io: avoid redundant waitpid in DESTROY

We shouldn't attempt to reap a process again after it's been
reaped asynchronously in the SIGCHLD handler.  Noticed while
working on changes to get lei/store to use checkpointing.

1 files changed, 5 insertions, 5 deletions

diff --git a/lib/PublicInbox/IO.pm b/lib/PublicInbox/IO.pm
index 02057600..8640f112 100644
--- a/lib/PublicInbox/IO.pm
+++ b/lib/PublicInbox/IO.pm
@@ -16,7 +16,7 @@ use PublicInbox::OnDestroy;
 
 sub waitcb { # awaitpid callback
         my ($pid, $errref, $cb, @args) = @_;
-        $$errref = $? if $errref; # sets .cerr for _close
+        $$errref = $?; # sets .cerr for _close
         $cb->($pid, @args) if $cb; # may clobber $?
 }
 
@@ -24,9 +24,9 @@ sub attach_pid {
         my ($io, $pid, @cb_arg) = @_;
         bless $io, __PACKAGE__;
         # we share $err (and not $self) with awaitpid to avoid a ref cycle
-        ${*$io}{pi_io_reap} = [ $PublicInbox::OnDestroy::fork_gen,
-                                $pid, \(my $err) ];
-        awaitpid($pid, \&waitcb, \$err, @cb_arg);
+        my $e = \(my $err);
+        ${*$io}{pi_io_reap} = [ $PublicInbox::OnDestroy::fork_gen, $pid, $e ];
+        awaitpid($pid, \&waitcb, $e, @cb_arg);
         $io;
 }
 
@@ -60,7 +60,7 @@ sub DESTROY {
         my $reap = delete ${*$io}{pi_io_reap};
         if (($reap->[0] // -1) == $PublicInbox::OnDestroy::fork_gen) {
                 $io->SUPER::close;
-                awaitpid($reap->[1]);
+                ${$reap->[2]} // awaitpid($reap->[1]);
         }
         $io->SUPER::DESTROY;
 }