samples: bpf: Convert xdp_router_ipv4 to XDP samples helper
authorLorenzo Bianconi <lorenzo@kernel.org>
Wed, 16 Mar 2022 07:13:23 +0000 (08:13 +0100)
committerAndrii Nakryiko <andrii@kernel.org>
Mon, 4 Apr 2022 00:12:09 +0000 (17:12 -0700)
Rely on the libbpf skeleton facility and other utilities provided by XDP
sample helpers in xdp_router_ipv4 sample.

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/7f4d98ee2c13c04d5eb924eebf79ced32fee8418.1647414711.git.lorenzo@kernel.org
samples/bpf/Makefile
samples/bpf/xdp_router_ipv4.bpf.c [new file with mode: 0644]
samples/bpf/xdp_router_ipv4_kern.c [deleted file]
samples/bpf/xdp_router_ipv4_user.c

index 38638845db9d7480ea0a218f56116ae19658d116..b4fa0e69aa14f943239860d9124be78eefb48f11 100644 (file)
@@ -96,7 +96,6 @@ test_cgrp2_sock2-objs := test_cgrp2_sock2.o
 xdp1-objs := xdp1_user.o
 # reuse xdp1 source intentionally
 xdp2-objs := xdp1_user.o
-xdp_router_ipv4-objs := xdp_router_ipv4_user.o
 test_current_task_under_cgroup-objs := $(CGROUP_HELPERS) \
                                       test_current_task_under_cgroup_user.o
 trace_event-objs := trace_event_user.o $(TRACE_HELPERS)
@@ -124,6 +123,7 @@ xdp_redirect_cpu-objs := xdp_redirect_cpu_user.o $(XDP_SAMPLE)
 xdp_redirect_map-objs := xdp_redirect_map_user.o $(XDP_SAMPLE)
 xdp_redirect-objs := xdp_redirect_user.o $(XDP_SAMPLE)
 xdp_monitor-objs := xdp_monitor_user.o $(XDP_SAMPLE)
+xdp_router_ipv4-objs := xdp_router_ipv4_user.o $(XDP_SAMPLE)
 
 # Tell kbuild to always build the programs
 always-y := $(tprogs-y)
@@ -153,7 +153,6 @@ always-y += parse_varlen.o parse_simple.o parse_ldabs.o
 always-y += test_cgrp2_tc_kern.o
 always-y += xdp1_kern.o
 always-y += xdp2_kern.o
-always-y += xdp_router_ipv4_kern.o
 always-y += test_current_task_under_cgroup_kern.o
 always-y += trace_event_kern.o
 always-y += sampleip_kern.o
@@ -342,6 +341,7 @@ $(obj)/xdp_redirect_map_multi_user.o: $(obj)/xdp_redirect_map_multi.skel.h
 $(obj)/xdp_redirect_map_user.o: $(obj)/xdp_redirect_map.skel.h
 $(obj)/xdp_redirect_user.o: $(obj)/xdp_redirect.skel.h
 $(obj)/xdp_monitor_user.o: $(obj)/xdp_monitor.skel.h
+$(obj)/xdp_router_ipv4_user.o: $(obj)/xdp_router_ipv4.skel.h
 
 $(obj)/tracex5_kern.o: $(obj)/syscall_nrs.h
 $(obj)/hbm_out_kern.o: $(src)/hbm.h $(src)/hbm_kern.h
@@ -399,6 +399,7 @@ $(obj)/xdp_redirect_map_multi.bpf.o: $(obj)/xdp_sample.bpf.o
 $(obj)/xdp_redirect_map.bpf.o: $(obj)/xdp_sample.bpf.o
 $(obj)/xdp_redirect.bpf.o: $(obj)/xdp_sample.bpf.o
 $(obj)/xdp_monitor.bpf.o: $(obj)/xdp_sample.bpf.o
+$(obj)/xdp_router_ipv4.bpf.o: $(obj)/xdp_sample.bpf.o
 
 $(obj)/%.bpf.o: $(src)/%.bpf.c $(obj)/vmlinux.h $(src)/xdp_sample.bpf.h $(src)/xdp_sample_shared.h
        @echo "  CLANG-BPF " $@
@@ -409,7 +410,8 @@ $(obj)/%.bpf.o: $(src)/%.bpf.c $(obj)/vmlinux.h $(src)/xdp_sample.bpf.h $(src)/x
                -c $(filter %.bpf.c,$^) -o $@
 
 LINKED_SKELS := xdp_redirect_cpu.skel.h xdp_redirect_map_multi.skel.h \
-               xdp_redirect_map.skel.h xdp_redirect.skel.h xdp_monitor.skel.h
+               xdp_redirect_map.skel.h xdp_redirect.skel.h xdp_monitor.skel.h \
+               xdp_router_ipv4.skel.h
 clean-files += $(LINKED_SKELS)
 
 xdp_redirect_cpu.skel.h-deps := xdp_redirect_cpu.bpf.o xdp_sample.bpf.o
@@ -417,6 +419,7 @@ xdp_redirect_map_multi.skel.h-deps := xdp_redirect_map_multi.bpf.o xdp_sample.bp
 xdp_redirect_map.skel.h-deps := xdp_redirect_map.bpf.o xdp_sample.bpf.o
 xdp_redirect.skel.h-deps := xdp_redirect.bpf.o xdp_sample.bpf.o
 xdp_monitor.skel.h-deps := xdp_monitor.bpf.o xdp_sample.bpf.o
+xdp_router_ipv4.skel.h-deps := xdp_router_ipv4.bpf.o xdp_sample.bpf.o
 
 LINKED_BPF_SRCS := $(patsubst %.bpf.o,%.bpf.c,$(foreach skel,$(LINKED_SKELS),$($(skel)-deps)))
 
diff --git a/samples/bpf/xdp_router_ipv4.bpf.c b/samples/bpf/xdp_router_ipv4.bpf.c
new file mode 100644 (file)
index 0000000..248119c
--- /dev/null
@@ -0,0 +1,180 @@
+/* Copyright (C) 2017 Cavium, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of version 2 of the GNU General Public License
+ * as published by the Free Software Foundation.
+ */
+
+#include "vmlinux.h"
+#include "xdp_sample.bpf.h"
+#include "xdp_sample_shared.h"
+
+#define ETH_ALEN       6
+#define ETH_P_8021Q    0x8100
+#define ETH_P_8021AD   0x88A8
+
+struct trie_value {
+       __u8 prefix[4];
+       __be64 value;
+       int ifindex;
+       int metric;
+       __be32 gw;
+};
+
+/* Key for lpm_trie */
+union key_4 {
+       u32 b32[2];
+       u8 b8[8];
+};
+
+struct arp_entry {
+       __be64 mac;
+       __be32 dst;
+};
+
+struct direct_map {
+       struct arp_entry arp;
+       int ifindex;
+       __be64 mac;
+};
+
+/* Map for trie implementation */
+struct {
+       __uint(type, BPF_MAP_TYPE_LPM_TRIE);
+       __uint(key_size, 8);
+       __uint(value_size, sizeof(struct trie_value));
+       __uint(max_entries, 50);
+       __uint(map_flags, BPF_F_NO_PREALLOC);
+} lpm_map SEC(".maps");
+
+/* Map for ARP table */
+struct {
+       __uint(type, BPF_MAP_TYPE_HASH);
+       __type(key, __be32);
+       __type(value, __be64);
+       __uint(max_entries, 50);
+} arp_table SEC(".maps");
+
+/* Map to keep the exact match entries in the route table */
+struct {
+       __uint(type, BPF_MAP_TYPE_HASH);
+       __type(key, __be32);
+       __type(value, struct direct_map);
+       __uint(max_entries, 50);
+} exact_match SEC(".maps");
+
+struct {
+       __uint(type, BPF_MAP_TYPE_DEVMAP);
+       __uint(key_size, sizeof(int));
+       __uint(value_size, sizeof(int));
+       __uint(max_entries, 100);
+} tx_port SEC(".maps");
+
+SEC("xdp")
+int xdp_router_ipv4_prog(struct xdp_md *ctx)
+{
+       void *data_end = (void *)(long)ctx->data_end;
+       void *data = (void *)(long)ctx->data;
+       struct ethhdr *eth = data;
+       u64 nh_off = sizeof(*eth);
+       struct datarec *rec;
+       __be16 h_proto;
+       u32 key = 0;
+
+       rec = bpf_map_lookup_elem(&rx_cnt, &key);
+       if (rec)
+               NO_TEAR_INC(rec->processed);
+
+       if (data + nh_off > data_end)
+               goto drop;
+
+       h_proto = eth->h_proto;
+       if (h_proto == bpf_htons(ETH_P_8021Q) ||
+           h_proto == bpf_htons(ETH_P_8021AD)) {
+               struct vlan_hdr *vhdr;
+
+               vhdr = data + nh_off;
+               nh_off += sizeof(struct vlan_hdr);
+               if (data + nh_off > data_end)
+                       goto drop;
+
+               h_proto = vhdr->h_vlan_encapsulated_proto;
+       }
+
+       switch (bpf_ntohs(h_proto)) {
+       case ETH_P_ARP:
+               if (rec)
+                       NO_TEAR_INC(rec->xdp_pass);
+               return XDP_PASS;
+       case ETH_P_IP: {
+               struct iphdr *iph = data + nh_off;
+               struct direct_map *direct_entry;
+               __be64 *dest_mac, *src_mac;
+               int forward_to;
+
+               if (iph + 1 > data_end)
+                       goto drop;
+
+               direct_entry = bpf_map_lookup_elem(&exact_match, &iph->daddr);
+
+               /* Check for exact match, this would give a faster lookup */
+               if (direct_entry && direct_entry->mac &&
+                   direct_entry->arp.mac) {
+                       src_mac = &direct_entry->mac;
+                       dest_mac = &direct_entry->arp.mac;
+                       forward_to = direct_entry->ifindex;
+               } else {
+                       struct trie_value *prefix_value;
+                       union key_4 key4;
+
+                       /* Look up in the trie for lpm */
+                       key4.b32[0] = 32;
+                       key4.b8[4] = iph->daddr & 0xff;
+                       key4.b8[5] = (iph->daddr >> 8) & 0xff;
+                       key4.b8[6] = (iph->daddr >> 16) & 0xff;
+                       key4.b8[7] = (iph->daddr >> 24) & 0xff;
+
+                       prefix_value = bpf_map_lookup_elem(&lpm_map, &key4);
+                       if (!prefix_value)
+                               goto drop;
+
+                       forward_to = prefix_value->ifindex;
+                       src_mac = &prefix_value->value;
+                       if (!src_mac)
+                               goto drop;
+
+                       dest_mac = bpf_map_lookup_elem(&arp_table, &iph->daddr);
+                       if (!dest_mac) {
+                               if (!prefix_value->gw)
+                                       goto drop;
+
+                               dest_mac = bpf_map_lookup_elem(&arp_table,
+                                                              &prefix_value->gw);
+                       }
+               }
+
+               if (src_mac && dest_mac) {
+                       int ret;
+
+                       __builtin_memcpy(eth->h_dest, dest_mac, ETH_ALEN);
+                       __builtin_memcpy(eth->h_source, src_mac, ETH_ALEN);
+
+                       ret = bpf_redirect_map(&tx_port, forward_to, 0);
+                       if (ret == XDP_REDIRECT) {
+                               if (rec)
+                                       NO_TEAR_INC(rec->xdp_redirect);
+                               return ret;
+                       }
+               }
+       }
+       default:
+               break;
+       }
+drop:
+       if (rec)
+               NO_TEAR_INC(rec->xdp_drop);
+
+       return XDP_DROP;
+}
+
+char _license[] SEC("license") = "GPL";
diff --git a/samples/bpf/xdp_router_ipv4_kern.c b/samples/bpf/xdp_router_ipv4_kern.c
deleted file mode 100644 (file)
index b37ca2b..0000000
+++ /dev/null
@@ -1,186 +0,0 @@
-/* Copyright (C) 2017 Cavium, Inc.
- *
- * This program is free software; you can redistribute it and/or modify it
- * under the terms of version 2 of the GNU General Public License
- * as published by the Free Software Foundation.
- */
-#define KBUILD_MODNAME "foo"
-#include <uapi/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 <bpf/bpf_helpers.h>
-#include <linux/slab.h>
-#include <net/ip_fib.h>
-
-struct trie_value {
-       __u8 prefix[4];
-       __be64 value;
-       int ifindex;
-       int metric;
-       __be32 gw;
-};
-
-/* Key for lpm_trie*/
-union key_4 {
-       u32 b32[2];
-       u8 b8[8];
-};
-
-struct arp_entry {
-       __be64 mac;
-       __be32 dst;
-};
-
-struct direct_map {
-       struct arp_entry arp;
-       int ifindex;
-       __be64 mac;
-};
-
-/* Map for trie implementation*/
-struct {
-       __uint(type, BPF_MAP_TYPE_LPM_TRIE);
-       __uint(key_size, 8);
-       __uint(value_size, sizeof(struct trie_value));
-       __uint(max_entries, 50);
-       __uint(map_flags, BPF_F_NO_PREALLOC);
-} lpm_map SEC(".maps");
-
-/* Map for counter*/
-struct {
-       __uint(type, BPF_MAP_TYPE_PERCPU_ARRAY);
-       __type(key, u32);
-       __type(value, u64);
-       __uint(max_entries, 256);
-} rxcnt SEC(".maps");
-
-/* Map for ARP table*/
-struct {
-       __uint(type, BPF_MAP_TYPE_HASH);
-       __type(key, __be32);
-       __type(value, __be64);
-       __uint(max_entries, 50);
-} arp_table SEC(".maps");
-
-/* Map to keep the exact match entries in the route table*/
-struct {
-       __uint(type, BPF_MAP_TYPE_HASH);
-       __type(key, __be32);
-       __type(value, struct direct_map);
-       __uint(max_entries, 50);
-} exact_match SEC(".maps");
-
-struct {
-       __uint(type, BPF_MAP_TYPE_DEVMAP);
-       __uint(key_size, sizeof(int));
-       __uint(value_size, sizeof(int));
-       __uint(max_entries, 100);
-} tx_port SEC(".maps");
-
-/* Function to set source and destination mac of the packet */
-static inline void set_src_dst_mac(void *data, void *src, void *dst)
-{
-       unsigned short *source = src;
-       unsigned short *dest  = dst;
-       unsigned short *p = data;
-
-       __builtin_memcpy(p, dest, 6);
-       __builtin_memcpy(p + 3, source, 6);
-}
-
-/* Parse IPV4 packet to get SRC, DST IP and protocol */
-static inline int parse_ipv4(void *data, u64 nh_off, void *data_end,
-                            __be32 *src, __be32 *dest)
-{
-       struct iphdr *iph = data + nh_off;
-
-       if (iph + 1 > data_end)
-               return 0;
-       *src = iph->saddr;
-       *dest = iph->daddr;
-       return iph->protocol;
-}
-
-SEC("xdp_router_ipv4")
-int xdp_router_ipv4_prog(struct xdp_md *ctx)
-{
-       void *data_end = (void *)(long)ctx->data_end;
-       __be64 *dest_mac = NULL, *src_mac = NULL;
-       void *data = (void *)(long)ctx->data;
-       struct trie_value *prefix_value;
-       int rc = XDP_DROP, forward_to;
-       struct ethhdr *eth = data;
-       union key_4 key4;
-       long *value;
-       u16 h_proto;
-       u32 ipproto;
-       u64 nh_off;
-
-       nh_off = sizeof(*eth);
-       if (data + nh_off > data_end)
-               return rc;
-
-       h_proto = eth->h_proto;
-
-       if (h_proto == htons(ETH_P_8021Q) || h_proto == 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 rc;
-               h_proto = vhdr->h_vlan_encapsulated_proto;
-       }
-       if (h_proto == htons(ETH_P_ARP)) {
-               return XDP_PASS;
-       } else if (h_proto == htons(ETH_P_IP)) {
-               struct direct_map *direct_entry;
-               __be32 src_ip = 0, dest_ip = 0;
-
-               ipproto = parse_ipv4(data, nh_off, data_end, &src_ip, &dest_ip);
-               direct_entry = bpf_map_lookup_elem(&exact_match, &dest_ip);
-               /* Check for exact match, this would give a faster lookup*/
-               if (direct_entry && direct_entry->mac && direct_entry->arp.mac) {
-                       src_mac = &direct_entry->mac;
-                       dest_mac = &direct_entry->arp.mac;
-                       forward_to = direct_entry->ifindex;
-               } else {
-                       /* Look up in the trie for lpm*/
-                       key4.b32[0] = 32;
-                       key4.b8[4] = dest_ip & 0xff;
-                       key4.b8[5] = (dest_ip >> 8) & 0xff;
-                       key4.b8[6] = (dest_ip >> 16) & 0xff;
-                       key4.b8[7] = (dest_ip >> 24) & 0xff;
-                       prefix_value = bpf_map_lookup_elem(&lpm_map, &key4);
-                       if (!prefix_value)
-                               return XDP_DROP;
-                       src_mac = &prefix_value->value;
-                       if (!src_mac)
-                               return XDP_DROP;
-                       dest_mac = bpf_map_lookup_elem(&arp_table, &dest_ip);
-                       if (!dest_mac) {
-                               if (!prefix_value->gw)
-                                       return XDP_DROP;
-                               dest_ip = prefix_value->gw;
-                               dest_mac = bpf_map_lookup_elem(&arp_table, &dest_ip);
-                       }
-                       forward_to = prefix_value->ifindex;
-               }
-       } else {
-               ipproto = 0;
-       }
-       if (src_mac && dest_mac) {
-               set_src_dst_mac(data, src_mac, dest_mac);
-               value = bpf_map_lookup_elem(&rxcnt, &ipproto);
-               if (value)
-                       *value += 1;
-               return  bpf_redirect_map(&tx_port, forward_to, 0);
-       }
-       return rc;
-}
-
-char _license[] SEC("license") = "GPL";
index 6dae87d83e1c738a68398af606c538f70ee3e9c6..7828784612ecbd6c62b7c9bb2e93f69c3075abdc 100644 (file)
 #include <bpf/libbpf.h>
 #include <sys/resource.h>
 #include <libgen.h>
+#include <getopt.h>
+#include "xdp_sample_user.h"
+#include "xdp_router_ipv4.skel.h"
 
-int sock, sock_arp, flags = XDP_FLAGS_UPDATE_IF_NOEXIST;
-static int total_ifindex;
-static int *ifindex_list;
-static __u32 *prog_id_list;
-char buf[8192];
+static const char *__doc__ =
+"XDP IPv4 router implementation\n"
+"Usage: xdp_router_ipv4 <IFNAME-0> ... <IFNAME-N>\n";
+
+static char buf[8192];
 static int lpm_map_fd;
-static int rxcnt_map_fd;
 static int arp_table_map_fd;
 static int exact_match_map_fd;
 static int tx_port_map_fd;
 
-static int get_route_table(int rtm_family);
-static void int_exit(int sig)
-{
-       __u32 prog_id = 0;
-       int i = 0;
+static int mask = SAMPLE_RX_CNT | SAMPLE_REDIRECT_ERR_MAP_CNT |
+                 SAMPLE_DEVMAP_XMIT_CNT_MULTI | SAMPLE_EXCEPTION_CNT;
 
-       for (i = 0; i < total_ifindex; i++) {
-               if (bpf_xdp_query_id(ifindex_list[i], flags, &prog_id)) {
-                       printf("bpf_xdp_query_id on iface %d failed\n",
-                              ifindex_list[i]);
-                       exit(1);
-               }
-               if (prog_id_list[i] == prog_id)
-                       bpf_xdp_detach(ifindex_list[i], flags, NULL);
-               else if (!prog_id)
-                       printf("couldn't find a prog id on iface %d\n",
-                              ifindex_list[i]);
-               else
-                       printf("program on iface %d changed, not removing\n",
-                              ifindex_list[i]);
-               prog_id = 0;
-       }
-       exit(0);
-}
+DEFINE_SAMPLE_INIT(xdp_router_ipv4);
 
-static void close_and_exit(int sig)
-{
-       close(sock);
-       close(sock_arp);
+static const struct option long_options[] = {
+       { "help", no_argument, NULL, 'h' },
+       { "skb-mode", no_argument, NULL, 'S' },
+       { "force", no_argument, NULL, 'F' },
+       { "interval", required_argument, NULL, 'i' },
+       { "verbose", no_argument, NULL, 'v' },
+       { "stats", no_argument, NULL, 's' },
+       {}
+};
 
-       int_exit(0);
-}
-
-/* Get the mac address of the interface given interface name */
-static __be64 getmac(char *iface)
-{
-       struct ifreq ifr;
-       __be64 mac = 0;
-       int fd, i;
-
-       fd = socket(AF_INET, SOCK_DGRAM, 0);
-       ifr.ifr_addr.sa_family = AF_INET;
-       strncpy(ifr.ifr_name, iface, IFNAMSIZ - 1);
-       if (ioctl(fd, SIOCGIFHWADDR, &ifr) < 0) {
-               printf("ioctl failed leaving....\n");
-               return -1;
-       }
-       for (i = 0; i < 6 ; i++)
-               *((__u8 *)&mac + i) = (__u8)ifr.ifr_hwaddr.sa_data[i];
-       close(fd);
-       return mac;
-}
+static int get_route_table(int rtm_family);
 
 static int recv_msg(struct sockaddr_nl sock_addr, int sock)
 {
@@ -130,7 +96,6 @@ static void read_route(struct nlmsghdr *nh, int nll)
        int i;
        struct route_table {
                int  dst_len, iface, metric;
-               char *iface_name;
                __be32 dst, gw;
                __be64 mac;
        } route;
@@ -145,17 +110,7 @@ static void read_route(struct nlmsghdr *nh, int nll)
                __be64 mac;
        } direct_entry;
 
-       if (nh->nlmsg_type == RTM_DELROUTE)
-               printf("DELETING Route entry\n");
-       else if (nh->nlmsg_type == RTM_GETROUTE)
-               printf("READING Route entry\n");
-       else if (nh->nlmsg_type == RTM_NEWROUTE)
-               printf("NEW Route entry\n");
-       else
-               printf("%d\n", nh->nlmsg_type);
-
        memset(&route, 0, sizeof(route));
-       printf("Destination     Gateway         Genmask         Metric Iface\n");
        for (; NLMSG_OK(nh, nll); nh = NLMSG_NEXT(nh, nll)) {
                rt_msg = (struct rtmsg *)NLMSG_DATA(nh);
                rtm_family = rt_msg->rtm_family;
@@ -192,11 +147,7 @@ static void read_route(struct nlmsghdr *nh, int nll)
                route.gw = atoi(gws);
                route.iface = atoi(ifs);
                route.metric = atoi(metrics);
-               route.iface_name = alloca(sizeof(char *) * IFNAMSIZ);
-               route.iface_name = if_indextoname(route.iface, route.iface_name);
-               route.mac = getmac(route.iface_name);
-               if (route.mac == -1)
-                       int_exit(0);
+               assert(get_mac_addr(route.iface, &route.mac) == 0);
                assert(bpf_map_update_elem(tx_port_map_fd,
                                           &route.iface, &route.iface, 0) == 0);
                if (rtm_family == AF_INET) {
@@ -207,7 +158,6 @@ static void read_route(struct nlmsghdr *nh, int nll)
                                int metric;
                                __be32 gw;
                        } *prefix_value;
-                       struct in_addr dst_addr, gw_addr, mask_addr;
 
                        prefix_key = alloca(sizeof(*prefix_key) + 3);
                        prefix_value = alloca(sizeof(*prefix_value));
@@ -235,17 +185,6 @@ static void read_route(struct nlmsghdr *nh, int nll)
                        for (i = 0; i < 4; i++)
                                prefix_key->data[i] = (route.dst >> i * 8) & 0xff;
 
-                       dst_addr.s_addr = route.dst;
-                       printf("%-16s", inet_ntoa(dst_addr));
-
-                       gw_addr.s_addr = route.gw;
-                       printf("%-16s", inet_ntoa(gw_addr));
-
-                       mask_addr.s_addr = htonl(~(0xffffffffU >> route.dst_len));
-                       printf("%-16s%-7d%s\n", inet_ntoa(mask_addr),
-                              route.metric,
-                              route.iface_name);
-
                        if (bpf_map_lookup_elem(lpm_map_fd, prefix_key,
                                                prefix_value) < 0) {
                                for (i = 0; i < 4; i++)
@@ -261,13 +200,6 @@ static void read_route(struct nlmsghdr *nh, int nll)
                                                           ) == 0);
                        } else {
                                if (nh->nlmsg_type == RTM_DELROUTE) {
-                                       printf("deleting entry\n");
-                                       printf("prefix key=%d.%d.%d.%d/%d",
-                                              prefix_key->data[0],
-                                              prefix_key->data[1],
-                                              prefix_key->data[2],
-                                              prefix_key->data[3],
-                                              prefix_key->prefixlen);
                                        assert(bpf_map_delete_elem(lpm_map_fd,
                                                                   prefix_key
                                                                   ) == 0);
@@ -331,14 +263,14 @@ static int get_route_table(int rtm_family)
 
        sock = socket(AF_NETLINK, SOCK_RAW, NETLINK_ROUTE);
        if (sock < 0) {
-               printf("open netlink socket: %s\n", strerror(errno));
-               return -1;
+               fprintf(stderr, "open netlink socket: %s\n", strerror(errno));
+               return -errno;
        }
        memset(&sa, 0, sizeof(sa));
        sa.nl_family = AF_NETLINK;
        if (bind(sock, (struct sockaddr *)&sa, sizeof(sa)) < 0) {
-               printf("bind to netlink: %s\n", strerror(errno));
-               ret = -1;
+               fprintf(stderr, "bind netlink socket: %s\n", strerror(errno));
+               ret = -errno;
                goto cleanup;
        }
        memset(&req, 0, sizeof(req));
@@ -357,15 +289,15 @@ static int get_route_table(int rtm_family)
        msg.msg_iovlen = 1;
        ret = sendmsg(sock, &msg, 0);
        if (ret < 0) {
-               printf("send to netlink: %s\n", strerror(errno));
-               ret = -1;
+               fprintf(stderr, "send to netlink: %s\n", strerror(errno));
+               ret = -errno;
                goto cleanup;
        }
        memset(buf, 0, sizeof(buf));
        nll = recv_msg(sa, sock);
        if (nll < 0) {
-               printf("recv from netlink: %s\n", strerror(nll));
-               ret = -1;
+               fprintf(stderr, "recv from netlink: %s\n", strerror(nll));
+               ret = nll;
                goto cleanup;
        }
        nh = (struct nlmsghdr *)buf;
@@ -395,14 +327,7 @@ static void read_arp(struct nlmsghdr *nh, int nll)
                __be64 mac;
        } direct_entry;
 
-       if (nh->nlmsg_type == RTM_GETNEIGH)
-               printf("READING arp entry\n");
-       printf("Address         HwAddress\n");
        for (; NLMSG_OK(nh, nll); nh = NLMSG_NEXT(nh, nll)) {
-               struct in_addr dst_addr;
-               char mac_str[18];
-               int len = 0, i;
-
                rt_msg = (struct ndmsg *)NLMSG_DATA(nh);
                rt_attr = (struct rtattr *)RTM_RTA(rt_msg);
                ndm_family = rt_msg->ndm_family;
@@ -424,13 +349,6 @@ static void read_arp(struct nlmsghdr *nh, int nll)
                arp_entry.dst = atoi(dsts);
                arp_entry.mac = atol(mac);
 
-               dst_addr.s_addr = arp_entry.dst;
-               for (i = 0; i < 6; i++)
-                       len += snprintf(mac_str + len, 18 - len, "%02llx%s",
-                                       ((arp_entry.mac >> i * 8) & 0xff),
-                                       i < 5 ? ":" : "");
-               printf("%-16s%s\n", inet_ntoa(dst_addr), mac_str);
-
                if (ndm_family == AF_INET) {
                        if (bpf_map_lookup_elem(exact_match_map_fd,
                                                &arp_entry.dst,
@@ -481,14 +399,14 @@ static int get_arp_table(int rtm_family)
 
        sock = socket(AF_NETLINK, SOCK_RAW, NETLINK_ROUTE);
        if (sock < 0) {
-               printf("open netlink socket: %s\n", strerror(errno));
-               return -1;
+               fprintf(stderr, "open netlink socket: %s\n", strerror(errno));
+               return -errno;
        }
        memset(&sa, 0, sizeof(sa));
        sa.nl_family = AF_NETLINK;
        if (bind(sock, (struct sockaddr *)&sa, sizeof(sa)) < 0) {
-               printf("bind to netlink: %s\n", strerror(errno));
-               ret = -1;
+               fprintf(stderr, "bind netlink socket: %s\n", strerror(errno));
+               ret = -errno;
                goto cleanup;
        }
        memset(&req, 0, sizeof(req));
@@ -506,15 +424,15 @@ static int get_arp_table(int rtm_family)
        msg.msg_iovlen = 1;
        ret = sendmsg(sock, &msg, 0);
        if (ret < 0) {
-               printf("send to netlink: %s\n", strerror(errno));
-               ret = -1;
+               fprintf(stderr, "send to netlink: %s\n", strerror(errno));
+               ret = -errno;
                goto cleanup;
        }
        memset(buf, 0, sizeof(buf));
        nll = recv_msg(sa, sock);
        if (nll < 0) {
-               printf("recv from netlink: %s\n", strerror(nll));
-               ret = -1;
+               fprintf(stderr, "recv from netlink: %s\n", strerror(nll));
+               ret = nll;
                goto cleanup;
        }
        nh = (struct nlmsghdr *)buf;
@@ -527,24 +445,17 @@ cleanup:
 /* Function to keep track and update changes in route and arp table
  * Give regular statistics of packets forwarded
  */
-static int monitor_route(void)
+static void monitor_route(void *ctx)
 {
-       unsigned int nr_cpus = bpf_num_possible_cpus();
-       const unsigned int nr_keys = 256;
        struct pollfd fds_route, fds_arp;
-       __u64 prev[nr_keys][nr_cpus];
        struct sockaddr_nl la, lr;
-       __u64 values[nr_cpus];
+       int sock, sock_arp, nll;
        struct nlmsghdr *nh;
-       int nll, ret = 0;
-       int interval = 5;
-       __u32 key;
-       int i;
 
        sock = socket(AF_NETLINK, SOCK_RAW, NETLINK_ROUTE);
        if (sock < 0) {
-               printf("open netlink socket: %s\n", strerror(errno));
-               return -1;
+               fprintf(stderr, "open netlink socket: %s\n", strerror(errno));
+               return;
        }
 
        fcntl(sock, F_SETFL, O_NONBLOCK);
@@ -552,17 +463,19 @@ static int monitor_route(void)
        lr.nl_family = AF_NETLINK;
        lr.nl_groups = RTMGRP_IPV6_ROUTE | RTMGRP_IPV4_ROUTE | RTMGRP_NOTIFY;
        if (bind(sock, (struct sockaddr *)&lr, sizeof(lr)) < 0) {
-               printf("bind to netlink: %s\n", strerror(errno));
-               ret = -1;
-               goto cleanup;
+               fprintf(stderr, "bind netlink socket: %s\n", strerror(errno));
+               close(sock);
+               return;
        }
+
        fds_route.fd = sock;
        fds_route.events = POLL_IN;
 
        sock_arp = socket(AF_NETLINK, SOCK_RAW, NETLINK_ROUTE);
        if (sock_arp < 0) {
-               printf("open netlink socket: %s\n", strerror(errno));
-               return -1;
+               fprintf(stderr, "open netlink socket: %s\n", strerror(errno));
+               close(sock);
+               return;
        }
 
        fcntl(sock_arp, F_SETFL, O_NONBLOCK);
@@ -570,184 +483,196 @@ static int monitor_route(void)
        la.nl_family = AF_NETLINK;
        la.nl_groups = RTMGRP_NEIGH | RTMGRP_NOTIFY;
        if (bind(sock_arp, (struct sockaddr *)&la, sizeof(la)) < 0) {
-               printf("bind to netlink: %s\n", strerror(errno));
-               ret = -1;
+               fprintf(stderr, "bind netlink socket: %s\n", strerror(errno));
                goto cleanup;
        }
+
        fds_arp.fd = sock_arp;
        fds_arp.events = POLL_IN;
 
-       memset(prev, 0, sizeof(prev));
-       do {
-               signal(SIGINT, close_and_exit);
-               signal(SIGTERM, close_and_exit);
-
-               sleep(interval);
-               for (key = 0; key < nr_keys; key++) {
-                       __u64 sum = 0;
-
-                       assert(bpf_map_lookup_elem(rxcnt_map_fd,
-                                                  &key, values) == 0);
-                       for (i = 0; i < nr_cpus; i++)
-                               sum += (values[i] - prev[key][i]);
-                       if (sum)
-                               printf("proto %u: %10llu pkt/s\n",
-                                      key, sum / interval);
-                       memcpy(prev[key], values, sizeof(values));
+       memset(buf, 0, sizeof(buf));
+       if (poll(&fds_route, 1, 3) == POLL_IN) {
+               nll = recv_msg(lr, sock);
+               if (nll < 0) {
+                       fprintf(stderr, "recv from netlink: %s\n",
+                               strerror(nll));
+                       goto cleanup;
                }
 
-               memset(buf, 0, sizeof(buf));
-               if (poll(&fds_route, 1, 3) == POLL_IN) {
-                       nll = recv_msg(lr, sock);
-                       if (nll < 0) {
-                               printf("recv from netlink: %s\n", strerror(nll));
-                               ret = -1;
-                               goto cleanup;
-                       }
+               nh = (struct nlmsghdr *)buf;
+               read_route(nh, nll);
+       }
 
-                       nh = (struct nlmsghdr *)buf;
-                       printf("Routing table updated.\n");
-                       read_route(nh, nll);
+       memset(buf, 0, sizeof(buf));
+       if (poll(&fds_arp, 1, 3) == POLL_IN) {
+               nll = recv_msg(la, sock_arp);
+               if (nll < 0) {
+                       fprintf(stderr, "recv from netlink: %s\n",
+                               strerror(nll));
+                       goto cleanup;
                }
-               memset(buf, 0, sizeof(buf));
-               if (poll(&fds_arp, 1, 3) == POLL_IN) {
-                       nll = recv_msg(la, sock_arp);
-                       if (nll < 0) {
-                               printf("recv from netlink: %s\n", strerror(nll));
-                               ret = -1;
-                               goto cleanup;
-                       }
 
-                       nh = (struct nlmsghdr *)buf;
-                       read_arp(nh, nll);
-               }
+               nh = (struct nlmsghdr *)buf;
+               read_arp(nh, nll);
+       }
 
-       } while (1);
 cleanup:
+       close(sock_arp);
        close(sock);
-       return ret;
 }
 
-static void usage(const char *prog)
+static void usage(char *argv[], const struct option *long_options,
+                 const char *doc, int mask, bool error,
+                 struct bpf_object *obj)
 {
-       fprintf(stderr,
-               "%s: %s [OPTS] interface name list\n\n"
-               "OPTS:\n"
-               "    -S    use skb-mode\n"
-               "    -F    force loading prog\n",
-               __func__, prog);
+       sample_usage(argv, long_options, doc, mask, error);
 }
 
-int main(int ac, char **argv)
+int main(int argc, char **argv)
 {
-       struct bpf_prog_info info = {};
-       __u32 info_len = sizeof(info);
-       const char *optstr = "SF";
-       struct bpf_program *prog;
-       struct bpf_object *obj;
-       char filename[256];
-       char **ifname_list;
-       int prog_fd, opt;
-       int err, i = 1;
-
-       snprintf(filename, sizeof(filename), "%s_kern.o", argv[0]);
-
-       total_ifindex = ac - 1;
-       ifname_list = (argv + 1);
-
-       while ((opt = getopt(ac, argv, optstr)) != -1) {
+       bool error = true, generic = false, force = false;
+       int opt, interval = 5, ret = EXIT_FAIL_BPF;
+       struct xdp_router_ipv4 *skel;
+       int i, total_ifindex = argc - 1;
+       char **ifname_list = argv + 1;
+       int longindex = 0;
+
+       if (libbpf_set_strict_mode(LIBBPF_STRICT_ALL) < 0) {
+               fprintf(stderr, "Failed to set libbpf strict mode: %s\n",
+                       strerror(errno));
+               goto end;
+       }
+
+       skel = xdp_router_ipv4__open();
+       if (!skel) {
+               fprintf(stderr, "Failed to xdp_router_ipv4__open: %s\n",
+                       strerror(errno));
+               goto end;
+       }
+
+       ret = sample_init_pre_load(skel);
+       if (ret < 0) {
+               fprintf(stderr, "Failed to sample_init_pre_load: %s\n",
+                       strerror(-ret));
+               ret = EXIT_FAIL_BPF;
+               goto end_destroy;
+       }
+
+       ret = xdp_router_ipv4__load(skel);
+       if (ret < 0) {
+               fprintf(stderr, "Failed to xdp_router_ipv4__load: %s\n",
+                       strerror(errno));
+               goto end_destroy;
+       }
+
+       ret = sample_init(skel, mask);
+       if (ret < 0) {
+               fprintf(stderr, "Failed to initialize sample: %s\n", strerror(-ret));
+               ret = EXIT_FAIL;
+               goto end_destroy;
+       }
+
+       while ((opt = getopt_long(argc, argv, "si:SFvh",
+                                 long_options, &longindex)) != -1) {
                switch (opt) {
+               case 's':
+                       mask |= SAMPLE_REDIRECT_MAP_CNT;
+                       total_ifindex--;
+                       ifname_list++;
+                       break;
+               case 'i':
+                       interval = strtoul(optarg, NULL, 0);
+                       total_ifindex -= 2;
+                       ifname_list += 2;
+                       break;
                case 'S':
-                       flags |= XDP_FLAGS_SKB_MODE;
+                       generic = true;
                        total_ifindex--;
                        ifname_list++;
                        break;
                case 'F':
-                       flags &= ~XDP_FLAGS_UPDATE_IF_NOEXIST;
+                       force = true;
+                       total_ifindex--;
+                       ifname_list++;
+                       break;
+               case 'v':
+                       sample_switch_mode();
                        total_ifindex--;
                        ifname_list++;
                        break;
+               case 'h':
+                       error = false;
                default:
-                       usage(basename(argv[0]));
-                       return 1;
+                       usage(argv, long_options, __doc__, mask, error, skel->obj);
+                       goto end_destroy;
                }
        }
 
-       if (!(flags & XDP_FLAGS_SKB_MODE))
-               flags |= XDP_FLAGS_DRV_MODE;
-
-       if (optind == ac) {
-               usage(basename(argv[0]));
-               return 1;
+       ret = EXIT_FAIL_OPTION;
+       if (optind == argc) {
+               usage(argv, long_options, __doc__, mask, true, skel->obj);
+               goto end_destroy;
        }
 
-       obj = bpf_object__open_file(filename, NULL);
-       if (libbpf_get_error(obj))
-               return 1;
-
-       prog = bpf_object__next_program(obj, NULL);
-       bpf_program__set_type(prog, BPF_PROG_TYPE_XDP);
-
-       printf("\n******************loading bpf file*********************\n");
-       err = bpf_object__load(obj);
-       if (err) {
-               printf("bpf_object__load(): %s\n", strerror(errno));
-               return 1;
+       lpm_map_fd = bpf_map__fd(skel->maps.lpm_map);
+       if (lpm_map_fd < 0) {
+               fprintf(stderr, "Failed loading lpm_map %s\n",
+                       strerror(-lpm_map_fd));
+               goto end_destroy;
        }
-       prog_fd = bpf_program__fd(prog);
-
-       lpm_map_fd = bpf_object__find_map_fd_by_name(obj, "lpm_map");
-       rxcnt_map_fd = bpf_object__find_map_fd_by_name(obj, "rxcnt");
-       arp_table_map_fd = bpf_object__find_map_fd_by_name(obj, "arp_table");
-       exact_match_map_fd = bpf_object__find_map_fd_by_name(obj,
-                                                            "exact_match");
-       tx_port_map_fd = bpf_object__find_map_fd_by_name(obj, "tx_port");
-       if (lpm_map_fd < 0 || rxcnt_map_fd < 0 || arp_table_map_fd < 0 ||
-           exact_match_map_fd < 0 || tx_port_map_fd < 0) {
-               printf("bpf_object__find_map_fd_by_name failed\n");
-               return 1;
+       arp_table_map_fd = bpf_map__fd(skel->maps.arp_table);
+       if (arp_table_map_fd < 0) {
+               fprintf(stderr, "Failed loading arp_table_map_fd %s\n",
+                       strerror(-arp_table_map_fd));
+               goto end_destroy;
+       }
+       exact_match_map_fd = bpf_map__fd(skel->maps.exact_match);
+       if (exact_match_map_fd < 0) {
+               fprintf(stderr, "Failed loading exact_match_map_fd %s\n",
+                       strerror(-exact_match_map_fd));
+               goto end_destroy;
+       }
+       tx_port_map_fd = bpf_map__fd(skel->maps.tx_port);
+       if (tx_port_map_fd < 0) {
+               fprintf(stderr, "Failed loading tx_port_map_fd %s\n",
+                       strerror(-tx_port_map_fd));
+               goto end_destroy;
        }
 
-       ifindex_list = (int *)calloc(total_ifindex, sizeof(int *));
+       ret = EXIT_FAIL_XDP;
        for (i = 0; i < total_ifindex; i++) {
-               ifindex_list[i] = if_nametoindex(ifname_list[i]);
-               if (!ifindex_list[i]) {
-                       printf("Couldn't translate interface name: %s",
-                              strerror(errno));
-                       return 1;
+               int index = if_nametoindex(ifname_list[i]);
+
+               if (!index) {
+                       fprintf(stderr, "Interface %s not found %s\n",
+                               ifname_list[i], strerror(-tx_port_map_fd));
+                       goto end_destroy;
                }
+               if (sample_install_xdp(skel->progs.xdp_router_ipv4_prog,
+                                      index, generic, force) < 0)
+                       goto end_destroy;
        }
-       prog_id_list = (__u32 *)calloc(total_ifindex, sizeof(__u32 *));
-       for (i = 0; i < total_ifindex; i++) {
-               if (bpf_xdp_attach(ifindex_list[i], prog_fd, flags, NULL) < 0) {
-                       printf("link set xdp fd failed\n");
-                       int recovery_index = i;
 
-                       for (i = 0; i < recovery_index; i++)
-                               bpf_xdp_detach(ifindex_list[i], flags, NULL);
+       if (get_route_table(AF_INET) < 0) {
+               fprintf(stderr, "Failed reading routing table\n");
+               goto end_destroy;
+       }
 
-                       return 1;
-               }
-               err = bpf_obj_get_info_by_fd(prog_fd, &info, &info_len);
-               if (err) {
-                       printf("can't get prog info - %s\n", strerror(errno));
-                       return err;
-               }
-               prog_id_list[i] = info.id;
-               memset(&info, 0, sizeof(info));
-               printf("Attached to %d\n", ifindex_list[i]);
+       if (get_arp_table(AF_INET) < 0) {
+               fprintf(stderr, "Failed reading arptable\n");
+               goto end_destroy;
        }
-       signal(SIGINT, int_exit);
-       signal(SIGTERM, int_exit);
 
-       printf("\n*******************ROUTE TABLE*************************\n");
-       get_route_table(AF_INET);
-       printf("\n*******************ARP TABLE***************************\n");
-       get_arp_table(AF_INET);
-       if (monitor_route() < 0) {
-               printf("Error in receiving route update");
-               return 1;
+       ret = sample_run(interval, monitor_route, NULL);
+       if (ret < 0) {
+               fprintf(stderr, "Failed during sample run: %s\n", strerror(-ret));
+               ret = EXIT_FAIL;
+               goto end_destroy;
        }
+       ret = EXIT_OK;
 
-       return 0;
+end_destroy:
+       xdp_router_ipv4__destroy(skel);
+end:
+       sample_exit(ret);
 }