)]}'
{
  "commit": "ceca3c193e73bb409d093515fae9e5ef02ffa2de",
  "tree": "b29f24fb0c280dd04bc4cdd0d68ffdec06e48ad7",
  "parents": [
    "dc0b124d8edc6c2f95fc3a689cd40ec05ad85108"
  ],
  "author": {
    "name": "Oleg Nesterov",
    "email": "oleg@redhat.com",
    "time": "Tue Apr 26 22:37:11 2011 +0200"
  },
  "committer": {
    "name": "Chris Metcalf",
    "email": "cmetcalf@tilera.com",
    "time": "Wed May 04 14:41:53 2011 -0400"
  },
  "message": "tile: do_hardwall_trap: do not play with task-\u003esighand\n\n1. do_hardwall_trap() checks -\u003esighand !\u003d NULL and then takes -\u003esiglock.\n\n   This is unsafe even if the task can\u0027t run (I assume it is pinned to\n   the same CPU), its parent can reap the task and set -\u003esighand \u003d NULL\n   right after this check. Even if the compiler dosn\u0027t read -\u003esighand\n   twice and this memory can\u0027t to away __group_send_sig_info() is wrong\n   after that. Use do_send_sig_info().\n\n2. Send SIGILL to the thread, not to the whole process. Unless it has\n   the handler or blocked this kills the whole thread-group as before.\n   IIUC, different threads can be bound to different rect\u0027s.\n\n3. Check PF_EXITING instead of -\u003esighand. A zombie thread can go away\n   but its -\u003esighand can be !NULL.\n\nReported-by: Matt Fleming \u003cmatt@console-pimps.org\u003e\nSigned-off-by: Oleg Nesterov \u003coleg@redhat.com\u003e\nSigned-off-by: Chris Metcalf \u003ccmetcalf@tilera.com\u003e\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "e910530436e64a4079ba65e22bdca13e6ad01bfa",
      "old_mode": 33188,
      "old_path": "arch/tile/kernel/hardwall.c",
      "new_id": "3bddef710de432b089d03d9ff52e928ea487d938",
      "new_mode": 33188,
      "new_path": "arch/tile/kernel/hardwall.c"
    }
  ]
}