|
From: Haizhu L. <hai...@ho...> - 2006-06-06 21:29:05
|
Hello, Can someone explain to me the first stack frames as below, the second to last frame is _dl_init, and then call_init, and then the /opt/hsc/lib/libtrace.so, that means dl_init calls call_init and then call_init calls something in /opt/hsc/lib/libtrace.so? That does not look right since /opt/hsc/lib/libtrace.so is not standard library. >>Example back traces: >> >>==3714== 218,712 bytes in 36 blocks are possibly lost in loss record 26 of >>26 >>==3714== at 0x401C9DC: operator new(unsigned) (in >>/usr/lib/valgrind/x86-linux/vgprelo >>ad_memcheck.so) >>==3714== by 0x4243689: (within /usr/lib/libstdc++.so.5.0.6) >>==3714== by 0x42433C0: std::__default_alloc_template<true, >>0>::_S_chunk_alloc(unsigne >>d, int&) (in /usr/lib/libstdc++.so.5.0.6) >>==3714== by 0x4243200: std::__default_alloc_template<true, >>0>::_S_refill(unsigned) (i >>n /usr/lib/libstdc++.so.5.0.6) >>==3714== by 0x402C0B2: std::_Deque_base<trace_rec_t, >>std::allocator<trace_rec_t> >::_ >>M_initialize_map(unsigned) (bits/stl_alloc.h:349) >>==3714== by 0x402B8C5: __static_initialization_and_destruction_0(int, >>int) (bits/stl_ >>deque.h:430) >>==3714== by 0x402BE1D: _GLOBAL__I_MAKE_DAEMON (bits/stl_queue.h:89) >>==3714== by 0x402F5E4: (within /opt/hsc/lib/libtrace.so) >>==3714== by 0x4028128: (within /opt/hsc/lib/libtrace.so) >>==3714== by 0x400D44B: call_init (in /lib/ld-2.3.3.so) >>==3714== by 0x400D531: _dl_init (in /lib/ld-2.3.3.so) >>==3714== by 0x40009C4: (within /lib/ld-2.3.3.so) >From: Brian Crowder <cr...@fi...> >CC: val...@li... >Subject: Re: [Valgrind-users] can num-callers be larger than 12 with >valgrind 3.1.1 >Date: Thu, 01 Jun 2006 13:36:51 -0700 > > >You're not in main() here, you're calling a CTOR on a globally declared >std::deque, _dl_init IS the very first frame of the stack, there are no >more beyond it. > >-- Brian > >Haizhu Liu wrote: >>Hello, >> >>Example back traces: >> >>==3714== 218,712 bytes in 36 blocks are possibly lost in loss record 26 of >>26 >>==3714== at 0x401C9DC: operator new(unsigned) (in >>/usr/lib/valgrind/x86-linux/vgprelo >>ad_memcheck.so) >>==3714== by 0x4243689: (within /usr/lib/libstdc++.so.5.0.6) >>==3714== by 0x42433C0: std::__default_alloc_template<true, >>0>::_S_chunk_alloc(unsigne >>d, int&) (in /usr/lib/libstdc++.so.5.0.6) >>==3714== by 0x4243200: std::__default_alloc_template<true, >>0>::_S_refill(unsigned) (i >>n /usr/lib/libstdc++.so.5.0.6) >>==3714== by 0x402C0B2: std::_Deque_base<trace_rec_t, >>std::allocator<trace_rec_t> >::_ >>M_initialize_map(unsigned) (bits/stl_alloc.h:349) >>==3714== by 0x402B8C5: __static_initialization_and_destruction_0(int, >>int) (bits/stl_ >>deque.h:430) >>==3714== by 0x402BE1D: _GLOBAL__I_MAKE_DAEMON (bits/stl_queue.h:89) >>==3714== by 0x402F5E4: (within /opt/hsc/lib/libtrace.so) >>==3714== by 0x4028128: (within /opt/hsc/lib/libtrace.so) >>==3714== by 0x400D44B: call_init (in /lib/ld-2.3.3.so) >>==3714== by 0x400D531: _dl_init (in /lib/ld-2.3.3.so) >>==3714== by 0x40009C4: (within /lib/ld-2.3.3.so) >> >>I ran valgrind on interl i686, and I ran valgrind with -v option, >> >>--3714-- Reading syms from /usr/ld-2.3.3.so (0x4000000) >>--3714-- Reading syms from /opt/hsc/bin/ibnmsmd (0x8048000) >>--3714-- Reading syms from /usr/lib/valgrind/x86-linux/memcheck >>(0xB0000000) >>--3714-- object doesn't have a symbol table >>--3714-- object doesn't have a dynamic symbol table >>--3714-- Reading suppressions file: /usr/lib/valgrind/default.supp >>--3714-- REDIR: 0x40129B0 (index) redirected to 0xB0021532 (???) >>--3714-- Reading syms from /usr/lib/valgrind/x86-linux/vgpreload_core.so >>(0x4018000) >>--3714-- object doesn't have a symbol table >>--3714-- Reading syms from >>/usr/lib/valgrind/x86-linux/vgpreload_memcheck.so (0x401A000) >>--3714-- object doesn't have a symbol table >>--3714-- REDIR: 0x4012B50 (strlen) redirected to 0x401CFD0 (strlen) >>--3714-- Reading syms from /opt/hsc/lib/libtrace.so (0x4025000) >>--3714-- Reading syms from /lib/tls/libpthread.so.0 (0x4033000) >>--3714-- Reading syms from /usr/lib/libnetsnmp.so.5.1.0 (0x4043000) >>--3714-- Reading syms from /usr/lib/libcrypto.so.0.9.6 (0x40C7000) >>--3714-- Reading syms from /usr/lib/libstdc++.so.5.0.6 (0x41B8000) >>--3714-- object doesn't have a symbol table >>--3714-- Reading syms from /lib/tls/libm.so.6 (0x4289000) >>--3714-- Reading syms from /lib/tls/libc.so.6 (0x42AB000 >>-3714-- object doesn't have a symbol table >> >>>From: Julian Seward <js...@ac...> >>>To: val...@li... >>>Subject: Re: [Valgrind-users] can num-callers be larger than 12 with >>>valgrind 3.1.1 >>>Date: Thu, 1 Jun 2006 16:39:34 +0100 >>> >>>On Thursday 01 June 2006 16:26, Haizhu Liu wrote: >>> > Hello, >>> > >>> > I tried to use the num-callers option to be larger than 12, but it >>>does not >>> > seem to give me more than that many frames. Is 12 the maximum? >>> >>>No, the max is 50 or 100 or some crazy number like that. >>> >>>Can you send some examples of the backtraces you are getting? >>> >>>What platform (cpu) are you using? >>> >>>If you run with -v do you get any warnings immediately following >>>the lines "Reading syms from ..." ? >>> >>>J >>> >>> >>>------------------------------------------------------- >>>All the advantages of Linux Managed Hosting--Without the Cost and Risk! >>>Fully trained technicians. The highest number of Red Hat certifications >>>in >>>the hosting industry. Fanatical Support. Click to learn more >>>http://sel.as-us.falkag.net/sel?cmd=lnk&kid=107521&bid=248729&dat=121642 >>>_______________________________________________ >>>Valgrind-users mailing list >>>Val...@li... >>>https://lists.sourceforge.net/lists/listinfo/valgrind-users >> >> >> >> >> >>------------------------------------------------------- >>All the advantages of Linux Managed Hosting--Without the Cost and Risk! >>Fully trained technicians. The highest number of Red Hat certifications in >>the hosting industry. Fanatical Support. Click to learn more >>http://sel.as-us.falkag.net/sel?cmd=lnk&kid=107521&bid=248729&dat=121642 >>_______________________________________________ >>Valgrind-users mailing list >>Val...@li... >>https://lists.sourceforge.net/lists/listinfo/valgrind-users >> >> >> > > >------------------------------------------------------- >All the advantages of Linux Managed Hosting--Without the Cost and Risk! >Fully trained technicians. The highest number of Red Hat certifications in >the hosting industry. Fanatical Support. Click to learn more >http://sel.as-us.falkag.net/sel?cmd=lnk&kid=107521&bid=248729&dat=121642 >_______________________________________________ >Valgrind-users mailing list >Val...@li... >https://lists.sourceforge.net/lists/listinfo/valgrind-users |