Xref: utzoo comp.unix.questions:14591 comp.sys.att:6843
Path: utzoo!attcan!uunet!cs.utexas.edu!rutgers!rochester!pt.cs.cmu.edu!cadre!pitt!darth!dehart
From: dehart@darth.PGH.PA.US (Ed DeHart)
Newsgroups: comp.unix.questions,comp.sys.att
Subject: Re: vfork is miniframe libraries?
Summary: vfork = fork
Message-ID: <7@darth.PGH.PA.US>
Date: 29 Jun 89 13:36:16 GMT
References: <22990@santra.UUCP>
Followup-To: poster
Organization: Darth Software, Pittsburgh Pa.
Lines: 36

In article <22990@santra.UUCP>, hsu@kampi.hut.fi (Heikki Suonsivu) writes:
> Convergent Miniframe libc has vfork. Does anyone know if it is real or
> just a wrapper for fork? No dis.. 
> 

I had thought that vfork was real since the CT folks used the BSD 4.2 virtual
memory model for CTIX.  Before posting something I decided to test vfork
just to make sure that it really was vfork.

My tests showed that my guess was incorrect and that vfork is just fork with
a different name.

Since the child process uses the parent's address space the first ground rule
is to not mess with anything, just exec out like a good child.  My test program
had the child change the value of something then have the parent print the
result.
--------------------

#define CHILD_PROC 0

main()
{
	int	proc, foo=0;

	if ((proc = vfork()) == CHILD_PROC) {
		foo = 1;
		execl ("/bin/date",0);
		}
	else
		printf ("proc -> %d, foo -> %d\n", proc, foo);
}
I would have expected foo to be 1 if vfork was real but foo was 0.  No vfork.

Thanks, Ed DeHart / wa3yoa
   Usenet:   {pitt | bellcore}!darth!dehart  -OR-  uunet!apexepa!darth!dehart
   Internet: dehart@darth.pgh.pa.us -OR- darth!dehart@vax.cs.pittsburgh.edu