Thanks for sharing Andrei. let me ask you a bit, this means you have
to save first before accessing your own stack and restore it after
you leave it, right? Or is it unnecessary?
If you want to reuse the stack you were using before switching to the new one, then yes, you have to save the stack pointer are restore it later. There are portable ways of doing these; look at getcontext/setcontext. People that would be interested in this sort of tricks are probably implementing a threads library of sorts, so, for example, for forking, you'd want to save the current context (savecontext), make a new one for the second thread with a new stack (makecontext), use that context for a while (setcontext) then revert to the original one (setcontext again).
Cheers,
- a_m
regards,
Mulyadi Santosa
Freelance Linux trainer and consultant
blog: the-hydra.blogspot.com
training: mulyaditraining.blogspot.com
