Re: Problem with TThread::Exists() in current svn rev 41804

Hi Joern,

The problem is fixed in trunk by revision 41815 and in the v5.32 release candidate branch.

Thanks,
Philippe.

On 11/8/11 6:13 AM, Joern Adamczewski-Musch wrote:
> Dear ROOTers,
>
> since most recent changes in root,
> my application crashes when it tries to use static method
> TThread::Exists():
>
>
> Program received signal SIGSEGV, Segmentation fault.
> 0x00007ffff3c447f7 in ThreadInternalLock() () from
> /usr/local/root/head/lib/libThread.so
> (gdb) bt
> #0 0x00007ffff3c447f7 in ThreadInternalLock() () from
> /usr/local/root/head/lib/libThread.so
> #1 0x00007ffff3c448e8 in TThread::Exists() () from
> /usr/local/root/head/lib/libThread.so
> #2 0x00007ffff2a1dd92 in TGo4LockGuard::TGo4LockGuard
> (this=0x7ffffffe42a0, mutex=0xc91f90,
> forcelock=false) at Go4LockGuard/TGo4LockGuard.cxx:44
> #3 0x00007ffff2a2d175 in TGo4CommandProtoList::AddCommand
> (this=0xc9ca10, com=0xc9cbf0)
> at Go4CommandsBase/TGo4CommandProtoList.cxx:111
> #4 ....
>
> Here I used revision 41804, gcc version 4.4.1. (x86_64)
> The same crash was encountered also on ubuntu 10.11 with gcc 4.6.1. (i686)
>
> Probably this is due to some changes concerning the ROOT library
> linkage, which was introduced last week to cover problems with gcc
> 4.6.1? I guess that maybe some static objects are not yet initialized
> when ThreadInternalLock is executed.
>
> Currently I can do some workaround in my code,
> but this may indicate some deeper problem in ROOT.
>
> Best regards,
>
> Joern
>
Received on Tue Nov 08 2011 - 19:15:15 CET