Bizarre copy of ARRAY in sassign at Carp/Heavy.pm #9282
Comments
From chris@heathens.co.nzThis is a bug report for perl from chris@heathens.co.nz, The following test program gives this error: Bizarre copy of ARRAY in sassign at /usr/lib/perl5/5.8.8/Carp/Heavy.pm line 45. Tested on the 5.8.x branch and on perl-current. use Carp; sub do_carp { sub call_with_args { my $h = {}; Flags: This perlbug was built using Perl v5.8.8 in the Red Hat build system. Site configuration information for perl v5.8.8: Configured by Red Hat, Inc. at Thu Jan 31 13:32:29 EST 2008. Summary of my perl5 (revision 5 version 8 subversion 8) configuration: Locally applied patches: @INC for perl v5.8.8: Environment for perl v5.8.8: |
From @nwc10On Tue, Apr 08, 2008 at 11:39:25AM -0700, Chris Heath wrote:
Thanks for this bug report. The error has been seen before, but not with this $ cat C.pm use Devel::Peek; Dump \@DB::args; foreach(@DB::args) { 1; sub do_carp { sub call_with_args { my $h = {}; [the line with Devel::Peek is optional] It looks like the address of something on Perl's stack ends up being re-used Nicholas Clark |
|
The RT System itself - Status changed from 'new' to 'open' |
From p5p@spam.wizbit.be
'# Deleting the undef makes it all work again!' [...]
Reduced the test case a bit more: Test case 1: Bizarre copy of ARRAY #!/usr/bin/perl my $args = [ "AAA" ]; $args = [ "BBB" ]; sub s2 { s1(@$args); Expected output: 'ARG: AAA' $ miniperl-5.000 rt-52610.pl $ miniperl-5.001 rt-52610.pl $ miniperl-perl-5.001n rt-52610.pl $ miniperl-5.002beta1 rt-52610.pl $ perl-5.8.8 rt-52610.pl => perl-5.000 outputs 'ARG: AAA' because closures were not yet Test case 2: copy $args before changing it #!/usr/bin/perl my $args = [ "AAA" ]; my $t = $args; sub s2 { s1(@$args); __END__ $ perl-5.8.8 rt-52610.pl Test case 3: Remove $h and $z #!/usr/bin/perl my $args = [ "AAA" ]; $args = [ "CCC" ]; sub s2 { s1(@$args); $ perl-5.8.0 rt-52610.pl $ perl-5.8.1 rt-52610.pl $ perl-5.8.8 rt-52610.pl Test case 4: use three hashes instead of a hash and a subroutine #!/usr/bin/perl my $args = [ "AAA" ]; $args = [ "BBB" ]; my $h = {}; s2(); sub s2 { s1(@$args); $ perl-5.8.8 rt-52610.pl Test case 5: Use: @foo2 = @DB::args #!/usr/bin/perl my $args = [ "AAA" ]; $args = [ "BBB" ]; sub s2 { s1(@$args); $ perl-5.8.6 rt-52610.pl $ perl-5.8.7 rt-52610.pl $ perl-5.8.8 rt-52610.pl Kind regards, Bram |
From p5p@spam.wizbit.beAnother test case: #!/usr/bin/perl my @args = 1 .. 10; sub s2 { $ perl-5.8.8 rt-52610.pl What happens is that the array is placed on the stack, then sub s1 is Then caller() access the stack and (I assume!) stores/aliasses all Kind regards, Bram |
1 similar comment
From p5p@spam.wizbit.beAnother test case: #!/usr/bin/perl my @args = 1 .. 10; sub s2 { $ perl-5.8.8 rt-52610.pl What happens is that the array is placed on the stack, then sub s1 is Then caller() access the stack and (I assume!) stores/aliasses all Kind regards, Bram |
From p5p@spam.wizbit.beOne more thought: caller and @DB::args aren't really nessesary to show #!/usr/bin/perl -l my @args = 1 .. 10; If nothing in sub s1() access @_ then this is a 'silent' bug. The real fix is to make the stack do ref counting.. This information would be more useful and allows for the problem to be Any thoughts? Kind regards, Bram |
From nadim@khemir.netCan also generate these error message: Bizarre copy of ARRAY in sassign at panic: attempt to copy freed scalar 1ad5ac0 to 1ccee98 at Bizarre copy of HASH in sassign at as reported in https://rt.cpan.org/Ticket/Display.html?id=55179 |
From rabbit@thesaurusCreated by rabbit@thesaurus.(none)I finally managed to reroduce the Test:: related failures that for pv in $(perlbrew installed | grep '^5') ; do perlbrew switch $pv ; perl perlfail.t &> $pv.log ; done Cheers! Perl Info |
From @ribasushiTest and logs |
From @ribasushi |
From @ribasushi |
From @ribasushi |
From @ribasushi#!/usr/bin/perl ############################################################### use strict; { use Carp; sub diaf { shift->d ('Croaking croaker croaked') } sub wiaf { shift->w ( 'Carping carper carped') } use Test::More qw/no_plan/; my $max = 8; for (1 .. $max) { # test without quoting labels warning_like ( for (1 .. $max) { # Just for debugging purposes, removing it doesn't END { |
From @ribasushi |
From @ribasushi |
From @ribasushiMore fail. I aplogize for the duplication, didn't realize there was an http://rt.perl.org/rt3/Ticket/Display.html?id=76496 |
From @ribasushiAnd of course it turns out bug is real and reported for quite some time. http://rt.perl.org/rt3/Ticket/Display.html?id=52610 |
|
The RT System itself - Status changed from 'new' to 'open' |
From @ribasushiOn Tue Jul 13 20:10:42 2010, ribasushi wrote:
Hmmm... after reading the explanations on all previous tickets, I |
From @ribasushi |
From @ribasushi |
From @ribasushi |
From @ribasushi |
From @ribasushi#!/usr/bin/perl use strict; { use Carp; sub diaf { croak ('Croaking croaker croaked') } use Test::More qw/no_plan/; for my $x (1 .. 5) { throws_ok (sub { Tst::Croaker->diaf }, qr/croaked/, "throws exception"); # Just for log-tagging purposes, removing it doesn't END { |
From @ribasushi |
From @nwc10On Fri, Jul 16, 2010 at 12:32:02AM -0700, rabbit+bugs@rabbit.us via RT wrote:
It's a shame that the tickets were merged (but I would have done it too). Bug 76496 is actually a bug in Test::Exception. Could you report it there, with this patch to fix it: Inline Patch--- lib/Test/Exception.pm~ 2010-01-12 06:31:23.000000000 +0000
+++ lib/Test/Exception.pm 2010-07-17 11:41:50.000000000 +0100
@@ -85,11 +85,23 @@
sub _quiet_caller (;$) { ## no critic Prototypes
my $height = $_[0];
$height++;
- if( wantarray and !@_ ) {
- return (CORE::caller($height))[0..2];
+ if ( CORE::caller() eq 'DB' ) {
+ # passthrough the @DB::args trick
+ package DB;
+ if( wantarray and !@_ ) {
+ return (CORE::caller($height))[0..2];
+ }
+ else {
+ return CORE::caller($height);
+ }
}
else {
- return CORE::caller($height);
+ if( wantarray and !@_ ) {
+ return (CORE::caller($height))[0..2];
+ }
+ else {
+ return CORE::caller($height);
+ }
}
}
Result - @DB::args is stale - it's referring to the state of @_ from some for $x (1 .. 1) { [which has been subsequently re-used for the string "Regexp" inside qr//, and I suspect that the core docs for caller need to be updated to warn that if you I have a feeling of déjà vu here. Did some other module that overrides caller Nicholas Clark |
From @xdgOn Sat, Jul 17, 2010 at 6:51 AM, Nicholas Clark <nick@ccl4.org> wrote:
Sub::Uplevel implemented support for DB about two years ago. -- David |
From @ribasushiOn Sat Jul 17 03:51:32 2010, nicholas wrote:
Nicholas++ Bug reported https://rt.cpan.org/Ticket/Display.html?id=55179#txn-805481 I also looked at Test::Warn but it looks *much* simpler and does nto Thank you all! |
From @ribasushiOn Sat Jul 17 03:51:32 2010, nicholas wrote:
Just for completeness sake I'm announcing that the latest |
From @autarchThis no longer manifests in the same way. The attached code now ends up This is somewhat of an improvement, perhaps. |
From douglasg.wilson@gmail.comOn Tue Apr 15 02:53:34 2008, nicholas wrote:
Please, please can we get Carp patched with: @args = map { Then at least Carp will report useful errors. Carp should be |
Migrated from rt.perl.org#52610 (status was 'open')
Searchable as RT52610$
The text was updated successfully, but these errors were encountered: