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;
Date:   Wed, 26 Aug 2020 15:20:00 +0000
From:   Mira Ressel <aranea@aixah.de>
To:     David Miller <davem@davemloft.net>
Cc:     kuba@kernel.org, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH 1/2] veth: Initialize dev->perm_addr
Message-ID: <20200826152000.ckxrcfyetdvuvqum@vega>
References: <20200824143828.5964-1-aranea@aixah.de>
 <20200824.102545.1450838041398463071.davem@davemloft.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20200824.102545.1450838041398463071.davem@davemloft.net>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Mon, Aug 24, 2020 at 10:25:45AM -0700, David Miller wrote:
> From: Mira Ressel <aranea@aixah.de>
> Date: Mon, 24 Aug 2020 14:38:26 +0000
> 
> > Set the perm_addr of veth devices to whatever MAC has been assigned to
> > the device. Otherwise, it remains all zero, with the consequence that
> > ipv6_generate_stable_address() (which is used if the sysctl
> > net.ipv6.conf.DEV.addr_gen_mode is set to 2 or 3) assigns every veth
> > interface on a host the same link-local address.
> > 
> > The new behaviour matches that of several other virtual interface types
> > (such as gre), and as far as I can tell, perm_addr isn't used by any
> > other code sites that are relevant to veth.
> > 
> > Signed-off-by: Mira Ressel <aranea@aixah.de>
>  ...
> > @@ -1342,6 +1342,8 @@ static int veth_newlink(struct net *src_net, struct net_device *dev,
> >  	if (!ifmp || !tbp[IFLA_ADDRESS])
> >  		eth_hw_addr_random(peer);
> >  
> > +	memcpy(peer->perm_addr, peer->dev_addr, peer->addr_len);
> 
> Semantically don't you want to copy over the peer->perm_addr?
> 
> Otherwise this loses the entire point of the permanent address, and
> what the ipv6 address generation facility expects.

I'm confused. Am I misinterpreting what you're saying here, or did you
make a typo?

I'm setting the peer->perm_addr, which would otherwise be zero, to its
dev_addr, which has been either generated randomly by the kernel or
provided by userland in a netlink attribute.

-- 
Regards,
Mira