icmp: Work around possible failure on bind() due to e.g. broken SELinux policy

If we can't bind() ping sockets, the echo identifier sent out from
the socket won't be the original one seen from the tap. Binding a
ping socket doesn't require any security capability, but it might
still fail due to a broken SELinux policy, see for example:
	https://bugzilla.redhat.com/show_bug.cgi?id=1848929

Track the ICMP echo identifier as part of the epoll reference for
the socket and replace it in the reply on mismatch. We won't send
out the original identifier as sent from the guest, but still better
than missing replies.

Signed-off-by: Stefano Brivio <sbrivio@redhat.com>

3 files changed, 22 insertions, 5 deletions

diff --git a/icmp.c b/icmp.c
index 57d954a..49fdf91 100644
--- a/icmp.c
+++ b/icmp.c
@@ -87,10 +87,17 @@ void icmp_sock_handler(struct ctx *c, union epoll_ref ref, uint32_t events,
 		struct sockaddr_in6 *sr6 = (struct sockaddr_in6 *)&sr;
 		struct icmp6hdr *ih = (struct icmp6hdr *)buf;
 
+		id = ntohs(ih->icmp6_identifier);
+
+		/* If bind() fails e.g. because of a broken SELinux policy, this
+		 * might happen. Fix up the identifier to match the sent one.
+		 */
+		if (id != ref.icmp.id)
+			ih->icmp6_identifier = htons(ref.icmp.id);
+
 		/* In PASTA mode, we'll get any reply we send, discard them. */
 		if (c->mode == MODE_PASTA) {
 			seq = ntohs(ih->icmp6_sequence);
-			id = ntohs(ih->icmp6_identifier);
 
 			if (icmp_id_map[V6][id].seq == seq)
 				return;
@@ -103,9 +110,12 @@ void icmp_sock_handler(struct ctx *c, union epoll_ref ref, uint32_t events,
 		struct sockaddr_in *sr4 = (struct sockaddr_in *)&sr;
 		struct icmphdr *ih = (struct icmphdr *)buf;
 
+		id = ntohs(ih->un.echo.id);
+		if (id != ref.icmp.id)
+			ih->un.echo.id = htons(ref.icmp.id);
+
 		if (c->mode == MODE_PASTA) {
 			seq = ntohs(ih->un.echo.sequence);
-			id = ntohs(ih->un.echo.id);
 
 			if (icmp_id_map[V4][id].seq == seq)
 				return;
@@ -148,7 +158,7 @@ int icmp_tap_handler(struct ctx *c, int af, void *addr,
 		if (msg[0].l4_len < sizeof(*ih) || ih->type != ICMP_ECHO)
 			return 1;
 
-		id = ntohs(ih->un.echo.id);
+		iref.id = id = ntohs(ih->un.echo.id);
 
 		if ((s = icmp_id_map[V4][id].sock) <= 0) {
 			s = sock_l4(c, AF_INET, IPPROTO_ICMP, id, 0, iref.u32);
@@ -177,7 +187,7 @@ int icmp_tap_handler(struct ctx *c, int af, void *addr,
 		    (ih->icmp6_type != 128 && ih->icmp6_type != 129))
 			return 1;
 
-		id = ntohs(ih->icmp6_identifier);
+		iref.id = id = ntohs(ih->icmp6_identifier);
 		if ((s = icmp_id_map[V6][id].sock) <= 0) {
 			s = sock_l4(c, AF_INET6, IPPROTO_ICMPV6, id, 0,
 				    iref.u32);
diff --git a/icmp.h b/icmp.h
index 12547b7..27f0a5c 100644
--- a/icmp.h
+++ b/icmp.h
@@ -15,10 +15,12 @@ void icmp_timer(struct ctx *c, struct timespec *ts);
  * union icmp_epoll_ref - epoll reference portion for ICMP tracking
  * @v6:			Set for IPv6 sockets or connections
  * @u32:		Opaque u32 value of reference
+ * @id:			Associated echo identifier, needed if bind() fails
  */
 union icmp_epoll_ref {
 	struct {
-		uint32_t	v6:1;
+		uint32_t	v6:1,
+				id:16;
 	};
 	uint32_t u32;
 };
diff --git a/tap.c b/tap.c
index e31a419..68d3633 100644
--- a/tap.c
+++ b/tap.c
@@ -134,6 +134,11 @@ void tap_ip_send(struct ctx *c, struct in6_addr *src, uint8_t proto,
 			struct udphdr *uh = (struct udphdr *)(iph + 1);
 
 			uh->check = 0;
+		} else if (iph->protocol == IPPROTO_ICMP) {
+			struct icmphdr *ih = (struct icmphdr *)(iph + 1);
+
+			ih->checksum = 0;
+			ih->checksum = csum_unaligned(ih, len, 0);
 		}
 
 		tap_send(c, buf, len + sizeof(*iph) + sizeof(*eh), 1);