> From: Darren Reed <darrenr@reed.wattle.id.au>
> Date: 1999-10-03 06:38:03 -0700
> To: tech-net@netbsd.org
> Subject: m_copy - doesn't copy if M_EXT is set.
> Delivered-to: tech-net@netbsd.org
> X-Mailer: ELM [version 2.4ME+ PL37 (25)]
>
>
> It would appear that the behaviour of m_copy()/m_copym() is changed, 
> depending on whether or not M_EXT is set.  If M_EXT is set, then the 
> buffer is *NOT* coppied, just another external reference is made and 
> the appropriate counter incremented.  If it is unset, then a new
> buffer is allocated (M_GET) and data copied from the orignal to the 
> copy.
   You're correct about the behavior of m_copy()/m_copym(), but not  
about it's having changed.  This is old behavior.  I was a bit  
surprised, initially, when trying to add support for the Mac OS  
networking stack in what is now Mac OS X Server.  The Open Transport  
stack kept getting TCP packets with the checksums set to zero :-}.

  I took the simple way out and implemented "m_dup()" [see the  
Darwin source; the code is somewhere in ".../net", not in  
".../kern/uipc_mbuf.c"].   This does a "real" copy.

Regards,

Justin

--
Justin C. Walker, Curmudgeon-At-Large *
Institute for General Semantics       |
Manager, CoreOS Networking            | When crypto is outlawed,
Apple Computer, Inc.                  | Only outlaws will have crypto.
2 Infinite Loop                       |
Cupertino, CA 95014                   |
*-------------------------------------*-------------------------------*