Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!iuvax!purdue!haven!adm!smoke!gwyn
From: gwyn@smoke.BRL.MIL (Doug Gwyn)
Newsgroups: comp.lang.c
Subject: Re: Buddy system allocator?
Message-ID: <11387@smoke.BRL.MIL>
Date: 23 Oct 89 14:40:40 GMT
References: <426@grenada.UUCP> <89Oct22.120811edt.3343@neat.cs.toronto.edu>
Reply-To: gwyn@brl.arpa (Doug Gwyn)
Organization: Ballistic Research Lab (BRL), APG, MD.
Lines: 12

In article <89Oct22.120811edt.3343@neat.cs.toronto.edu> moraes@cs.toronto.edu (Mark Moraes) writes:
>I usually prefer the boundary tags algorithm with some of the mods
>[Kunth] describes suggests in the Exercises ...

That's also the one I usually choose when I have to implement a general
memory allocator.  If anyone is going to implement this, note that the
solution to exercise 2.5-19 given in the Second Edition has a bug in it:
upon allocation failure, one must not leave ROVER pointing to the
collapsed tail; set it to LOC(AVAIL).  Also, the algorithm given in the
solution to exercise 1.5-12 can be improved upon.  I won't say how,
since you should study this stuff before implementing it, to the point
that you should be able to figure it out yourself.