From owner-freebsd-arch  Wed May 16 14:48:25 2001
Delivered-To: freebsd-arch@freebsd.org
Received: from earth.backplane.com (earth-nat-cw.backplane.com [208.161.114.67])
	by hub.freebsd.org (Postfix) with ESMTP id 67EF537B423
	for <freebsd-arch@FreeBSD.ORG>; Wed, 16 May 2001 14:48:23 -0700 (PDT)
	(envelope-from dillon@earth.backplane.com)
Received: (from dillon@localhost)
	by earth.backplane.com (8.11.3/8.11.2) id f4GLmNm79465;
	Wed, 16 May 2001 14:48:23 -0700 (PDT)
	(envelope-from dillon)
Date: Wed, 16 May 2001 14:48:23 -0700 (PDT)
From: Matt Dillon <dillon@earth.backplane.com>
Message-Id: <200105162148.f4GLmNm79465@earth.backplane.com>
To: freebsd-arch@FreeBSD.ORG
Subject: Upgrading u_short cr_ref to u_int cr_ref in ucred on -stable
References:  <14721.48065.766815.376959@grasshopper.cs.duke.edu>
Sender: owner-freebsd-arch@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

    As per conversations on this list 5-10 April (see
    'Eliminate crget() from...') I intend to MFC the change 
    Alfred made in -current to -stable for the ucred reference count.
    This will occur tonight.

    The change will turn cr_ref from a u_short to a u_int.  Due to
    alignment issues, the other fields in the structure will not
    change their location and a quick run through the driver code
    seems to show that device drivers do not access cr_ref.
    That is, only crhold() seems to really reference the cr_ref
    field directly and devices do not appear to call crhold(), so I
    think we are plenty safe enough in regards to binary compatibility.

    This being between releases, if this change creates a serious issue
    somewhere along the line we can always rip it out and throw in
    Terry's stop-gap fix (that allocates a new copy when the ref count
    would otherwise overflow).  I do not think that will be necessary
    though.


					    -Matt

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-arch" in the body of the message