]> www.infradead.org Git - users/jedix/linux-maple.git/commit
ipmi: Remove smi_msg from waiting_rcv_msgs list before handle_one_recv_msg()
authorJunichi Nomura <j-nomura@ce.jp.nec.com>
Fri, 10 Jun 2016 04:31:52 +0000 (04:31 +0000)
committerBrian Maly <brian.maly@oracle.com>
Tue, 21 Aug 2018 17:57:24 +0000 (13:57 -0400)
commit2e2f1866e555d022b13d25fe0832700333ae1e40
treec2f8e68f3bcf7e4c1eee08006de309d5d5658380
parentb77ff7760f482c6e277f1cea6f2687b39af6cf1a
ipmi: Remove smi_msg from waiting_rcv_msgs list before handle_one_recv_msg()

Commit 7ea0ed2b5be8 ("ipmi: Make the message handler easier to use for
SMI interfaces") changed handle_new_recv_msgs() to call handle_one_recv_msg()
for a smi_msg while the smi_msg is still connected to waiting_rcv_msgs list.
That could lead to following list corruption problems:

1) low-level function treats smi_msg as not connected to list

  handle_one_recv_msg() could end up calling smi_send(), which
  assumes the msg is not connected to list.

  For example, the following sequence could corrupt list by
  doing list_add_tail() for the entry still connected to other list.

    handle_new_recv_msgs()
      msg = list_entry(waiting_rcv_msgs)
      handle_one_recv_msg(msg)
        handle_ipmb_get_msg_cmd(msg)
          smi_send(msg)
            spin_lock(xmit_msgs_lock)
            list_add_tail(msg)
            spin_unlock(xmit_msgs_lock)

2) race between multiple handle_new_recv_msgs() instances

  handle_new_recv_msgs() once releases waiting_rcv_msgs_lock before calling
  handle_one_recv_msg() then retakes the lock and list_del() it.

  If others call handle_new_recv_msgs() during the window shown below
  list_del() will be done twice for the same smi_msg.

  handle_new_recv_msgs()
    spin_lock(waiting_rcv_msgs_lock)
    msg = list_entry(waiting_rcv_msgs)
    spin_unlock(waiting_rcv_msgs_lock)
  |
  | handle_one_recv_msg(msg)
  |
    spin_lock(waiting_rcv_msgs_lock)
    list_del(msg)
    spin_unlock(waiting_rcv_msgs_lock)

Fixes: 7ea0ed2b5be8 ("ipmi: Make the message handler easier to use for SMI interfaces")
Signed-off-by: Jun'ichi Nomura <j-nomura@ce.jp.nec.com>
[Added a comment to describe why this works.]
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Cc: stable@vger.kernel.org # 3.19
Tested-by: Ye Feng <yefeng.yl@alibaba-inc.com>
(cherry picked from commit ae4ea9a2460c7fee2ae8feeb4dfe96f5f6c3e562)

Orabug:28281455

Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Signed-off-by: Brian Maly <brian.maly@oracle.com>
drivers/char/ipmi/ipmi_msghandler.c