On Fri, Feb 10, 2012 at 1:55 PM, Michael G Schwern via RT
<bug-Test-Simple@rt.cpan.org> wrote:
Show quoted text> <URL:
https://rt.cpan.org/Ticket/Display.html?id=74858 >
>
> On 2012.2.10 10:16 AM, Reini Urban via RT wrote:
>> This is really a CORE bug with threads, but I keep it here until we find
>> the real cause.
>> ribasushi found the leak in Devel::PeekPoke, I limited it down to
>> Test::More.
>>
>> valgrind --leak-check=full /usr/local/bin/perl5.14.2d -MTest::More -e0
>>
>> ==26633== LEAK SUMMARY:
>> ==26633== definitely lost: 7,760 bytes in 182 blocks
>> ==26633== indirectly lost: 726,750 bytes in 15,247 blocks
>>
>> The more tests, the more leaks.
>> -Uuseithreads does not leak.
>>
>> Looks like some cloned OPs have the wrong refcnt or are not properly
>> perl_destruct'ed. valgrind only reports OPs to leak.
>
> If I'm reading that right, there are leaks just loading Test::More on a
> threaded Perl? Not even running tests nor loading threads.pm?
right.
The leaks are observable with
-MTest::More -e0 but grow larger the more tests you run.
This is really a bug for CORE, I just wanted to a ticket here also, and
refer to it in the CORE perlbug.
Should be easy to find with this info, but too hard for me. I'm not
familiar with perl_clone.
Something for Nick, Dave or Zefram.
--
Reini Urban
http://cpanel.net/ http://www.perl-compiler.org/