Re: [Freeipmi-users] memory leak on ipmi_monitoring_sensor_readings_by_record_id?

[Albert Chu]

I believe I found a few mem-leaks, one on an error path that you're
unlikely to hit.  Another more generic/possible mem-leak, which is likely
the one you're hitting.

I can definitely patch upstream and FreeIPMI 1.5.X, but FreeIPMI 1.2.9 is
well past support.  Can you use 1.5.X?

Al

On Mon, Jul 17, 2017 at 2:40 PM, Albert Chu <address@hidden> wrote:

> Hi,
>
> Skimming through code, it seems I may not be freeing memory after the
> call to ipmi_get_event_messages().  I'll have to look at this a bit more
> carefully, but I think you're right.
>
> Al
>
> On Mon, 2017-07-17 at 09:00 +0000, Yun Sangwon wrote:
> > Hello,
> >
> > I'm using the ipmimonitoring library of freeipmi 1.2.9-8.el7.x86_64 on
> RHEL 7.3 to monitor values of sensors using IPMI in a program.
> > For continued monitoring, the program calls the API
> ipmi_monitoring_sensor_readings_by_record_id() every period (in my case
> 10 seconds).
> >
> > It works fine but for me there is a memory leak issue.
> >
> > Through some tests, I think the issue is being happened on the API
> ipmi_monitoring_sensor_readings_by_record_id().
> > For the information, valgrind report is same as below:
> >
> > $ valgrind --trace-children=yes --leak-check=full
> --log-file=/tmp/valgrind-ipmimon.txt ./ipmimonitoring-sensors
> >
> > $ cat /tmp/valgrind-ipmimon.txt
> > ==10568== Memcheck, a memory error detector
> > ==10568== Copyright (C) 2002-2015, and GNU GPL'd, by Julian Seward et al.
> > ==10568== Using Valgrind-3.11.0 and LibVEX; rerun with -h for copyright
> info
> > ==10568== Command: ./ipmimonitoring-sensors
> > ==10568== Parent PID: 25490
> > ==10568==
> > ==10568==
> > ==10568== Process terminating with default action of signal 2 (SIGINT)
> > ==10568==    at 0x592F650: __nanosleep_nocancel (in /usr/lib64/
> libc-2.17.so)
> > ==10568==    by 0x592F503: sleep (in /usr/lib64/libc-2.17.so)
> > ==10568==    by 0x401F9F: main (in /home/test/ipmimonitoring-sensors)
> > ==10568==
> > ==10568== HEAP SUMMARY:
> > ==10568==     in use at exit: 3,060 bytes in 180 blocks
> > ==10568==   total heap usage: 125,321 allocs, 125,141 frees, 264,313,716
> bytes allocated
> > ==10568==
> > ==10568== 3,060 (1,440 direct, 1,620 indirect) bytes in 90 blocks are
> definitely lost in loss record 2 of 2
> > ==10568==    at 0x4C28BE3: malloc (vg_replace_malloc.c:299)
> > ==10568==    by 0x5DC25F1: ipmi_get_event_messages (in
> /usr/lib64/libfreeipmi.so.13.1.7)
> > ==10568==    by 0x4E3F704: ??? (in /usr/lib64/libipmimonitoring.
> so.5.0.5)
> > ==10568==    by 0x4E41626: ??? (in /usr/lib64/libipmimonitoring.
> so.5.0.5)
> > ==10568==    by 0x4E3904F: ??? (in /usr/lib64/libipmimonitoring.
> so.5.0.5)
> > ==10568==    by 0x5DA44B9: ipmi_sdr_cache_iterate (in
> /usr/lib64/libfreeipmi.so.13.1.7)
> > ==10568==    by 0x4E3A853: ??? (in /usr/lib64/libipmimonitoring.
> so.5.0.5)
> > ==10568==    by 0x4E3AA3B: ipmi_monitoring_sensor_readings_by_record_id
> (in /usr/lib64/libipmimonitoring.so.5.0.5)
> > ==10568==    by 0x40169F: _ipmimonitoring (in /home/test/ipmimonitoring-
> sensors)
> > ==10568==    by 0x401F82: main (in /home/test/ipmimonitoring-sensors)
> > ==10568==
> > ==10568== LEAK SUMMARY:
> > ==10568==    definitely lost: 1,440 bytes in 90 blocks
> > ==10568==    indirectly lost: 1,620 bytes in 90 blocks
> > ==10568==      possibly lost: 0 bytes in 0 blocks
> > ==10568==    still reachable: 0 bytes in 0 blocks
> > ==10568==         suppressed: 0 bytes in 0 blocks
> > ==10568==
> > ==10568== For counts of detected and suppressed errors, rerun with: -v
> > ==10568== ERROR SUMMARY: 1 errors from 1 contexts (suppressed: 0 from 0)
> >
> > For the information, it is same in case of using the example
> http://www.gnu.org/software/freeipmi/ipmimonitoring-sensors.c.
> >
> > I don't know whether it is my fault or freeipmi issue.
> > For the moment I didn't test using latest version as I want to use the
> freeipmi version 1.2.9-8 which is bundled on RHEL v7.3.
> >
> > Could you please guide or share any information?
> >
> > Thank you very much in advance.
> > Sangwon
> > _______________________________________________
> > Freeipmi-users mailing list
> > address@hidden
> > https://lists.gnu.org/mailman/listinfo/freeipmi-users
>
> --
> Albert Chu
> address@hidden
> Computer Scientist
> High Performance Systems Division
> Lawrence Livermore National Laboratory
>
>
>
> _______________________________________________
> Freeipmi-users mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/freeipmi-users
>