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();
107 $self->ipc_atfork_prepare;
108 defined(my $pid = fork) or die "fork: $!";
110 eval { PublicInbox::DS->Reset };
111 delete @$self{qw(-wq_s1 -wq_workers -wq_ppid)};
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, $pid) = @_;
133 # SIGTERM (15) is our default exit signal
134 warn "PID:$pid died with \$?=$?\n" if $? && ($? & 127) != 15;
139 my $pids = delete $self->{"-wq_old_pids.$$"} or return;
140 dwaitpid($_, \&ipc_worker_reap, $self) for @$pids;
143 # for base class, override in sub classes
144 sub ipc_atfork_prepare {}
146 sub ipc_atfork_child {
148 my $io = delete($self->{-ipc_atfork_child_close}) or return;
153 # idempotent, can be called regardless of whether worker is active or not
154 sub ipc_worker_stop {
156 my ($pid, $ppid) = delete(@$self{qw(-ipc_pid -ipc_ppid)});
157 my ($w_req, $r_res) = delete(@$self{qw(-ipc_req -ipc_res)});
158 if (!$w_req && !$r_res) {
159 die "unexpected PID:$pid without IPC pipes" if $pid;
162 die 'no PID with IPC pipes' unless $pid;
163 $w_req = $r_res = undef;
165 return if $$ != $ppid;
166 dwaitpid($pid, \&ipc_worker_reap, $self);
169 # use this if we have multiple readers reading curl or "pigz -dc"
170 # and writing to the same store
173 require PublicInbox::Lock;
174 $self->{-ipc_lock} //= bless { lock_path => $f }, 'PublicInbox::Lock'
177 sub ipc_async_wait ($$) {
178 my ($self, $max) = @_; # max == -1 to wait for all
179 my $aif = $self->{-async_inflight} or return;
180 my $r_res = $self->{-ipc_res} or die 'BUG: no ipc_res';
181 while (my ($sub, $bytes, $cb, $cb_arg) = splice(@$aif, 0, 4)) {
182 my $ret = _get_rec($r_res) //
183 die "no response on $sub (req.size=$bytes)";
184 $self->{-async_inflight_bytes} -= $bytes;
186 eval { $cb->($cb_arg, $ret) };
187 warn "E: $sub callback error: $@\n" if $@;
188 return if --$max == 0;
192 # call $self->$sub(@args), on a worker if ipc_worker_spawn was used
194 my ($self, $sub, @args) = @_;
195 if (my $w_req = $self->{-ipc_req}) { # run in worker
196 my $ipc_lock = $self->{-ipc_lock};
197 my $lock = $ipc_lock ? $ipc_lock->lock_for_scope : undef;
198 if (defined(wantarray)) {
199 my $r_res = $self->{-ipc_res} or die 'BUG: no ipc_res';
200 ipc_async_wait($self, -1);
201 _send_rec($w_req, [ wantarray, $sub, @args ]);
202 my $ret = _get_rec($r_res) // die "no response on $sub";
203 die $$ret if ref($ret) eq 'PublicInbox::IPC::Die';
204 wantarray ? @$ret : $$ret;
205 } else { # likely, fire-and-forget into pipe
206 _send_rec($w_req, [ undef , $sub, @args ]);
208 } else { # run locally
214 my ($self, $sub, $sub_args, $cb, $cb_arg) = @_;
215 if (my $w_req = $self->{-ipc_req}) { # run in worker
216 my $rec = _pack_rec([ 1, $sub, @$sub_args ]);
217 my $cur_bytes = \($self->{-async_inflight_bytes} //= 0);
218 while (($$cur_bytes + length($rec)) > PIPE_BUF) {
219 ipc_async_wait($self, 1);
221 my $ipc_lock = $self->{-ipc_lock};
222 my $lock = $ipc_lock ? $ipc_lock->lock_for_scope : undef;
223 print $w_req $rec or croak "print: $!";
224 $$cur_bytes += length($rec);
225 push @{$self->{-async_inflight}},
226 $sub, length($rec), $cb, $cb_arg;
228 my $ret = [ eval { $self->$sub(@$sub_args) } ];
230 $ret = ( bless(\$exc, 'PublicInbox::IPC::Die') );
232 eval { $cb->($cb_arg, $ret) };
233 warn "E: $sub callback error: $@\n" if $@;
237 # needed when there's multiple IPC workers and the parent forking
238 # causes newer siblings to inherit older siblings sockets
239 sub ipc_sibling_atfork_child {
241 my ($pid, undef) = delete(@$self{qw(-ipc_pid -ipc_ppid)});
242 delete(@$self{qw(-ipc_req -ipc_res)});
243 $pid == $$ and die "BUG: $$ ipc_atfork_child called on itself";
246 sub wq_worker_loop ($) {
248 my $len = $self->{wq_req_len} // (4096 * 33);
249 my $s2 = $self->{-wq_s2} // die 'BUG: no -wq_s2';
251 my @fds = $recv_cmd->($s2, my $buf, $len) or return; # EOF
252 my @m = @{$self->{-wq_recv_modes} // [qw( +<&= >&= >&= )]};
255 my $mode = shift(@m);
256 if (open(my $cmdfh, $mode, $fd)) {
257 $self->{$nfd++} = $cmdfh;
258 $cmdfh->autoflush(1);
260 die "$$ open($mode$fd) (FD:$nfd): $!";
263 # Sereal dies on truncated data, Storable returns undef
264 my $args = thaw($buf) //
265 die "thaw error on buffer of size:".length($buf);
266 my $sub = shift @$args;
267 eval { $self->$sub(@$args) };
268 warn "$$ wq_worker: $@" if $@ &&
269 ref($@) ne 'PublicInbox::SIGPIPE';
270 delete @$self{0..($nfd-1)};
274 sub wq_do { # always async
275 my ($self, $sub, $ios, @args) = @_;
276 if (my $s1 = $self->{-wq_s1}) { # run in worker
277 my $fds = [ map { fileno($_) } @$ios ];
278 $send_cmd->($s1, $fds, freeze([$sub, @args]), MSG_EOR);
280 @$self{0..$#$ios} = @$ios;
281 eval { $self->$sub(@args) };
282 warn "wq_do: $@" if $@ && ref($@) ne 'PublicInbox::SIGPIPE';
283 delete @$self{0..$#$ios}; # don't close
287 sub _wq_worker_start ($$) {
288 my ($self, $oldset) = @_;
289 my $pid = fork // die "fork: $!";
291 eval { PublicInbox::DS->Reset };
292 delete @$self{qw(-wq_s1 -wq_workers -wq_ppid)};
293 $SIG{$_} = 'IGNORE' for (qw(PIPE TTOU TTIN));
294 $SIG{$_} = 'DEFAULT' for (qw(TERM QUIT INT CHLD));
295 local $0 = $self->{-wq_ident};
296 PublicInbox::DS::sig_setmask($oldset);
297 # ensure we properly exit even if warn() dies:
298 my $end = PublicInbox::OnDestroy->new($$, sub { exit(!!$@) });
299 my $on_destroy = $self->ipc_atfork_child;
300 eval { wq_worker_loop($self) };
301 warn "worker $self->{-wq_ident} PID:$$ died: $@" if $@;
303 undef $end; # trigger exit
305 $self->{-wq_workers}->{$pid} = \undef;
309 # starts workqueue workers if Sereal or Storable is installed
310 sub wq_workers_start {
311 my ($self, $ident, $nr_workers, $oldset) = @_;
312 ($enc && $send_cmd && $recv_cmd && defined($SEQPACKET)) or return;
313 return if $self->{-wq_s1}; # idempotent
314 $self->{-wq_s1} = $self->{-wq_s2} = undef;
315 socketpair($self->{-wq_s1}, $self->{-wq_s2}, AF_UNIX, $SEQPACKET, 0) or
316 die "socketpair: $!";
317 $self->ipc_atfork_prepare;
319 $nr_workers = $WQ_MAX_WORKERS if $nr_workers > $WQ_MAX_WORKERS;
320 my $sigset = $oldset // PublicInbox::DS::block_signals();
321 $self->{-wq_workers} = {};
322 $self->{-wq_ident} = $ident;
323 _wq_worker_start($self, $sigset) for (1..$nr_workers);
324 PublicInbox::DS::sig_setmask($sigset) unless $oldset;
325 $self->{-wq_ppid} = $$;
328 sub wq_worker_incr { # SIGTTIN handler
329 my ($self, $oldset) = @_;
330 $self->{-wq_s2} or return;
331 return if wq_workers($self) >= $WQ_MAX_WORKERS;
332 $self->ipc_atfork_prepare;
333 my $sigset = $oldset // PublicInbox::DS::block_signals();
334 _wq_worker_start($self, $sigset);
335 PublicInbox::DS::sig_setmask($sigset) unless $oldset;
338 sub wq_exit { # wakes up wq_worker_decr_wait
339 send($_[0]->{-wq_s2}, $$, MSG_EOR) // die "$$ send: $!";
343 sub wq_worker_decr { # SIGTTOU handler, kills first idle worker
345 return unless wq_workers($self);
346 my $s2 = $self->{-wq_s2} // die 'BUG: no wq_s2';
347 $self->wq_do('wq_exit', [ $s2, $s2, $s2 ]);
348 # caller must call wq_worker_decr_wait in main loop
351 sub wq_worker_decr_wait {
352 my ($self, $timeout) = @_;
353 return if $self->{-wq_ppid} != $$; # can't reap siblings or parents
354 my $s1 = $self->{-wq_s1} // croak 'BUG: no wq_s1';
355 vec(my $rin = '', fileno($s1), 1) = 1;
356 select(my $rout = $rin, undef, undef, $timeout) or
357 croak 'timed out waiting for wq_exit';
358 recv($s1, my $pid, 64, 0) // croak "recv: $!";
359 my $workers = $self->{-wq_workers} // croak 'BUG: no wq_workers';
360 delete $workers->{$pid} // croak "BUG: PID:$pid invalid";
361 dwaitpid($pid, \&ipc_worker_reap, $self);
364 # set or retrieve number of workers
366 my ($self, $nr) = @_;
367 my $cur = $self->{-wq_workers} or return;
369 while (scalar(keys(%$cur)) > $nr) {
370 $self->wq_worker_decr;
371 $self->wq_worker_decr_wait;
373 $self->wq_worker_incr while scalar(keys(%$cur)) < $nr;
379 my ($self, $nohang) = @_;
380 delete @$self{qw(-wq_s1 -wq_s2)} or return;
381 my $ppid = delete $self->{-wq_ppid} or return;
382 my $workers = delete $self->{-wq_workers} // die 'BUG: no wq_workers';
383 return if $ppid != $$; # can't reap siblings or parents
384 my @pids = map { $_ + 0 } keys %$workers;
386 push @{$self->{"-wq_old_pids.$$"}}, @pids;
388 dwaitpid($_, \&ipc_worker_reap, $self) for @pids;
394 my $pids = $self->{"-wq_old_pids.$$"} or return;
399 my ($self, $sig) = @_;
400 my $workers = $self->{-wq_workers} or return;
401 kill($sig // 'TERM', keys %$workers);
404 sub WQ_MAX_WORKERS { $WQ_MAX_WORKERS }
408 my $ppid = $self->{-wq_ppid};
409 wq_kill($self) if $ppid && $ppid == $$;
411 ipc_worker_stop($self);
414 # Sereal doesn't have dclone
415 sub deep_clone { thaw(freeze($_[-1])) }