lei_mail_sync: rely on flock(2), avoid IPC

Since 44917fdd24a8bec1 ("lei_mail_sync: do not use transactions"),
relying on lei/store to serialize access was a pointless endeavor.

Rely on flock(2) to serialize multiple writers since (in my
experience) it's the easiest way to deal with parallel writers
when using SQLite.  This allows us to simplify existing callers
while speeding up 'lei refresh-mail-sync --all=local' by 5% or
so.

1 files changed, 12 insertions, 14 deletions

diff --git a/lib/PublicInbox/LeiNoteEvent.pm b/lib/PublicInbox/LeiNoteEvent.pm
index 41415346..c03c5319 100644
--- a/lib/PublicInbox/LeiNoteEvent.pm
+++ b/lib/PublicInbox/LeiNoteEvent.pm
@@ -2,6 +2,7 @@
 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
 
 # internal command for dealing with inotify, kqueue vnodes, etc
+# it is a semi-persistent worker
 package PublicInbox::LeiNoteEvent;
 use strict;
 use v5.10.1;
@@ -12,11 +13,8 @@ our $to_flush; # { cfgpath => $lei }
 
 sub flush_lei ($) {
         my ($lei) = @_;
-        if (my $lne = delete $lei->{cfg}->{-lei_note_event}) {
-                $lne->wq_close(1, undef, $lei); # runs _lei_wq_eof;
-        } elsif ($lei->{sto}) { # lms_clear_src calls only:
-                $lei->sto_done_request;
-        }
+        my $lne = delete $lei->{cfg}->{-lei_note_event};
+        $lne->wq_close(1, undef, $lei) if $lne; # runs _lei_wq_eof;
 }
 
 # we batch up writes and flush every 5s (matching Linux default
@@ -38,14 +36,14 @@ sub note_event_arm_done ($) {
 sub eml_event ($$$$) {
         my ($self, $eml, $vmd, $state) = @_;
         my $sto = $self->{lei}->{sto};
-        my $lse = $self->{lse} //= $sto->search;
         if ($state =~ /\Aimport-(?:rw|ro)\z/) {
                 $sto->ipc_do('set_eml', $eml, $vmd);
         } elsif ($state =~ /\Aindex-(?:rw|ro)\z/) {
                 my $xoids = $self->{lei}->ale->xoids_for($eml);
                 $sto->ipc_do('index_eml_only', $eml, $vmd, $xoids);
         } elsif ($state =~ /\Atag-(?:rw|ro)\z/) {
-                my $c = $lse->kw_changed($eml, $vmd->{kw}, my $docids = []);
+                my $docids = [];
+                my $c = $self->{lse}->kw_changed($eml, $vmd->{kw}, $docids);
                 if (scalar @$docids) { # already in lei/store
                         $sto->ipc_do('set_eml_vmd', undef, $vmd, $docids) if $c;
                 } elsif (my $xoids = $self->{lei}->ale->xoids_for($eml)) {
@@ -69,21 +67,19 @@ sub lei_note_event {
         my $cfg = $lei->_lei_cfg or return; # gone (race)
         my $sto = $lei->_lei_store or return; # gone
         return flush_lei($lei) if $folder eq 'done'; # special case
-        my $lms = $sto->search->lms or return;
+        my $lms = $lei->lms or return;
+        $lms->lms_write_prepare if $new_cur eq ''; # for ->clear_src below
         my $err = $lms->arg2folder($lei, [ $folder ]);
         return if $err->{fail};
-        undef $lms;
         my $state = $cfg->get_1("watch.$folder", 'state') // 'tag-rw';
         return if $state eq 'pause';
+        return $lms->clear_src($folder, \$bn) if $new_cur eq '';
+        $lms->lms_pause;
         $lei->ale; # prepare
         $sto->write_prepare($lei);
-        if ($new_cur eq '') {
-                $sto->ipc_do('lms_clear_src', $folder, \$bn);
-                return note_event_arm_done($lei);
-        }
         require PublicInbox::MdirReader;
         my $self = $cfg->{-lei_note_event} //= do {
-                my $wq = bless {}, __PACKAGE__;
+                my $wq = bless { lms => $lms }, __PACKAGE__;
                 # MUAs such as mutt can trigger massive rename() storms so
                 # use all CPU power available:
                 my $jobs = $wq->detect_nproc // 1;
@@ -105,6 +101,8 @@ sub lei_note_event {
 sub ipc_atfork_child {
         my ($self) = @_;
         $self->{lei}->_lei_atfork_child(1); # persistent, for a while
+        $self->{lms}->lms_write_prepare;
+        $self->{lse} = $self->{lei}->{sto}->search;
         $self->SUPER::ipc_atfork_child;
 }