When is the stack pointer unreliable?

[Jeffrey A Law]

  In message < 199809070123.VAA03584@jwlab.FEITH.COM >you write:
  > >> Since reloading the stack pointer causes a stall on some processors
  > >> it should be avoided when possible.  Under what situations is sp
  > >> unreliable?
  > > alloca calls,
  > 
  > In which case current_function_calls_alloca is set.
Right.

  > > variable sized objects in the stack come immediately to mind.
  > 
  > I assume that this is always handled by allocate_dynamic_stack_space
  > in which case current_function_calls_alloca is set.
Correct.


  > > You might also need to check if the compiler has optimized away
  > > popping of args after a function call if we unconditionally reach the
  > > epilogue after performing the call.
  > 
  > Good point.  That hadn't occurred to me.
I think this is controlled by EXIT_IGNORE_STACK or something like
that.  I don't know if any state is kept about whether or not that
optimization was performed.

jeff