Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!uakari.primate.wisc.edu!pikes!aspen.craycos.com!pmk
From: pmk@craycos.com (Peter Klausler)
Newsgroups: comp.lang.c
Subject: Re: Finding NULL byte in a long
Message-ID: <1990Dec13.153552.1862@craycos.com>
Date: 13 Dec 90 15:35:52 GMT
References: <1990Dec5.033206.10463@nimbus3.uucp> <198@nazgul.UUCP>
Organization: Cray Computer Corporation
Lines: 25

In article <198@nazgul.UUCP> bright@nazgul.UUCP (Walter Bright) writes:
>In article <1990Dec5.033206.10463@nimbus3.uucp> djs@nimbus3.UUCP (Doug) writes:
>/I know this has been on the net before, since I thought I saved it,
>/but can't find it now.  Anyway, could someone tell me what the
>/C expression is that tells you if a long has a NULL byte in it.
>/This is without masking each byte and testing it for 0.  It is very
>/clever and non-obvious.  Thanks.
>
>Hmmm, how about:
>	p = memchr(&x,0,sizeof(x));

Performance, however, is sometimes important.

If one can be unportable and depend on two's-complement integer arithmetic,
eight-bit bytes, and the size of a long being, say, 64, give this a try:

	#define UPPERS	   0x8080808080808080
	#define ONES	   0x0101010101010101
	#define	HASNULL(x) (~((((x) | UPPERS) - ONES) | (x)) & UPPERS)

The solution for a 32-bit microcomputer is similar.

Finding the *position* of the null byte is a little trickier. It helps a lot
if your machine has a "leading zero count"/"find first set bit" instruction
and your compiler lets you use it.