Message-ID:	<334CEC36.76D9@cs.com>
Date:	Thu, 10 Apr 1997 06:33:42 -0700
From:	"John M. Aldrich" <fighteer AT cs DOT com>
Reply-To:	fighteer AT cs DOT com
Organization:	Two pounds of chaos and a pinch of salt
MIME-Version:	1.0
To:	Oberhumer Markus <k3040e4 AT c210 DOT edvz DOT uni-linz DOT ac DOT at>
CC:	djgpp-workers <djgpp-workers AT delorie DOT com>
Subject:	Re: stackavail() ?
References:	<199704102250 DOT SAA19142 AT delorie DOT com>

Oberhumer Markus wrote:
> 
> I'm searching for a fast and reliable implementation to determine the
> available stack space, e.g.
> 
> unsigned stackavail( void );

It's an extremely difficult thing to do in DPMI, because the DPMI spec
does not support any direct method of detecting the available stack
space.  There was a thread about it several months ago on the newsgroup
in which several suggestions were made by myself or others, but nothing
definite was decided.  If you do want to attempt to determine the
available stack, you'll need to know the following things:

 - The size of the stack.  This may be possible to find by examining the
value of the global variable _stklen, but I don't know if this has a
reliable value unless you set it yourself in a program.  I also have no
idea what happens if the stack size is modified externally in the stub
(using stubedit).  There may also be a value that the DPMI host stores
that you can access via DPMI functions.
 - You need to record the virtual address of the top of the stack.  To
do this, you either have to do a detailed examination of the stack
itself (the way a debugger would), or find a way to store the value of
the stack pointer after it's been allocated but before anything is
pushed onto it.  I guess you could fudge this by examining the stack at
the beginning of main(), where only a tiny amount has been used.
 - If you can determine those two things, then you can make a function
which detects its own position on the stack and performs simple math to
determine the remaining space.

Gcc may want to make things difficult for you by optimizing the
variables that you are using to test the stack.  You may need to declare
anything you use for this purpose 'volatile'.  Grabbing the virtual
address of the stack may also require hacking into the startup code in
'crt0.s'.  What's more, there's no guarantee that any of these methods
will work in a debugger. It's an uphill climb, but if you can manage it,
a lot of people would probably be very grateful.  ;)

-- 
John M. Aldrich <fighteer AT cs DOT com>                      

* Anything that happens, happens.
* Anything that, in happening, causes something else to happen,
  causes something else to happen.
* Anything that, in happening, causes itself to happen again, happens
  again.
* It doesn't necessarily do it in chronological order, though.
 
                                       --- Douglas Adams

- Raw text -