2
0
mirror of https://github.com/xcat2/xcat-core.git synced 2026-11-20 09:22:17 +00:00

fix(xcat-core): --nproc=0 exits 127 before any package work

Parallel::ForkManager does not fork when max_procs is 0: start()
returns 0 in the original process. start_build_worker took that 0 as a
forked worker. It closed the only writer of the gate pipe, read
end-of-file and exited 127, so both worker loops in buildrpms.pl
stopped before they built a package or updated a repository.

start_build_worker now compares $$ before and after the fork callback.
When the callback returned 0 in the same process, it closes the gate and
returns 0, and the caller runs the work under the locks it already
holds. Both loops call start_build_worker, so both are covered.
build_worker_nofork.t passes with this change.

Signed-off-by: Daniel Hilst <392820+dhilst@users.noreply.github.com>
This commit is contained in:
Daniel Hilst
2026-11-06 16:26:07 -03:00
parent 54e8357588
commit bdebb7d4cf
+11 -1
View File
@@ -765,6 +765,8 @@ Returns:
failed and the worker exits with status 127, and () when the fork failed. In scalar
context: the pid, or undef.
In the child: 0, after the parent released it. The child never returns otherwise.
When $fork returns 0 without forking, as Parallel::ForkManager does with max_procs 0: 0
(list context: (0, 1)) in the calling process, which then runs the work itself.
=cut
@@ -774,7 +776,8 @@ sub start_build_worker {
require POSIX;
$fork //= sub { fork() };
pipe(my $gate_r, my $gate_w) or return;
my $pid = $fork->();
my $self = $$;
my $pid = $fork->();
unless (defined $pid) {
my $error = $!;
close($gate_r);
@@ -782,6 +785,13 @@ sub start_build_worker {
$! = $error;
return;
}
# Parallel::ForkManager->new(0) does not fork, and start() returns 0 in this process. The
# work then runs here, under the locks this process already holds.
if (!$pid && $$ == $self) {
close($gate_r);
close($gate_w);
return wantarray ? (0, 1) : 0;
}
unless ($pid) {
close($gate_w);
POSIX::setpgid(0, 0) or POSIX::_exit(127);