Hi Richard,
I have just reinvestigated the long lasting rose module
unregister_netdevice issue with kernel 4.10.0.
Here is the context: when removing rose module
rmmod rose
NET: Unregistered protocol family 11
is followed by a message looping indefinitely with a random xx count
number :
unregistered_netdevice: waiting for rose0 to become free. Usage count = xx
unregistered_netdevice: waiting for rose0 to become free. Usage count = xx
unregistered_netdevice: waiting for rose0 to become free. Usage count = xx
.....
I tried to apply the patches you sent to linux-netdev on 2016-07-16
[PATCH 1/6]NET:AX25:ROSE Add device use count
First I applied it and find out that it was successfully removing the
unregister bug.
Then I cut it into 5 parts and unpatched the parts one by one.
Then I applied different patch combinations in order to find out which
ones were sufficient to cure the unregister issue.
Here is the result : parts 1, 2 and 5 are necessary all together.
Part 3 and 4 of your original patch did not add anything according to
unregister issue (same results as with 5 parts).
According to the result, I am not sure parts 3 and 4 are absolutely
necessary here.
When applied, rmmod rose is correctly removing rose module and no error
message occurs.
Here is your patch reduced to parts 1, 2 and 5.
index 36dbc2d..89745aa 100644
--- a/net/rose/af_rose.c
+++ b/net/rose/af_rose.c
@@ -687,8 +687,10 @@ static int rose_bind(struct socket *sock, struct
sockaddr *uaddr, int addr_len)
rose->source_call = user->call;
ax25_uid_put(user);
} else {
- if (ax25_uid_policy && !capable(CAP_NET_BIND_SERVICE))
+ if (ax25_uid_policy && !capable(CAP_NET_BIND_SERVICE)) {
+ dev_put(dev);
return -EACCES;
+ }
rose->source_call = *source;
}
index 36dbc2d..89745aa 100644
--- a/net/rose/af_rose.c
+++ b/net/rose/af_rose.c
@@ -709,6 +711,7 @@ static int rose_bind(struct socket *sock, struct
sockaddr *uaddr, int addr_len)
rose_insert_socket(sk);
sock_reset_flag(sk, SOCK_ZAPPED);
+ dev_put(dev);
return 0;
}
diff --git a/net/rose/rose_loopback.c b/net/rose/rose_loopback.c
index 3444562..ea48cee 100644
--- a/net/rose/rose_loopback.c
+++ b/net/rose/rose_loopback.c
@@ -102,6 +102,7 @@ static void rose_loopback_timer(unsigned long param)
if ((dev = rose_dev_get(dest)) != NULL) {
if (rose_rx_call_request(skb, dev,
rose_loopback_neigh, lci_o) == 0)
kfree_skb(skb);
+ dev_put(dev);
} else {
kfree_skb(skb);
}
I encourage you to publish a new commit limited to this set of three
dev_put(dev) that are obviously missing to allow a correct removal of
rose module.
Bernard
--
To unsubscribe from this list: send the line "unsubscribe linux-hams" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
