There's no reason why all users of L2CAP would need to worry about
initializing chan->nesting to L2CAP_NESTING_NORMAL (which is important
since 0 is the same as NESTING_SMP). This patch moves the initialization
to the common place that's used to create all new channels, i.e. the
l2cap_chan_create() function.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

The teardown callback for L2CAP channels is problematic in that it is
explicitly called for all types of channels from l2cap_chan_del(),
meaning it's not possible to hard-code a nesting level when taking the
socket lock. The simplest way to have a correct nesting level for the
socket locking is to use the same value as for the chan. This also means
that the other places trying to lock parent sockets need to be update to
use the chan value (since L2CAP_NESTING_PARENT is defined as 2 whereas
SINGLE_DEPTH_NESTING has the value 1).
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

By default lockdep considers all L2CAP channels equal. This would mean
that we get warnings if a channel is locked when another one's lock is
tried to be acquired in the same thread. This kind of inter-channel
locking dependencies exist in the form of parent-child channels as well
as any channel wishing to elevate the security by requesting procedures
on the SMP channel.

To eliminate the chance for these lockdep warnings we introduce a
nesting level for each channel and use that when acquiring the channel
lock. For now there exists the earlier mentioned three identified
categories: SMP, "normal" channels and parent channels (i.e. those in
BT_LISTEN state). The nesting level is defined as atomic_t since we need
access to it before the lock is actually acquired.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

The mgmt_user_passkey_request and related functions do not do anything
else except read access to hdev->id. This member never changes after the
hdev creation so there is no need to acquire a lock to read it.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

Any code calling bt_accept_dequeue() to get a new child socket from a
server socket should use lock_sock_nested to avoid lockdep warnings due
to the parent and child sockets being locked at the same time. The
l2cap_sock_accept() function is already doing this correctly but a
second place calling bt_accept_dequeue() is the code path from
l2cap_sock_teardown_cb() that calls l2cap_sock_cleanup_listen().

This patch fixes the proper nested locking annotation and thereby avoids
the following style of lockdep warning.

[  +0.000224] [ INFO: possible recursive locking detected ]
[  +0.000222] 3.17.0+ #1153 Not tainted
[  +0.000130] ---------------------------------------------
[  +0.000227] l2cap-tester/562 is trying to acquire lock:
[  +0.000210]  (sk_lock-AF_BLUETOOTH-BTPROTO_L2CAP){+.+...}, at: [<c1393f47>] bt_accept_dequeue+0x68/0x11b
[  +0.000467]
but task is already holding lock:
[  +0.000186]  (sk_lock-AF_BLUETOOTH-BTPROTO_L2CAP){+.+...}, at: [<c13b949a>] lock_sock+0xa/0xc
[  +0.000421]
other info that might help us debug this:
[  +0.000199]  Possible unsafe locking scenario:

[  +0.000117]        CPU0
[  +0.000000]        ----
[  +0.000000]   lock(sk_lock-AF_BLUETOOTH-BTPROTO_L2CAP);
[  +0.000000]   lock(sk_lock-AF_BLUETOOTH-BTPROTO_L2CAP);
[  +0.000000]
 *** DEADLOCK ***
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

When kfree() is all that's needed to free an object protected by RCU
there's a kfree_rcu() convenience function that can be used. This patch
updates the 6lowpan code to use this, thereby eliminating the need for
the separate peer_free() function.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

This patch fixes a regression that was introduced by commit
cb77c3ec. In addition to BT_CONFIG,
BT_CONNECTED is also a state in which we may get a remote name and need
to indicate over mgmt the connection status. This scenario is
particularly likely to happen for incoming connections that do not need
authentication since there the hci_conn state will reach BT_CONNECTED
before the remote name is received.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

This fixes the following sparse warning:

net/bluetooth/amp.c:152:53: warning: Variable length array is used.

The warning itself is probably harmless since this kind of usage of
shash_desc is present also in other places in the kernel (there's even a
convenience macro SHASH_DESC_ON_STACK available for defining such stack
variables). However, dynamically allocated versions are also used in
several places of the kernel (e.g. kernel/kexec.c and lib/digsig.c)
which have the benefit of not exhibiting the sparse warning.

Since there are no more sparse warnings in the Bluetooth subsystem after
fixing this one it is now easier to spot whenever new ones might get
introduced by future patches.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

This patch reverts commit:

a7807d73 ("Bluetooth: 6lowpan: Avoid memory leak if memory allocation
fails")

which was wrong suggested by Alexander Aring. The function skb_unshare
run also kfree_skb on failure.
Signed-off-by: NAlexander Aring <alex.aring@gmail.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Cc: stable@vger.kernel.org # 3.18.x

If a remote name request is initiated while acl connection is going on,
and if it fails then mgmt_connected will be sent. Evetually after acl
connection, authentication will not be initiated and userspace will
never get pairing reply.

< HCI Command: Create Connection (0x01|0x0005) plen 13
    bdaddr AA:BB:CC:DD:EE:FF ptype 0xcc18 rswitch 0x01 clkoffset 0x2306 (valid)
    Packet type: DM1 DM3 DM5 DH1 DH3 DH5
> HCI Event: Command Status (0x0f) plen 4
    Create Connection (0x01|0x0005) status 0x00 ncmd 1
> HCI Event: Inquiry Complete (0x01) plen 1
    status 0x00
< HCI Command: Remote Name Request (0x01|0x0019) plen 10
    bdaddr AA:BB:CC:DD:EE:FF mode 1 clkoffset 0x2306
> HCI Event: Command Status (0x0f) plen 4
    Remote Name Request (0x01|0x0019) status 0x0c ncmd 1
    Error: Command Disallowed
> HCI Event: Connect Complete (0x03) plen 11
    status 0x00 handle 50 bdaddr 00:0D:FD:47:53:B2 type ACL encrypt 0x00
< HCI Command: Read Remote Supported Features (0x01|0x001b) plen 2
    handle 50
> HCI Event: Command Status (0x0f) plen 4
    Read Remote Supported Features (0x01|0x001b) status 0x00 ncmd 1
> HCI Event: Max Slots Change (0x1b) plen 3
    handle 50 slots 5
> HCI Event: Read Remote Supported Features (0x0b) plen 11
    status 0x00 handle 50
    Features: 0xff 0xff 0x8f 0xfe 0x9b 0xff 0x59 0x83
< HCI Command: Read Remote Extended Features (0x01|0x001c) plen 3
    handle 50 page 1
> HCI Event: Command Status (0x0f) plen 4
    Read Remote Extended Features (0x01|0x001c) status 0x00 ncmd 1
> HCI Event: Read Remote Extended Features (0x23) plen 13
    status 0x00 handle 50 page 1 max 1
    Features: 0x01 0x00 0x00 0x00 0x00 0x00 0x00 0x00

This patch sends mgmt_connected in remote name command status only if
conn->state is BT_CONFIG
Signed-off-by: NJaganath Kanakkassery <jaganath.k@samsung.com>
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>

Currently we ensure that the skb is freed on every error path in IPHC
decompression which makes it easy to introduce skb leaks.  By centralising
the skb_free into the receive function it makes future decompression routines
easier to maintain.  It does come at the expense of ensuring that the skb
passed into the decompression routine must not be copied.
Signed-off-by: NMartin Townsend <mtownsend1973@gmail.com>
Acked-by: NJukka Rissanen <jukka.rissanen@linux.intel.com>
Acked-by: NAlexander Aring <alex.aring@gmail.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

Opcodes in switch/case in hci_cmd_status_evt are not sorted
by value. This patch restores proper ordering.
Signed-off-by: NKuba Pawlak <kubax.t.pawlak@intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

If role switch was rejected by the controller and HCI Event: Command Status
returned with status "Command Disallowed" (0x0C) the flag
HCI_CONN_RSWITCH_PEND remains set. No further role switches are
possible as this flag prevents us from sending any new HCI Switch Role
requests and the only way to clear it is to receive a valid
HCI Event Switch Role.

This patch clears the flag if command was rejected.

2013-01-01 00:03:44.209913 < HCI Command: Switch Role (0x02|0x000b) plen 7
    bdaddr BC:C6:DB:C4:6F:79 role 0x00
    Role: Master
2013-01-01 00:03:44.210867 > HCI Event: Command Status (0x0f) plen 4
    Switch Role (0x02|0x000b) status 0x0c ncmd 1
    Error: Command Disallowed
Signed-off-by: NKuba Pawlak <kubax.t.pawlak@intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

This patch fixes the following sparse warnings in rfcomm/core.c:

net/bluetooth/rfcomm/core.c:391:16: warning: dubious: x | !y
net/bluetooth/rfcomm/core.c:546:24: warning: dubious: x | !y
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

According to Management Interface API 'Start Discovery' command should
generate a Command Complete event on failure. Currently kernel is
sending Command Status on early errors. This results in userspace
ignoring such event due to invalid size.

bluetoothd[28499]: src/adapter.c:trigger_start_discovery()
bluetoothd[28499]: src/adapter.c:cancel_passive_scanning()
bluetoothd[28499]: src/adapter.c:start_discovery_timeout()
bluetoothd[28499]: src/adapter.c:start_discovery_complete() status 0x0a
bluetoothd[28499]: Wrong size of start discovery return parameters
Reported-by: NJukka Taimisto <jtt@codenomicon.com>
Signed-off-by: NSzymon Janc <szymon.janc@tieto.com>
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>

The debufs entry for the BR/EDR whitelist is confusing since there is
a controller debugfs entry with the name white_list and both are two
different things.

With the BR/EDR whitelist, the actual interface in use is the device
list and thus just include all values from the internal BR/EDR whitelist
in the device_list debugfs entry.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>

Some Bluetooth drivers require to reset the upper stack. To avoid having
all drivers send HCI Hardware Error events, provide a generic function
to wrap the reset functionality.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>

The current kernel options do not make it clear which modules are for
Bluetooth Classic (BR/EDR) and which are for Bluetooth Low Energy (LE).

To make it really clear, introduce BT_BREDR and BT_LE options with
proper dependencies into the different modules. Both new options
default to y to not create a regression with previous kernel config
files.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>

When the HCI_Hardware_Error event is send by the controller or
injected by the driver, then at least print an error message.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>

When the HCI_Reset command returns, the status needs to be checked. It
is unlikely that HCI_Reset actually fails, but when it fails, it is a
bad idea to reset all values since the controller will have not reset
its values in that case.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>

__hci_cmd_sync_ev(), __hci_req_sync() could miss wake_up_interrupt from
hci_req_sync_complete() because hci_cmd_work() workqueue and its response
could be completed before they are ready to get the signal through
add_wait_queue(), set_current_state(TASK_INTERRUPTIBLE).
Signed-off-by: NChan-yeol Park <chanyeol.park@samsung.com>
Signed-off-by: NKyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>

The internal representation of the LE white list needs to be cleared
when receiving a successful HCI_Reset command. A reset of the controller
is expected to start with an empty LE white list.

When the LE white list is not cleared on controller reset, the passive
background scanning might skip programming the remote devices. Only
changes to the LE white list are programmed when passive background
is started.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Cc: stable@vger.kernel.org # 3.17.x

This was accidentally changed from list_for_each_entry_safe() to
list_for_each_entry() so now it has a use after free bug.  I've changed
it back.

Fixes: 90305829 ('Bluetooth: 6lowpan: Converting rwlocks to use RCU')
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Acked-by: NJukka Rissanen <jukka.rissanen@linux.intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

Use spin_lock_bh() as the code is called from softirq in networking subsystem.
This is needed to prevent deadlocks when 6lowpan link is in use.
Signed-off-by: NJukka Rissanen <jukka.rissanen@linux.intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

These days we allow simultaneous LE scanning and advertising. Checking
for whether advertising is enabled or not is therefore not a reliable
way to determine whether directed advertising was used to trigger the
connection creation. The appropriate place to check (instead of the hdev
context) is the connection role that's stored in the hci_conn. This
patch fixes such a check in le_conn_timeout() which could otherwise lead
to incorrect HCI commands being sent.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Cc: stable@vger.kernel.org # 3.16.x

The le_conn_timeout() may call hci_le_conn_failed() which in turn may
call hci_conn_del(). Trying to use the _sync variant for cancelling the
conn timeout from hci_conn_del() could therefore result in a deadlock.
This patch converts hci_conn_del() to use the non-sync variant so the
deadlock is not possible.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Cc: stable@vger.kernel.org # 3.16.x

Fix spelling errors in comments.
Signed-off-by: NStephen Hemminger <stephen@networkplumber.org>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

When a device ndo_start_xmit() calls again dev_queue_xmit(),
lockdep can complain because dev_queue_xmit() is re-entered and the
spinlocks protecting tx queues share a common lockdep class.

Same issue was fixed for ieee802154 in commit "20e7c4e8"
Signed-off-by: NJukka Rissanen <jukka.rissanen@linux.intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

The rwlocks are converted to use RCU. This helps performance as the
irq locks are not needed any more.
Signed-off-by: NJukka Rissanen <jukka.rissanen@linux.intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

This reverts commits c6992e9e and
4cd3362d.

The reason for the revert is that we cannot have more than one module
initialization function and the SMP one breaks the build with modular
kernels. As the proper fix for this is right now looking non-trivial
it's better to simply revert the problematic patches in order to keep
the upstream tree compilable.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

As we have decouple decompression from data delivery we can now rename all
occurences of process_data in receive path.
Signed-off-by: NMartin Townsend <mtownsend1973@gmail.com>
Acked-by: NAlexander Aring <alex.aring@gmail.com>
Acked-by: NJukka Rissanen <jukka.rissanen@linux.intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

Signed-off-by: NMartin Townsend <mtownsend1973@gmail.com>
Acked-by: NAlexander Aring <alex.aring@gmail.com>
Acked-by: NJukka Rissanen <jukka.rissanen@linux.intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

As process_data now returns just error codes fix up the calls to this
function to only drop the skb if an error code is returned.
Signed-off-by: NMartin Townsend <mtownsend1973@gmail.com>
Acked-by: NAlexander Aring <alex.aring@gmail.com>
Acked-by: NJukka Rissanen <jukka.rissanen@linux.intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

Separating skb delivery from decompression ensures that we can support further
decompression schemes and removes the mixed return value of error codes with
NET_RX_FOO.
Signed-off-by: NMartin Townsend <mtownsend1973@gmail.com>
Acked-by: NAlexander Aring <alex.aring@gmail.com>
Acked-by: NJukka Rissanen <jukka.rissanen@linux.intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

This patch adds self-tests for the c1 and s1 crypto functions used for
SMP pairing. The data used is the sample data from the core
specification.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

This patch adds a basic skeleton for SMP self-tests. The tests are put
behind a new configuration option since running them will slow down the
boot process. For now there are no actual tests defined but those will
come in a subsequent patch.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

In order to make unit testing possible we need to make the SMP crypto
functions only take the crypto context instead of the full SMP context
(the latter would require having hci_dev, hci_conn, l2cap_chan,
l2cap_conn, etc around). The drawback is that we no-longer get the
involved hdev in the debug logs, but this is really the only way to make
simple unit tests for the code.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

use clkoff_cp for hci_cp_read_clock_offset instead of cp
(already defined above).
Suggested-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NFabian Frederick <fabf@skynet.be>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

In the error case where credits is greater than max_credits there
is a missing l2cap_chan_unlock before returning.
Signed-off-by: NMartin Townsend <mtownsend1973@gmail.com>
Tested-by: NJukka Rissanen <jukka.rissanen@linux.intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>

Currently there are potentially 2 skb_copy_expand calls in IPHC
decompression.  This patch replaces this with one call to
skb_cow which will check to see if there is enough headroom
first to ensure it's only done if necessary and will handle
alignment issues for cache.
As skb_cow uses pskb_expand_head we ensure the skb isn't shared from
bluetooth and ieee802.15.4 code that use the IPHC decompression.
Signed-off-by: NMartin Townsend <martin.townsend@xsilon.com>
Acked-by: NAlexander Aring <alex.aring@gmail.com>
Acked-by: NJukka Rissanen <jukka.rissanen@linux.intel.com>
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>