diff options
Diffstat (limited to 'ext/ExtUtils-Miniperl/lib/ExtUtils/Miniperl.pm')
| -rw-r--r-- | ext/ExtUtils-Miniperl/lib/ExtUtils/Miniperl.pm | 25 |
1 files changed, 23 insertions, 2 deletions
diff --git a/ext/ExtUtils-Miniperl/lib/ExtUtils/Miniperl.pm b/ext/ExtUtils-Miniperl/lib/ExtUtils/Miniperl.pm index 38a37244e2..18627f8a5e 100644 --- a/ext/ExtUtils-Miniperl/lib/ExtUtils/Miniperl.pm +++ b/ext/ExtUtils-Miniperl/lib/ExtUtils/Miniperl.pm @@ -5,7 +5,7 @@ use Exporter 'import'; use ExtUtils::Embed 1.31, qw(xsi_header xsi_protos xsi_body); our @EXPORT = qw(writemain); -our $VERSION = '1.12'; +our $VERSION = '1.13'; # blead will run this with miniperl, hence we can't use autodie or File::Temp my $temp; @@ -135,8 +135,29 @@ main(int argc, char **argv, char **env) PL_perl_destruct_level = 0; } PL_exit_flags |= PERL_EXIT_DESTRUCT_END; - if (!perl_parse(my_perl, xs_init, argc, argv, (char **)NULL)) + if (!perl_parse(my_perl, xs_init, argc, argv, (char **)NULL)) { + + /* perl_parse() may end up starting its own run loops, which + * might end up "leaking" PL_restartop from the parse phase into + * the run phase which then ends up confusing run_body(). This + * leakage shouldn't happen and if it does its a bug. + * + * Note we do not do this assert in perl_run() or perl_parse() + * as there are modules out there which explicitly set + * PL_restartop before calling perl_run() directly from XS code + * (Coro), and it is conceivable PL_restartop could be set prior + * to calling perl_parse() by XS code as well. + * + * What we want to check is that the top level perl_parse(), + * perl_run() pairing does not allow a leaking PL_restartop, as + * that indicates a bug in perl. By putting the assert here we + * can validate that Perl itself is operating correctly without + * risking breakage to XS code under DEBUGGING. - Yves + */ + assert(!PL_restartop); + perl_run(my_perl); + } #ifndef PERL_MICRO /* Unregister our signal handler before destroying my_perl */ |