Hi, x86_64 plt entries weren't so different as I thought. At least not for those things needed to make our current tests pass. Like x86 plt entries we take advantage of the fact that they are called through a CALL instruction and that the stack contains the return address (and possibly a selector and the GOT table address). frysk-core/frysk/stack/ChangeLog 2008-03-19 Mark Wielaard * TestLibFunctionStepFrame.java: No longer unresolved on x86_64. frysk-imports/libunwind/ChangeLog 2007-03-19 Mark Wielaard * src/x86_64/Gstep.c (is_call_instr_at): New function. (init_stack_based_ret): New function. (unw_step): Try stack based unwind with call instr, before fallback to frame pointer. Looking through the abi and instruction manuals however I see there are other possibilities that aren't matches by the above heuristics. I'll try and find some examples and add testcases for those later. Cheers, Mark