Re: valgrind suppressions for 64bit for ROOT 5.22/00a?

From: Axel Naumann <Axel.Naumann_at_cern.ch>
Date: Sat, 6 Jun 2009 19:05:13 +0200


Hi,

the output of
nm -C -S $ROOTSYS/lib/lib{Core,Cint}.so would allow us to determine which symbols the addresses correspond to.

Cheers, Axel.

On 2009-06-06 00:27, Philippe Canal wrote:
> Hi Pete,
>
> I can not reproduce those problems.
>
> It is a bit odd to have so little information with a debug build.
> I don't remember if valgrind 3.4.1 already has this feature, but newer
> version of valgrind
> have a new option:
> --track-origins=yes
> that tracks down the 'origin' of the uninitialized value (in addition to
> the attempted use).
>
> Cheers,
> Philippe.
>
> Peter Elmer wrote:

>>   Hi,
>>
>>   When running with valgrind (memcheck) with ROOT 5.22/00a for
>> gcc4.3/64bit,
>> on both SLC4 and SLC5, I see a tremendous number of errors (>1k) like
>> those
>> in (a) below, with varying depths of the nesting of libCint/libCore
>> calls. It is sufficient to simply dynamically load the libraries to
>> see the errors.
>>
>>   Currently these aren't suppressed by the suppression file shipped with
>> the ROOT sources. Are they known and/or understood?
>>
>>   [The traceback I got below is unfortunately already _after_ building
>> with
>> --build=debug, so I don't have any additional information as to where to
>> look. I had to use valgrind 3.4.1, too, as earlier versions had problems
>> providing any information at all about the stacktrace except the
>> "strlen".]
>>
>>                                  thanks,
>>                                    Pete
>>
>> (a)
>>
>> ==23974== Conditional jump or move depends on uninitialised value(s)
>> ==23974==    at 0x3897A78D8D: strlen (in /lib64/libc-2.5.so)
>> ==23974==    by 0x496180F: (within
>> /build/364/slc5_amd64_gcc432/lcg/root/5.22.00a-cms2/lib/libCint.so)
>> ==23974==    by 0x49F4BE4: (within
>> /build/364/slc5_amd64_gcc432/lcg/root/5.22.00a-cms2/lib/libCint.so)
>> ==23974==    by 0x495E5B6: (within
>> /build/364/slc5_amd64_gcc432/lcg/root/5.22.00a-cms2/lib/libCint.so)
>> ==23974==    by 0x590446E: (within
>> /build/364/slc5_amd64_gcc432/lcg/root/5.22.00a-cms2/lib/libCore.so)
>> ==23974==    by 0x5901E7F: (within
>> /build/364/slc5_amd64_gcc432/lcg/root/5.22.00a-cms2/lib/libCore.so)
>> ==23974==    by 0x5901EB1: (within
>> /build/364/slc5_amd64_gcc432/lcg/root/5.22.00a-cms2/lib/libCore.so)
>> ==23974==    by 0x59044D5: (within
>> /build/364/slc5_amd64_gcc432/lcg/root/5.22.00a-cms2/lib/libCore.so)
>> ==23974==    by 0x5455BD2: (within
>> /build/364/slc5_amd64_gcc432/lcg/root/5.22.00a-cms2/lib/libCore.so)
>> ==23974==    by 0x5C2BFFF: (within
>> /build/364/slc5_amd64_gcc432/external/gcc/4.3.2-cms/lib64/libgcc_s.so.1)
>> ==23974==    by 0x389760D22A: call_init (in /lib64/ld-2.5.so)
>> ==23974==    by 0x389760D334: _dl_init (in /lib64/ld-2.5.so)
>> ==23974==    by 0x3897600AA9: (within /lib64/ld-2.5.so)
>>
>>
>> -------------------------------------------------------------------------
>> Peter Elmer     E-mail: Peter.Elmer_at_cern.ch      Phone: +41 (22) 767-4644
>> Address: CERN Division PPE, Bat. 32 2C-14, CH-1211 Geneva 23, Switzerland
>> -------------------------------------------------------------------------
>>
>>   

>
>
Received on Sat Jun 06 2009 - 19:05:46 CEST

This archive was generated by hypermail 2.2.0 : Sat Jun 06 2009 - 23:50:01 CEST