On Thu, Feb 22, 2024 at 10:05 PM Richard Gobert
<richardbgobert@xxxxxxxxx> wrote:
>
> Commits a602456 ("udp: Add GRO functions to UDP socket") and 57c67ff ("udp:
> additional GRO support") introduce incorrect usage of {ip,ipv6}_hdr in the
> complete phase of gro. The functions always return skb->network_header,
> which in the case of encapsulated packets at the gro complete phase, is
> always set to the innermost L3 of the packet. That means that calling
> {ip,ipv6}_hdr for skbs which completed the GRO receive phase (both in
> gro_list and *_gro_complete) when parsing an encapsulated packet's _outer_
> L3/L4 may return an unexpected value.
>
> This incorrect usage leads to a bug in GRO's UDP socket lookup.
> udp{4,6}_lib_lookup_skb functions use ip_hdr/ipv6_hdr respectively. These
> *_hdr functions return network_header which will point to the innermost L3,
> resulting in the wrong offset being used in __udp{4,6}_lib_lookup with
> encapsulated packets.
>
> Reproduction example:
>
> Endpoint configuration example (fou + local address bind)
>
> # ip fou add port 6666 ipproto 4
> # ip link add name tun1 type ipip remote 2.2.2.1 local 2.2.2.2 encap fou encap-dport 5555 encap-sport 6666 mode ipip
> # ip link set tun1 up
> # ip a add 1.1.1.2/24 dev tun1
>
> Netperf TCP_STREAM result on net-next before patch is applied:
>
> net-next main, GRO enabled:
> $ netperf -H 1.1.1.2 -t TCP_STREAM -l 5
> Recv Send Send
> Socket Socket Message Elapsed
> Size Size Size Time Throughput
> bytes bytes bytes secs. 10^6bits/sec
>
> 131072 16384 16384 5.28 2.37
>
> net-next main, GRO disabled:
> $ netperf -H 1.1.1.2 -t TCP_STREAM -l 5
> Recv Send Send
> Socket Socket Message Elapsed
> Size Size Size Time Throughput
> bytes bytes bytes secs. 10^6bits/sec
>
> 131072 16384 16384 5.01 2745.06
>
> patch applied, GRO enabled:
> $ netperf -H 1.1.1.2 -t TCP_STREAM -l 5
> Recv Send Send
> Socket Socket Message Elapsed
> Size Size Size Time Throughput
> bytes bytes bytes secs. 10^6bits/sec
>
> 131072 16384 16384 5.01 2877.38
>
> This patch fixes this bug and prevents similar future misuse of
> network_header by setting network_header and inner_network_header to their
> respective values during the receive phase of GRO. This results in
> more coherent {inner_,}network_header values for every skb in gro_list,
> which also means there's no need to set/fix these values before passing
> the packet forward.
>
> network_header is already set in dev_gro_receive and under encapsulation we
> set inner_network_header. *_gro_complete functions use a new helper
> function - skb_gro_complete_network_header, which returns the
> network_header/inner_network_header offset during the GRO complete phase,
> depending on skb->encapsulation.
>
> Fixes: 57c67ff4bd92 ("udp: additional GRO support")
> Signed-off-by: Richard Gobert <richardbgobert@xxxxxxxxx>
> ---
> include/net/gro.h | 14 +++++++++++++-
> net/8021q/vlan_core.c | 3 +++
> net/ipv4/af_inet.c | 8 ++++----
> net/ipv4/tcp_offload.c | 2 +-
> net/ipv4/udp_offload.c | 2 +-
> net/ipv6/ip6_offload.c | 11 +++++------
> net/ipv6/tcpv6_offload.c | 2 +-
> net/ipv6/udp_offload.c | 2 +-
> 8 files changed, 29 insertions(+), 15 deletions(-)
>
> diff --git a/include/net/gro.h b/include/net/gro.h
> index b435f0ddbf64..89502a7e35ed 100644
> --- a/include/net/gro.h
> +++ b/include/net/gro.h
> @@ -177,10 +177,22 @@ static inline void *skb_gro_header(struct sk_buff *skb,
> return ptr;
> }
>
> +static inline int skb_gro_network_offset(struct sk_buff *skb)
> +{
> + return NAPI_GRO_CB(skb)->encap_mark ? skb_inner_network_offset(skb) :
> + skb_network_offset(skb);
> +}
> +
> static inline void *skb_gro_network_header(struct sk_buff *skb)
> {
> return (NAPI_GRO_CB(skb)->frag0 ?: skb->data) +
> - skb_network_offset(skb);
> + skb_gro_network_offset(skb);
> +}
> +
> +static inline void *skb_gro_complete_network_header(struct sk_buff *skb)
> +{
> + return skb->encapsulation ? skb_inner_network_header(skb) :
> + skb_network_header(skb);
> }
>
> static inline __wsum inet_gro_compute_pseudo(struct sk_buff *skb, int proto)
> diff --git a/net/8021q/vlan_core.c b/net/8021q/vlan_core.c
> index f00158234505..8bc871397e47 100644
> --- a/net/8021q/vlan_core.c
> +++ b/net/8021q/vlan_core.c
> @@ -478,6 +478,9 @@ static struct sk_buff *vlan_gro_receive(struct list_head *head,
> if (unlikely(!vhdr))
> goto out;
>
> + if (!NAPI_GRO_CB(skb)->encap_mark)
> + skb_set_network_header(skb, hlen);
> +
> type = vhdr->h_vlan_encapsulated_proto;
>
> ptype = gro_find_receive_by_type(type);
> diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
> index 835f4f9d98d2..c0f3c162bf73 100644
> --- a/net/ipv4/af_inet.c
> +++ b/net/ipv4/af_inet.c
> @@ -1564,7 +1564,9 @@ struct sk_buff *inet_gro_receive(struct list_head *head, struct sk_buff *skb)
>
> NAPI_GRO_CB(skb)->is_atomic = !!(iph->frag_off & htons(IP_DF));
> NAPI_GRO_CB(skb)->flush |= flush;
> - skb_set_network_header(skb, off);
> + if (NAPI_GRO_CB(skb)->encap_mark)
> + skb_set_inner_network_header(skb, off);
> +
> /* The above will be needed by the transport layer if there is one
> * immediately following this IP hdr.
> */
> @@ -1643,10 +1645,8 @@ int inet_gro_complete(struct sk_buff *skb, int nhoff)
> int proto = iph->protocol;
> int err = -ENOSYS;
>
> - if (skb->encapsulation) {
> + if (skb->encapsulation)
> skb_set_inner_protocol(skb, cpu_to_be16(ETH_P_IP));
> - skb_set_inner_network_header(skb, nhoff);
> - }
>
> iph_set_totlen(iph, skb->len - nhoff);
> csum_replace2(&iph->check, totlen, iph->tot_len);
> diff --git a/net/ipv4/tcp_offload.c b/net/ipv4/tcp_offload.c
> index 8311c38267b5..8bbcd3f502ac 100644
> --- a/net/ipv4/tcp_offload.c
> +++ b/net/ipv4/tcp_offload.c
> @@ -330,7 +330,7 @@ struct sk_buff *tcp4_gro_receive(struct list_head *head, struct sk_buff *skb)
>
> INDIRECT_CALLABLE_SCOPE int tcp4_gro_complete(struct sk_buff *skb, int thoff)
> {
> - const struct iphdr *iph = ip_hdr(skb);
> + const struct iphdr *iph = skb_gro_complete_network_header(skb);
> struct tcphdr *th = tcp_hdr(skb);
>
> th->check = ~tcp_v4_check(skb->len - thoff, iph->saddr,
> diff --git a/net/ipv4/udp_offload.c b/net/ipv4/udp_offload.c
> index 6c95d28d0c4a..7f59cede67f5 100644
> --- a/net/ipv4/udp_offload.c
> +++ b/net/ipv4/udp_offload.c
> @@ -709,7 +709,7 @@ EXPORT_SYMBOL(udp_gro_complete);
>
> INDIRECT_CALLABLE_SCOPE int udp4_gro_complete(struct sk_buff *skb, int nhoff)
> {
> - const struct iphdr *iph = ip_hdr(skb);
> + const struct iphdr *iph = skb_gro_complete_network_header(skb);
> struct udphdr *uh = (struct udphdr *)(skb->data + nhoff);
>
> /* do fraglist only if there is no outer UDP encap (or we already processed it) */
> diff --git a/net/ipv6/ip6_offload.c b/net/ipv6/ip6_offload.c
> index cca64c7809be..db7e3db587b9 100644
> --- a/net/ipv6/ip6_offload.c
> +++ b/net/ipv6/ip6_offload.c
> @@ -67,7 +67,7 @@ static int ipv6_gro_pull_exthdrs(struct sk_buff *skb, int off, int proto)
> off += len;
> }
>
> - skb_gro_pull(skb, off - skb_network_offset(skb));
> + skb_gro_pull(skb, off - skb_gro_network_offset(skb));
> return proto;
> }
>
> @@ -236,7 +236,8 @@ INDIRECT_CALLABLE_SCOPE struct sk_buff *ipv6_gro_receive(struct list_head *head,
> if (unlikely(!iph))
> goto out;
>
> - skb_set_network_header(skb, off);
> + if (NAPI_GRO_CB(skb)->encap_mark)
> + skb_set_inner_network_header(skb, off);
>
> flush += ntohs(iph->payload_len) != skb->len - hlen;
>
> @@ -259,7 +260,7 @@ INDIRECT_CALLABLE_SCOPE struct sk_buff *ipv6_gro_receive(struct list_head *head,
> NAPI_GRO_CB(skb)->proto = proto;
>
> flush--;
> - nlen = skb_network_header_len(skb);
> + nlen = skb_gro_offset(skb) - off;
>
> list_for_each_entry(p, head, list) {
> const struct ipv6hdr *iph2;
> @@ -353,10 +354,8 @@ INDIRECT_CALLABLE_SCOPE int ipv6_gro_complete(struct sk_buff *skb, int nhoff)
> int err = -ENOSYS;
> u32 payload_len;
>
> - if (skb->encapsulation) {
> + if (skb->encapsulation)
> skb_set_inner_protocol(skb, cpu_to_be16(ETH_P_IPV6));
> - skb_set_inner_network_header(skb, nhoff);
> - }
>
> payload_len = skb->len - nhoff - sizeof(*iph);
> if (unlikely(payload_len > IPV6_MAXPLEN)) {
> diff --git a/net/ipv6/tcpv6_offload.c b/net/ipv6/tcpv6_offload.c
> index bf0c957e4b5e..79eeaced2834 100644
> --- a/net/ipv6/tcpv6_offload.c
> +++ b/net/ipv6/tcpv6_offload.c
> @@ -29,7 +29,7 @@ struct sk_buff *tcp6_gro_receive(struct list_head *head, struct sk_buff *skb)
>
> INDIRECT_CALLABLE_SCOPE int tcp6_gro_complete(struct sk_buff *skb, int thoff)
> {
> - const struct ipv6hdr *iph = ipv6_hdr(skb);
> + const struct ipv6hdr *iph = skb_gro_complete_network_header(skb);
> struct tcphdr *th = tcp_hdr(skb);
>
> th->check = ~tcp_v6_check(skb->len - thoff, &iph->saddr,
> diff --git a/net/ipv6/udp_offload.c b/net/ipv6/udp_offload.c
> index 6b95ba241ebe..897caa2e39fb 100644
> --- a/net/ipv6/udp_offload.c
> +++ b/net/ipv6/udp_offload.c
> @@ -164,7 +164,7 @@ struct sk_buff *udp6_gro_receive(struct list_head *head, struct sk_buff *skb)
>
> INDIRECT_CALLABLE_SCOPE int udp6_gro_complete(struct sk_buff *skb, int nhoff)
> {
> - const struct ipv6hdr *ipv6h = ipv6_hdr(skb);
> + const struct ipv6hdr *ipv6h = skb_gro_complete_network_header(skb);
> struct udphdr *uh = (struct udphdr *)(skb->data + nhoff);
My intuition is that this patch has a high cost for normal GRO processing.
SW-GRO is already a bottleneck on ARM cores in smart NICS.
I would suggest instead using parameters to give both the nhoff and thoff values
this would avoid many conditionals in the fast path.
->
INDIRECT_CALLABLE_SCOPE int udp6_gro_complete(struct sk_buff *skb, int
nhoff, int thoff)
{
const struct ipv6hdr *ipv6h = (const struct ipv6hdr *)(skb->data + nhoff);
struct udphdr *uh = (struct udphdr *)(skb->data + thoff);
...
}
INDIRECT_CALLABLE_SCOPE int tcp6_gro_complete(struct sk_buff *skb, int
nhoff, int thoff)
{
const struct ipv6hdr *iph = (const struct ipv6hdr *)(skb->data + nhoff);
struct tcphdr *th = (struct tcphdr *)(skb->data + thoff);
Why storing in skb fields things that really could be propagated more
efficiently as function parameters ?
