The "terminate" flag is guaranteed to be set before the session terminates
and the handlers are woken up. Hence, we need to add it to the
sleep-condition.
Note that testing the flags is not enough as nothing prevents us from
setting the flags again after the session-handler terminated.
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
This field is always BT_CONNECTED. Remove it and set it to BT_CONNECTED in
hidp_copy_session() unconditionally.
Also note that this field is totally bogus. Userspace can query an
hidp-session for its state. However, whenever user-space queries us, this
field should be BT_CONNECTED. If it wasn't BT_CONNECTED, then we would be
currently cleaning up the session and the session itself would exit in the
next few milliseconds. Hence, there is no reason to let user-space know
that the session will exit now if they cannot make _any_ use of that.
Thus, remove the field and let user-space think that a session is always
BT_CONNECTED as long as they can query it.
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
We currently do not allow using hci_conn from outside of HCI-core.
However, several other users could make great use of it. This includes
HIDP, rfcomm and all other sub-protocols that rely on an active
connection.
Hence, we now introduce hci_conn ref-counting. We currently never call
get_device(). put_device() is exclusively used in hci_conn_del_sysfs().
Hence, we currently never have a greater device-refcnt than 1.
Therefore, it is safe to move the put_device() call from
hci_conn_del_sysfs() to hci_conn_del() (it's the only caller). In fact,
this even fixes a "use-after-free" bug as we access hci_conn after calling
hci_conn_del_sysfs() in hci_conn_del().
From now on we can add references to hci_conn objects in other layers
(like l2cap_sock, HIDP, rfcomm, ...) and grab a reference via
hci_conn_get(). This does _not_ guarantee, that the connection is still
alive. But, this isn't what we want. We can simply lock the hci_conn
device and use "device_is_registered(hci_conn->dev)" to test that.
However, this is hardly necessary as outside users should never rely on
the HCI connection to be alive, anyway. Instead, they should solely rely
on the device-object to be available.
But if sub-devices want the hci_conn object as sysfs parent, they need to
be notified when the connection drops. This will be introduced in later
patches with l2cap_users.
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
hci_conn_hold/put_device() is used to control when hci_conn->dev is no
longer needed and can be deleted from the system. Lets first look how they
are currently used throughout the code (excluding HIDP!).
All code that uses hci_conn_hold_device() looks like this:
...
hci_conn_hold_device();
hci_conn_add_sysfs();
...
On the other side, hci_conn_put_device() is exclusively used in
hci_conn_del().
So, considering that hci_conn_del() must not be called twice (which would
fail horribly), we know that hci_conn_put_device() is only called _once_
(which is in hci_conn_del()).
On the other hand, hci_conn_add_sysfs() must not be called twice, either
(it would call device_add twice, which breaks the device, see
drivers/base/core.c). So we know that hci_conn_hold_device() is also
called only once (it's only called directly before hci_conn_add_sysfs()).
So hold and put are known to be called only once. That means we can safely
remove them and directly call hci_conn_del_sysfs() in hci_conn_del().
But there is one issue left: HIDP also uses hci_conn_hold/put_device().
However, this case can be ignored and simply removed as it is totally
broken. The issue is, the only thing HIDP delays with
hci_conn_hold_device() is the removal of the hci_conn->dev from sysfs.
But, the hci_conn device has no mechanism to get notified when its own
parent (hci_dev) gets removed from sysfs. hci_dev_hold/put() does _not_
control when it is removed but only when the device object is created
and destroyed.
And hci_dev calls hci_conn_flush_*() when it removes itself from sysfs,
which itself causes hci_conn_del() to be called, but it does _not_ cause
hci_conn_del_sysfs() to be called, which is wrong.
Hence, we fix it to call hci_conn_del_sysfs() in hci_conn_del(). This
guarantees that a hci_conn object is removed from sysfs _before_ its
parent hci_dev is removed.
The changes to HIDP look scary, wrong and broken. However, if you look at
the HIDP session management, you will notice they're already broken in the
exact _same_ way (ever tried "unplugging" HIDP devices? Breaks _all_ the
time).
So this patch only makes HIDP look _scary_ and _obviously broken_. It does
not break HIDP itself, it already is!
See later patches in this series which fix HIDP to use proper
session-management.
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
This patch sends Reject Synchronous Connection Request Command when
hci_conn_timeout is triggered, and the SCO connection is in BT_CONNECT2
state. It prevents inconsistency if the remote host doesn't implement
properly the timeout for the connection request, and it removes the
connection reference left when the socket is closed for incoming SCO
connections.
[ 2650.129080] sco_sock_release: sock ffff8801ca417400, sk ffff88020c408800
[ 2650.129092] sco_sock_clear_timer: sock ffff88020c408800 state 6
[ 2650.129101] __sco_sock_close: sk ffff88020c408800 state 6 socket
ffff8801ca417400
[ 2650.129108] sco_chan_del: sk ffff88020c408800, conn ffff8801c650ea20,
err 104
[ 2650.129114] hci_conn_put: hcon ffff88020c40a800 orig refcnt 1
[ 2650.129128] sco_sock_kill: sk ffff88020c408800 state 9
[ 2650.129135] sco_sock_destruct: sk ffff88020c408800
[ 2650.138468] hci_conn_timeout: hcon ffff88020c40a800 state BT_CONNECT2
Signed-off-by: Claudio Takahasi <claudio.takahasi@openbossa.org>
Signed-off-by: Vinicius Costa Gomes <vinicius.gomes@openbossa.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
This patch changes the memory allocation flags in the sco_conn_add
function, replacing the type to GFP_KERNEL. This function is executed
in process context and it is not called inside an atomic section.
Signed-off-by: Claudio Takahasi <claudio.takahasi@openbossa.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
We use _get() and _put() for device ref-counting in the kernel. However,
hci_conn_put() is _not_ used for ref-counting, hence, rename it to
hci_conn_drop() so we can later fix ref-counting and introduce
hci_conn_put().
hci_conn_hold() and hci_conn_put() are currently used to manage how long a
connection should be held alive. When the last user drops the connection,
we spawn a delayed work that performs the disconnect. Obviously, this has
nothing to do with ref-counting for the _object_ but rather for the
keep-alive of the connection.
But we really _need_ proper ref-counting for the _object_ to allow
connection-users like rfcomm-tty, HIDP or others.
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Trying to squeeze every single vendor setup routine into the same function
and have it assigned all the time is actually a bad idea. Especially since
the core can handle the absence of a setup routine perfectly fine.
To make this a lot simpler for future additions of vendor setup code,
split the BCM92035 setup into its own function and only assign it when
this specific device has been detected.
Doing it like this has the nice side benefit that we do not have to keep
a copy of the driver_info around.
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
We need to verify that the given sockets actually are l2cap sockets. If
they aren't, we are not supposed to access bt_sk(sock) and we shouldn't
start the session if the offsets turn out to be valid local BT addresses.
That is, if someone passes a TCP socket to HIDCONNADD, then we access some
random offset in the TCP socket (which isn't even guaranteed to be valid).
Fix this by checking that the socket is an l2cap socket.
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
The driver init queue is no longer needed. This can be all handled
inside the drivers now. So remove it.
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
With the early init stage during setup, this quirk can be simplified
and kept fully inside the driver.
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Some drivers require a special stage for their early init. This is
always specific to the driver or transport. So call back into driver to
allow bringing up the device.
The advantage with this stage is that the Bluetooth core is actually
handling the HCI layer now. This means that command and event processing
is available.
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
This patch adds a __hci_cmd_sync_ev function, analogous to
__hci_cmd_sync except that it also takes an event parameter to indicate
that the command completes with a special event instead of command
complete. Internally this new function takes advantage of the
hci_req_add_ev function introduced in the previous patch.
The primary expected user of this new function are the setup routines of
HCI drivers which may want to send custom commands and return only when
they have completed.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
This patch adds support for having commands within HCI requests that do
not result in a command complete but some other event. This is at least
needed for some vendor specific commands to be issued in the
hdev->setup() procecure, but might also be useful for other commands.
The way that the support is implemented is by extending the skb control
buffer to have a field to indicate that the command is expected to
terminate with a special event. After sending the command each received
event can then be compared against this field through hdev->sent_cmd.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
This patch adds a helper function for sending a single HCI command
waiting for its completion and then returning back the parameters in the
resulting command complete event (if there was one).
The implementation is very similar to that of hci_req_sync() except that
instead of invocing a callback for sending HCI commands the function
constructs and sends one itself and after being woken up picks the last
received event from hdev->recv_evt (if it matches the right criteria)
and returns it.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
This patch adds tracking of received HCI events to the hci_dev struct.
This is necessary so that a subsequent patch can implement a function
for sending a single command synchronously and returning the resulting
command complete parameters in the function return value.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
This patch adds a NULL check for the HCI UART ldisc driver because some
of HCI UART drivers allow hci_uart_tty_receive function to be called
even though the HCI device hasn't been registered yet.
Signed-off-by: Chan-yeol Park <chanyeol.park@samsung.com>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
This patch adds a check HCI_UART_REGISTERED before reading UART data in
the HCI UART H4 driver. UART data could arrive when inside the
hci_uart_tty_ioctl function after calling test_and_set_bit for
HCI_UART_PROTO_SET but before the hci_uart_set_proto function has
returned.
Backtrace:
[<c05f27ec>] (hci_recv_stream_fragment+0x0/0x74) from [<c04126f4>] (h4_recv+0x18/0x40)
r7:eb1d4d1c r6:eb7683b0 r5:eae8e800 r4:0000000c
[<c04126dc>] (h4_recv+0x0/0x40) from [<c0411870>] (hci_uart_tty_receive+0x6c/0x94)
r5:eae8e800 r4:eb768380
[<c0411804>] (hci_uart_tty_receive+0x0/0x94) from [<c027be88>] (flush_to_ldisc+0x16c/0x17c)
r6:eae8e8d8 r5:eae8e800 r4:eae8e8c8
[<c027bd1c>] (flush_to_ldisc+0x0/0x17c) from [<c0050ae8>] (process_one_work+0x144/0x4d4)
[<c00509a4>] (process_one_work+0x0/0x4d4) from [<c0051208>] (worker_thread+0x180/0x370)
[<c0051088>] (worker_thread+0x0/0x370) from [<c005617c>] (kthread+0x90/0x9c)
[<c00560ec>] (kthread+0x0/0x9c) from [<c003a3a0>] (do_exit+0x0/0x7ec)
Signed-off-by: Chan-yeol Park <chanyeol.park@samsung.com>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
This patch removes the hci_req_cmd_status function since it is not
used anymore. The HCI request framework now considers the HCI command
has complete once the Command Status or Command Complete Event is
received.
Signed-off-by: Andre Guedes <andre.guedes@openbossa.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
David Herrmann
Claudio Takahasi
Marcel Holtmann
Johan Hedberg
Chan-yeol Park
Andre Guedes