g2gps opened a new issue, #9246:
URL: https://github.com/apache/nuttx/issues/9246
I'm currently porting porting and existing application to Nuttx which
expects to be able to set the IPv6 unicast and multicast hop limit to 255. From
what I can see in `net/inet/ipv6_setsockopt.c`, these socket options are
currently not implemented, with the default value being set to 64 in
`nuttx/nuttx/include/nuttx/net/netconfig.h`:
```c
/* The IP TTL (time to live) of IP packets sent by the network stack.
*
* This should normally not be changed.
*/
#define IP_TTL_DEFAULT 64
```
and subsequently fixed as a parameter of a UDP connection in
`nuttx/nuttx/net/udp/udp_conn.c`:
```c
#if defined(CONFIG_NET_IPv4) && defined(CONFIG_NET_IPv6)
conn->domain = domain;
#endif
conn->lport = 0;
conn->ttl = IP_TTL_DEFAULT;
#if CONFIG_NET_RECV_BUFSIZE > 0
conn->rcvbufs = CONFIG_NET_RECV_BUFSIZE;
#endif
```
It is possible to implement the socket options, for UDP only, with the
following:
```diff
@@ -35,6 +35,7 @@
#include "mld/mld.h"
#include "inet/inet.h"
#include "socket/socket.h"
+#include "udp/udp.h"
#if defined(CONFIG_NET_IPv6) && defined(CONFIG_NET_SOCKOPTS)
@@ -112,21 +113,57 @@ int ipv6_setsockopt(FAR struct socket *psock, int
option,
}
break;
+#ifdef NET_UDP_HAVE_STACK
+ case IPV6_MULTICAST_HOPS: /* Multicast hop limit */
+ {
+ FAR struct udp_conn_s *conn;
+ uint8_t ttl;
+ if (psock->s_type != SOCK_DGRAM ||
+ value == NULL || value_len == 0)
+ {
+ ret = -EINVAL;
+ break;
+ }
+
+ ttl = *(FAR uint8_t *)value;
+ conn = (FAR struct udp_conn_s *)psock->s_conn;
+ conn->ttl = ttl;
+ ret = OK;
+ }
+ break;
+#endif
/* The following IPv6 socket options are defined, but not implemented
*/
- case IPV6_MULTICAST_HOPS: /* Multicast hop limit */
case IPV6_MULTICAST_IF: /* Interface to use for outgoing multicast
* packets */
case IPV6_MULTICAST_LOOP: /* Multicast packets are delivered back to
* the local application */
#endif
- case IPV6_UNICAST_HOPS: /* Unicast hop limit */
case IPV6_V6ONLY: /* Restrict AF_INET6 socket to IPv6
* communications only */
nwarn("WARNING: Unimplemented IPv6 option: %d\n", option);
ret = -ENOSYS;
break;
+#ifdef NET_UDP_HAVE_STACK
+ case IPV6_UNICAST_HOPS:
+ {
+ FAR struct udp_conn_s *conn;
+ uint8_t ttl;
+ if (psock->s_type != SOCK_DGRAM ||
+ value == NULL || value_len == 0)
+ {
+ ret = -EINVAL;
+ break;
+ }
+
+ ttl = *(FAR uint8_t *)value;
+ conn = (FAR struct udp_conn_s *)psock->s_conn;
+ conn->ttl = ttl;
+ ret = OK;
+ }
+ break;
+#endif
case IPV6_RECVPKTINFO:
case IPV6_RECVHOPLIMIT:
{
```
This works for our application, however, I do not think it's a great
solutions as:
- It's only suitable for UDP, and adds protocol dependent characteristics
back into the more generic IVP6 socket options. I noticed this was just cleaned
up in #9214.
- The limit is applied to multicast and unicast. I'm not sure on the
implications here.
- If the ttl, or hop limit socket option was to be added for other
transports, a case would need to be added for each underlying connection type.
I'm not overly familiar with networking stack typologies, so I'm not sure
what the best approach is in this case. Should the ttl, or hop limit be an
attribute of the socket, with the transport layer using its value when it
constructs the IP header?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
