@@ -22,7 +22,7 @@
struct {
__uint(type, BPF_MAP_TYPE_DEVMAP);
__uint(key_size, sizeof(int));
- __uint(value_size, sizeof(int));
+ __uint(value_size, sizeof(struct bpf_devmap_val));
__uint(max_entries, 100);
} tx_port SEC(".maps");
@@ -52,6 +52,48 @@ static void swap_src_dst_mac(void *data)
p[5] = dst[2];
}
+static __always_inline __u16 csum_fold_helper(__u32 csum)
+{
+ __u32 sum;
+ sum = (csum & 0xffff) + (csum >> 16);
+ sum += (sum >> 16);
+ return ~sum;
+}
+
+static __always_inline __u16 ipv4_csum(__u16 seed, struct iphdr *iphdr_new,
+ struct iphdr *iphdr_old)
+{
+ __u32 csum, size = sizeof(struct iphdr);
+ csum = bpf_csum_diff((__be32 *)iphdr_old, size,
+ (__be32 *)iphdr_new, size, seed);
+ return csum_fold_helper(csum);
+}
+
+static void parse_ipv4(void *data, u64 nh_off, void *data_end, u8 ttl)
+{
+ struct iphdr *iph = data + nh_off;
+ struct iphdr iph_old;
+ __u16 csum_old;
+
+ if (iph + 1 > data_end)
+ return;
+
+ iph_old = *iph;
+ csum_old = iph->check;
+ iph->ttl = ttl;
+ iph->check = ipv4_csum(~csum_old, iph, &iph_old);
+}
+
+static void parse_ipv6(void *data, u64 nh_off, void *data_end, u8 hop_limit)
+{
+ struct ipv6hdr *ip6h = data + nh_off;
+
+ if (ip6h + 1 > data_end)
+ return;
+
+ ip6h->hop_limit = hop_limit;
+}
+
SEC("xdp_redirect_map")
int xdp_redirect_map_prog(struct xdp_md *ctx)
{
@@ -82,6 +124,36 @@ int xdp_redirect_map_prog(struct xdp_md *ctx)
return bpf_redirect_map(&tx_port, vport, 0);
}
+/* This map prog will set new IP ttl based on egress ifindex */
+SEC("xdp_devmap/map_prog")
+int xdp_devmap_prog(struct xdp_md *ctx)
+{
+ char fmt[] = "devmap redirect: egress dev %u with new ttl %u\n";
+ void *data_end = (void *)(long)ctx->data_end;
+ void *data = (void *)(long)ctx->data;
+ struct ethhdr *eth = data;
+ u16 h_proto;
+ u64 nh_off;
+ u8 ttl;
+
+ nh_off = sizeof(struct ethhdr);
+ if (data + nh_off > data_end)
+ return XDP_DROP;
+
+ /* set new ttl based on egress ifindex */
+ ttl = ctx->egress_ifindex % 64;
+
+ h_proto = eth->h_proto;
+ if (h_proto == htons(ETH_P_IP))
+ parse_ipv4(data, nh_off, data_end, ttl);
+ else if (h_proto == htons(ETH_P_IPV6))
+ parse_ipv6(data, nh_off, data_end, ttl);
+
+ bpf_trace_printk(fmt, sizeof(fmt), ctx->egress_ifindex, ttl);
+
+ return XDP_PASS;
+}
+
/* Redirect require an XDP bpf_prog loaded on the TX device */
SEC("xdp_redirect_dummy")
int xdp_redirect_dummy_prog(struct xdp_md *ctx)
@@ -98,12 +98,13 @@ int main(int argc, char **argv)
{
struct rlimit r = {RLIM_INFINITY, RLIM_INFINITY};
struct bpf_prog_load_attr prog_load_attr = {
- .prog_type = BPF_PROG_TYPE_XDP,
+ .prog_type = BPF_PROG_TYPE_UNSPEC,
};
- struct bpf_program *prog, *dummy_prog;
+ struct bpf_program *prog, *dummy_prog, *devmap_prog;
+ int prog_fd, dummy_prog_fd, devmap_prog_fd;
+ struct bpf_devmap_val devmap_val;
struct bpf_prog_info info = {};
__u32 info_len = sizeof(info);
- int prog_fd, dummy_prog_fd;
const char *optstr = "FSN";
struct bpf_object *obj;
int ret, opt, key = 0;
@@ -157,16 +158,18 @@ int main(int argc, char **argv)
return 1;
prog = bpf_program__next(NULL, obj);
- dummy_prog = bpf_program__next(prog, obj);
- if (!prog || !dummy_prog) {
+ devmap_prog = bpf_object__find_program_by_title(obj, "xdp_devmap/map_prog");
+ dummy_prog = bpf_object__find_program_by_title(obj, "xdp_redirect_dummy");
+ if (!prog || !devmap_prog || !dummy_prog) {
printf("finding a prog in obj file failed\n");
return 1;
}
/* bpf_prog_load_xattr gives us the pointer to first prog's fd,
- * so we're missing only the fd for dummy prog
+ * so we're missing the fd for devmap and dummy prog
*/
+ devmap_prog_fd = bpf_program__fd(devmap_prog);
dummy_prog_fd = bpf_program__fd(dummy_prog);
- if (prog_fd < 0 || dummy_prog_fd < 0) {
+ if (prog_fd < 0 || devmap_prog_fd < 0 || dummy_prog_fd < 0) {
printf("bpf_prog_load_xattr: %s\n", strerror(errno));
return 1;
}
@@ -209,7 +212,9 @@ int main(int argc, char **argv)
signal(SIGTERM, int_exit);
/* populate virtual to physical port map */
- ret = bpf_map_update_elem(tx_port_map_fd, &key, &ifindex_out, 0);
+ devmap_val.bpf_prog.fd = devmap_prog_fd;
+ devmap_val.ifindex = ifindex_out;
+ ret = bpf_map_update_elem(tx_port_map_fd, &key, &devmap_val, 0);
if (ret) {
perror("bpf_update_elem");
goto out;
This patch add running xdp program on egress interface support for xdp_redirect_map sample. The new prog will change the IP ttl based on egress ifindex. Signed-off-by: Hangbin Liu <liuhangbin@gmail.com> --- samples/bpf/xdp_redirect_map_kern.c | 74 ++++++++++++++++++++++++++++- samples/bpf/xdp_redirect_map_user.c | 21 ++++---- 2 files changed, 86 insertions(+), 9 deletions(-)