Re: [PATCH RFC] DWARF2 CFI exploitation for Linux on S/390

[Andrew Cagney <ac131313@redhat.com>]

> What needs to be done for this in general?

The theory is that the function:

static void
set_unwind_by_pc (CORE_ADDR pc, CORE_ADDR fp,
                   frame_register_unwind_ftype **unwind)
{
   if (!USE_GENERIC_DUMMY_FRAMES)
     /* Still need to set this to something.  The ``info frame'' code
        calls this function to find out where the saved registers are.
        Hopefully this is robust enough to stop any core dumps and
        return vaguely correct values..  */
     *unwind = frame_saved_regs_register_unwind;
   else if (PC_IN_CALL_DUMMY (pc, fp, fp))
     *unwind = generic_call_dummy_register_unwind;
   else
     *unwind = frame_saved_regs_register_unwind;
}

is modified so that it contains a clause like:

	else if (this frame's block contains dwarf2cfi
	         && this target things it works with dwarf2cfi)
	  *unwind = dwarf2cfi_register_unwind;

where dwarf2cfi_register_unwind() is strictly *recursive* - it uses 
functions defined in frame.h when it needs a read a more inner register.
Variation on the theme might be:

	else if (architecture things its got a custom unwind function for this 
frame's block)
	  *unwind = architecture's custom unwind function

There are warts in the theory - also need to add a per-frame 
frame_saved_pc() and (probably) some sort of per-frame frame_chain(). 
(see recent post to gdb@).

Andrew