Bluetooth: Fix NULL pointer dereference when sending data
authorAndrzej Kaczmarek <andrzej.kaczmarek@tieto.com>
Tue, 25 Feb 2014 16:16:22 +0000 (17:16 +0100)
committerMarcel Holtmann <marcel@holtmann.org>
Tue, 25 Feb 2014 18:02:53 +0000 (10:02 -0800)
When trying to allocate skb for new PDU, l2cap_chan is unlocked so we
can sleep waiting for memory as otherwise there's possible deadlock as
fixed in e454c84464. However, in a6a5568c03 lock was moved from socket
to channel level and it's no longer safe to just unlock and lock again
without checking l2cap_chan state since channel can be disconnected
when lock is not held.

This patch adds missing checks for l2cap_chan state when returning from
call which allocates skb.

Scenario is easily reproducible by running rfcomm-tester in a loop.

BUG: unable to handle kernel NULL pointer dereference at         (null)
IP: [<ffffffffa0442169>] l2cap_do_send+0x29/0x120 [bluetooth]
PGD 0
Oops: 0000 [#1] SMP
Modules linked in:
CPU: 7 PID: 4038 Comm: krfcommd Not tainted 3.14.0-rc2+ #15
Hardware name: Dell Inc. OptiPlex 790/0HY9JP, BIOS A10 11/24/2011
task: ffff8802bdd731c0 ti: ffff8801ec986000 task.ti: ffff8801ec986000
RIP: 0010:[<ffffffffa0442169>]  [<ffffffffa0442169>] l2cap_do_send+0x29/0x120
RSP: 0018:ffff8801ec987ad8  EFLAGS: 00010202
RAX: 0000000000000000 RBX: ffff8800c5796800 RCX: 0000000000000000
RDX: ffff880410e7a800 RSI: ffff8802b6c1da00 RDI: ffff8800c5796800
RBP: ffff8801ec987af8 R08: 00000000000000c0 R09: 0000000000000300
R10: 000000000000573b R11: 000000000000573a R12: ffff8802b6c1da00
R13: 0000000000000000 R14: ffff8802b6c1da00 R15: 0000000000000000
FS:  0000000000000000(0000) GS:ffff88042dce0000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000000000 CR3: 000000041257c000 CR4: 00000000000407e0
Stack:
 ffff8801ec987d78 ffff8800c5796800 ffff8801ec987d78 0000000000000000
 ffff8801ec987ba8 ffffffffa0449e37 0000000000000004 ffff8801ec987af0
 ffff8801ec987d40 0000000000000282 0000000000000000 ffffffff00000004
Call Trace:
 [<ffffffffa0449e37>] l2cap_chan_send+0xaa7/0x1120 [bluetooth]
 [<ffffffff81770100>] ? _raw_spin_unlock_bh+0x20/0x40
 [<ffffffffa045188b>] l2cap_sock_sendmsg+0xcb/0x110 [bluetooth]
 [<ffffffff81652b0f>] sock_sendmsg+0xaf/0xc0
 [<ffffffff810a8381>] ? update_curr+0x141/0x200
 [<ffffffff810a8961>] ? dequeue_entity+0x181/0x520
 [<ffffffff81652b60>] kernel_sendmsg+0x40/0x60
 [<ffffffffa04a8505>] rfcomm_send_frame+0x45/0x70 [rfcomm]
 [<ffffffff810766f0>] ? internal_add_timer+0x20/0x50
 [<ffffffffa04a8564>] rfcomm_send_cmd+0x34/0x60 [rfcomm]
 [<ffffffffa04a8605>] rfcomm_send_disc+0x75/0xa0 [rfcomm]
 [<ffffffffa04aacec>] rfcomm_run+0x8cc/0x1a30 [rfcomm]
 [<ffffffffa04aa420>] ? rfcomm_check_accept+0xc0/0xc0 [rfcomm]
 [<ffffffff8108e3a9>] kthread+0xc9/0xe0
 [<ffffffff8108e2e0>] ? flush_kthread_worker+0xb0/0xb0
 [<ffffffff817795fc>] ret_from_fork+0x7c/0xb0
 [<ffffffff8108e2e0>] ? flush_kthread_worker+0xb0/0xb0
Code: 00 00 66 66 66 66 90 55 48 89 e5 48 83 ec 20 f6 05 d6 a3 02 00 04
RIP  [<ffffffffa0442169>] l2cap_do_send+0x29/0x120 [bluetooth]
 RSP <ffff8801ec987ad8>
CR2: 0000000000000000

Signed-off-by: Andrzej Kaczmarek <andrzej.kaczmarek@tieto.com>
Acked-by: Johan Hedberg <johan.hedberg@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
net/bluetooth/l2cap_core.c

index 6ace116f3b394ec8b83e5751694318f736254da1..7bd78c5487fbe420a7971d2ec13ae345827fa761 100644 (file)
@@ -2434,6 +2434,14 @@ int l2cap_chan_send(struct l2cap_chan *chan, struct msghdr *msg, size_t len,
                if (IS_ERR(skb))
                        return PTR_ERR(skb);
 
+               /* Channel lock is released before requesting new skb and then
+                * reacquired thus we need to recheck channel state.
+                */
+               if (chan->state != BT_CONNECTED) {
+                       kfree_skb(skb);
+                       return -ENOTCONN;
+               }
+
                l2cap_do_send(chan, skb);
                return len;
        }
@@ -2483,6 +2491,14 @@ int l2cap_chan_send(struct l2cap_chan *chan, struct msghdr *msg, size_t len,
                if (IS_ERR(skb))
                        return PTR_ERR(skb);
 
+               /* Channel lock is released before requesting new skb and then
+                * reacquired thus we need to recheck channel state.
+                */
+               if (chan->state != BT_CONNECTED) {
+                       kfree_skb(skb);
+                       return -ENOTCONN;
+               }
+
                l2cap_do_send(chan, skb);
                err = len;
                break;