Message-ID:	<3EEFDD57.8838C657@yahoo.com>
Date:	Tue, 17 Jun 2003 23:32:39 -0400
From:	CBFalconer <cbfalconer AT yahoo DOT com>
Organization:	Ched Research
X-Mailer:	Mozilla 4.75 [en] (Win98; U)
X-Accept-Language:	en
MIME-Version:	1.0
To:	djgpp-workers AT delorie DOT com
Subject:	Re: nmalloc integration issues: memalign, names
References:	<3EEF6725 DOT 29821 DOT 7BD30 AT localhost>
Reply-To:	djgpp-workers AT delorie DOT com

"Mark E." wrote:
> 
> > memalign is supported by glibc. So is the function valloc,
> > which uses memalign. I'm assuming that they were added to DJGPP
> > for compatibility with
> 
> valloc and memalloc were added because at the time g++'s internal
> garbage collector required either valloc or mmap. I also seem to
> remember that around that time valloc or mmap became a requirement
> for gnu smalltalk.

Here is *another* new item or three.  What are valloc and
memalloc?  What is mmap?  If I knew these things in advance it
might be easier to allow for them.  For example, nmalloc has a
compile time equate to install a guard word.  This could be
reorganized to provide 32 bits of flags, usage counters, etc.
without changing anything but the overhead.  I have no idea
whether this would facilitate all those things, but it was cut out
in order to reduce the usual memory load.  A heavily reworked
system might even supply 31 bits without increasing overhead.

The specification sands are sifting.

-- 
Chuck F (cbfalconer AT yahoo DOT com) (cbfalconer AT worldnet DOT att DOT net)
   Available for consulting/temporary embedded and systems.
   <http://cbfalconer.home.att.net>  USE worldnet address!

- Raw text -