Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!brl-adm!adm!lcc.rich-wiz@CS.UCLA.EDU
From: lcc.rich-wiz@CS.UCLA.EDU (Richard Mathews)
Newsgroups: comp.lang.c
Subject: Re: String Processing Instruction
Message-ID: <6972@brl-adm.ARPA>
Date: Sun, 19-Apr-87 22:58:32 EST
Article-I.D.: brl-adm.6972
Posted: Sun Apr 19 22:58:32 1987
Date-Received: Mon, 20-Apr-87 01:44:55 EST
Sender: news@brl-adm.ARPA
Lines: 38

> In article <693@jenny.cl.cam.ac.uk> am@cl.cam.ac.uk (Alan Mycroft) writes:
> >You might be interested to know that such detection of null bytes in words
> >can be done in 3 or 4 instructions on almost any hardware (nay even in C).
> >(Code that follows relies on x being a 32 bit unsigned (or 2's complement
> >int with overflow ignored)...)
> >   #define has_nullbyte_(x) ((x - 0x01010101) & ~x & 0x80808080)
> >Then if e is an expression without side effects (e.g. variable)
> >   has_nullbyte_(e)
> >is nonzero iff the value of e has a null byte.

> I was so impressed by this new trick (well, to *me* it is new:-)
> that I immedeately decided to try it. my Whitechapel MG-1,
> a 32016 based machine, the results were impressive.

Be careful if you use this.  It does not work correctly for strings near
the end of memory.  Consider an (unaligned) string which is 21 bytes long
and is right at the end of data space (or any data segment on machines
which have such nonsense).  The posted code will fault on the last read.
On machines where data space is always a multiple of 4 bytes long, this
method will work if you first copy enough bytes to get the source pointer
to a longword boundary and then start copying 4 bytes at a time.

Someone else claimed that the has_nullbyte macro above does not work
correctly if it contains a byte equal to 0x80.  I do not see this.
Consider x==0x17801717.
	x - 0x01010101  == 0x167f1616
	~x		== 0xe87fe8e8
	(x - ONES) & ~x == 0x007f0000
	has_nullbyte(x) == 0x00000000
The claim that has_nullbyte(x) will be TRUE appears to be incorrect.
(I was able to prove to myself that the macro always gives the correct
answer, but I won't bother to write it up and post it.)

Richard M. Mathews
Locus Computing Corporation		       lcc.richard@LOCUS.UCLA.EDU
					       lcc.richard@UCLA-CS
				 {ihnp4,trwrb}!lcc!richard
       {randvax,sdcrdcf,ucbvax,trwspp}!ucla-cs!lcc!richard