Stacktracing going forward.

[Nurdin Premji <npremji at redhat dot com>]

Stacktracing work is mostly complete in the cross-platform case of 
unwinding an x86 binary running on an x86_64. The only missing part is 
the name of the inner most frame of the binary. Which appears to be 
inside the vdso.

In order to get elfutils to properly determine the name of this inner 
most vdso frame some changes to the elfutils bindings will need to be 
made. This will also be needed to get core files to be unwound.

Inside frysk-imports/lib/dw/cni/Dwfl.cxx, the method dwfl_begin creates 
a Dwfl given a pid. There is a call to the function 
dwfl_linux_proc_report(dwfl, pid). Which is defined in 
elfutils/libdwfl/linux-proc-maps.c.
This method creates all the dwfl_modules by reading /proc/pic/maps.

There are two limitations with this method. The first is that it only 
works on live processes (takes a pid). The second is that it calls a 
method find_sysinfo_ehdr (same file), which has a switch statement based 
on sizeof(long int). This means it assumes the host architecture and the 
target architecture are the same.

We will need to create a dwfl_frysk_proc_report(dwfl, Proc proc) or 
something similar which will do twhat dwfl_linux_proc_report does for 
any type of frysk proc.

(Alternatively we can create several dwfl_frysk_ptrace_proc, 
dwfl_frysk_core_proc methods with different parameters.)

We will probably also need to implement our own Dwfl_Callbacks callbacks 
which are set up just before the call to dwfl_linux_proc_report.