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(mkfifo WNOHANG);
12 use Socket qw(AF_UNIX MSG_EOR SOCK_STREAM);
13 use Errno qw(EMSGSIZE);
14 use File::Temp 0.19 (); # 0.19 for ->newdir
15 my $SEQPACKET = eval { Socket::SOCK_SEQPACKET() }; # portable enough?
16 use constant PIPE_BUF => $^O eq 'linux' ? 4096 : POSIX::_POSIX_PIPE_BUF();
17 my $WQ_MAX_WORKERS = 4096;
19 # ->imports at BEGIN turns sereal_*_with_object into custom ops on 5.14+
20 # and eliminate method call overhead
23 require Sereal::Encoder;
24 require Sereal::Decoder;
25 Sereal::Encoder->import('sereal_encode_with_object');
26 Sereal::Decoder->import('sereal_decode_with_object');
27 ($enc, $dec) = (Sereal::Encoder->new, Sereal::Decoder->new);
31 if ($enc && $dec) { # should be custom ops
32 *freeze = sub ($) { sereal_encode_with_object $enc, $_[0] };
33 *thaw = sub ($) { sereal_decode_with_object $dec, $_[0], my $ret };
35 eval { # some distros have Storable as a separate package from Perl
37 Storable->import(qw(freeze thaw));
39 } // warn("Storable (part of Perl) missing: $@\n");
42 my $recv_cmd = PublicInbox::Spawn->can('recv_cmd4');
43 my $send_cmd = PublicInbox::Spawn->can('send_cmd4') // do {
44 require PublicInbox::CmdIPC4;
45 $recv_cmd //= PublicInbox::CmdIPC4->can('recv_cmd4');
46 PublicInbox::CmdIPC4->can('send_cmd4');
49 sub wq_set_recv_modes {
50 my ($self, @modes) = @_;
51 $self->{-wq_recv_modes} = \@modes;
56 defined(my $len = <$r>) or return;
57 chop($len) eq "\n" or croak "no LF byte in $len";
58 defined(my $n = read($r, my $buf, $len)) or croak "read error: $!";
59 $n == $len or croak "short read: $n != $len";
65 my $buf = freeze($ref);
66 length($buf) . "\n" . $buf;
71 print $w _pack_rec($ref) or croak "print: $!";
74 sub ipc_return ($$$) {
75 my ($w, $ret, $exc) = @_;
76 _send_rec($w, $exc ? bless(\$exc, 'PublicInbox::IPC::Die') : $ret);
79 sub ipc_worker_loop ($$$) {
80 my ($self, $r_req, $w_res) = @_;
81 my ($rec, $wantarray, $sub, @args);
83 while ($rec = _get_rec($r_req)) {
84 ($wantarray, $sub, @args) = @$rec;
85 # no waiting if client doesn't care,
86 # this is the overwhelmingly likely case
87 if (!defined($wantarray)) {
88 eval { $self->$sub(@args) };
89 warn "$$ die: $@ (from nowait $sub)\n" if $@;
90 } elsif ($wantarray) {
91 my @ret = eval { $self->$sub(@args) };
92 ipc_return($w_res, \@ret, $@);
93 } else { # '' => wantscalar
94 my $ret = eval { $self->$sub(@args) };
95 ipc_return($w_res, \$ret, $@);
100 # starts a worker if Sereal or Storable is installed
101 sub ipc_worker_spawn {
102 my ($self, $ident, $oldset) = @_;
103 return unless $enc; # no Sereal or Storable
104 return if ($self->{-ipc_ppid} // -1) == $$; # idempotent
105 delete(@$self{qw(-ipc_req -ipc_res -ipc_ppid -ipc_pid)});
106 pipe(my ($r_req, $w_req)) or die "pipe: $!";
107 pipe(my ($r_res, $w_res)) or die "pipe: $!";
108 my $sigset = $oldset // PublicInbox::DS::block_signals();
109 $self->ipc_atfork_prepare;
110 my $seed = rand(0xffffffff);
111 my $pid = fork // die "fork: $!";
114 eval { PublicInbox::DS->Reset };
115 delete @$self{qw(-wq_s1 -wq_workers -wq_ppid)};
116 $w_req = $r_res = undef;
117 $w_res->autoflush(1);
118 $SIG{$_} = 'IGNORE' for (qw(TERM INT QUIT));
120 PublicInbox::DS::sig_setmask($sigset);
121 my $on_destroy = $self->ipc_atfork_child;
122 eval { ipc_worker_loop($self, $r_req, $w_res) };
123 die "worker $ident PID:$$ died: $@\n" if $@;
126 PublicInbox::DS::sig_setmask($sigset) unless $oldset;
127 $r_req = $w_res = undef;
128 $w_req->autoflush(1);
129 $self->{-ipc_req} = $w_req;
130 $self->{-ipc_res} = $r_res;
131 $self->{-ipc_ppid} = $$;
132 $self->{-ipc_pid} = $pid;
135 sub ipc_worker_reap { # dwaitpid callback
136 my ($self, $pid) = @_;
137 # SIGTERM (15) is our default exit signal
138 warn "PID:$pid died with \$?=$?\n" if $? && ($? & 127) != 15;
143 my $pids = delete $self->{"-wq_old_pids.$$"} or return;
144 dwaitpid($_, \&ipc_worker_reap, $self) for @$pids;
147 # for base class, override in sub classes
148 sub ipc_atfork_prepare {}
150 sub ipc_atfork_child {
152 my $io = delete($self->{-ipc_atfork_child_close}) or return;
157 # idempotent, can be called regardless of whether worker is active or not
158 sub ipc_worker_stop {
160 my ($pid, $ppid) = delete(@$self{qw(-ipc_pid -ipc_ppid)});
161 my ($w_req, $r_res) = delete(@$self{qw(-ipc_req -ipc_res)});
162 if (!$w_req && !$r_res) {
163 die "unexpected PID:$pid without IPC pipes" if $pid;
166 die 'no PID with IPC pipes' unless $pid;
167 $w_req = $r_res = undef;
169 return if $$ != $ppid;
170 dwaitpid($pid, \&ipc_worker_reap, $self);
173 # use this if we have multiple readers reading curl or "pigz -dc"
174 # and writing to the same store
177 require PublicInbox::Lock;
178 $self->{-ipc_lock} //= bless { lock_path => $f }, 'PublicInbox::Lock'
181 sub ipc_async_wait ($$) {
182 my ($self, $max) = @_; # max == -1 to wait for all
183 my $aif = $self->{-async_inflight} or return;
184 my $r_res = $self->{-ipc_res} or die 'BUG: no ipc_res';
185 while (my ($sub, $bytes, $cb, $cb_arg) = splice(@$aif, 0, 4)) {
186 my $ret = _get_rec($r_res) //
187 die "no response on $sub (req.size=$bytes)";
188 $self->{-async_inflight_bytes} -= $bytes;
190 eval { $cb->($cb_arg, $ret) };
191 warn "E: $sub callback error: $@\n" if $@;
192 return if --$max == 0;
196 # call $self->$sub(@args), on a worker if ipc_worker_spawn was used
198 my ($self, $sub, @args) = @_;
199 if (my $w_req = $self->{-ipc_req}) { # run in worker
200 my $ipc_lock = $self->{-ipc_lock};
201 my $lock = $ipc_lock ? $ipc_lock->lock_for_scope : undef;
202 if (defined(wantarray)) {
203 my $r_res = $self->{-ipc_res} or die 'BUG: no ipc_res';
204 ipc_async_wait($self, -1);
205 _send_rec($w_req, [ wantarray, $sub, @args ]);
206 my $ret = _get_rec($r_res) // die "no response on $sub";
207 die $$ret if ref($ret) eq 'PublicInbox::IPC::Die';
208 wantarray ? @$ret : $$ret;
209 } else { # likely, fire-and-forget into pipe
210 _send_rec($w_req, [ undef , $sub, @args ]);
212 } else { # run locally
218 my ($self, $sub, $sub_args, $cb, $cb_arg) = @_;
219 if (my $w_req = $self->{-ipc_req}) { # run in worker
220 my $rec = _pack_rec([ 1, $sub, @$sub_args ]);
221 my $cur_bytes = \($self->{-async_inflight_bytes} //= 0);
222 while (($$cur_bytes + length($rec)) > PIPE_BUF) {
223 ipc_async_wait($self, 1);
225 my $ipc_lock = $self->{-ipc_lock};
226 my $lock = $ipc_lock ? $ipc_lock->lock_for_scope : undef;
227 print $w_req $rec or croak "print: $!";
228 $$cur_bytes += length($rec);
229 push @{$self->{-async_inflight}},
230 $sub, length($rec), $cb, $cb_arg;
232 my $ret = [ eval { $self->$sub(@$sub_args) } ];
234 $ret = ( bless(\$exc, 'PublicInbox::IPC::Die') );
236 eval { $cb->($cb_arg, $ret) };
237 warn "E: $sub callback error: $@\n" if $@;
241 # needed when there's multiple IPC workers and the parent forking
242 # causes newer siblings to inherit older siblings sockets
243 sub ipc_sibling_atfork_child {
245 my ($pid, undef) = delete(@$self{qw(-ipc_pid -ipc_ppid)});
246 delete(@$self{qw(-ipc_req -ipc_res)});
247 $pid == $$ and die "BUG: $$ ipc_atfork_child called on itself";
251 my ($self, $s2, $len, $full_stream) = @_;
252 my @fds = $recv_cmd->($s2, my $buf, $len);
253 my $n = length($buf // '') or return;
254 my @m = @{$self->{-wq_recv_modes} // [qw( +<&= >&= >&= )]};
257 my $mode = shift(@m);
258 if (open(my $cmdfh, $mode, $fd)) {
259 $self->{$nfd++} = $cmdfh;
260 $cmdfh->autoflush(1);
262 die "$$ open($mode$fd) (FD:$nfd): $!";
265 while ($full_stream && $n < $len) {
266 my $r = sysread($s2, $buf, $len - $n, $n) // croak "read: $!";
267 croak "read EOF after $n/$len bytes" if $r == 0;
270 # Sereal dies on truncated data, Storable returns undef
271 my $args = thaw($buf) // die "thaw error on buffer of size: $n";
273 my $sub = shift @$args;
274 eval { $self->$sub(@$args) };
275 warn "$$ wq_worker: $@" if $@ && ref($@) ne 'PublicInbox::SIGPIPE';
276 delete @$self{0..($nfd-1)};
280 sub wq_worker_loop ($) {
282 my $len = $self->{wq_req_len} // (4096 * 33);
283 my $s2 = $self->{-wq_s2} // die 'BUG: no -wq_s2';
284 1 while (_recv_and_run($self, $s2, $len));
287 sub do_sock_stream { # via wq_do, for big requests
288 my ($self, $len) = @_;
289 _recv_and_run($self, delete $self->{0}, $len, 1);
292 sub wq_do { # always async
293 my ($self, $sub, $ios, @args) = @_;
294 if (my $s1 = $self->{-wq_s1}) { # run in worker
295 my $fds = [ map { fileno($_) } @$ios ];
296 my $n = $send_cmd->($s1, $fds, freeze([$sub, @args]), MSG_EOR);
297 return if defined($n);
298 croak "sendmsg error: $!" if $! != EMSGSIZE;
299 socketpair(my $r, my $w, AF_UNIX, SOCK_STREAM, 0) or
300 croak "socketpair: $!";
301 my $buf = freeze([$sub, @args]);
302 $n = $send_cmd->($s1, [ fileno($r) ],
303 freeze(['do_sock_stream', length($buf)]),
304 MSG_EOR) // croak "sendmsg: $!";
306 $n = $send_cmd->($w, $fds, $buf, 0) // croak "sendmsg: $!";
307 while ($n < length($buf)) {
308 my $x = syswrite($w, $buf, length($buf) - $n, $n) //
309 croak "syswrite: $!";
310 croak "syswrite wrote 0 bytes" if $x == 0;
314 @$self{0..$#$ios} = @$ios;
315 eval { $self->$sub(@args) };
316 warn "wq_do: $@" if $@ && ref($@) ne 'PublicInbox::SIGPIPE';
317 delete @$self{0..$#$ios}; # don't close
321 sub _wq_worker_start ($$) {
322 my ($self, $oldset) = @_;
323 my $seed = rand(0xffffffff);
324 my $pid = fork // die "fork: $!";
327 eval { PublicInbox::DS->Reset };
328 delete @$self{qw(-wq_s1 -wq_workers -wq_ppid)};
329 $SIG{$_} = 'IGNORE' for (qw(PIPE TTOU TTIN));
330 $SIG{$_} = 'DEFAULT' for (qw(TERM QUIT INT CHLD));
331 local $0 = $self->{-wq_ident};
332 PublicInbox::DS::sig_setmask($oldset);
333 # ensure we properly exit even if warn() dies:
334 my $end = PublicInbox::OnDestroy->new($$, sub { exit(!!$@) });
335 my $on_destroy = $self->ipc_atfork_child;
336 eval { wq_worker_loop($self) };
337 warn "worker $self->{-wq_ident} PID:$$ died: $@" if $@;
339 undef $end; # trigger exit
341 $self->{-wq_workers}->{$pid} = \undef;
345 # starts workqueue workers if Sereal or Storable is installed
346 sub wq_workers_start {
347 my ($self, $ident, $nr_workers, $oldset) = @_;
348 ($enc && $send_cmd && $recv_cmd && defined($SEQPACKET)) or return;
349 return if $self->{-wq_s1}; # idempotent
350 $self->{-wq_s1} = $self->{-wq_s2} = undef;
351 socketpair($self->{-wq_s1}, $self->{-wq_s2}, AF_UNIX, $SEQPACKET, 0) or
352 die "socketpair: $!";
353 $self->ipc_atfork_prepare;
355 $nr_workers = $WQ_MAX_WORKERS if $nr_workers > $WQ_MAX_WORKERS;
356 my $sigset = $oldset // PublicInbox::DS::block_signals();
357 $self->{-wq_workers} = {};
358 $self->{-wq_ident} = $ident;
359 _wq_worker_start($self, $sigset) for (1..$nr_workers);
360 PublicInbox::DS::sig_setmask($sigset) unless $oldset;
361 $self->{-wq_ppid} = $$;
364 sub wq_worker_incr { # SIGTTIN handler
365 my ($self, $oldset) = @_;
366 $self->{-wq_s2} or return;
367 return if wq_workers($self) >= $WQ_MAX_WORKERS;
368 $self->ipc_atfork_prepare;
369 my $sigset = $oldset // PublicInbox::DS::block_signals();
370 _wq_worker_start($self, $sigset);
371 PublicInbox::DS::sig_setmask($sigset) unless $oldset;
374 sub wq_exit { # wakes up wq_worker_decr_wait
375 send($_[0]->{-wq_s2}, $$, MSG_EOR) // die "$$ send: $!";
379 sub wq_worker_decr { # SIGTTOU handler, kills first idle worker
381 return unless wq_workers($self);
382 my $s2 = $self->{-wq_s2} // die 'BUG: no wq_s2';
383 $self->wq_do('wq_exit', [ $s2, $s2, $s2 ]);
384 # caller must call wq_worker_decr_wait in main loop
387 sub wq_worker_decr_wait {
388 my ($self, $timeout) = @_;
389 return if $self->{-wq_ppid} != $$; # can't reap siblings or parents
390 my $s1 = $self->{-wq_s1} // croak 'BUG: no wq_s1';
391 vec(my $rin = '', fileno($s1), 1) = 1;
392 select(my $rout = $rin, undef, undef, $timeout) or
393 croak 'timed out waiting for wq_exit';
394 recv($s1, my $pid, 64, 0) // croak "recv: $!";
395 my $workers = $self->{-wq_workers} // croak 'BUG: no wq_workers';
396 delete $workers->{$pid} // croak "BUG: PID:$pid invalid";
397 dwaitpid($pid, \&ipc_worker_reap, $self);
400 # set or retrieve number of workers
402 my ($self, $nr) = @_;
403 my $cur = $self->{-wq_workers} or return;
405 while (scalar(keys(%$cur)) > $nr) {
406 $self->wq_worker_decr;
407 $self->wq_worker_decr_wait;
409 $self->wq_worker_incr while scalar(keys(%$cur)) < $nr;
415 my ($self, $nohang) = @_;
416 delete @$self{qw(-wq_s1 -wq_s2)} or return;
417 my $ppid = delete $self->{-wq_ppid} or return;
418 my $workers = delete $self->{-wq_workers} // die 'BUG: no wq_workers';
419 return if $ppid != $$; # can't reap siblings or parents
420 my @pids = map { $_ + 0 } keys %$workers;
422 push @{$self->{"-wq_old_pids.$$"}}, @pids;
424 dwaitpid($_, \&ipc_worker_reap, $self) for @pids;
430 my $pids = $self->{"-wq_old_pids.$$"} or return;
435 my ($self, $sig) = @_;
436 my $workers = $self->{-wq_workers} or return;
437 kill($sig // 'TERM', keys %$workers);
440 sub WQ_MAX_WORKERS { $WQ_MAX_WORKERS }
444 my $ppid = $self->{-wq_ppid};
445 wq_kill($self) if $ppid && $ppid == $$;
448 ipc_worker_stop($self);
451 # Sereal doesn't have dclone
452 sub deep_clone { thaw(freeze($_[-1])) }