1. 02 4月, 2020 2 次提交
    • J
      ipv6: don't auto-add link-local address to lag ports · 744fdc82
      Jarod Wilson 提交于
      Bonding slave and team port devices should not have link-local addresses
      automatically added to them, as it can interfere with openvswitch being
      able to properly add tc ingress.
      
      Basic reproducer, courtesy of Marcelo:
      
      $ ip link add name bond0 type bond
      $ ip link set dev ens2f0np0 master bond0
      $ ip link set dev ens2f1np2 master bond0
      $ ip link set dev bond0 up
      $ ip a s
      1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
      group default qlen 1000
          link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
          inet 127.0.0.1/8 scope host lo
             valid_lft forever preferred_lft forever
          inet6 ::1/128 scope host
             valid_lft forever preferred_lft forever
      2: ens2f0np0: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc
      mq master bond0 state UP group default qlen 1000
          link/ether 00:0f:53:2f:ea:40 brd ff:ff:ff:ff:ff:ff
      5: ens2f1np2: <NO-CARRIER,BROADCAST,MULTICAST,SLAVE,UP> mtu 1500 qdisc
      mq master bond0 state DOWN group default qlen 1000
          link/ether 00:0f:53:2f:ea:40 brd ff:ff:ff:ff:ff:ff
      11: bond0: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc
      noqueue state UP group default qlen 1000
          link/ether 00:0f:53:2f:ea:40 brd ff:ff:ff:ff:ff:ff
          inet6 fe80::20f:53ff:fe2f:ea40/64 scope link
             valid_lft forever preferred_lft forever
      
      (above trimmed to relevant entries, obviously)
      
      $ sysctl net.ipv6.conf.ens2f0np0.addr_gen_mode=0
      net.ipv6.conf.ens2f0np0.addr_gen_mode = 0
      $ sysctl net.ipv6.conf.ens2f1np2.addr_gen_mode=0
      net.ipv6.conf.ens2f1np2.addr_gen_mode = 0
      
      $ ip a l ens2f0np0
      2: ens2f0np0: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc
      mq master bond0 state UP group default qlen 1000
          link/ether 00:0f:53:2f:ea:40 brd ff:ff:ff:ff:ff:ff
          inet6 fe80::20f:53ff:fe2f:ea40/64 scope link tentative
             valid_lft forever preferred_lft forever
      $ ip a l ens2f1np2
      5: ens2f1np2: <NO-CARRIER,BROADCAST,MULTICAST,SLAVE,UP> mtu 1500 qdisc
      mq master bond0 state DOWN group default qlen 1000
          link/ether 00:0f:53:2f:ea:40 brd ff:ff:ff:ff:ff:ff
          inet6 fe80::20f:53ff:fe2f:ea40/64 scope link tentative
             valid_lft forever preferred_lft forever
      
      Looks like addrconf_sysctl_addr_gen_mode() bypasses the original "is
      this a slave interface?" check added by commit c2edacf8, and
      results in an address getting added, while w/the proposed patch added,
      no address gets added. This simply adds the same gating check to another
      code path, and thus should prevent the same devices from erroneously
      obtaining an ipv6 link-local address.
      
      Fixes: d35a00b8 ("net/ipv6: allow sysctl to change link-local address generation mode")
      Reported-by: NMoshe Levi <moshele@mellanox.com>
      CC: Stephen Hemminger <stephen@networkplumber.org>
      CC: Marcelo Ricardo Leitner <mleitner@redhat.com>
      CC: netdev@vger.kernel.org
      Signed-off-by: NJarod Wilson <jarod@redhat.com>
      Signed-off-by: NDavid S. Miller <davem@davemloft.net>
      744fdc82
    • C
      net_sched: add a temporary refcnt for struct tcindex_data · 304e0242
      Cong Wang 提交于
      Although we intentionally use an ordered workqueue for all tc
      filter works, the ordering is not guaranteed by RCU work,
      given that tcf_queue_work() is esstenially a call_rcu().
      
      This problem is demostrated by Thomas:
      
        CPU 0:
          tcf_queue_work()
            tcf_queue_work(&r->rwork, tcindex_destroy_rexts_work);
      
        -> Migration to CPU 1
      
        CPU 1:
           tcf_queue_work(&p->rwork, tcindex_destroy_work);
      
      so the 2nd work could be queued before the 1st one, which leads
      to a free-after-free.
      
      Enforcing this order in RCU work is hard as it requires to change
      RCU code too. Fortunately we can workaround this problem in tcindex
      filter by taking a temporary refcnt, we only refcnt it right before
      we begin to destroy it. This simplifies the code a lot as a full
      refcnt requires much more changes in tcindex_set_parms().
      
      Reported-by: syzbot+46f513c3033d592409d2@syzkaller.appspotmail.com
      Fixes: 3d210534 ("net_sched: fix a race condition in tcindex_destroy()")
      Cc: Thomas Gleixner <tglx@linutronix.de>
      Cc: Paul E. McKenney <paulmck@kernel.org>
      Cc: Jamal Hadi Salim <jhs@mojatatu.com>
      Cc: Jiri Pirko <jiri@resnulli.us>
      Signed-off-by: NCong Wang <xiyou.wangcong@gmail.com>
      Reviewed-by: NPaul E. McKenney <paulmck@kernel.org>
      Signed-off-by: NDavid S. Miller <davem@davemloft.net>
      304e0242
  2. 01 4月, 2020 3 次提交
  3. 31 3月, 2020 18 次提交
  4. 30 3月, 2020 17 次提交
新手
引导
客服 返回
顶部