[PATCH] SSP double paring fix: init auth_type with HCI_AT_GENERAL_BONDING

[Andrei Emeltchenko <andrei.emeltchenko.news@gmail.com>]

Hi,
I have a simple patch which fixes issues with SSP mentioned below.

---
 net/bluetooth/hci_conn.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/net/bluetooth/hci_conn.c b/net/bluetooth/hci_conn.c
index fa47d5d..d863bda 100644
--- a/net/bluetooth/hci_conn.c
+++ b/net/bluetooth/hci_conn.c
@@ -211,6 +211,7 @@ struct hci_conn *hci_conn_add(struct hci_dev
*hdev, int type, bdaddr_t *dst)
 	conn->type  = type;
 	conn->mode  = HCI_CM_ACTIVE;
 	conn->state = BT_OPEN;
+	conn->auth_type = HCI_AT_GENERAL_BONDING;

 	conn->power_save = 1;
 	conn->disc_timeout = HCI_DISCONN_TIMEOUT;
-- 
1.6.0.4

Regards,
Andrei

On Wed, Jun 17, 2009 at 3:03 PM, Andrei
Emeltchenko<andrei.emeltchenko.news@gmail.com> wrote:
> Hi,
>
> We have error with SSP paring. Originally we could not connect some
> devices and the problem was reproduced on my PC with 2 USB Bluetooth
> adapters.
>
> bluez asks user paring confirmation 2 times.
>
> Test:
>
> hci1:   Type: USB
>        BD Address: 00:02:72:16:4E:D5 ACL MTU: 1021:8 SCO MTU: 64:1
>        UP RUNNING PSCAN
>        RX bytes:12122 acl:78 sco:0 events:449 errors:0
>        TX bytes:22547 acl:82 sco:0 commands:294 errors:0
>
> hci2:   Type: USB
>        BD Address: 00:02:72:16:19:39 ACL MTU: 1021:8 SCO MTU: 64:1
>        UP RUNNING PSCAN
>        RX bytes:12593 acl:82 sco:0 events:475 errors:0
>        TX bytes:28404 acl:78 sco:0 commands:329 errors:0
>
>
> hci1:
> sudo ./l2test -i hci1 -r -E -S   # server
>
> hci2:
> sudo ./l2test -i hci2 -n 00:02:72:16:4E:D5 -b 10 -N 1 -E -S -s   # client
>
>  ./simple-agent hci1
> Agent registered
> RequestConfirmation (/org/bluez/6255/hci1/dev_00_02_72_16_19_39, 533464)
> Confirm passkey (yes/no): yes
> RequestConfirmation (/org/bluez/6255/hci1/dev_00_02_72_16_19_39, 152200)
> Confirm passkey (yes/no): yes
>
>
> ./simple-agent hci2
> Agent registered
> RequestConfirmation (/org/bluez/6255/hci2/dev_00_02_72_16_4E_D5, 533464)
> Confirm passkey (yes/no): yes
> RequestConfirmation (/org/bluez/6255/hci2/dev_00_02_72_16_4E_D5, 152200)
> Confirm passkey (yes/no): yes
>
> For the single connection passkey is asked 2 times.
>
> The reason looks to be because bluetoothd does not store link keys
> since ioctl  HCIGETAUTHINFO returns 0.
> Kernel in function hci_get_auth_info returns "conn->auth_type" which
> is 0 at this stage. As I understood
> "auth_type" initialized later when "l2cap_connect_req" comes in
> "l2cap_check_security".
>
> What is the best way to solv this? In kernel or in user-space? Is it an error?
>
> Regards,
> Andrei
>