Bizzare copy of **** in stack args

[autarch]

Migrated from rt.cpan.org #50447 (status was 'open')

Requestors:

• [email protected]

Attachments:

• bizarre

From [email protected] on 2009-10-13 14:12:36:

Simple (but awful) code, that raise error:

use Devel::StackTrace;

sub do_work {
my $s = Devel::StackTrace->new(
'no_refs' => 1,
);
}

sub call_with_args {
my ($arg_hash, $func) = @_;
$func->( @{$arg_hash->{'args'}} );
}

my $h = {};

my $arg_hash = { 'args' => [ undef ] };

call_with_args(
$arg_hash, sub {
$arg_hash->{'args'} = [];
do_work( sub { $h; } );
}
);

I propose this solution (in StackTrace.pm)

sub _record_caller_data
{
my $self = shift;

# We exclude this method by starting one frame back.
my $x = 1;
while ( my @c =
        do { package DB; @DB::args = (); caller($x++) } )
{
    my @args;

    foreach ( @DB::args )
    {
        my $arg;

        local $@;

        eval 
        {
            $arg = $self->{'no_refs'} && ref $_ ? $self->_ref_to_string( $_ ) : $_;
        };

        if ( $@ =~ /^Bizarre copy of \w+/ )
        {
            $arg = $&;
        }

        push @args, $arg;
    }

    push @{ $self->{raw} },
        { 'caller' => \@c,
          'args'   => \@args,
        };
}

}

[autarch]

From [email protected] (@autarch) on 2010-06-26 19:55:57:

Sorry for the slow reply.

This is really a bug in the Perl core, and should probably be reported
there.

As for a hackaround in Devel::StackTrace, any such thing would need to
not use $& to be useful.

[autarch]

From [email protected] on 2010-10-03 17:22:30:

On Sat Jun 26 15:55:57 2010, DROLSKY wrote:

This is really a bug in the Perl core, and should probably be
reported
there.

I've just submitted a report to [email protected]. I've written a
stripped-down amalgamation of Alexandre's test case and
Devel::StackTrace, which I attach here.

Graeme Hewson

[autarch]

From [email protected] on 2010-10-03 18:03:02:

My report is now at
http://rt.perl.org/rt3/Public/Bug/Display.html?id=78186

[autarch]

From [email protected] on 2010-10-04 17:52:07:

OK, so the core problem is not realistically fixable. It seems a
workaround is the best we could do.

I don't know if this has any relevance, which I happened to notice in
perl5134delta ("what is new for perl v5.13.4"):

"Carp"
Upgraded from version 1.16 to 1.18.

Carp now detects incomplete caller() overrides and avoids using
bogus @DB::args. To provide backtraces, Carp relies on particular
behaviour of the caller built-in. Carp now detects if other code
has overridden this with an incomplete implementation, and modifies
its backtrace accordingly. Previously incomplete overrides would
cause incorrect values in backtraces (best case), or obscure fatal
errors (worst case)

This fixes certain cases of "Bizarre copy of ARRAY" caused by
modules overriding "caller()" incorrectly.

[autarch]

From [email protected] (@SineSwiper) on 2012-04-23 00:56:54:

Getting hit by this, too, via Perl 5.14.2. So, it's not really fixed in
CORE, and it might never be. Since this thing is used for
Exception::Class, I think whatever workaround that can be implemented
should be. We certainly don't want stuff to crash on us DURING the
Exception, before we even get to print it.

[autarch]

From [email protected] (@autarch) on 2012-04-23 01:36:22:

On Sun Apr 22 20:56:54 2012, BBYRD wrote:

Getting hit by this, too, via Perl 5.14.2. So, it's not really fixed
in
CORE, and it might never be. Since this thing is used for
Exception::Class, I think whatever workaround that can be implemented
should be. We certainly don't want stuff to crash on us DURING the
Exception, before we even get to print it.

A good patch would be welcome. Bonus points for stealing the code from
Carp that fixes some of these bugs (but not all, I'm sure).

I think the linked rt.perl ticket might provide a good test case.