Path: utzoo!mnetor!uunet!lll-winken!lll-lcc!ames!pasteur!ucbvax!hoser.berkeley.edu!bryce
From: bryce@hoser.berkeley.edu (Bryce Nesbitt)
Newsgroups: comp.sys.amiga
Subject: Re: SetStack
Message-ID: <22692@ucbvax.BERKELEY.EDU>
Date: 21 Jan 88 04:15:43 GMT
References: <8801202127.AA10280@cory.Berkeley.EDU>
Sender: usenet@ucbvax.BERKELEY.EDU
Lines: 38

In article <> dillon@CORY.BERKELEY.EDU (Matt Dillon) writes:
|
|>;    SetStack(newstack,size)
|>;FUNCTION
|>;    This function points the current process' stack to a new memory
|>;    location. The active part of the old stack is duplicated, so no
|>;    context is lost.  SetStack() also tickles the proper fields in the task
|>;    structure to tell the system about the change.  At this time it does
|>;    not deallocate the old stack.
|
|	You CAN'T deallocate the old stack,  Because it will be deallocated
|by the parent task...

That is exactly why it does not.   It would be desirable if it could.
An unused stack is wasted memory.


|Also, the parent may
|depend on the task structure not changing, so it is probably better NOT to
|diddle the pointers... at least they should be the same on program exit.

pr_ReturnAddr always points to the top-4 of the *currently* executing
stack.  The size of the currently executing stack is stored in
the long at top-4.  At top-8 is the return address used during process 
removal.
 
This is the only thing SetStack() does or should diddle with.


Not keeping pr_ReturnAddr as a pointer to the *currently executing*
stack destroys stack checking algorithms.  With it you can have a nice
clean stack check every time the process is switched out.
Matter of fact, I'm running that right now.

|\ /|  . Ack! (NAK, SOH, EOT)
{o O} . bryce@hoser.berkeley.EDU -or- ucbvax!hoser!bryce (or try "cogsci")
 (")
  U	"As an engineer, I only set the value of a product... not the cost."
	-Bryce Nesbitt