net/ipfwd: limit number of ipforward struct to be less than number of IOB

If ipforward consumes all the IOB, devif_poll will not poll any more. Without polling, ipforward will not release any IOB, then all the network stack may fail with no IOB available.

Signed-off-by: Zhe Weng <wengzhe@xiaomi.com>

net/devif/devif_poll.c

@@ -925,6 +925,7 @@ int devif_poll(FAR struct net_driver_s *dev, devif_poll_callback_t callback)
 
       if (netdev_iob_prepare(dev, false, 0) != OK)
         {
+          nwarn("WARNING: IOB Prepare failed for dev %s!\n", dev->d_ifname);
           bstop = true;
           break;
         }

net/ipforward/Kconfig

@@ -36,3 +36,7 @@ config NET_IPFORWARD_NSTRUCT
 		packets that may be waiting to be forwarded from one network device
 		to another.  CONFIG_IOB_NBUFFERS also limits the forward because the
 		payload of the packet (up to the MSS) is retain in IOBs.
+
+		WARNING: DO NOT set this setting to a value greater than or equal to
+		CONFIG_IOB_NBUFFERS, otherwise it may consume all the IOB and let
+		netdev fail to work.

net/ipforward/ipforward.h

@@ -27,6 +27,7 @@
 
 #include <nuttx/config.h>
 
+#include <assert.h>
 #include <stdint.h>
 
 #undef HAVE_FWDALLOC
@@ -42,6 +43,9 @@
 #  define CONFIG_NET_IPFORWARD_NSTRUCT 4
 #endif
 
+static_assert(CONFIG_IOB_NBUFFERS > CONFIG_NET_IPFORWARD_NSTRUCT,
+              "IP forward may consume all the IOB and break netdev logic");
+
 /* Allocate a new IP forwarding data callback */
 
 #define ipfwd_callback_alloc(dev)   devif_callback_alloc(dev, \