Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20210724010117.GA633665@minyard.net>
In-Reply-To: <20210724010117.GA633665@minyard.net>
From:   Eyal Birger <eyal.birger@gmail.com>
Date:   Sat, 24 Jul 2021 05:40:50 +0300
Message-ID: <CAHsH6Gu12oTqeE_TV6suU8Qoa9tET-XCUDc_x+TkFrZuWMrhJQ@mail.gmail.com>
Subject: Re: IPSec questions and comments
To:     Corey Minyard <minyard@acm.org>
Cc:     Steffen Klassert <steffen.klassert@secunet.com>,
        Herbert Xu <herbert@gondor.apana.org.au>,
        "David S. Miller" <davem@davemloft.net>,
        linux-kernel <linux-kernel@vger.kernel.org>,
        Linux Kernel Network Developers <netdev@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Ho Corey,

On Sat, Jul 24, 2021 at 4:02 AM Corey Minyard <minyard@acm.org> wrote:
> 1) In struct xfrm_dst, what is the difference between the route and path
> fields?  From what I can tell, in the first element of a bundle they
> will both point the route the packet will take after it has been
> transformed.  In the other elements of a bundle, route is the same as in
> the first element and path will be NULL.  Is this really the intent?
> Can path just be eliminated?

For non-transport modes - such as tunnel - 'route' and 'path' won't be the
same in the first element (xdst0): 'route' will be the original dst and
'path' will be the route the transformed packet will take. the dst is
overridden in the xfrm_dst_lookup() call within xfrm_bundle_create(), after
xdst->route had been set.

AFAICT, the intent for the 'path' member is described in commit
0f6c480f23f4 ("xfrm: Move dst->path into struct xfrm_dst") - essentially
'path' contains the reference to the underlay route from the topmost bundle
member avoiding a walk through the child chain when needed.

Hope this helps.
Eyal.