X-Authentication-Warning: delorie.com: mailnull set sender to djgpp-bounces using -f
Message-ID: <3C68525B.476118E@yahoo.com>
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
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Alignment problem
References: <Pine DOT SUN DOT 3 DOT 91 DOT 1020211133619 DOT 25095H-100000 AT is> <3C67DE6D DOT DEE8305C AT yahoo DOT com> <3028-Mon11Feb2002220350+0200-eliz AT is DOT elta DOT co DOT il>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Lines: 46
Date: Mon, 11 Feb 2002 23:32:44 GMT
NNTP-Posting-Host: 12.90.168.45
X-Complaints-To: abuse AT worldnet DOT att DOT net
X-Trace: bgtnsc04-news.ops.worldnet.att.net 1013470364 12.90.168.45 (Mon, 11 Feb 2002 23:32:44 GMT)
NNTP-Posting-Date: Mon, 11 Feb 2002 23:32:44 GMT
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

Eli Zaretskii wrote:
> 
> > From: CBFalconer <cbfalconer AT yahoo DOT com>
> > Newsgroups: comp.os.msdos.djgpp
> > Date: Mon, 11 Feb 2002 15:10:03 GMT
> >
> > Flat profile:
> >
> > Each sample counts as 0.0555556 seconds.
> >   %   cumulative   self              self     total
> >  time   seconds   seconds    calls  ns/call  ns/call  name
> >  82.04      9.39     9.39    30001 312952.53 314921.80  merge
> [...]
> > which puts the problem (going by name) precisely where it was
> > expected.
> 
> Yep, pretty much so.  It had to be merge.
> 
> > I still don't know how you are organizing the free list,
> 
> In a nutshell, they are kept in linked lists arranged in buckets by
> block sizes, each sublist for a size that is a different power of 2.
> 
> Try stepping through malloc and free for several allocations and
> deallocations of vastly different sizes, and I think you will see the
> pattern clearly.
> 
> > If something identifies such blocks as being free or not, independant
> > of the free list, this could be made linear.
> 
> You will see in the code that the LSB of the size/next pointer header
> is used as the allocated/free marker.
> 
> > This implies that
> > all blocks contain fields for next, size, and one bit for flags.
> 
> Yes, there's a header and a trailer, each one 4-byte long.  See teh
> definition of struct BLOCK.

What defines the sizes for which you keep separate lists?  I hate
reading OPs source :-) at least without a clear statement of
algorithms.

BTW, your messages are suddenly supporting references and getting
threaded properly! Did you change newsreaders?