OSDN Git Service

rtnelink: Move link dump consistency check out of the loop
authorJakub Sitnicki <jkbs@redhat.com>
Wed, 9 Aug 2017 15:39:12 +0000 (17:39 +0200)
committerDavid S. Miller <davem@davemloft.net>
Mon, 14 Aug 2017 02:43:57 +0000 (19:43 -0700)
Calls to rtnl_dump_ifinfo() are protected by RTNL lock. So are the
{list,unlist}_netdevice() calls where we bump the net->dev_base_seq
number.

For this reason net->dev_base_seq can't change under out feet while
we're looping over links in rtnl_dump_ifinfo(). So move the check for
net->dev_base_seq change (since the last time we were called) out of the
loop.

This way we avoid giving a wrong impression that there are concurrent
updates to the link list going on while we're iterating over them.

Signed-off-by: Jakub Sitnicki <jkbs@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/core/rtnetlink.c

index 9e9f141..a78fd61 100644 (file)
@@ -1621,8 +1621,6 @@ static int rtnl_dump_ifinfo(struct sk_buff *skb, struct netlink_callback *cb)
        s_h = cb->args[0];
        s_idx = cb->args[1];
 
-       cb->seq = net->dev_base_seq;
-
        /* A hack to preserve kernel<->userspace interface.
         * The correct header is ifinfomsg. It is consistent with rtnl_getlink.
         * However, before Linux v3.9 the code here assumed rtgenmsg and that's
@@ -1668,8 +1666,6 @@ static int rtnl_dump_ifinfo(struct sk_buff *skb, struct netlink_callback *cb)
 
                                goto out_err;
                        }
-
-                       nl_dump_check_consistent(cb, nlmsg_hdr(skb));
 cont:
                        idx++;
                }
@@ -1679,6 +1675,8 @@ out:
 out_err:
        cb->args[1] = idx;
        cb->args[0] = h;
+       cb->seq = net->dev_base_seq;
+       nl_dump_check_consistent(cb, nlmsg_hdr(skb));
 
        return err;
 }