ebpf: add XDP load balancing code

This patch uses CPU redirect map to do load balancing. This is a
simplified version of xdp_filter that includes code for bypass.

diff --git a/ebpf/Makefile.am b/ebpf/Makefile.am
index bc5d7c61e437458b993ac7558ae0452d671b24c7..82bab6f08c1658a1ade6609ed633ab2f63c7bf5d 100644 (file)
--- a/ebpf/Makefile.am
+++ b/ebpf/Makefile.am
@@ -7,11 +7,12 @@ BPF_TARGETS  = lb.bpf
 BPF_TARGETS += filter.bpf
 BPF_TARGETS += bypass_filter.bpf
 BPF_TARGETS += xdp_filter.bpf
+BPF_TARGETS += xdp_lb.bpf
 BPF_TARGETS += vlan_filter.bpf
 
 all: $(BPF_TARGETS)
 
-EXTRA_DIST= include bypass_filter.c filter.c lb.c vlan_filter.c xdp_filter.c
+EXTRA_DIST= include bypass_filter.c filter.c lb.c vlan_filter.c xdp_filter.c xdp_lb.c
 
 $(BPF_TARGETS): %.bpf: %.c
 #      From C-code to LLVM-IR format suffix .ll (clang -S -emit-llvm)

diff --git a/ebpf/xdp_lb.c b/ebpf/xdp_lb.c
new file mode 100644 (file)
index 0000000..f2a9997
--- /dev/null
+++ b/ebpf/xdp_lb.c
@@ -0,0 +1,173 @@
+/* Copyright (C) 2019 Open Information Security Foundation
+ *
+ * You can copy, redistribute or modify this Program under the terms of
+ * the GNU General Public License version 2 as published by the Free
+ * Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * version 2 along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#define KBUILD_MODNAME "foo"
+#include <stddef.h>
+#include <linux/bpf.h>
+
+#include <linux/in.h>
+#include <linux/if_ether.h>
+#include <linux/if_packet.h>
+#include <linux/if_vlan.h>
+#include <linux/ip.h>
+#include <linux/ipv6.h>
+#include <linux/tcp.h>
+#include <linux/udp.h>
+#include "bpf_helpers.h"
+
+#include "hash_func01.h"
+
+#define LINUX_VERSION_CODE 263682
+
+/* Hashing initval */
+#define INITVAL 15485863
+
+/* Increase CPUMAP_MAX_CPUS if ever you have more than 64 CPUs */
+#define CPUMAP_MAX_CPUS     64
+
+struct vlan_hdr {
+    __u16      h_vlan_TCI;
+    __u16      h_vlan_encapsulated_proto;
+};
+
+/* Special map type that can XDP_REDIRECT frames to another CPU */
+struct bpf_map_def SEC("maps") cpu_map = {
+    .type              = BPF_MAP_TYPE_CPUMAP,
+    .key_size  = sizeof(__u32),
+    .value_size        = sizeof(__u32),
+    .max_entries       = CPUMAP_MAX_CPUS,
+};
+
+struct bpf_map_def SEC("maps") cpus_available = {
+    .type              = BPF_MAP_TYPE_ARRAY,
+    .key_size  = sizeof(__u32),
+    .value_size        = sizeof(__u32),
+    .max_entries       = CPUMAP_MAX_CPUS,
+};
+
+struct bpf_map_def SEC("maps") cpus_count = {
+    .type              = BPF_MAP_TYPE_ARRAY,
+    .key_size  = sizeof(__u32),
+    .value_size        = sizeof(__u32),
+    .max_entries       = 1,
+};
+
+static int __always_inline filter_ipv4(struct xdp_md *ctx, void *data, __u64 nh_off, void *data_end)
+{
+    struct iphdr *iph = data + nh_off;
+    __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);
+
+    if (cpu_max && *cpu_max) {
+        cpu_dest = cpu_hash % *cpu_max;
+        cpu_selected = bpf_map_lookup_elem(&cpus_available, &cpu_dest);
+        if (!cpu_selected)
+            return XDP_ABORTED;
+        cpu_dest = *cpu_selected;
+        return bpf_redirect_map(&cpu_map, cpu_dest, 0);
+    } else {
+        return XDP_PASS;
+    }
+}
+
+static int __always_inline filter_ipv6(struct xdp_md *ctx, void *data, __u64 nh_off, void *data_end)
+{
+    struct ipv6hdr *ip6h = data + nh_off;
+    __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];
+    cpu_hash += ip6h->saddr.s6_addr32[2] + ip6h->daddr.s6_addr32[2];
+    cpu_hash += ip6h->saddr.s6_addr32[3] + ip6h->daddr.s6_addr32[3];
+    cpu_hash = SuperFastHash((char *)&cpu_hash, 4, INITVAL);
+
+    if (cpu_max && *cpu_max) {
+        cpu_dest = cpu_hash % *cpu_max;
+        cpu_selected = bpf_map_lookup_elem(&cpus_available, &cpu_dest);
+        if (!cpu_selected)
+            return XDP_ABORTED;
+        cpu_dest = *cpu_selected;
+        return bpf_redirect_map(&cpu_map, cpu_dest, 0);
+    } else {
+        return XDP_PASS;
+    }
+
+    return XDP_PASS;
+}
+
+int SEC("xdp") xdp_loadfilter(struct xdp_md *ctx)
+{
+    void *data_end = (void *)(long)ctx->data_end;
+    void *data = (void *)(long)ctx->data;
+    struct ethhdr *eth = data;
+    __u16 h_proto;
+    __u64 nh_off;
+
+    nh_off = sizeof(*eth);
+    if (data + nh_off > data_end)
+        return XDP_PASS;
+
+    h_proto = eth->h_proto;
+
+    if (h_proto == __constant_htons(ETH_P_8021Q) || h_proto == __constant_htons(ETH_P_8021AD)) {
+        struct vlan_hdr *vhdr;
+
+        vhdr = data + nh_off;
+        nh_off += sizeof(struct vlan_hdr);
+        if (data + nh_off > data_end)
+            return XDP_PASS;
+        h_proto = vhdr->h_vlan_encapsulated_proto;
+    }
+    if (h_proto == __constant_htons(ETH_P_8021Q) || h_proto == __constant_htons(ETH_P_8021AD)) {
+        struct vlan_hdr *vhdr;
+
+        vhdr = data + nh_off;
+        nh_off += sizeof(struct vlan_hdr);
+        if (data + nh_off > data_end)
+            return XDP_PASS;
+        h_proto = vhdr->h_vlan_encapsulated_proto;
+    }
+
+    if (h_proto == __constant_htons(ETH_P_IP))
+        return filter_ipv4(ctx, data, nh_off, data_end);
+    else if (h_proto == __constant_htons(ETH_P_IPV6))
+        return filter_ipv6(ctx, data, nh_off, data_end);
+
+    return XDP_PASS;
+}
+
+char __license[] SEC("license") = "GPL";
+
+__u32 __version SEC("version") = LINUX_VERSION_CODE;