target/linux/generic/backport-4.14/600-ipv6-addrconf-call-ipv6_mc_up-for-non-Ethernet-inter.patch

   1 From 82afdcd4ec3c8ca6551cbf7c43c09e2fd240487a Mon Sep 17 00:00:00 2001
   2 From: Hangbin Liu <liuhangbin@gmail.com>
   3 Date: Tue, 10 Mar 2020 15:27:37 +0800
   4 Subject: [PATCH] ipv6/addrconf: call ipv6_mc_up() for non-Ethernet interface
   5 MIME-Version: 1.0
   6 Content-Type: text/plain; charset=UTF-8
   7 Content-Transfer-Encoding: 8bit
   8
   9 Rafał found an issue that for non-Ethernet interface, if we down and up
  10 frequently, the memory will be consumed slowly.
  11
  12 The reason is we add allnodes/allrouters addressed in multicast list in
  13 ipv6_add_dev(). When link down, we call ipv6_mc_down(), store all multicast
  14 addresses via mld_add_delrec(). But when link up, we don't call ipv6_mc_up()
  15 for non-Ethernet interface to remove the addresses. This makes idev->mc_tomb
  16 getting bigger and bigger. The call stack looks like:
  17
  18 addrconf_notify(NETDEV_REGISTER)
  19         ipv6_add_dev
  20                 ipv6_dev_mc_inc(ff01::1)
  21                 ipv6_dev_mc_inc(ff02::1)
  22                 ipv6_dev_mc_inc(ff02::2)
  23
  24 addrconf_notify(NETDEV_UP)
  25         addrconf_dev_config
  26                 /* Alas, we support only Ethernet autoconfiguration. */
  27                 return;
  28
  29 addrconf_notify(NETDEV_DOWN)
  30         addrconf_ifdown
  31                 ipv6_mc_down
  32                         igmp6_group_dropped(ff02::2)
  33                                 mld_add_delrec(ff02::2)
  34                         igmp6_group_dropped(ff02::1)
  35                         igmp6_group_dropped(ff01::1)
  36
  37 After investigating, I can't found a rule to disable multicast on
  38 non-Ethernet interface. In RFC2460, the link could be Ethernet, PPP, ATM,
  39 tunnels, etc. In IPv4, it doesn't check the dev type when calls ip_mc_up()
  40 in inetdev_event(). Even for IPv6, we don't check the dev type and call
  41 ipv6_add_dev(), ipv6_dev_mc_inc() after register device.
  42
  43 So I think it's OK to fix this memory consumer by calling ipv6_mc_up() for
  44 non-Ethernet interface.
  45
  46 v2: Also check IFF_MULTICAST flag to make sure the interface supports
  47     multicast
  48
  49 Reported-by: Rafał Miłecki <zajec5@gmail.com>
  50 Tested-by: Rafał Miłecki <zajec5@gmail.com>
  51 Fixes: 74235a25c673 ("[IPV6] addrconf: Fix IPv6 on tuntap tunnels")
  52 Fixes: 1666d49e1d41 ("mld: do not remove mld souce list info when set link down")
  53 Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
  54 Signed-off-by: David S. Miller <davem@davemloft.net>
  55 ---
  56  net/ipv6/addrconf.c | 4 ++++
  57  1 file changed, 4 insertions(+)
  58
  59 --- a/net/ipv6/addrconf.c
  60 +++ b/net/ipv6/addrconf.c
  61 @@ -3223,6 +3223,10 @@ static void addrconf_dev_config(struct n
  62             (dev->type != ARPHRD_TUNNEL) &&
  63             (dev->type != ARPHRD_NONE)) {
  64                 /* Alas, we support only Ethernet autoconfiguration. */
  65 +               idev = __in6_dev_get(dev);
  66 +               if (!IS_ERR_OR_NULL(idev) && dev->flags & IFF_UP &&
  67 +                   dev->flags & IFF_MULTICAST)
  68 +                       ipv6_mc_up(idev);
  69                 return;
  70         }
  71