net: pppoe: implement GRO support

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 <[email protected]>
Signed-off-by: NipaLocal <nipa@local>

Author: nbd168

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 +1174,105 @@ 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;
+	__be16 type;
+	int flush = 1;
+
+	off_pppoe = skb_gro_offset(skb);
+	hlen = off_pppoe + sizeof(*phdr) + 2;
+	phdr = skb_gro_header(skb, hlen, 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 - 2)
+		goto out;
+
+	NAPI_GRO_CB(skb)->network_offsets[NAPI_GRO_CB(skb)->encap_mark] = hlen;
+
+	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 = ptype->callbacks.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 err = -ENOENT;
+
+	ptype = gro_find_complete_by_type(type);
+	if (ptype)
+		err = ptype->callbacks.gro_complete(skb, nhoff +
+						    sizeof(*phdr) + 2);
+
+	return err;
+}
+
+static struct packet_offload pppoe_packet_offload __read_mostly = {
+	.type = cpu_to_be16(ETH_P_PPP_SES),
+	.priority = 10,
+	.callbacks = {
+		.gro_receive = pppoe_gro_receive,
+		.gro_complete = pppoe_gro_complete,
+	},
+};
+
 static int __init pppoe_init(void)
 {
 	int err;
@@ -1189,6 +1289,7 @@ 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 +1309,7 @@ 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);