netdev - Re: [PATCH] [PATCH ipsec] xfrm: Store ipsec interface index

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Message-ID: <20240321093248.GC14887@unreal>
Date: Thu, 21 Mar 2024 11:32:48 +0200
From: Leon Romanovsky <leon@...nel.org>
To: Feng Wang <wangfe@...gle.com>
Cc: Steffen Klassert <steffen.klassert@...unet.com>, netdev@...r.kernel.org,
	herbert@...dor.apana.org.au, davem@...emloft.net
Subject: Re: [PATCH] [PATCH ipsec] xfrm: Store ipsec interface index

On Wed, Mar 20, 2024 at 11:05:13AM -0700, Feng Wang wrote:
> Hi Steffen,
> 
> Thanks for your comment.  Firstly,  the patch is using the xfrm interface
> ID instead of network interface ID. Secondly, would you please point me to
> the 'packet offload drivers' in the kernel tree?

First, please don't reply to emails in top-post format.
Second, did you try to search for "packet offload drivers" in the kernel?
https://elixir.bootlin.com/linux/v6.8.1/source/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c#L1152

Thanks

> I want to understand how the offload driver can distinguish 2 ipsec
> sessions if two sessions accidentally have the same address/mask and proto
> values(same xfrm_selector)? The offload driver needs to find the
> corresponding encryption parameters to do the work.
> 
> Thank you for your help,
> 
> Feng
> 
> 
> 
> On Tue, Mar 19, 2024 at 9:33 PM Steffen Klassert <
> steffen.klassert@...unet.com> wrote:
> 
> > On Tue, Mar 19, 2024 at 10:15:13AM -0700, Feng Wang wrote:
> > > Hi Leon,
> > >
> > > There is no "packet offload driver" in the current kernel tree.  The
> > packet
> > > offload driver mostly is vendor specific, it implements hardware packet
> > > offload.
> >
> > There are 'packet offload drivers' in the kernel, that's why we
> > support this kind of offload. We don't add code for proprietary
> > drivers.
> >
> > > On Tue, Mar 19, 2024 at 1:42 AM Leon Romanovsky <leon@...nel.org> wrote:
> > >
> > > > On Mon, Mar 18, 2024 at 04:13:28PM -0700, Feng Wang wrote:
> > > > > From: wangfe <wangfe@...gle.com>
> > > > >
> > > > > When there are multiple ipsec sessions, packet offload driver
> > > > > can use the index to distinguish the packets from the different
> > > > > sessions even though xfrm_selector are same.
> > > >
> > > > Do we have such "packet offload driver" in the kernel tree?
> > > >
> > > > Thanks
> > > >
> > > > > Thus each packet is handled corresponding to its session parameter.
> > > > >
> > > > > Signed-off-by: wangfe <wangfe@...gle.com>
> > > > > ---
> > > > >  net/xfrm/xfrm_interface_core.c | 4 +++-
> > > > >  1 file changed, 3 insertions(+), 1 deletion(-)
> > > > >
> > > > > diff --git a/net/xfrm/xfrm_interface_core.c
> > > > b/net/xfrm/xfrm_interface_core.c
> > > > > index 21d50d75c260..996571af53e5 100644
> > > > > --- a/net/xfrm/xfrm_interface_core.c
> > > > > +++ b/net/xfrm/xfrm_interface_core.c
> > > > > @@ -506,7 +506,9 @@ xfrmi_xmit2(struct sk_buff *skb, struct
> > net_device
> > > > *dev, struct flowi *fl)
> > > > >       xfrmi_scrub_packet(skb, !net_eq(xi->net, dev_net(dev)));
> > > > >       skb_dst_set(skb, dst);
> > > > >       skb->dev = tdev;
> > > > > -
> > > > > +#ifdef CONFIG_XFRM_OFFLOAD
> > > > > +     skb->skb_iif = if_id;
> > > > > +#endif
> >
> > This looks wrong. The network interface ID is not the same as the xfrm
> > interface ID.
> >