Bug #80400 for autobox: Segmentation fault with END block

Thu Oct 25 20:16:04 2012 tokuhirom+cpan [...] gmail.com - Ticket created

Subject:

Segmentation fault with END block

This code cause SEGV. use strict; use warnings; use utf8; { package M; my $X; sub register { $X = $_[0] } END { $X->() } } use autobox; M::register(sub { 1->OMG(); });

Thu Oct 25 21:00:39 2012 CHOCOLATE [...] cpan.org - Taken

Thu Oct 25 21:02:21 2012 CHOCOLATE [...] cpan.org - Correspondence added

Thanks for reporting this. Simpler test case: my $X; END { $X->() } use autobox; $X = sub { 1->OMG() };

Thu Oct 25 21:02:22 2012 The RT System itself - Status changed from 'new' to 'open'

Fri Feb 01 04:41:38 2013 tokuhirom+cpan [...] gmail.com - Correspondence added

I guess there is no reason to free resources at END state. Because OS will free memory at process end. Here is a propsed patch. index 7506b08..9456ecd 100644 --- a/autobox.xs +++ b/autobox.xs @@ -343,8 +343,8 @@ END() PL_check[OP_ENTERSUB] = autobox_old_ck_subr; } - PTABLE_free(AUTOBOX_OP_MAP); - AUTOBOX_OP_MAP = NULL; + /* PTABLE_free(AUTOBOX_OP_MAP); */ + /* AUTOBOX_OP_MAP = NULL; */ MODULE = autobox PACKAGE = autobox::universal

Tue Apr 30 03:09:44 2013 tokuhirom+cpan [...] gmail.com - Correspondence added

How do you think about this patch? On Fri Feb 01 04:41:38 2013, TOKUHIROM wrote: Show quoted text

> I guess there is no reason to free resources at END state. Because OS > will free memory at process > end. > > Here is a propsed patch. > > index 7506b08..9456ecd 100644 > --- a/autobox.xs > +++ b/autobox.xs > @@ -343,8 +343,8 @@ END() > PL_check[OP_ENTERSUB] = autobox_old_ck_subr; > } > > - PTABLE_free(AUTOBOX_OP_MAP); > - AUTOBOX_OP_MAP = NULL; > + /* PTABLE_free(AUTOBOX_OP_MAP); */ > + /* AUTOBOX_OP_MAP = NULL; */ > > MODULE = autobox PACKAGE = autobox::universal

Tue Apr 30 07:26:18 2013 CHOCOLATE [...] cpan.org - Correspondence added

Thanks for the patch and sorry for not replying. Obviously it's better than a segfault, but "let the OS take care of it" feels like a workaround rather than a fix. Cleaning up an allocated resource is not something to be ashamed of :-) Ideally, I'd rather know why the segfault occurs there e.g. is there another/better/safer (later?) hook for destructors/cleanup? I've spotted ancient, unfixed (but reported) bugs in perl's special-block handling in the past, so admittedly this may be the only solution. On Tue Apr 30 03:09:44 2013, TOKUHIROM wrote: Show quoted text

> How do you think about this patch?

Tue Apr 30 14:34:10 2013 CHOCOLATE [...] cpan.org - Correspondence added

This should be fixed in 2.78. Thanks again.

Tue Apr 30 15:40:22 2013 CHOCOLATE [...] cpan.org - Severity Normal added

Tue Apr 30 15:40:22 2013 CHOCOLATE [...] cpan.org - Status changed from 'open' to 'resolved'

Tue Apr 30 19:59:21 2013 CHOCOLATE [...] cpan.org - Fixed in 2.78 added

Bug #80400 for autobox: Segmentation fault with END block

Preferred bug tracker