Re: MPE Port

[Jeffrey A Law <law@cygnus.com>]

  In message < 4.2.0.58.19991026065301.00c68310@garfield.dis.com >you write:

  > There is a table called the XRT (if I understand it correctly on HP-UX, it
  > is similar in function to the PLT there) that is fixed up at execution time
  > with the space, offset of the code, and the DP and SP of the called
  > procedure.
  > There are support routines that can dynamically load a program and create
  > an XRT for it. All "procedure labels" on MPE are really pointers to the XRT
  > entry. There are entry and exit stubs to set and restore those registers
  > and to promote privilege as needed.
Not exactly what I was looking for.

I'm interested in how/where are dynamic libraries loaded, does the dynamic
linker make a distinction between code & data in the dynamic library, if
so, does it arrange to load them in different spaces (which would be a
significant divergence from the hpux dynamic linker).


More along the lines of how does mapping of dynamic executables work.  THe


  > to start looking at this as soon as I have time. I don't know the purpose
  > of a "trampoline" and what the __gcc_plt_call does, yet. Can you
  > enlighten me?
You'd have to read the gdb code for all the grody details.

When gdb wants to call a function in the inferior program, it writes a bunch of
code onto the stack (ie the stub/trampoline) which is responsible for doing
argument setup, dealing with cross space calls/returns, etc etc.).  It then
transfers execution control to that code on the stack, which in turn calls
the final target.

__gcc_plt_call is necessary because the hpux dynamic linker will not allow
the return address for a call to be in a non-text, non-shlib space when the
target function is lazily bound.
jeff