- 13 Feb, 2012 40 commits
-
-
David Herrmann authored
Recently we changed the way how we release driver-data. A driver is now responsible of destroying the data so we no longer need the device lock on deinitialization in btusb driver. Signed-off-by:
David Herrmann <dh.herrmann@googlemail.com> Acked-by:
Marcel Holtmann <marcel@holtmann.org> Signed-off-by:
Johan Hedberg <johan.hedberg@intel.com>
-
David Herrmann authored
This enhances code readability a lot and avoids using void* even though we know the type of the variable. Signed-off-by:
-
David Herrmann authored
This avoids using the dev_set/get_drvdata() functions to retrieve a pointer to our own structure. We can use simple pointer arithmetic here. The drvdata field is actually not needed by any other code-path but this makes the code more consistent with hci_dev. Signed-off-by:
-
David Herrmann authored
The linux device model provides dev_set/get_drvdata so we can use this to save private driver data. This also removes several unnecessary casts. Signed-off-by:
-
David Herrmann authored
We currently use dev_set_drvdata to keep a pointer to ourself. This doesn't make sense as we are the bus and not a driver. Therefore, introduce to_hci_dev() so we can get a struct hci_dev pointer from a struct device pointer. dev_set/get_drvdata() is reserved for drivers that provide a device and not for the bus using the device. The bus can use simple pointer arithmetic to retrieve its private data. Signed-off-by:
-
Johan Hedberg authored
This patch add a new Device Unpaired mgmt event. This will be sent to all mgmt sockets except the one that requested unpairing (that socket will get a command complete instead). The event is also reserved for future SMP updates where a remote device will be able to request pairing revocation from us. Signed-off-by:
-
Johan Hedberg authored
The default response status to unpair_device should be set as 0 instead of a generic failure value. When disconnection is not needed (i.e. we can reply imediately) we should return success and not failure. Signed-off-by:
-
Johan Hedberg authored
This patch updates the Authentication Failed mgmt event to match the latest API specification by adding an address type to it. Signed-off-by:
-
Johan Hedberg authored
This patch updates the implmentation for mgmt_block_device and mgmt_unblock_device and their corresponding events to match the latest API specification. Signed-off-by:
-
Johan Hedberg authored
This patch updates the implementation for these mgmt to be up to date with the latest API specification. Right now the address type isn't actually used for anything but that might change in the future. Signed-off-by:
-
Johan Hedberg authored
This patch upadate the user confirm and user passkey mgmt messages to match the latest API specification by adding an address type parameter to them. Signed-off-by:
-
Johan Hedberg authored
This patch adds an address type parameter to the disconnect command and response in order to match the latest mgmt API specification. Signed-off-by:
-
Johan Hedberg authored
This patch renames the mgmt_remove_keys command to mgmt_unpair_device and updates its parameters to match the latest API (specifically, it adds an address type parameter to the command and its response). Signed-off-by:
-
Szymon Janc authored
Pending I-Frame(s) are considered as acknowledgement. To void double acking (via I-Frame and later via RR) clear ack timer when sending first pending I-Frame. Signed-off-by:
Szymon Janc <szymon.janc@tieto.com> Signed-off-by:
Luiz Augusto von Dentz <luiz.von.dentz@intel.com> Acked-by:
-
Szymon Janc authored
Make l2cap_ertm_send return number of pending I-Frames transmitted instead of all (pending + retransmitted) I-Frames transmitted. As only pending I-Frames are considered as acknowledgement, this could lead to situation when no ACK was sent in __l2cap_send_ack (if only already transmitted I-Frames were retransmitted). Signed-off-by:
Gustavo F. Padovan <padovan@profusion.mobi> Acked-by:
-
Andrei Emeltchenko authored
In debug use chan %p instead of sk. Signed-off-by:
Andrei Emeltchenko <andrei.emeltchenko@intel.com> Acked-by:
-
Andrei Emeltchenko authored
Change elseif to switch. This make sense even more with following patches which otherwise have to add more elseifs statements. Signed-off-by:
-
Andrei Emeltchenko authored
Change is needed to remove dependency on sk when possible before introducing l2cap channel lock. Signed-off-by:
-
Andre Guedes authored
This is helpful for device discovery implementation & debuging. Signed-off-by:
Andre Guedes <andre.guedes@openbossa.org> Acked-by:
-
Johan Hedberg authored
The correct length of the event is the size of the ev struct (not size of the pointer like the code was previously using) plus the length of the variable-sized EIR data at the end of the struct. Signed-off-by:
-
Andre Guedes authored
This patch fixes a #define indentation in mgmt.c. Signed-off-by:
-
Andre Guedes authored
This patch adds LE-Only discovery procedure support to MGMT Start Discovery command. Signed-off-by:
-
Andre Guedes authored
We are not supposed to block in start_discovery() because start_discovery code is running in write() syscall context and this would block the write operation on the mgmt socket. This way, we cannot directly call hci_do_le_scan() to scan LE devices in start_discovery(). To overcome this issue a derefered work (hdev->le_scan) was created so we can properly call hci_do_le_scan(). The helper function hci_le_scan() simply set LE scan parameters and queue hdev->le_scan work. The work is queued on system_long_wq since it can sleep for a few seconds in the worst case (timeout). Signed-off-by:
-
Andre Guedes authored
This patch adds to hci_core the hci_do_le_scan function which should be used to scan LE devices. In order to enable LE scan, hci_do_le_scan() sends commands (Set LE Scan Parameters and Set LE Scan Enable) to the controller and waits for its results. If commands were executed successfully a delayed work is scheduled to disable the ongoing scanning after some amount of time. This function blocks. Signed-off-by:
-
Andre Guedes authored
This patch does a trivial code refacting in hci_discovery_active. Signed-off-by:
-
Andre Guedes authored
Send MGMT Discovering events once LE scan starts/stops so the userspace can track when local adapters are discovering LE devices. This way, we also keep the same behavior of inquiry which sends MGMT Discovering events once inquiry starts/stops even if it is triggered by an external tool (e.g. hcitool). Signed-off-by:
-
Johan Hedberg authored
We should only perform a reset in hci_dev_do_close if the HCI_QUIRK_NO_RESET flag is set (since in such a case a reset will not be performed when initializing the device). Signed-off-by:
-
Andrei Emeltchenko authored
Use __check_timout helper to remove duplicated code Signed-off-by:
Ulisses Furquim <ulisses@profusion.mobi> Acked-by:
-
Andrei Emeltchenko authored
Split HCI scheduling for block and packet flow control. Signed-off-by:
-
Vinicius Costa Gomes authored
If we want to have proper pairing support over LE we need to inform userspace that a new LTK is available, so userspace can store that key permanently. Signed-off-by:
Vinicius Costa Gomes <vinicius.gomes@openbossa.org> Acked-by:
-
Vinicius Costa Gomes authored
With the use of the new structures and lists for the SMP LTK's we may remove some code that is now unused. No need to have extra fields of information inside link_key now that it is only used for Link Keys. Signed-off-by:
-
Vinicius Costa Gomes authored
Instead of having a separated command for removing SMP keys, we use the Remove Keys command to remove *all* keys. Signed-off-by:
-
Vinicius Costa Gomes authored
This adds a method to notify that a new LTK is available and a handler to store keys coming from userspace into the kernel LTK list. Signed-off-by:
-
Vinicius Costa Gomes authored
This updates all the users of the older way, that was using the link_keys list to store the SMP keys, to use the new way. This includes defining new types for the keys, we have a type for each combination of STK/LTK and Master/Slave. Signed-off-by:
-
Vinicius Costa Gomes authored
This includes a new list for storing the keys and a new structure used to represent each key. Some notes: authenticated is used to identify that the key may be used to setup a HIGH security link. As the same list is used to store both the STK's and the LTK's the type field is used so we can separate between those two types of keys and if the key should be used when in the master or slave role. Signed-off-by:
-
Vinicius Costa Gomes authored
Every command handler of mgmt does a cast to the command structure so it can properly interpreted. So we can avoid that cast if we make those functions receive a void * directly. Signed-off-by:
-
Manoj Iyer authored
T: Bus=01 Lev=02 Prnt=02 Port=03 Cnt=03 Dev#= 5 Spd=12 MxCh= 0 D: Ver= 2.00 Cls=ff(vend.) Sub=01 Prot=01 MxPS=64 #Cfgs= 1 P: Vendor=0a5c ProdID=21f3 Rev=01.12 S: Manufacturer=Broadcom Corp S: Product=BCM20702A0 S: SerialNumber=74DE2B344A7B C: #Ifs= 4 Cfg#= 1 Atr=e0 MxPwr=0mA I: If#= 0 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=01 Prot=01 Driver=(none) I: If#= 1 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=01 Prot=01 Driver=(none) I: If#= 2 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=(none) I: If#= 3 Alt= 0 #EPs= 0 Cls=fe(app. ) Sub=01 Prot=01 Driver=(none) Signed-off-by:
Manoj Iyer <manoj.iyer@canonical.com> Tested-by:
Dennis Chua <dennis.chua@canonical.com> Acked-by:
-
Andrei Emeltchenko authored
Fixes possible bug when deleting element from the list in function hci_chan_list_flush. list_for_each_entry_rcu is used and after deleting element from the list we also free pointer and then list_entry_rcu is taken from freed pointer. Signed-off-by:
-
Andrei Emeltchenko authored
Use list_for_each_entry_safe which is safe version against removal of list entry. Otherwise we remove hci_conn element and reference next element which result in accessing LIST_POISON. [ 95.571834] Bluetooth: unknown link type 127 [ 95.578349] BUG: unable to handle kernel paging request at 20002000 [ 95.580236] IP: [<20002000>] 0x20001fff [ 95.580763] *pde = 00000000 [ 95.581196] Oops: 0000 [#1] SMP ... [ 95.582298] Pid: 3355, comm: hciconfig Tainted: G O 3.2.0-VirttualBox [ 95.582298] EIP: 0060:[<20002000>] EFLAGS: 00210206 CPU: 0 [ 95.582298] EIP is at 0x20002000 ... [ 95.582298] Call Trace: [ 95.582298] [<f8231ab6>] ? hci_conn_hash_flush+0x76/0xf0 [bluetooth] [ 95.582298] [<f822bcb1>] hci_dev_do_close+0xc1/0x2e0 [bluetooth] [ 95.582298] [<f822d679>] ? hci_dev_get+0x69/0xb0 [bluetooth] [ 95.582298] [<f822e1da>] hci_dev_close+0x2a/0x50 [bluetooth] [ 95.582298] [<f824102f>] hci_sock_ioctl+0x1af/0x3f0 [bluetooth] [ 95.582298] [<c11153ea>] ? handle_pte_fault+0x8a/0x8f0 [ 95.582298] [<c146becf>] sock_ioctl+0x5f/0x260 [ 95.582298] [<c146be70>] ? sock_fasync+0x90/0x90 [ 95.582298] [<c1152b33>] do_vfs_ioctl+0x83/0x5b0 [ 95.582298] [<c1563f87>] ? do_page_fault+0x297/0x500 [ 95.582298] [<c1563cf0>] ? spurious_fault+0xd0/0xd0 [ 95.582298] [<c107165b>] ? up_read+0x1b/0x30 [ 95.582298] [<c1563f87>] ? do_page_fault+0x297/0x500 [ 95.582298] [<c100aa9f>] ? init_fpu+0xef/0x160 [ 95.582298] [<c15617c0>] ? do_debug+0x180/0x180 [ 95.582298] [<c100a958>] ? fpu_finit+0x28/0x80 [ 95.582298] [<c11530e7>] sys_ioctl+0x87/0x90 [ 95.582298] [<c156795f>] sysenter_do_call+0x12/0x38 ... Signed-off-by:
-
Johan Hedberg authored
This patch implements the Cancel Pair Device command for mgmt. It's used by user space to cancel an ongoing pairing attempt which was triggered by the Pair Device command. Signed-off-by:
-
