ebpf: add tunnel aware load balancing

This patch decapsulates GRE tunnel in xdp_lb

ebpf/xdp_lb.c

@@ -23,6 +23,7 @@
 #include <linux/if_ether.h>
 #include <linux/if_packet.h>
 #include <linux/if_vlan.h>
+#include <linux/if_tunnel.h>
 #include <linux/ip.h>
 #include <linux/ipv6.h>
 #include <linux/tcp.h>
@@ -66,18 +67,18 @@ struct bpf_map_def SEC("maps") cpus_count = {
     .max_entries	= 1,
 };
 
-static int __always_inline filter_ipv4(struct xdp_md *ctx, void *data, __u64 nh_off, void *data_end)
+static int __always_inline hash_ipv4(void *data, void *data_end)
 {
-    struct iphdr *iph = data + nh_off;
+    struct iphdr *iph = data;
+    if ((void *)(iph + 1) > data_end)
+        return XDP_PASS;
+
     __u32 key0 = 0;
     __u32 cpu_dest;
     __u32 *cpu_max = bpf_map_lookup_elem(&cpus_count, &key0);
     __u32 *cpu_selected;
     __u32 cpu_hash;
 
-    if ((void *)(iph + 1) > data_end)
-        return XDP_PASS;
-
     /* IP-pairs hit same CPU */
     cpu_hash = iph->saddr + iph->daddr;
     cpu_hash = SuperFastHash((char *)&cpu_hash, 4, INITVAL);
@@ -94,18 +95,18 @@ static int __always_inline filter_ipv4(struct xdp_md *ctx, void *data, __u64 nh_
     }
 }
 
-static int __always_inline filter_ipv6(struct xdp_md *ctx, void *data, __u64 nh_off, void *data_end)
+static int __always_inline hash_ipv6(void *data, void *data_end)
 {
-    struct ipv6hdr *ip6h = data + nh_off;
+    struct ipv6hdr *ip6h = data;
+    if ((void *)(ip6h + 1) > data_end)
+        return XDP_PASS;
+
     __u32 key0 = 0;
     __u32 cpu_dest;
     __u32 *cpu_max = bpf_map_lookup_elem(&cpus_count, &key0);
     __u32 *cpu_selected;
     __u32 cpu_hash;
 
-    if ((void *)(ip6h + 1) > data_end)
-        return XDP_PASS;
-
     /* IP-pairs hit same CPU */
     cpu_hash  = ip6h->saddr.s6_addr32[0] + ip6h->daddr.s6_addr32[0];
     cpu_hash += ip6h->saddr.s6_addr32[1] + ip6h->daddr.s6_addr32[1];
@@ -127,6 +128,75 @@ static int __always_inline filter_ipv6(struct xdp_md *ctx, void *data, __u64 nh_
     return XDP_PASS;
 }
 
+static int __always_inline filter_gre(struct xdp_md *ctx, void *data, __u64 nh_off, void *data_end)
+{
+    struct iphdr *iph = data + nh_off;
+    __be16 proto;
+    struct gre_hdr {
+        __be16 flags;
+        __be16 proto;
+    };
+
+    nh_off += sizeof(struct iphdr);
+    struct gre_hdr *grhdr = (struct gre_hdr *)(iph + 1);
+
+    if ((void *)(grhdr + 1) > data_end)
+        return XDP_PASS;
+
+    if (grhdr->flags & (GRE_VERSION|GRE_ROUTING))
+        return XDP_PASS;
+
+    nh_off += 4;
+    proto = grhdr->proto;
+    if (grhdr->flags & GRE_CSUM)
+        nh_off += 4;
+    if (grhdr->flags & GRE_KEY)
+        nh_off += 4;
+    if (grhdr->flags & GRE_SEQ)
+        nh_off += 4;
+
+    if (data + nh_off > data_end)
+        return XDP_PASS;
+    if (bpf_xdp_adjust_head(ctx, 0 + nh_off))
+        return XDP_PASS;
+
+    data = (void *)(long)ctx->data;
+    data_end = (void *)(long)ctx->data_end;
+
+    if (proto == __constant_htons(ETH_P_8021Q)) {
+        struct vlan_hdr *vhdr = (struct vlan_hdr *)(data);
+        if ((void *)(vhdr + 1) > data_end)
+            return XDP_PASS;
+        proto = vhdr->h_vlan_encapsulated_proto;
+        nh_off += sizeof(struct vlan_hdr);
+    }
+
+    if (proto == __constant_htons(ETH_P_IP)) {
+        return hash_ipv4(data, data_end);
+    } else if (proto == __constant_htons(ETH_P_IPV6)) {
+        return hash_ipv6(data, data_end);
+    } else
+        return XDP_PASS;
+}
+
+static int __always_inline filter_ipv4(struct xdp_md *ctx, void *data, __u64 nh_off, void *data_end)
+{
+    struct iphdr *iph = data + nh_off;
+    if ((void *)(iph + 1) > data_end)
+        return XDP_PASS;
+
+    if (iph->protocol == IPPROTO_GRE) {
+        return filter_gre(ctx, data, nh_off, data_end);
+    }
+    return hash_ipv4(data + nh_off, data_end);
+}
+
+static int __always_inline filter_ipv6(struct xdp_md *ctx, void *data, __u64 nh_off, void *data_end)
+{
+    struct ipv6hdr *ip6h = data + nh_off;
+    return hash_ipv6((void *)ip6h, data_end);
+}
+
 int SEC("xdp") xdp_loadfilter(struct xdp_md *ctx)
 {
     void *data_end = (void *)(long)ctx->data_end;
@@ -141,6 +211,12 @@ int SEC("xdp") xdp_loadfilter(struct xdp_md *ctx)
 
     h_proto = eth->h_proto;
 
+#if 0
+    if (h_proto != __constant_htons(ETH_P_IP)) {
+        char fmt[] = "Current proto: %u\n";
+        bpf_trace_printk(fmt, sizeof(fmt), h_proto);
+    }
+#endif
     if (h_proto == __constant_htons(ETH_P_8021Q) || h_proto == __constant_htons(ETH_P_8021AD)) {
         struct vlan_hdr *vhdr;