Received-SPF: pass (google.com: domain of linux-kernel+bounces-188765-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Precedence: bulk
MIME-Version: 1.0
References: <20240520070348.26725-1-chengen.du@canonical.com> <664f5938d2bef_1b5d2429467@willemb.c.googlers.com.notmuch>
In-Reply-To: <664f5938d2bef_1b5d2429467@willemb.c.googlers.com.notmuch>
From: Chengen Du <chengen.du@canonical.com>
Date: Fri, 24 May 2024 22:04:18 +0800
Message-ID: <CAPza5qc+m6aK0hOn8m1OxnmNVibJQn-VFXBAnjrca+UrcmEW4g@mail.gmail.com>
Subject: Re: [PATCH] af_packet: Handle outgoing VLAN packets without hardware offloading
To: Willem de Bruijn <willemdebruijn.kernel@gmail.com>
Cc: davem@davemloft.net, edumazet@google.com, kuba@kernel.org, 
	pabeni@redhat.com, netdev@vger.kernel.org, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Hi Willem,

Thank you for your reply.

Please allow me to briefly summarize to ensure we are in agreement on
the solution.

Firstly, I will submit a patch to add a new bit in the status to
indicate the presence of VLAN information in the payload and modify
the header's entry accordingly.
A new member will be added to the sockaddr_ll struct to represent the
VLAN-encapsulated protocol, thus avoiding direct modification of the
sll_protocol.

Following this patch, I will work on enabling the link layer header
via a socket option.

If there are no ambiguities, I will submit the patch next week.
Your assistance and suggestions are highly appreciated.

Best regards,
Chengen Du

On Thu, May 23, 2024 at 10:57=E2=80=AFPM Willem de Bruijn
<willemdebruijn.kernel@gmail.com> wrote:
>
> Chengen Du wrote:
> > In the outbound packet path, if hardware VLAN offloading is unavailable=
,
> > the VLAN tag is inserted into the payload but then cleared from the
> > metadata. Consequently, this could lead to a false negative result when
> > checking for the presence of a VLAN tag using skb_vlan_tag_present(),
> > causing the packet sniffing outcome to lack VLAN tag information. As a
> > result, the packet capturing tool may be unable to parse packets as
> > expected.
> >
> > Signed-off-by: Chengen Du <chengen.du@canonical.com>
>
> Fixes tag and Cc: stable.
>
> As discussed please add more detail to the commit message that
> explains the bug. And/or add a Link: for instance to the github
> issue.
>
> > ---
> >  net/packet/af_packet.c | 25 +++++++++++++++++++------
> >  1 file changed, 19 insertions(+), 6 deletions(-)
> >
> > diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
> > index ea3ebc160e25..73e9acb1875b 100644
> > --- a/net/packet/af_packet.c
> > +++ b/net/packet/af_packet.c
> > @@ -1010,12 +1010,15 @@ static void prb_fill_vlan_info(struct tpacket_k=
bdq_core *pkc,
> >       if (skb_vlan_tag_present(pkc->skb)) {
> >               ppd->hv1.tp_vlan_tci =3D skb_vlan_tag_get(pkc->skb);
> >               ppd->hv1.tp_vlan_tpid =3D ntohs(pkc->skb->vlan_proto);
> > -             ppd->tp_status =3D TP_STATUS_VLAN_VALID | TP_STATUS_VLAN_=
TPID_VALID;
> > +     } else if (eth_type_vlan(pkc->skb->protocol)) {
> > +             ppd->hv1.tp_vlan_tci =3D ntohs(vlan_eth_hdr(pkc->skb)->h_=
vlan_TCI);
> > +             ppd->hv1.tp_vlan_tpid =3D ntohs(pkc->skb->protocol);
> >       } else {
> >               ppd->hv1.tp_vlan_tci =3D 0;
> >               ppd->hv1.tp_vlan_tpid =3D 0;
> > -             ppd->tp_status =3D TP_STATUS_AVAILABLE;
> >       }
> > +     ppd->tp_status =3D (ppd->hv1.tp_vlan_tci || ppd->hv1.tp_vlan_tpid=
) ?
> > +             TP_STATUS_VLAN_VALID | TP_STATUS_VLAN_TPID_VALID : TP_STA=
TUS_AVAILABLE;
>
> Don't move this out of the original branch and don't make the valid
> conditional on the value of tp_vlan_tci. Just duplicating the line
> to both branches is fine. Here and below.
>
> >  }
> >
> >  static void prb_run_all_ft_ops(struct tpacket_kbdq_core *pkc,
> > @@ -2427,11 +2430,15 @@ static int tpacket_rcv(struct sk_buff *skb, str=
uct net_device *dev,
> >               if (skb_vlan_tag_present(skb)) {
> >                       h.h2->tp_vlan_tci =3D skb_vlan_tag_get(skb);
> >                       h.h2->tp_vlan_tpid =3D ntohs(skb->vlan_proto);
> > -                     status |=3D TP_STATUS_VLAN_VALID | TP_STATUS_VLAN=
_TPID_VALID;
> > +             } else if (eth_type_vlan(skb->protocol)) {
> > +                     h.h2->tp_vlan_tci =3D ntohs(vlan_eth_hdr(skb)->h_=
vlan_TCI);
> > +                     h.h2->tp_vlan_tpid =3D ntohs(skb->protocol);
> >               } else {
> >                       h.h2->tp_vlan_tci =3D 0;
> >                       h.h2->tp_vlan_tpid =3D 0;
> >               }
> > +             if (h.h2->tp_vlan_tci || h.h2->tp_vlan_tpid)
> > +                     status |=3D TP_STATUS_VLAN_VALID | TP_STATUS_VLAN=
_TPID_VALID;
> >               memset(h.h2->tp_padding, 0, sizeof(h.h2->tp_padding));
> >               hdrlen =3D sizeof(*h.h2);
> >               break;
> > @@ -2457,7 +2464,8 @@ static int tpacket_rcv(struct sk_buff *skb, struc=
t net_device *dev,
> >       sll->sll_halen =3D dev_parse_header(skb, sll->sll_addr);
> >       sll->sll_family =3D AF_PACKET;
> >       sll->sll_hatype =3D dev->type;
> > -     sll->sll_protocol =3D skb->protocol;
> > +     sll->sll_protocol =3D eth_type_vlan(skb->protocol) ?
> > +             vlan_eth_hdr(skb)->h_vlan_encapsulated_proto : skb->proto=
col;
>
> For QinQ you probably want the true network protocol, not the inner
> VLAN tag.
> >       sll->sll_pkttype =3D skb->pkt_type;
> >       if (unlikely(packet_sock_flag(po, PACKET_SOCK_ORIGDEV)))
> >               sll->sll_ifindex =3D orig_dev->ifindex;
> > @@ -3482,7 +3490,8 @@ static int packet_recvmsg(struct socket *sock, st=
ruct msghdr *msg, size_t len,
> >               /* Original length was stored in sockaddr_ll fields */
> >               origlen =3D PACKET_SKB_CB(skb)->sa.origlen;
> >               sll->sll_family =3D AF_PACKET;
> > -             sll->sll_protocol =3D skb->protocol;
> > +             sll->sll_protocol =3D eth_type_vlan(skb->protocol) ?
> > +                     vlan_eth_hdr(skb)->h_vlan_encapsulated_proto : sk=
b->protocol;
> >       }
> >
> >       sock_recv_cmsgs(msg, sk, skb);
> > @@ -3538,11 +3547,15 @@ static int packet_recvmsg(struct socket *sock, =
struct msghdr *msg, size_t len,
> >               if (skb_vlan_tag_present(skb)) {
> >                       aux.tp_vlan_tci =3D skb_vlan_tag_get(skb);
> >                       aux.tp_vlan_tpid =3D ntohs(skb->vlan_proto);
> > -                     aux.tp_status |=3D TP_STATUS_VLAN_VALID | TP_STAT=
US_VLAN_TPID_VALID;
> > +             } else if (eth_type_vlan(skb->protocol)) {
> > +                     aux.tp_vlan_tci =3D ntohs(vlan_eth_hdr(skb)->h_vl=
an_TCI);
> > +                     aux.tp_vlan_tpid =3D ntohs(skb->protocol);
> >               } else {
> >                       aux.tp_vlan_tci =3D 0;
> >                       aux.tp_vlan_tpid =3D 0;
> >               }
> > +             if (aux.tp_vlan_tci || aux.tp_vlan_tpid)
> > +                     aux.tp_status |=3D TP_STATUS_VLAN_VALID | TP_STAT=
US_VLAN_TPID_VALID;
> >               put_cmsg(msg, SOL_PACKET, PACKET_AUXDATA, sizeof(aux), &a=
ux);
> >       }
> >
> > --
> > 2.40.1
> >
>
>