kernel-packages team mailing list archive

[Atzm Watanabe <atzm@xxxxxxxxx>]

Hi,

Thank you for the reply.

This problem can be reproduced on 14.04 LTS (3.13.0-76-generic).
But because the bug has fixed in upstream kernel since 4.2, the problem is not caused on 15.10 (4.2.0-30-generic).
I have not try to reproduce on 12.04 LTS yet, but if the kernel version is earlier than 4.2, the problem will be reproduced probably.

-- 
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/1540886

Title:
  xfrm and fwmark do not work on VXLAN xmit

Status in linux package in Ubuntu:
  Incomplete

Bug description:
  On Ubuntu 15.04, kernel 3.19.0-49-generic has known issue that xfrm and fwmark do not work on VXLAN xmit.
  This issue was fixed on upstream kernel: https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=239fb791d4ee194740e69fe9694f58ec404d1689

  I think the above patch should be backported because this may cause serious problems including security issues.
  For example, outgoing VXLAN packet will be sent without encryption even if IPsec security policy is configured properly.
  As the result, the packet which should be encrypted can be snooped.

  How to reproduce:

  When using ipsec-tools (for minimum reproducing steps):

  --- Node-A
  # modprobe esp4
  # modprobe af_key
  # modprobe xfrm4_mode_transport
  # setkey -c <<EOL
  > flush;
  > spdflush;
  > add <Node-A> <Node-B> esp 0x201 -E 3des-cbc 0x7aeaca3f87d060a12f4a4487d5a5c3355920fae69a96c831;
  > add <Node-B> <Node-A> esp 0x301 -E 3des-cbc 0xf6ddb555acfd9d77b03ea3843f2653255afe8eb5573965df;
  > spdadd <Node-A> <Node-B> udp -P out ipsec esp/transport//require;
  > spdadd <Node-B> <Node-A> udp -P in  ipsec esp/transport//require;
  > EOL
  # ip link add vxlan100 type vxlan id 100 remote <Node-B>
  # ip addr add 1.1.1.1/24 dev vxlan100
  # ip link set vxlan100 up

  --- Node-B
  # modprobe esp4
  # modprobe af_key
  # modprobe xfrm4_mode_transport
  # setkey -c <<EOL
  > flush;
  > spdflush;
  > add <Node-A> <Node-B> esp 0x201 -E 3des-cbc 0x7aeaca3f87d060a12f4a4487d5a5c3355920fae69a96c831;
  > add <Node-B> <Node-A> esp 0x301 -E 3des-cbc 0xf6ddb555acfd9d77b03ea3843f2653255afe8eb5573965df;
  > spdadd <Node-B> <Node-A> udp -P out ipsec esp/transport//require;
  > spdadd <Node-A> <Node-B> udp -P in  ipsec esp/transport//require;
  > EOL
  # ip link add vxlan100 type vxlan id 100 remote <Node-A>
  # ip addr add 1.1.1.2/24 dev vxlan100
  # ip link set vxlan100 up
  # ping 1.1.1.1

  Then packets which is encapsulated with VXLAN header will be shown in
  tcpdump, but they must be ESP packets.

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1540886/+subscriptions