Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <17b4e630c63657249a7268943f8806004de4cdca.camel@redhat.com>
Subject: Re: [PATCH v2 1/1] net: gro: fix misuse of CB in udp socket lookup
From:   Paolo Abeni <pabeni@redhat.com>
To:     Richard Gobert <richardbgobert@gmail.com>, davem@davemloft.net,
        edumazet@google.com, kuba@kernel.org,
        willemdebruijn.kernel@gmail.com, dsahern@kernel.org,
        tom@herbertland.com, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org, gal@nvidia.com
Date:   Tue, 25 Jul 2023 09:15:25 +0200
In-Reply-To: <20230720162624.GA16428@debian>
References: <20230720161322.GA16323@debian> <20230720162624.GA16428@debian>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.46.4 (3.46.4-1.fc37) 
MIME-Version: 1.0
Precedence: bulk

Hi Richard,

On Thu, 2023-07-20 at 18:26 +0200, Richard Gobert wrote:
> This patch fixes a misuse of IP{6}CB(skb) in GRO, while calling to
> `udp6_lib_lookup2` when handling udp tunnels. `udp6_lib_lookup2` fetch th=
e
> device from CB. The fix changes it to fetch the device from `skb->dev`.
> l3mdev case requires special attention since it has a master and a slave
> device.
>=20
> Fixes: a6024562ffd7 ("udp: Add GRO functions to UDP socket")
> Reported-by: Gal Pressman <gal@nvidia.com>
> Signed-off-by: Richard Gobert <richardbgobert@gmail.com>
> ---
>  include/net/udp.h      |  2 ++
>  net/ipv4/udp.c         | 28 ++++++++++++++++++++++++++--
>  net/ipv4/udp_offload.c |  7 +++++--
>  net/ipv6/udp.c         | 29 +++++++++++++++++++++++++++--
>  net/ipv6/udp_offload.c |  7 +++++--
>  5 files changed, 65 insertions(+), 8 deletions(-)
>=20
> diff --git a/include/net/udp.h b/include/net/udp.h
> index 4d13424f8f72..48af1479882f 100644
> --- a/include/net/udp.h
> +++ b/include/net/udp.h
> @@ -299,6 +299,7 @@ int udp_lib_getsockopt(struct sock *sk, int level, in=
t optname,
>  int udp_lib_setsockopt(struct sock *sk, int level, int optname,
>  		       sockptr_t optval, unsigned int optlen,
>  		       int (*push_pending_frames)(struct sock *));
> +void udp4_get_iif_sdif(const struct sk_buff *skb, int *iif, int *sdif);
>  struct sock *udp4_lib_lookup(struct net *net, __be32 saddr, __be16 sport=
,
>  			     __be32 daddr, __be16 dport, int dif);
>  struct sock *__udp4_lib_lookup(struct net *net, __be32 saddr, __be16 spo=
rt,
> @@ -310,6 +311,7 @@ struct sock *udp6_lib_lookup(struct net *net,
>  			     const struct in6_addr *saddr, __be16 sport,
>  			     const struct in6_addr *daddr, __be16 dport,
>  			     int dif);
> +void udp6_get_iif_sdif(const struct sk_buff *skb, int *iif, int *sdif);
>  struct sock *__udp6_lib_lookup(struct net *net,
>  			       const struct in6_addr *saddr, __be16 sport,
>  			       const struct in6_addr *daddr, __be16 dport,
> diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c
> index 8c3ebd95f5b9..85eb9977db2c 100644
> --- a/net/ipv4/udp.c
> +++ b/net/ipv4/udp.c
> @@ -550,15 +550,39 @@ static inline struct sock *__udp4_lib_lookup_skb(st=
ruct sk_buff *skb,
>  				 inet_sdif(skb), udptable, skb);
>  }
> =20
> +/* This function is the alternative to 'inet_iif' and 'inet_sdif'
> + * functions in case we can not rely on fields of IPCB.
> + *
> + * The caller must verify skb_valid_dst(skb) is false and skb->dev is in=
itialized.
> + * The caller must hold the RCU read lock.
> + */
> +inline void udp4_get_iif_sdif(const struct sk_buff *skb, int *iif, int *=
sdif)

I think you misread David Ahern's suggestion on v1. The idea would be
to move this function (and udp6_get_iif_sdif) in an header file, as
'static inline'[1]. Additionally there is nothing specific about UDP
here so I would rename them inet{,6}_gro_iif_sdif and place them in
include/net/gro.h.

Otherwise LGTM.

Thanks,

Paolo

[1] the usage of the "inline" keyword is basically allowed only in
header files