Re: [PATCH] Bluetooth: Fix potential NULL pointer dereference in L2CAP

[Johan Hedberg <johan.hedberg@gmail.com>]

Hi Andrei,

On Tue, Jun 05, 2012, Andrei Emeltchenko wrote:
> > In the following scenario conn->hchan is not set when l2cap_conn_del is
> > called:
> > 
> > < HCI Command: LE Create Connection (0x08|0x000d) plen 25
> >     bdaddr xx:xx:xx:xx:xx:xx type 0
> > > HCI Event: Command Status (0x0f) plen 4
> >     LE Create Connection (0x08|0x000d) status 0x00 ncmd 1
> > > HCI Event: LE Meta Event (0x3e) plen 19
> >     LE Connection Complete
> >       status 0x00 handle 64, role master
> >       bdaddr xx:xx:xx:xx:xx:xx (Public)
> > < ACL data: handle 64 flags 0x00 dlen 11
> >     SMP: Pairing Request (0x01)
> >       capability 0x04 oob 0x00 auth req 0x01
> >       max key size 0x10 init key dist 0x00 resp key dist 0x01
> >       Capability: KeyboardDisplay (OOB data not present)
> >       Authentication: Bonding (No MITM Protection)
> >       Initiator Key Distribution:
> >       Responder Key Distribution:  LTK
> > > HCI Event: Number of Completed Packets (0x13) plen 5
> >     handle 64 packets 1
> > < HCI Command: Disconnect (0x01|0x0006) plen 3
> >     handle 64 reason 0x13
> >     Reason: Remote User Terminated Connection
> > > HCI Event: Command Status (0x0f) plen 4
> >     Disconnect (0x01|0x0006) status 0x00 ncmd 1
> > > HCI Event: Disconn Complete (0x05) plen 4
> >     status 0x00 handle 64 reason 0x22
> >     Reason: LMP Response Timeout
> > 
> > This results in the following crash:
> > 
> > BUG: unable to handle kernel NULL pointer dereference at 00000004
> > IP: [<c123f9e5>] hci_chan_del+0x41/0x6e
> > *pde = 00000000
> > Oops: 0002 [#1] SMP
> > Modules linked in:
> > 
> > Pid: 32, comm: kworker/u:3 Not tainted 3.5.0-rc1+ #322 Bochs Bochs
> > EIP: 0060:[<c123f9e5>] EFLAGS: 00010246 CPU: 1
> > EIP is at hci_chan_del+0x41/0x6e
> > EAX: 00200909 EBX: f5dd2280 ECX: 00000000 EDX: 00000000
> > ESI: f5d5edc4 EDI: f6201e60 EBP: f6201e50 ESP: f6201e48
> >  DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068
> > CR0: 8005003b CR2: 00000004 CR3: 35f12000 CR4: 00000690
> > DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
> > DR6: ffff0ff0 DR7: 00000400
> > Process kworker/u:3 (pid: 32, ti=f6200000 task=f686e730 task.ti=f6200000)
> > Stack:
> >  f5d5ef00 f6201e60 f6201e88 c12507b3 f62e6000 c12507b3 c1245fad 0000000c
> >  f5d5ef94 00000026 f5d5ef9c f5d5edc4 f6917800 f62e6000 00000022 f6201e98
> >  f6201ea8 c125440b f6201eb8 c125440b c124963d f6254622 f62e6000 f6201eb8
> > Call Trace:
> >  [<c12507b3>] l2cap_conn_del+0xef/0x135
> >  [<c12507b3>] ? l2cap_conn_del+0xef/0x135
> >  [<c1245fad>] ? mgmt_event+0x95/0xa6
> >  [<c125440b>] l2cap_disconn_cfm+0x49/0x57
> >  [<c125440b>] ? l2cap_disconn_cfm+0x49/0x57
> >  [<c124963d>] ? user_confirm_reply+0x7d/0x7d
> >  [<c124446c>] hci_event_packet+0x33e/0x1cce
> >  [<c124446c>] ? hci_event_packet+0x33e/0x1cce
> >  [<c11e1aaa>] ? __kfree_skb+0x6a/0x6d
> >  [<c11e1af9>] ? kfree_skb+0x25/0x27
> >  [<c124be14>] ? hci_send_to_sock+0x168/0x174
> >  [<c123b66c>] hci_rx_work+0xf3/0x347
> >  [<c123b66c>] ? hci_rx_work+0xf3/0x347
> >  [<c123b9a7>] ? hci_cmd_work+0xb4/0xd8
> > 
> > This patch fixes this issue by adding a NULL check for conn->hchan.
> > 
> > Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
> > ---
> >  net/bluetooth/l2cap_core.c |    3 ++-
> >  1 file changed, 2 insertions(+), 1 deletion(-)
> > 
> > diff --git a/net/bluetooth/l2cap_core.c b/net/bluetooth/l2cap_core.c
> > index f9bffe3..62df491 100644
> > --- a/net/bluetooth/l2cap_core.c
> > +++ b/net/bluetooth/l2cap_core.c
> > @@ -1295,7 +1295,8 @@ static void l2cap_conn_del(struct hci_conn *hcon, int err)
> >  
> >  	mutex_unlock(&conn->chan_lock);
> >  
> > -	hci_chan_del(conn->hchan);
> > +	if (conn->hchan)
> > +		hci_chan_del(conn->hchan);
> 
> Is this SMP - related bug? Shall we check rather for LE/SMP stuff?

This happened with LE/SMP but if you look at the backlog it's quite
clear that this place is the code where a NULL pointer was passed to
hci_chan_del. After applying the patch I no-longer got any crash and the
connection was cleaned up like it should.

The whole L2CAP part of the kernel is something I'm not very familiar
with so I don't have any "deep" analysis I could make about the issue.
So I'm expecting people who are (e.g. Gustavo) to take a look at the
hcidump, the backtrace, the fact that the issue went away with the
patch, and determine whether the patch is appropriate or not.

Johan