Re: [Libunwind-devel] Backtrace performance on x86-64 based on Dwarf info

[Milian Wolff]

On Monday 02 June 2014 09:24:21 Arun Sharma wrote:
> On Mon, Jun 2, 2014 at 3:23 AM, Milian Wolff <address@hidden> wrote:
> > Looking at the backtrace for the return values I go
> > ...
> > 
> > 0x400fff main in /home/milian/projects/kde4/mallocinfo/tests/test.cpp:66
> > /home/milian/projects/.build/kde4/mallocinfo/tests/test_cpp
> > 
> > 0x7f0b689a1000 __libc_start_main /usr/lib/libc.so.6
> > 
> > 0x400d29 ? /home/milian/projects/.build/kde4/mallocinfo/tests/test_cpp
> > 
> > 0x0 <unknown module>
> 
> On a system with recent glibc, you the unwind should terminate around
> __libc_start_main because of RBP=0.
> http://koala.cs.pub.ro/lxr/#glibc+2.9/sysdeps/x86_64/elf/start.S#L65
> 
> What version of glibc are you using?

I'm using glibc 2.19-5 as supplied by the ArchLinux repositories. Where in the 
Doing a objdump -D on a test executable yields:

Disassembly of section .text:

000000000043e110 <_start>:
  43e110:       31 ed                   xor    %ebp,%ebp
  43e112:       49 89 d1                mov    %rdx,%r9
  43e115:       5e                      pop    %rsi
  43e116:       48 89 e2                mov    %rsp,%rdx
  43e119:       48 83 e4 f0             and    $0xfffffffffffffff0,%rsp
  43e11d:       50                      push   %rax
  43e11e:       54                      push   %rsp
  43e11f:       49 c7 c0 60 97 44 00    mov    $0x449760,%r8
  43e126:       48 c7 c1 f0 96 44 00    mov    $0x4496f0,%rcx
  43e12d:       48 c7 c7 9d e4 43 00    mov    $0x43e49d,%rdi
  43e134:       e8 87 fd ff ff          callq  43dec0 <address@hidden>
  43e139:       f4                      hlt    
  43e13a:       00 00                   add    %al,(%rax)
  43e13c:       00 00                   add    %al,(%rax)
        ...

So... what now?

Bye
-- 
Milian Wolff
address@hidden
http://milianw.de