Subject: Re: Real vfork() (was: third results)
To: None <tech-kern@NetBSD.ORG>
From: Greg A. Woods <>
List: tech-kern
Date: 04/16/1998 02:07:34
[ On Wed, April 15, 1998 at 22:25:35 (-0700), Jason Thorpe wrote: ]
> Subject: Re: Real vfork() (was: third results) 
> 	* The original pre-4.4BSD vfork(2) semantics are now specified
> 	  in XPG4.2.  This means that new systems that want to carry the
> 	  UNIX(tm) trademark must implement the address space-sharing
> 	  semantics.  I.e. the interface is compatible with the rest of
> 	  the industry.

But that's wrong.  See my previous post quoting the specification from's on-line documents.  kre's rule of thumb would meet a
specification that's much closer to the truth.

> 	* There is a noticeable performance improvement, especially
> 	  for large processes, e.g. make(1) bulding the C library or
> 	  kernel.

I'm not sure so many people would actually notice.  Certainly not on a
modern machine with several active users.  We'd need actual numbers to
prove this one way or another, and I'd like to see throughput numbers in
the face of lots of concurrent activity, which means using a much more
scientific benchmark than just one (or more concurrent) "make build"

							Greg A. Woods

+1 416 443-1734      VE3TCP      <>      <robohack!woods>
Planix, Inc. <>; Secrets of the Weird <>