#802778 False positive mem leak

Package:
valgrind
Source:
valgrind
Description:
instrumentation framework for building dynamic analysis tools
Submitter:
Mathieu Malaterre
Date:
2025-11-17 22:47:01 UTC
Severity:
normal
Tags:
#802778#5
Date:
2015-10-23 13:25:53 UTC
From:
To:
Seems like gcc 5 is doing something funky
(/usr/lib/x86_64-linux-gnu/libstdc++.so.6.0.21):

==3674==
==3674== HEAP SUMMARY:
==3674==     in use at exit: 72,704 bytes in 1 blocks
==3674==   total heap usage: 306,736 allocs, 306,735 frees, 22,184,464
bytes allocated
==3674==
==3674== 72,704 bytes in 1 blocks are still reachable in loss record 1 of 1
==3674==    at 0x4C28C4F: malloc (vg_replace_malloc.c:299)
==3674==    by 0x72ED11F: pool (eh_alloc.cc:117)
==3674==    by 0x72ED11F: __static_initialization_and_destruction_0
(eh_alloc.cc:244)
==3674==    by 0x72ED11F: _GLOBAL__sub_I_eh_alloc.cc (eh_alloc.cc:307)
==3674==    by 0x400EA09: call_init.part.0 (dl-init.c:78)
==3674==    by 0x400EAF2: call_init (dl-init.c:36)
==3674==    by 0x400EAF2: _dl_init (dl-init.c:126)
==3674==    by 0x40011C9: ??? (in /lib/x86_64-linux-gnu/ld-2.19.so)
==3674==    by 0x1: ???
==3674==    by 0xFFF00049A: ???
==3674==    by 0xFFF0004A9: ???
==3674==
==3674== LEAK SUMMARY:
==3674==    definitely lost: 0 bytes in 0 blocks
==3674==    indirectly lost: 0 bytes in 0 blocks
==3674==      possibly lost: 0 bytes in 0 blocks
==3674==    still reachable: 72,704 bytes in 1 blocks
==3674==         suppressed: 0 bytes in 0 blocks

I do have the dbg library:

$ apt-cache policy libstdc++6
libstdc++6:
  Installed: 5.2.1-22
  Candidate: 5.2.1-22
  Version table:
 *** 5.2.1-22 0
        500 http://ftp.fr.debian.org/debian/ stretch/main amd64 Packages
        100 /var/lib/dpkg/status
#802778#10
Date:
2016-01-09 14:12:58 UTC
From:
To:
This is a false positive and should be suppressed. It may take me some time to
look into this though, so if you want to provide a patch (to the debian.supp
file) feel free to do so.

Thanks
#802778#15
Date:
2025-11-17 12:19:41 UTC
From:
To:
On Sat, 9 Jan 2016 14:12:58 +0000 Alessandro Ghedini <ghedo@debian.org>
wrote:
 > On Fri, Oct 23, 2015 at 03:25:53PM +0200, Mathieu Malaterre wrote:
 > > Package: valgrind
 > > Version: 1:3.11.0-1
 > > Tags: upstream
 > >
 > > Seems like gcc 5 is doing something funky
 > > (/usr/lib/x86_64-linux-gnu/libstdc++.so.6.0.21):
 > >
 > > ==3674==
 > > ==3674== HEAP SUMMARY:
 > > ==3674== in use at exit: 72,704 bytes in 1 blocks
 > > ==3674== total heap usage: 306,736 allocs, 306,735 frees, 22,184,464
 > > bytes allocated
 > > ==3674==
 > > ==3674== 72,704 bytes in 1 blocks are still reachable in loss
record 1 of 1
 > > ==3674== at 0x4C28C4F: malloc (vg_replace_malloc.c:299)
 > > ==3674== by 0x72ED11F: pool (eh_alloc.cc:117)
 > > ==3674== by 0x72ED11F: __static_initialization_and_destruction_0
 > > (eh_alloc.cc:244)
 > > ==3674== by 0x72ED11F: _GLOBAL__sub_I_eh_alloc.cc (eh_alloc.cc:307)
 > > ==3674== by 0x400EA09: call_init.part.0 (dl-init.c:78)
 > > ==3674== by 0x400EAF2: call_init (dl-init.c:36)
 > > ==3674== by 0x400EAF2: _dl_init (dl-init.c:126)
 > > ==3674== by 0x40011C9: ??? (in /lib/x86_64-linux-gnu/ld-2.19.so)
 > > ==3674== by 0x1: ???
 > > ==3674== by 0xFFF00049A: ???
 > > ==3674== by 0xFFF0004A9: ???
 > > ==3674==
 > > ==3674== LEAK SUMMARY:
 > > ==3674== definitely lost: 0 bytes in 0 blocks
 > > ==3674== indirectly lost: 0 bytes in 0 blocks
 > > ==3674== possibly lost: 0 bytes in 0 blocks
 > > ==3674== still reachable: 72,704 bytes in 1 blocks
 > > ==3674== suppressed: 0 bytes in 0 blocks
 >
 > This is a false positive and should be suppressed. It may take me
some time to
 > look into this though, so if you want to provide a patch (to the
debian.supp
 > file) feel free to do so.


1. This never was a false positive (Valgrind neither imagined an
inexistent call to malloc nor missed an existing call to free).

2. It also should not be suppressed.

3. This was fixed decades ago. 2002 according to git. Valgrind should
call __libc_freeres() so that memory like this gets freed.


A+

Paul