[Bug binutils/947] New: Extension to BFD and addr2line to print inlining chain

[fnf at specifixinc dot com]

Consider the following test case:

  #include <vector>

  using namespace std;

  int main ()
  {
    vector<int> V(10);
    V[10] = 1;
  }

Compile it like:

  g++ -O2 -ggdb -o t t.cc

Run gdb and examine the first few instructions in main:

  (gdb) x/10i main
  0x8048880 <main>:       push   %ebp
  0x8048881 <main+1>:     mov    %esp,%ebp
  0x8048883 <main+3>:     push   %ebx
  0x8048884 <main+4>:     sub    $0x34,%esp
  0x8048887 <main+7>:     and    $0xfffffff0,%esp
  0x804888a <main+10>:    sub    $0x10,%esp
  0x804888d <main+13>:    cmpb   $0x0,0x804a7c0
  0x8048894 <main+20>:    je     0x8048964 <main+228>
  0x804889a <main+26>:    xor    %eax,%eax
  0x804889c <main+28>:    lea    0xffffffe8(%ebp),%ebx

Run addr2line on the address where the cmpb instruction is and you
get (if you use the latest versions of gcc and binutils):

  $ addr2line -s -f -e t 0804888d
  _S_get_pool
  mt_allocator.h:450

The patch (which I'll attach separately) allows addr2line to print
additional information about how we arrived at that address via several
levels of inlining:

  $ addr2line -i -s -f -e t 0804888d
  _S_get_pool
  mt_allocator.h:450
  __mt_alloc
  mt_allocator.h:656
  allocator
  allocator.h:97
  vector
  stl_vector.h:217
  main
  t.cc:7

One possible use of this is in environments which trap various
exceptions and want to print a traceback for the user, by calling
addr2line on the current PC and each saved PC in the stack.  In the
case of functions that inline other functions, that traceback is
incomplete because it misses all the inlined functions and the caller
of inlined functions.  I.E.  for the current example, where the caller
of main is probably _start(), it would appear that the current
function (_S_get_pool) was called directly by _start, skipping main
and all the other intermediate inlined functions.

The patch is somewhat large because it adds another function to the
BFD target structure, which currently returns FALSE for all non-DWARF2
environments.  The meat of the changes is actually in bfd/dwarf2.c and
binutils/addr2line.c.

-- 
           Summary: Extension to BFD and addr2line to print inlining chain
           Product: binutils
           Version: 2.17 (HEAD)
            Status: NEW
          Severity: normal
          Priority: P2
         Component: binutils
        AssignedTo: unassigned at sources dot redhat dot com
        ReportedBy: fnf at specifixinc dot com
                CC: bug-binutils at gnu dot org,fnf at specifixinc dot com
 GCC build triplet: i686-pc-linux-gnu
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu


http://sources.redhat.com/bugzilla/show_bug.cgi?id=947

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.