)]}'
{
  "commit": "e4e364e865b382f9d99c7fc230ec2ce7df21257a",
  "tree": "9ff5ab54a0e40d7ad2b55d3ec48c6e175ebf50c7",
  "parents": [
    "2741a559a01e1ba9bf87285569dc1a104d134ecf"
  ],
  "author": {
    "name": "Paul Jackson",
    "email": "pj@sgi.com",
    "time": "Fri Mar 31 02:30:52 2006 -0800"
  },
  "committer": {
    "name": "Linus Torvalds",
    "email": "torvalds@g5.osdl.org",
    "time": "Fri Mar 31 12:18:55 2006 -0800"
  },
  "message": "[PATCH] cpuset: memory migration interaction fix\n\nFix memory migration so that it works regardless of what cpuset the invoking\ntask is in.\n\nIf a task invoked a memory migration, by doing one of:\n\n       1) writing a different nodemask to a cpuset \u0027mems\u0027 file, or\n\n       2) writing a tasks pid to a different cpuset\u0027s \u0027tasks\u0027 file,\n          where the cpuset had its \u0027memory_migrate\u0027 option turned on, then the\n          allocation of the new pages for the migrated task(s) was constrained\n          by the invoking tasks cpuset.\n\nIf this task wasn\u0027t in a cpuset that allowed the requested memory nodes, the\nmemory migration would happen to some other nodes that were in that invoking\ntasks cpuset.  This was usually surprising and puzzling behaviour: Why didn\u0027t\nthe pages move?  Why did the pages move -there-?\n\nTo fix this, temporarilly change the invoking tasks \u0027mems_allowed\u0027 task_struct\nfield to the nodes the migrating tasks is moving to, so that new pages can be\nallocated there.\n\nSigned-off-by: Paul Jackson \u003cpj@sgi.com\u003e\nAcked-by: Christoph Lameter \u003cclameter@sgi.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "bf42381a41953233e5415b29c0053e8959346516",
      "old_mode": 33188,
      "old_path": "kernel/cpuset.c",
      "new_id": "72248d1b9e3f7a18198796ad1134c9abb2fd4b4e",
      "new_mode": 33188,
      "new_path": "kernel/cpuset.c"
    }
  ]
}