Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 13 Feb 2009 23:54:27 +0100
From:      Luigi Rizzo <rizzo@iet.unipi.it>
To:        Robert Watson <rwatson@freebsd.org>
Cc:        svn-src-stable@freebsd.org, Luigi Rizzo <luigi@freebsd.org>, src-committers@freebsd.org, svn-src-stable-7@freebsd.org, svn-src-all@freebsd.org
Subject:   Re: svn commit: r188587 - stable/7/sys/netinet
Message-ID:  <20090213225427.GA69192@onelab2.iet.unipi.it>
In-Reply-To: <alpine.BSF.2.00.0902131853550.55346@fledge.watson.org>
References:  <200902131809.n1DI9cXb095457@svn.freebsd.org> <alpine.BSF.2.00.0902131853550.55346@fledge.watson.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Feb 13, 2009 at 06:54:37PM +0000, Robert Watson wrote:
> 
> On Fri, 13 Feb 2009, Luigi Rizzo wrote:
> 
> >Author: luigi
> >Date: Fri Feb 13 18:09:38 2009
> >New Revision: 188587
> >URL: http://svn.freebsd.org/changeset/base/188587
> >
> >Log:
> > MFC - replace usage of n_* types with uint*_t , to help removing
> > unnecessary dependencies on in_systm.h
> 
> As harmless as these changes may seem, insta-MFCs often come with 
> unexpected side effects.  A three day MFC wait will also never hurt for 
> what are definitely not critical bug fixes :-).

To put this commit in the right context:

I know the recommendation on MFC. But this is actually not an MFC
but an MFStable as I will explain below.

I started this work to remove useless header dependencies from ipfw
and dummynet (and, as a side effect, the network stack as well).
You may have seen userland changes being committed 1-2 weeks ago,
in that case only in HEAD because that's was the appropriate
approach in that context.

But the kernel side is different.  As I mentioned in some other
commits today, HEAD and RELENG_7 are largely different in this
respect after the VIMAGE import: in HEAD, several netinet/ files
#include vnet.h and vinet.h which in turn bring in almost everything
related to networking: if.h, route.h, <*_var.h> for the stats
records, locks and so on.
As a result, the cleanup work needs to be done differently in the
two branches.

In fact, the real reference is RELENG_7 which is clean from nested
#include's: it's there that I can find out which headers are not
necessary, and then I use the information to drive the changes to
HEAD keeping in mind that eventually vinet.h should be cleaned up
as well. That's why I say that these changes are closer to Merge
>From Stable than to MFC.

The n_* cleanup just happens to be part of both changesets, but
that's only a minor part.

	cheers
	luigi



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20090213225427.GA69192>