NetNews Usenet Archive 1992 #19/NN_1992_19.iso/spool/comp/os/msdos/programm/8954

Internet Message Format | 1992-09-01 | 1.8 KB

Xref: sparky comp.os.msdos.programmer:8954 comp.os.msdos.misc:4989
Newsgroups: comp.os.msdos.programmer,comp.os.msdos.misc
Path: sparky!uunet!destroyer!sol.ctr.columbia.edu!eff!world!jkenton
From: [e-mail] (Jeff Kenton)
Subject: Re: Why ms-dos is non reentrant
Message-ID: <[e-mail]>
Organization: Kenton Systems Corporation, Weston MA
References: <[e-mail]> <[e-mail]> <[e-mail]>
Date: Tue, 1 Sep 1992 18:10:46 GMT
Lines: 30

In article <[e-mail]> [e-mail] (Kenneth Falck) writes:
>
>So, the question remains, why don't the Microsoft programmers bang
>their brains together and rewrite MSDOS to be fully re-entrant and
>all the routines to use dynamic memory allocation instead of static?
>
>If this really is so difficult for them, I must have understood
>something wrong about this.
>


Because some of the data is naturally not sharable between routines
which can interrupt each other.  Consider the FAT (allocation table)
for a disk.  You don't want a routine to start allocating a new disk
sector for a file, and have it interrupted by another routine which
allocates the same sector.

Now, you can put software locks around critical data structures to
avoid this, but it's a lot of work to add this to a system and it
won't buy you much in DOS.  In a system that supports multi-processing,
it's a requirement, and all such systems do it.  You won't find it
added to DOS until it supports multi-processing.  That's what NT is
all about (that, and real 32-bit software).


--
-------------------------------------------------------------------------
=            Jeff Kenton    [phone]            =
=                [e-mail]            =
-------------------------------------------------------------------------