Xref: news2.mv.net comp.os.msdos.djgpp:3698
From: Shawn Hargreaves <slh100 AT york DOT ac DOT uk>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: more sizeof questions
Date: Thu, 9 May 1996 10:52:13 +0100
Organization: The University of York, UK
Lines: 30
Message-ID: <Pine.SGI.3.91.960509104849.8470A-100000@tower.york.ac.uk>
References: <199605090048 DOT UAA09442 AT eelab DOT newpaltz DOT edu>
NNTP-Posting-Host: tower.york.ac.uk
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: <199605090048.UAA09442@eelab.newpaltz.edu>
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp

On Wed, 8 May 1996, John Fortin wrote:

> OK, I'm a bit confused..(nothing new i'm afraid)... The structure 
> below was taken from the VESA 1.2 spec. sizeof(_VGAInfoBlock) should 
> be 256.  DJGPP gives 260.  What's up with this.  A previous posting 
> mentioned 4 byte boundaries, but 256 is divisable by 4.  Any Ideas??
> 
> typedef struct 
> {
>     DWORD	VESASignature;			4
>     WORD	VESAVersion;			2	
>     DWORD    	OEMStringPointer;		4

The OEMStringPointer is 32 bits, but follows a 16 bit value. This would 
put it at a non-dword address, so gcc adds two bytes between the two. 
Normally this is a Good Thing (makes your code faster). If you are 
working with an already specified structure (eg. the VESA spec) it is a 
pain, and you have to add __attribute__ ((packed)) after each of your 
variables.

btw. this is one good reason not to dump structures into files just by 
calling fwrite(&struct, sizeof(struct). It's tempting to do, but it tends 
to break code when you move from one compiler to another...


/*
 *	Shawn Hargreaves.        Why is 'phonetic' spelt with a ph?
 *	Check out Allegro and FED on http://www.york.ac.uk/~slh100/
 */