From:	leathm AT solwarra (Leath Muller)
Message-Id:	<199609170101.LAA14759@gbrmpa.gov.au>
Subject:	Re: Possible static data zeroisation bug in DJGPP v2?
To:	csrabak AT dce03 DOT ipt DOT br (Cesar Scarpini Rabak)
Date:	Tue, 17 Sep 1996 11:01:08 +1000 (EST)
Cc:	djgpp AT delorie DOT com
In-Reply-To:	<1.5.4.16.19960916173527.29e7e9f6@dmeasc.rc.ipt.br> from "Cesar Scarpini Rabak" at Sep 16, 96 08:33:51 pm

> >You must forgive me for my disbelief, but how can an "external" variable
> >be initialized at all?  It's initialized elsewhere,  where it is
> >defined, not declared (externally declared or not...)
 
> Have you heard about a piece of code that runs BEFORE the very fisrt line of
> your program runs (in DJGPP is in crt0.o...) the C Run Time start up code?
 
> One of its functions is precisely to do this chore.

Correct me if I'm wrong, but wasn't this changed so that it _did not_ zero
the code for performance reasons? (I actually think this was part of Quake
ie: id needed some more performance, so taking out the zeroing of allocated
memory was used to help speed things up a bit). I think you now specify that
you want it zeroed as part of the __crt0_startup_flags (I cant verify this,
I have never needed to test it)
 
> >As far as static variables go, I don't have that info available, but
> >it would make sense to zero it yourself anyway.  I believe ANSI says
> >absolutely nothing is initialized unless you do it yourself.
 
> Sorry, but the Standard says the converse.

Whatever. It doesn't really matter. (No flames...this is a comment.)
Zeroing would HAVE to occur on UNIX machines because of security issues...
and because UNIX is a multi user machine. A PC running DOS is
generally a single user machine, so its for this reason that security
isn't a problem, so taking out the zeroing is not a problem.

Moral of the story: If you want something zeroed, either zero it yourself,
or stick it in the startup flags (if you can stick it in the startup flags ;)

Leathal.

- Raw text -