On Fri, May 11, 2012 at 04:52:33PM +0100, Dave Mitchell wrote:
> On Fri, May 11, 2012 at 04:22:18PM +0100, Tim Bunce wrote:
> > Dave, I'd be grateful if you could take a look at this since it
> > seems to involve our recent changes and standard perl modules.
>
> I can reproduce. Am looking into it.
now fixed with the attached patch.
Basically creating a thread from within another thread would cause a
crash, because mg_ptr wasn't being updated during CV duplication: this
field contains a pointer back to the CV to allow mg_dup to access it;
after duplication it still pointed back to the original, so duping a dup
caused the crash.
--
Any [programming] language that doesn't occasionally surprise the
novice will pay for it by continually surprising the expert.
-- Larry Wall
Index: DBI.xs
===================================================================
--- DBI.xs (revision 15303)
+++ DBI.xs (working copy)
@@ -283,6 +283,7 @@
CV *ncv = (CV*)ptr_table_fetch(PL_ptr_table, (cv));
(void)param; /* avoid 'unused variable' warning */
+ mg->mg_ptr = (char *)ncv;
ima = (dbi_ima_t*) CvXSUBANY(cv).any_ptr;
Newx(nima, 1, dbi_ima_t);
*nima = *ima; /* structure copy */
Index: t/35thrclone.t
===================================================================
--- t/35thrclone.t (revision 15303)
+++ t/35thrclone.t (working copy)
@@ -52,6 +52,11 @@
cmp_ok($dbh->{Driver}->{Kids}, '==', 1, '... the Driver has one Kid')
unless $DBI::PurePerl && ok(1);
}
+
+ # RT #77137: a thread created from a thread was crashing the
+ # interpreter
+
+ threads->new(sub {})->join();
}
# load up the threads
