Re: [PATCH net-next v3] net: pppoe: implement GRO/GSO support

[Richard Gobert <richardbgobert@gmail.com>]

Felix Fietkau wrote:
> Only handles packets where the pppoe header length field matches the exact
> packet length. Significantly improves rx throughput.
> 
> When running NAT traffic through a MediaTek MT7621 devices from a host
> behind PPPoE to a host directly connected via ethernet, the TCP throughput
> that the device is able to handle improves from ~130 Mbit/s to ~630 Mbit/s,
> using fraglist GRO.
> 
> Signed-off-by: Felix Fietkau <nbd@nbd.name>
> ---
> v2: fix compile error
> v3:
>   - increase priority value
>   - implement GSO support
>   - use INDIRECT_CALL_INET
>   - update pppoe length field
>   - remove unnecessary network_offsets update
> 
>  drivers/net/ppp/pppoe.c | 160 +++++++++++++++++++++++++++++++++++++++-
>  net/ipv4/af_inet.c      |   2 +
>  net/ipv6/ip6_offload.c  |   2 +
>  3 files changed, 163 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/ppp/pppoe.c b/drivers/net/ppp/pppoe.c
> index 410effa42ade..a8d8eb870bce 100644
> --- a/drivers/net/ppp/pppoe.c
> +++ b/drivers/net/ppp/pppoe.c
> @@ -77,6 +77,7 @@
>  #include <net/net_namespace.h>
>  #include <net/netns/generic.h>
>  #include <net/sock.h>
> +#include <net/gro.h>
>  
>  #include <linux/uaccess.h>
>  
> @@ -435,7 +436,7 @@ static int pppoe_rcv(struct sk_buff *skb, struct net_device *dev,
>  	if (skb->len < len)
>  		goto drop;
>  
> -	if (pskb_trim_rcsum(skb, len))
> +	if (!skb_is_gso(skb) && pskb_trim_rcsum(skb, len))
>  		goto drop;
>  
>  	ph = pppoe_hdr(skb);
> @@ -1173,6 +1174,161 @@ static struct pernet_operations pppoe_net_ops = {
>  	.size = sizeof(struct pppoe_net),
>  };
>  
> +static u16
> +compare_pppoe_header(struct pppoe_hdr *phdr, struct pppoe_hdr *phdr2)
> +{
> +	return (__force __u16)((phdr->sid ^ phdr2->sid) |
> +			       (phdr->tag[0].tag_type ^ phdr2->tag[0].tag_type));
> +}
> +
> +static __be16 pppoe_hdr_proto(struct pppoe_hdr *phdr)
> +{
> +	switch (phdr->tag[0].tag_type) {
> +	case cpu_to_be16(PPP_IP):
> +		return cpu_to_be16(ETH_P_IP);
> +	case cpu_to_be16(PPP_IPV6):
> +		return cpu_to_be16(ETH_P_IPV6);
> +	default:
> +		return 0;
> +	}
> +
> +}
> +
> +static struct sk_buff *pppoe_gro_receive(struct list_head *head,
> +					 struct sk_buff *skb)
> +{
> +	const struct packet_offload *ptype;
> +	unsigned int hlen, off_pppoe;
> +	struct sk_buff *pp = NULL;
> +	struct pppoe_hdr *phdr;
> +	struct sk_buff *p;
> +	int flush = 1;
> +	__be16 type;
> +
> +	off_pppoe = skb_gro_offset(skb);
> +	hlen = off_pppoe + sizeof(*phdr);
> +	phdr = skb_gro_header(skb, hlen + 2, off_pppoe);
> +	if (unlikely(!phdr))
> +		goto out;
> +
> +	/* ignore packets with padding or invalid length */
> +	if (skb_gro_len(skb) != be16_to_cpu(phdr->length) + hlen)
> +		goto out;
> +
> +	type = pppoe_hdr_proto(phdr);
> +	if (!type)
> +		goto out;
> +
> +	ptype = gro_find_receive_by_type(type);
> +	if (!ptype)
> +		goto out;
> +
> +	flush = 0;
> +
> +	list_for_each_entry(p, head, list) {
> +		struct pppoe_hdr *phdr2;
> +
> +		if (!NAPI_GRO_CB(p)->same_flow)
> +			continue;
> +
> +		phdr2 = (struct pppoe_hdr *)(p->data + off_pppoe);
> +		if (compare_pppoe_header(phdr, phdr2))
> +			NAPI_GRO_CB(p)->same_flow = 0;
> +	}
> +
> +	skb_gro_pull(skb, sizeof(*phdr) + 2);
> +	skb_gro_postpull_rcsum(skb, phdr, sizeof(*phdr) + 2);
> +
> +	pp = indirect_call_gro_receive_inet(ptype->callbacks.gro_receive,
> +					    ipv6_gro_receive, inet_gro_receive,
> +					    head, skb);
> +
> +out:
> +	skb_gro_flush_final(skb, pp, flush);
> +
> +	return pp;
> +}
> +
> +static int pppoe_gro_complete(struct sk_buff *skb, int nhoff)
> +{
> +	struct pppoe_hdr *phdr = (struct pppoe_hdr *)(skb->data + nhoff);
> +	__be16 type = pppoe_hdr_proto(phdr);
> +	struct packet_offload *ptype;
> +	int len, err;
> +
> +	ptype = gro_find_complete_by_type(type);
> +	if (!ptype)
> +		return -ENOENT;
> +
> +	err = INDIRECT_CALL_INET(ptype->callbacks.gro_complete,
> +				 ipv6_gro_complete, inet_gro_complete,
> +				 skb, nhoff + sizeof(*phdr) + 2);
> +	if (err)
> +		return err;
> +
> +	len = skb->len - (nhoff + sizeof(*phdr));
> +	phdr->length = cpu_to_be16(len);
> +
> +	return 0;
> +}
> +
> +static struct sk_buff *pppoe_gso_segment(struct sk_buff *skb,
> +					 netdev_features_t features)
> +{

I don't think this will be called for PPPoE over GRE packets,
since gre_gso_segment skips everything up to the network header.

> +	unsigned int pppoe_hlen = sizeof(struct pppoe_hdr) + 2;
> +	struct sk_buff *segs = ERR_PTR(-EINVAL);
> +	u16 mac_offset = skb->mac_header;
> +	struct packet_offload *ptype;
> +	u16 mac_len = skb->mac_len;
> +	struct pppoe_hdr *phdr;
> +	__be16 orig_type, type;
> +	int len, nhoff;
> +
> +	skb_reset_network_header(skb);
> +	nhoff = skb_network_header(skb) - skb_mac_header(skb);
> +
> +	if (unlikely(!pskb_may_pull(skb, pppoe_hlen)))
> +		goto out;
> +
> +	phdr = (struct pppoe_hdr *)skb_network_header(skb);
> +	type = pppoe_hdr_proto(phdr);
> +	ptype = gro_find_complete_by_type(type);
> +	if (!ptype)
> +		goto out;
> +
> +	orig_type = skb->protocol;
> +	__skb_pull(skb, pppoe_hlen);
> +	segs = ptype->callbacks.gso_segment(skb, features);
> +	if (IS_ERR_OR_NULL(segs)) {
> +		skb_gso_error_unwind(skb, orig_type, pppoe_hlen, mac_offset,
> +				     mac_len);
> +		goto out;
> +	}
> +
> +	skb = segs;
> +	do {
> +		phdr = (struct pppoe_hdr *)(skb_mac_header(skb) + nhoff);
> +		len = skb->len - (nhoff + sizeof(*phdr));
> +		phdr->length = cpu_to_be16(len);
> +		skb->network_header = (u8 *)phdr - skb->head;
> +		skb->protocol = orig_type;
> +		skb_reset_mac_len(skb);
> +	} while ((skb = skb->next));
> +
> +out:
> +	return segs;
> +}
> +
> +static struct packet_offload pppoe_packet_offload __read_mostly = {
> +	.type = cpu_to_be16(ETH_P_PPP_SES),
> +	.priority = 20,
> +	.callbacks = {
> +		.gro_receive = pppoe_gro_receive,
> +		.gro_complete = pppoe_gro_complete,
> +		.gso_segment = pppoe_gso_segment,
> +	},
> +};
> +
>  static int __init pppoe_init(void)
>  {
>  	int err;
> @@ -1189,6 +1345,7 @@ static int __init pppoe_init(void)
>  	if (err)
>  		goto out_unregister_pppoe_proto;
>  
> +	dev_add_offload(&pppoe_packet_offload);
>  	dev_add_pack(&pppoes_ptype);
>  	dev_add_pack(&pppoed_ptype);
>  	register_netdevice_notifier(&pppoe_notifier);
> @@ -1208,6 +1365,7 @@ static void __exit pppoe_exit(void)
>  	unregister_netdevice_notifier(&pppoe_notifier);
>  	dev_remove_pack(&pppoed_ptype);
>  	dev_remove_pack(&pppoes_ptype);
> +	dev_remove_offload(&pppoe_packet_offload);
>  	unregister_pppox_proto(PX_PROTO_OE);
>  	proto_unregister(&pppoe_sk_proto);
>  	unregister_pernet_device(&pppoe_net_ops);
> diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
> index 76e38092cd8a..0480a6d4f203 100644
> --- a/net/ipv4/af_inet.c
> +++ b/net/ipv4/af_inet.c
> @@ -1533,6 +1533,7 @@ struct sk_buff *inet_gro_receive(struct list_head *head, struct sk_buff *skb)
>  
>  	return pp;
>  }
> +EXPORT_INDIRECT_CALLABLE(inet_gro_receive);
>  
>  static struct sk_buff *ipip_gro_receive(struct list_head *head,
>  					struct sk_buff *skb)
> @@ -1618,6 +1619,7 @@ int inet_gro_complete(struct sk_buff *skb, int nhoff)
>  out:
>  	return err;
>  }
> +EXPORT_INDIRECT_CALLABLE(inet_gro_complete);
>  
>  static int ipip_gro_complete(struct sk_buff *skb, int nhoff)
>  {
> diff --git a/net/ipv6/ip6_offload.c b/net/ipv6/ip6_offload.c
> index fce91183797a..9e3640b018a4 100644
> --- a/net/ipv6/ip6_offload.c
> +++ b/net/ipv6/ip6_offload.c
> @@ -306,6 +306,7 @@ INDIRECT_CALLABLE_SCOPE struct sk_buff *ipv6_gro_receive(struct list_head *head,
>  
>  	return pp;
>  }
> +EXPORT_INDIRECT_CALLABLE(ipv6_gro_receive);
>  
>  static struct sk_buff *sit_ip6ip6_gro_receive(struct list_head *head,
>  					      struct sk_buff *skb)
> @@ -388,6 +389,7 @@ INDIRECT_CALLABLE_SCOPE int ipv6_gro_complete(struct sk_buff *skb, int nhoff)
>  out:
>  	return err;
>  }
> +EXPORT_INDIRECT_CALLABLE(ipv6_gro_complete);
>  
>  static int sit_gro_complete(struct sk_buff *skb, int nhoff)
>  {