Path: utzoo!attcan!uunet!cs.utexas.edu!rutgers!rochester!rit!ultb!jed4885
From: jed4885@ultb.UUCP (J.E. Dyer)
Newsgroups: comp.arch
Subject: Re: 80486 vs. 68040 code size (Really: No-Access bytes)
Message-ID: <814@ultb.UUCP>
Date: 12 May 89 14:49:45 GMT
References: <3150@tank.uchicago.edu> <1921@se-sd.sandiego.ncr.com>
Reply-To: jed4885@ultb.UUCP (J.E. Dyer (713ICS))
Distribution: eunet,world
Organization: Rochester Institute of Technology, Information Systems
Lines: 28

In article <1921@se-sd.sandiego.ncr.com> lord@se-sd.sandiego.NCR.COM (Dave Lord(SSP)) writes:
>In article <3150@tank.uchicago.edu> phd_ivo@gsbacd.uchicago.edu writes:
>
>>byte? This way, one can put a single No-ACCESS byte between datastructures at
>>link/load time, which should detect out of pointer ranges much more quickly
>>than the wait-and-see in use in our systems. (At least I have spent days of my
>
>	while (a[index] != 0 and index++ <= A_BOUNDS){ ... whatever ...};

	The problem I see with this method is that not all array access
	is sequential.  For example, I might have another array which
	contains the index I want to use.  For example:

	for (i = 0; i < A_Bounds; i++) {
		x = a[b[i]];
		....
	}

	In this case, if the value in b[i] is invalid, the chances are
	that it's not going to be equal to A_Bounds, which is the only
	error that a N0-access byte would detect.  IMHO, parity checking
	is much more valuable.

						--Jason

-sig-of-the-day-
"Kludge??? That's not a Kludge, it's _ART_!" 
BITNET: JED4885@RITVAX			UUCP: jed4885@ultb