Re: Wrong timeout when connecting to device with removed pairing

["Gustavo F. Padovan" <padovan@profusion.mobi>]

Hi Luiz,

* Luiz Augusto von Dentz <luiz.dentz@gmail.com> [2011-04-12 14:36:01 +0300]:

> Apparently this is a regression caused by commit
> 769be974d0c7b4fe1a52f9cdaad22259b60953f7 - [Bluetooth] Use ACL config
> stage to retrieve remote features, when receiving pincode request in a
> new connection the state will be BT_CONFIG so HCI_PAIRING_TIMEOUT is
> never set to the connection, this could be reproduced by removing a
> link key in the remote side and trying to connect:
> 
> 2011-04-05 12:11:37.070953 < HCI Command: Authentication Requested
> (0x01|0x0011) plen 2
>     handle 1
> 2011-04-05 12:11:37.071686 > HCI Event: Command Status (0x0f) plen 4
>     Authentication Requested (0x01|0x0011) status 0x00 ncmd 1
> 2011-04-05 12:11:37.071838 > HCI Event: Link Key Request (0x17) plen 6
>     bdaddr 00:1F:E1:EB:B9:CB
> 2011-04-05 12:11:37.075073 < HCI Command: Link Key Request Reply
> (0x01|0x000b) plen 22
>     bdaddr 00:1F:E1:EB:B9:CB key 92BA9EAD06542B0600457C609B8F5395
> 2011-04-05 12:11:37.075775 > HCI Event: Command Complete (0x0e) plen 10
>     Link Key Request Reply (0x01|0x000b) ncmd 1
>     status 0x00 bdaddr 00:1F:E1:EB:B9:CB
> 2011-04-05 12:11:37.087005 > HCI Event: PIN Code Request (0x16) plen 6
>     bdaddr 00:1F:E1:EB:B9:CB
> 2011-04-05 12:11:39.038849 < ACL data: handle 1 flags 0x00 dlen 12
>     L2CAP(s): Disconn req: dcid 0x00a9 scid 0x0040
> 2011-04-05 12:11:39.100647 > HCI Event: Number of Completed Packets
> (0x13) plen 5
>     handle 1 packets 1
> 2011-04-05 12:11:39.100830 > ACL data: handle 1 flags 0x02 dlen 12
>     L2CAP(s): Disconn rsp: dcid 0x00a9 scid 0x0040
> 2011-04-05 12:11:40.433624 > HCI Event: Auth Complete (0x06) plen 3
>     status 0x13 handle 1
>     Error: Remote User Terminated Connection
> 2011-04-05 12:11:40.437774 > HCI Event: Disconn Complete (0x05) plen 4
>     status 0x00 handle 1 reason 0x13
>     Reason: Remote User Terminated Connection
> 
> I think the following patch might fix it:
> 
> diff --git a/net/bluetooth/hci_event.c b/net/bluetooth/hci_event.c
> index 3fbfa50..abdc4de 100644
> --- a/net/bluetooth/hci_event.c
> +++ b/net/bluetooth/hci_event.c
> @@ -1973,10 +1973,14 @@ static inline void hci_pin_code_request_evt(struct
> hci_dev *hdev, struct sk_buff
>         hci_dev_lock(hdev);
> 
>         conn = hci_conn_hash_lookup_ba(hdev, ACL_LINK, &ev->bdaddr);
> -       if (conn && conn->state == BT_CONNECTED) {
> -               hci_conn_hold(conn);
> +       if (conn) {
>                 conn->disc_timeout = HCI_PAIRING_TIMEOUT;
> -               hci_conn_put(conn);
> +
> +               /* Update disconnect timer */
> +               if (conn->state == BT_CONNECTED) {
> +                       hci_conn_hold(conn);
> +                       hci_conn_put(conn);

I think these two line can be:

		mod_timer(&conn->disc_timer, jiffies + HCI_PAIRING_TIMEOUT);


The two main difference are that here we don't delete the idle_timer, but it
isn't used anywhere, so I think we can delete it. Also we don't double the
timeout value when !conn->out

from hci_conn_put():

                timeo = msecs_to_jiffies(conn->disc_timeout);
		if (!conn->out)
			timeo *= 2; 

I still don't know why we double it.

-- 
Gustavo F. Padovan
http://profusion.mobi