)]}'
{
  "commit": "c6742b14fe7352059cd4954a356a8105757af31b",
  "tree": "eae15d976e0f23a28cc004a44ca863ed440cb888",
  "parents": [
    "24b41d66c8ad8f77839fca777b92e365dad0cf5c"
  ],
  "author": {
    "name": "Paolo Bonzini",
    "email": "pbonzini@redhat.com",
    "time": "Tue Jul 14 13:45:34 2015 +0200"
  },
  "committer": {
    "name": "Paolo Bonzini",
    "email": "pbonzini@redhat.com",
    "time": "Thu Jul 16 20:00:20 2015 +0200"
  },
  "message": "memory: fix refcount leak in memory_region_present\n\nmemory_region_present() leaks a reference to a MemoryRegion in the\ncase \"mr \u003d\u003d container\".  While fixing it, avoid reference counting\naltogether for memory_region_present(), by using RCU only.\n\nThe return value could in principle be already invalid immediately\nafter memory_region_present returns, but presumably the caller knows\nthat and it\u0027s using memory_region_present to probe for devices that\nare unpluggable, or something like that.  The RCU critical section\nis needed anyway, because it protects as-\u003ecurrent_map.\n\nReported-by: Peter Maydell \u003cpeter.maydell@linaro.org\u003e\nSigned-off-by: Paolo Bonzini \u003cpbonzini@redhat.com\u003e\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "5a0cc66982dc8c165a40aa8abaac1c8016339381",
      "old_mode": 33188,
      "old_path": "memory.c",
      "new_id": "0acebb1f22f040649a7c055dddb33310e9179de4",
      "new_mode": 33188,
      "new_path": "memory.c"
    }
  ]
}