netdev - Re: [PATCH 2/5] xfrm: Pass UDP encapsulation in TX packet offload

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

Message-ID: <20240312112630.GU12921@unreal>
Date: Tue, 12 Mar 2024 13:26:30 +0200
From: Leon Romanovsky <leon@...nel.org>
To: Steffen Klassert <steffen.klassert@...unet.com>
Cc: Paolo Abeni <pabeni@...hat.com>, David Miller <davem@...emloft.net>,
	Jakub Kicinski <kuba@...nel.org>,
	Herbert Xu <herbert@...dor.apana.org.au>, netdev@...r.kernel.org
Subject: Re: [PATCH 2/5] xfrm: Pass UDP encapsulation in TX packet offload

On Tue, Mar 12, 2024 at 12:20:49PM +0100, Steffen Klassert wrote:
> On Tue, Mar 12, 2024 at 01:15:28PM +0200, Leon Romanovsky wrote:
> > On Tue, Mar 12, 2024 at 07:20:06AM +0100, Steffen Klassert wrote:
> > > On Mon, Mar 11, 2024 at 05:25:03PM +0100, Paolo Abeni wrote:
> > > > Hi,
> > > > 
> > > > On Wed, 2024-03-06 at 11:04 +0100, Steffen Klassert wrote:
> > > > > From: Leon Romanovsky <leonro@...dia.com>
> > > > > 
> > > > > In addition to citied commit in Fixes line, allow UDP encapsulation in
> > > > > TX path too.
> > > > > 
> > > > > Fixes: 89edf40220be ("xfrm: Support UDP encapsulation in packet offload mode")
> > > > > CC: Steffen Klassert <steffen.klassert@...unet.com>
> > > > > Reported-by: Mike Yu <yumike@...gle.com>
> > > > > Signed-off-by: Leon Romanovsky <leonro@...dia.com>
> > > > > Signed-off-by: Saeed Mahameed <saeedm@...dia.com>
> > > > > Signed-off-by: Steffen Klassert <steffen.klassert@...unet.com>
> > > > 
> > > > This is causing self-test failures:
> > > > 
> > > > https://netdev.bots.linux.dev/flakes.html?tn-needle=pmtu-sh
> > > > 
> > > > reverting this change locally resolves the issue.
> > > > 
> > > > @Leon, @Steffen: could you please have a look?
> > > 
> > > Looks like the check for x->encap was removed unconditionally.
> > > I should just be removed when XFRM_DEV_OFFLOAD_PACKET is set,
> > > otherwise we might create a GSO packet with UPD encapsulation.
> > > 
> > > Leon?
> > 
> > Right, I missed IPsec SW path, that x->encap check can be removed
> > in packet offload because HW supports it and in crypto offload, because
> > there is a check in xfrm_dev_state_add() to prevent it.
> > 
> > What about this fix?
> > 
> > diff --git a/net/xfrm/xfrm_device.c b/net/xfrm/xfrm_device.c
> > index 653e51ae3964..6e3e5a09cfeb 100644
> > --- a/net/xfrm/xfrm_device.c
> > +++ b/net/xfrm/xfrm_device.c
> > @@ -407,7 +407,7 @@ bool xfrm_dev_offload_ok(struct sk_buff *skb, struct xfrm_state *x)
> >         struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
> >         struct net_device *dev = x->xso.dev;
> > 
> > -       if (!x->type_offload)
> > +       if (!x->type_offload || x->xso.type == XFRM_DEV_OFFLOAD_UNSPECIFIED)
> >                 return false;
> 
> Then we can't generate GSO packets for the SW path anymore. We just need
> to reject UDP enacpsulation in SW here.

Is it better?

diff --git a/net/xfrm/xfrm_device.c b/net/xfrm/xfrm_device.c
index 653e51ae3964..6346690d5c69 100644
--- a/net/xfrm/xfrm_device.c
+++ b/net/xfrm/xfrm_device.c
@@ -407,7 +407,8 @@ bool xfrm_dev_offload_ok(struct sk_buff *skb, struct xfrm_state *x)
        struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
        struct net_device *dev = x->xso.dev;

-       if (!x->type_offload)
+       if (!x->type_offload ||
+           (x->xso.type == XFRM_DEV_OFFLOAD_UNSPECIFIED && x->encap))
                return false;

        if (x->xso.type == XFRM_DEV_OFFLOAD_PACKET ||

>