)]}'
{
  "commit": "14c998f034bdc9a5bfa53bca18fbd0738cbc65e8",
  "tree": "161263012d022bddcdb7a0389f0ab367c013ce66",
  "parents": [
    "2506ece0c0bbd2fc19a4827b96dc52ea47e2ce4a"
  ],
  "author": {
    "name": "Mark McLoughlin",
    "email": "markmc@redhat.com",
    "time": "Sun Jun 08 20:50:56 2008 +1000"
  },
  "committer": {
    "name": "Jeff Garzik",
    "email": "jgarzik@redhat.com",
    "time": "Tue Jun 10 18:20:31 2008 -0400"
  },
  "message": "virtio: virtio_net free transmit skbs in a timer\n\nvirtio_net currently only frees old transmit skbs just\nbefore queueing new ones. If the queue is full, it then\nenables interrupts and waits for notification that more\nwork has been performed.\n\nHowever, a side-effect of this scheme is that there are\nalways xmit skbs left dangling when no new packets are\nsent, against the Documentation/networking/driver.txt\nguideline:\n\n  \"... it is not allowed for your TX mitigation scheme\n   to let TX packets \"hang out\" in the TX ring unreclaimed\n   forever if no new TX packets are sent.\"\n\nAdd a timer to ensure that any time we queue new TX\nskbs, we will shortly free them again.\n\nThis fixes an easily reproduced hang at shutdown where\niptables attempts to unload nf_conntrack and nf_conntrack\nwaits for an skb it is tracking to be freed, but virtio_net\nnever frees it.\n\nSigned-off-by: Mark McLoughlin \u003cmarkmc@redhat.com\u003e\nSigned-off-by: Rusty Russell \u003crusty@rustcorp.com.au\u003e\nSigned-off-by: Jeff Garzik \u003cjgarzik@redhat.com\u003e\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "9a3b85e55ccc4fa62c671b7f3dc7835e225e6e7a",
      "old_mode": 33188,
      "old_path": "drivers/net/virtio_net.c",
      "new_id": "156d76fee1646de6f3260d176f300218734dc246",
      "new_mode": 33188,
      "new_path": "drivers/net/virtio_net.c"
    }
  ]
}