Eric Wong
2019-06-24  2:52 ` [PATCH 45/57] nntp: reduce allocations for greeting Eric Wong
2019-06-24  2:52 ` [PATCH 46/57] ds: always use EV_ADD with EV_SET Eric Wong
2019-06-24  2:52 ` [PATCH 47/57] nntp: simplify long response logic and fix nesting Eric Wong
2019-06-24  2:52 ` [PATCH 48/57] ds: flush_write runs ->write callbacks even if closed Eric Wong
2019-06-24  2:52 ` [PATCH 49/57] nntp: lazily allocate and stash rbuf Eric Wong
2019-06-24  2:52 ` [PATCH 50/57] ci: require IO::KQueue on FreeBSD, for now Eric Wong
2019-06-24  2:52 ` [PATCH 51/57] nntp: send greeting immediately for plain sockets Eric Wong
2019-06-24  2:52 ` [PATCH 52/57] daemon: set TCP_DEFER_ACCEPT on everything but NNTP Eric Wong
2019-06-24  2:52 ` [PATCH 53/57] daemon: use FreeBSD accept filters on non-NNTP Eric Wong
2019-06-24  2:52 ` [PATCH 54/57] ds: split out IO::KQueue-specific code Eric Wong
2019-06-24  5:24   ` Eric Wong
2019-06-24  2:52 ` [PATCH 55/57] ds: reimplement IO::Poll support to look like epoll Eric Wong
2019-06-24  2:52 ` [PATCH 56/57] Revert "ci: require IO::KQueue on FreeBSD, for now" Eric Wong
2019-06-24  2:52 ` [PATCH 57/57] ds: reduce overhead of tempfile creation Eric Wong
2019-06-24  5:25 ` [PATCH 58/57] Makefile: skip DSKQXS in global syntax check Eric Wong
2019-06-24 18:28 ` [PATCH 59/57] ds: ->write must not clobber empty wbuf array Eric Wong

Reply instructions:

You may reply publically to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: https://public-inbox.org/README

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20190624025258.25592-34-e@80x24.org \
    --to=e@80x24.org \
    --cc=meta@public-inbox.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

user/dev discussion of public-inbox itself

Archives are clonable:
	git clone --mirror https://public-inbox.org/meta
	git clone --mirror http://czquwvybam4bgbro.onion/meta
	git clone --mirror http://hjrcffqmbrq6wope.onion/meta
	git clone --mirror http://ou63pmih66umazou.onion/meta

Example config snippet for mirrors

Newsgroups are available over NNTP:
	nntp://news.public-inbox.org/inbox.comp.mail.public-inbox.meta
	nntp://ou63pmih66umazou.onion/inbox.comp.mail.public-inbox.meta
	nntp://czquwvybam4bgbro.onion/inbox.comp.mail.public-inbox.meta
	nntp://hjrcffqmbrq6wope.onion/inbox.comp.mail.public-inbox.meta
	nntp://news.gmane.org/gmane.mail.public-inbox.general

 note: .onion URLs require Tor: https://www.torproject.org/

AGPL code for this site: git clone https://public-inbox.org/public-inbox.git