1 # Copyright (C) 2020-2021 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # base class for remote IPC calls and workqueues, requires Storable or Sereal
5 package PublicInbox::IPC;
8 use Carp qw(confess croak);
9 use PublicInbox::DS qw(dwaitpid);
10 use PublicInbox::Spawn;
11 use POSIX qw(WNOHANG);
12 use Socket qw(AF_UNIX MSG_EOR);
13 my $SEQPACKET = eval { Socket::SOCK_SEQPACKET() }; # portable enough?
14 use constant PIPE_BUF => $^O eq 'linux' ? 4096 : POSIX::_POSIX_PIPE_BUF();
15 my $WQ_MAX_WORKERS = 4096;
17 # ->imports at BEGIN turns sereal_*_with_object into custom ops on 5.14+
18 # and eliminate method call overhead
21 require Sereal::Encoder;
22 require Sereal::Decoder;
23 Sereal::Encoder->import('sereal_encode_with_object');
24 Sereal::Decoder->import('sereal_decode_with_object');
25 ($enc, $dec) = (Sereal::Encoder->new, Sereal::Decoder->new);
29 if ($enc && $dec) { # should be custom ops
30 *freeze = sub ($) { sereal_encode_with_object $enc, $_[0] };
31 *thaw = sub ($) { sereal_decode_with_object $dec, $_[0], my $ret };
33 eval { # some distros have Storable as a separate package from Perl
35 Storable->import(qw(freeze thaw));
37 } // warn("Storable (part of Perl) missing: $@\n");
40 my $recv_cmd = PublicInbox::Spawn->can('recv_cmd4');
41 my $send_cmd = PublicInbox::Spawn->can('send_cmd4') // do {
42 require PublicInbox::CmdIPC4;
43 $recv_cmd //= PublicInbox::CmdIPC4->can('recv_cmd4');
44 PublicInbox::CmdIPC4->can('send_cmd4');
47 sub wq_set_recv_modes {
48 my ($self, @modes) = @_;
49 $self->{-wq_recv_modes} = \@modes;
54 defined(my $len = <$r>) or return;
55 chop($len) eq "\n" or croak "no LF byte in $len";
56 defined(my $n = read($r, my $buf, $len)) or croak "read error: $!";
57 $n == $len or croak "short read: $n != $len";
63 my $buf = freeze($ref);
64 length($buf) . "\n" . $buf;
69 print $w _pack_rec($ref) or croak "print: $!";
72 sub ipc_return ($$$) {
73 my ($w, $ret, $exc) = @_;
74 _send_rec($w, $exc ? bless(\$exc, 'PublicInbox::IPC::Die') : $ret);
77 sub ipc_worker_loop ($$$) {
78 my ($self, $r_req, $w_res) = @_;
79 my ($rec, $wantarray, $sub, @args);
81 while ($rec = _get_rec($r_req)) {
82 ($wantarray, $sub, @args) = @$rec;
83 # no waiting if client doesn't care,
84 # this is the overwhelmingly likely case
85 if (!defined($wantarray)) {
86 eval { $self->$sub(@args) };
87 warn "$$ die: $@ (from nowait $sub)\n" if $@;
88 } elsif ($wantarray) {
89 my @ret = eval { $self->$sub(@args) };
90 ipc_return($w_res, \@ret, $@);
91 } else { # '' => wantscalar
92 my $ret = eval { $self->$sub(@args) };
93 ipc_return($w_res, \$ret, $@);
98 # starts a worker if Sereal or Storable is installed
99 sub ipc_worker_spawn {
100 my ($self, $ident, $oldset) = @_;
101 return unless $enc; # no Sereal or Storable
102 return if ($self->{-ipc_ppid} // -1) == $$; # idempotent
103 delete(@$self{qw(-ipc_req -ipc_res -ipc_ppid -ipc_pid)});
104 pipe(my ($r_req, $w_req)) or die "pipe: $!";
105 pipe(my ($r_res, $w_res)) or die "pipe: $!";
106 my $sigset = $oldset // PublicInbox::DS::block_signals();
108 $self->ipc_atfork_prepare;
109 defined(my $pid = fork) or die "fork: $!";
111 eval { PublicInbox::DS->Reset };
112 $w_req = $r_res = undef;
113 $w_res->autoflush(1);
114 $SIG{$_} = 'IGNORE' for (qw(TERM INT QUIT));
116 PublicInbox::DS::sig_setmask($sigset);
117 my $on_destroy = $self->ipc_atfork_child;
118 eval { ipc_worker_loop($self, $r_req, $w_res) };
119 die "worker $ident PID:$$ died: $@\n" if $@;
122 PublicInbox::DS::sig_setmask($sigset) unless $oldset;
123 $r_req = $w_res = undef;
124 $w_req->autoflush(1);
125 $self->{-ipc_req} = $w_req;
126 $self->{-ipc_res} = $r_res;
127 $self->{-ipc_ppid} = $$;
128 $self->{-ipc_pid} = $pid;
131 sub ipc_worker_reap { # dwaitpid callback
132 my ($self, $pid) = @_;
133 warn "PID:$pid died with \$?=$?\n" if $?;
136 # for base class, override in sub classes
137 sub ipc_atfork_prepare {}
139 sub ipc_atfork_child {
141 my $io = delete($self->{-ipc_atfork_child_close}) or return;
146 # idempotent, can be called regardless of whether worker is active or not
147 sub ipc_worker_stop {
149 my ($pid, $ppid) = delete(@$self{qw(-ipc_pid -ipc_ppid)});
150 my ($w_req, $r_res) = delete(@$self{qw(-ipc_req -ipc_res)});
151 if (!$w_req && !$r_res) {
152 die "unexpected PID:$pid without IPC pipes" if $pid;
155 die 'no PID with IPC pipes' unless $pid;
156 $w_req = $r_res = undef;
158 return if $$ != $ppid;
159 dwaitpid($pid, \&ipc_worker_reap, $self);
162 # use this if we have multiple readers reading curl or "pigz -dc"
163 # and writing to the same store
166 require PublicInbox::Lock;
167 $self->{-ipc_lock} //= bless { lock_path => $f }, 'PublicInbox::Lock'
170 sub ipc_async_wait ($$) {
171 my ($self, $max) = @_; # max == -1 to wait for all
172 my $aif = $self->{-async_inflight} or return;
173 my $r_res = $self->{-ipc_res} or die 'BUG: no ipc_res';
174 while (my ($sub, $bytes, $cb, $cb_arg) = splice(@$aif, 0, 4)) {
175 my $ret = _get_rec($r_res) //
176 die "no response on $sub (req.size=$bytes)";
177 $self->{-async_inflight_bytes} -= $bytes;
179 eval { $cb->($cb_arg, $ret) };
180 warn "E: $sub callback error: $@\n" if $@;
181 return if --$max == 0;
185 # call $self->$sub(@args), on a worker if ipc_worker_spawn was used
187 my ($self, $sub, @args) = @_;
188 if (my $w_req = $self->{-ipc_req}) { # run in worker
189 my $ipc_lock = $self->{-ipc_lock};
190 my $lock = $ipc_lock ? $ipc_lock->lock_for_scope : undef;
191 if (defined(wantarray)) {
192 my $r_res = $self->{-ipc_res} or die 'BUG: no ipc_res';
193 ipc_async_wait($self, -1);
194 _send_rec($w_req, [ wantarray, $sub, @args ]);
195 my $ret = _get_rec($r_res) // die "no response on $sub";
196 die $$ret if ref($ret) eq 'PublicInbox::IPC::Die';
197 wantarray ? @$ret : $$ret;
198 } else { # likely, fire-and-forget into pipe
199 _send_rec($w_req, [ undef , $sub, @args ]);
201 } else { # run locally
207 my ($self, $sub, $sub_args, $cb, $cb_arg) = @_;
208 if (my $w_req = $self->{-ipc_req}) { # run in worker
209 my $rec = _pack_rec([ 1, $sub, @$sub_args ]);
210 my $cur_bytes = \($self->{-async_inflight_bytes} //= 0);
211 while (($$cur_bytes + length($rec)) > PIPE_BUF) {
212 ipc_async_wait($self, 1);
214 my $ipc_lock = $self->{-ipc_lock};
215 my $lock = $ipc_lock ? $ipc_lock->lock_for_scope : undef;
216 print $w_req $rec or croak "print: $!";
217 $$cur_bytes += length($rec);
218 push @{$self->{-async_inflight}},
219 $sub, length($rec), $cb, $cb_arg;
221 my $ret = [ eval { $self->$sub(@$sub_args) } ];
223 $ret = ( bless(\$exc, 'PublicInbox::IPC::Die') );
225 eval { $cb->($cb_arg, $ret) };
226 warn "E: $sub callback error: $@\n" if $@;
230 # needed when there's multiple IPC workers and the parent forking
231 # causes newer siblings to inherit older siblings sockets
232 sub ipc_sibling_atfork_child {
234 my ($pid, undef) = delete(@$self{qw(-ipc_pid -ipc_ppid)});
235 delete(@$self{qw(-ipc_req -ipc_res)});
236 $pid == $$ and die "BUG: $$ ipc_atfork_child called on itself";
239 sub _close_recvd ($) {
241 my $x = $self->{-wq_recv_modes};
242 my $end = $x ? $#$x : 2;
243 close($_) for (grep { defined } (delete @$self{0..$end}));
246 sub wq_worker_loop ($) {
249 my $len = $self->{wq_req_len} // (4096 * 33);
251 my $s2 = $self->{-wq_s2} // die 'BUG: no -wq_s2';
252 local $SIG{PIPE} = sub {
255 die(bless(\$cur_sub, 'PublicInbox::SIGPIPE')) if $cur_sub;
258 my (@fds) = $recv_cmd->($s2, $buf, $len) or return; # EOF
260 my @m = @{$self->{-wq_recv_modes} // [qw( +<&= >&= >&= )]};
262 my $mode = shift(@m);
263 if (open(my $cmdfh, $mode, $fd)) {
264 $self->{$i++} = $cmdfh;
265 $cmdfh->autoflush(1);
267 die "$$ open($mode$fd) (FD:$i): $!";
270 # Sereal dies on truncated data, Storable returns undef
271 $args = thaw($buf) //
272 die "thaw error on buffer of size:".length($buf);
275 eval { $self->$sub(@$args) };
276 undef $sub; # quiet SIG{PIPE} handler
279 warn "$$ wq_worker: $@" if $@ &&
280 ref($@) ne 'PublicInbox::SIGPIPE';
281 # need to close explicitly to avoid warnings after SIGPIPE
286 sub wq_do { # always async
287 my ($self, $sub, $ios, @args) = @_;
288 if (my $s1 = $self->{-wq_s1}) { # run in worker
289 my $fds = [ map { fileno($_) } @$ios ];
290 $send_cmd->($s1, $fds, freeze([$sub, @args]), MSG_EOR);
292 @$self{0..$#$ios} = @$ios;
293 eval { $self->$sub(@args) };
294 warn "wq_do: $@" if $@ && ref($@) ne 'PublicInbox::SIGPIPE';
295 delete @$self{0..$#$ios}; # don't close
299 sub _wq_worker_start ($$) {
300 my ($self, $oldset) = @_;
301 my $pid = fork // die "fork: $!";
303 eval { PublicInbox::DS->Reset };
304 close(delete $self->{-wq_s1});
305 delete $self->{qw(-wq_workers -wq_ppid)};
306 $SIG{$_} = 'IGNORE' for (qw(PIPE TTOU TTIN));
307 $SIG{$_} = 'DEFAULT' for (qw(TERM QUIT INT CHLD));
308 local $0 = $self->{-wq_ident};
309 PublicInbox::DS::sig_setmask($oldset);
310 # ensure we properly exit even if warn() dies:
311 my $end = PublicInbox::OnDestroy->new($$, sub { exit(!!$@) });
312 my $on_destroy = $self->ipc_atfork_child;
313 eval { wq_worker_loop($self) };
314 warn "worker $self->{-wq_ident} PID:$$ died: $@" if $@;
316 undef $end; # trigger exit
318 $self->{-wq_workers}->{$pid} = \undef;
322 # starts workqueue workers if Sereal or Storable is installed
323 sub wq_workers_start {
324 my ($self, $ident, $nr_workers, $oldset) = @_;
325 ($enc && $send_cmd && $recv_cmd && defined($SEQPACKET)) or return;
326 return if $self->{-wq_s1}; # idempotent
328 socketpair($s1, $s2, AF_UNIX, $SEQPACKET, 0) or die "socketpair: $!";
329 $self->ipc_atfork_prepare;
331 $nr_workers = $WQ_MAX_WORKERS if $nr_workers > $WQ_MAX_WORKERS;
332 my $sigset = $oldset // PublicInbox::DS::block_signals();
333 $self->{-wq_workers} = {};
334 $self->{-wq_ident} = $ident;
335 $self->{-wq_s1} = $s1;
336 $self->{-wq_s2} = $s2;
337 _wq_worker_start($self, $sigset) for (1..$nr_workers);
338 PublicInbox::DS::sig_setmask($sigset) unless $oldset;
339 $self->{-wq_ppid} = $$;
342 sub wq_worker_incr { # SIGTTIN handler
343 my ($self, $oldset) = @_;
344 $self->{-wq_s2} or return;
345 return if wq_workers($self) >= $WQ_MAX_WORKERS;
346 $self->ipc_atfork_prepare;
347 my $sigset = $oldset // PublicInbox::DS::block_signals();
348 _wq_worker_start($self, $sigset);
349 PublicInbox::DS::sig_setmask($sigset) unless $oldset;
352 sub wq_exit { # wakes up wq_worker_decr_wait
353 send($_[0]->{-wq_s2}, $$, MSG_EOR) // die "$$ send: $!";
357 sub wq_worker_decr { # SIGTTOU handler, kills first idle worker
359 return unless wq_workers($self);
360 my $s2 = $self->{-wq_s2} // die 'BUG: no wq_s2';
361 $self->wq_do('wq_exit', [ $s2, $s2, $s2 ]);
362 # caller must call wq_worker_decr_wait in main loop
365 sub wq_worker_decr_wait {
366 my ($self, $timeout) = @_;
367 return if $self->{-wq_ppid} != $$; # can't reap siblings or parents
368 my $s1 = $self->{-wq_s1} // croak 'BUG: no wq_s1';
369 vec(my $rin = '', fileno($s1), 1) = 1;
370 select(my $rout = $rin, undef, undef, $timeout) or
371 croak 'timed out waiting for wq_exit';
372 recv($s1, my $pid, 64, 0) // croak "recv: $!";
373 my $workers = $self->{-wq_workers} // croak 'BUG: no wq_workers';
374 delete $workers->{$pid} // croak "BUG: PID:$pid invalid";
375 dwaitpid($pid, \&ipc_worker_reap, $self);
378 # set or retrieve number of workers
380 my ($self, $nr) = @_;
381 my $cur = $self->{-wq_workers} or return;
383 while (scalar(keys(%$cur)) > $nr) {
384 $self->wq_worker_decr;
385 $self->wq_worker_decr_wait;
387 $self->wq_worker_incr while scalar(keys(%$cur)) < $nr;
394 delete @$self{qw(-wq_s1 -wq_s2)} or return;
395 my $ppid = delete $self->{-wq_ppid} or return;
396 my $workers = delete $self->{-wq_workers} // die 'BUG: no wq_workers';
397 return if $ppid != $$; # can't reap siblings or parents
398 for my $pid (keys %$workers) {
399 dwaitpid($pid, \&ipc_worker_reap, $self);
403 sub WQ_MAX_WORKERS { $WQ_MAX_WORKERS }
407 ipc_worker_stop($_[0]);