l2ping only works for 10 seconds with 2.6.27

l2ping only works for 10 seconds with 2.6.27

[Nick Pelly]

Hi,

I have noticed with 2.6.27 that a bluez receiving side of l2ping will
detch the ACL connection (preventing further l2cap echos) after 10
seconds. HCI log below.

Is this by design, or was it unintentional?

I notice the OSX stack will detach after 60 seconds, so maybe it is by
design, after all the receiving side may not want to deal with these
unsolicited pings forever. In that case do you know the right place to
revert this behavior (l2ping is useful for testing).

Nick

2008-11-25 17:32:06.310607 > HCI Event: Connect Request (0x04) plen 10
    bdaddr 00:17:E8:2E:6B:E3 class 0x00020c type ACL
2008-11-25 17:32:06.310821 < HCI Command: Accept Connection Request
(0x01|0x0009) plen 7
    bdaddr 00:17:E8:2E:6B:E3 role 0x00
    Role: Master
2008-11-25 17:32:06.315826 > HCI Event: Command Status (0x0f) plen 4
    Accept Connection Request (0x01|0x0009) status 0x00 ncmd 1
2008-11-25 17:32:06.484832 > HCI Event: Role Change (0x12) plen 8
    status 0x00 bdaddr 00:17:E8:2E:6B:E3 role 0x00
    Role: Master
2008-11-25 17:32:06.517578 > HCI Event: Connect Complete (0x03) plen 11
    status 0x00 handle 1 bdaddr 00:17:E8:2E:6B:E3 type ACL encrypt 0x00
2008-11-25 17:32:06.517639 > HCI Event: Page Scan Repetition Mode
Change (0x20) plen 7
    bdaddr 00:17:E8:2E:6B:E3 mode 1
2008-11-25 17:32:06.517883 < HCI Command: Read Remote Supported
Features (0x01|0x001b) plen 2
    handle 1
2008-11-25 17:32:06.523284 > HCI Event: Command Status (0x0f) plen 4
    Read Remote Supported Features (0x01|0x001b) status 0x00 ncmd 1
2008-11-25 17:32:06.530517 > HCI Event: Max Slots Change (0x1b) plen 3
    handle 1 slots 5
2008-11-25 17:32:06.530548 > ACL data: handle 1 flags 0x02 dlen 10
    L2CAP(s): Info req: type 2
2008-11-25 17:32:06.530670 < ACL data: handle 1 flags 0x02 dlen 16
    L2CAP(s): Info rsp: type 2 result 0
      Extended feature mask 0x0000
2008-11-25 17:32:06.540283 < HCI Command: Remote Name Request
(0x01|0x0019) plen 10
    bdaddr 00:17:E8:2E:6B:E3 mode 2 clkoffset 0x0000
2008-11-25 17:32:06.545532 > HCI Event: Command Status (0x0f) plen 4
    Remote Name Request (0x01|0x0019) status 0x00 ncmd 1
2008-11-25 17:32:06.577911 > HCI Event: Read Remote Supported Features
(0x0b) plen 11
    status 0x00 handle 1
    Features: 0xff 0xff 0x2d 0xfe 0x9b 0xf9 0x00 0x80
2008-11-25 17:32:06.577941 > HCI Event: Remote Name Req Complete (0x07) plen 255
    status 0x00 bdaddr 00:17:E8:2E:6B:E3 name 'dream'
2008-11-25 17:32:06.577941 > HCI Event: Number of Completed Packets
(0x13) plen 5
    handle 1 packets 1
2008-11-25 17:32:06.610351 > ACL data: handle 1 flags 0x02 dlen 52
    L2CAP(s): Echo req: dlen 44
    0000: 41 42 43 44 45 46 47 48  49 4a 4b 4c 4d 4e 4f 50  ABCDEFGHIJKLMNOP
    0010: 51 52 53 54 55 56 57 58  59 5a 5b 5c 5d 5e 5f 60  QRSTUVWXYZ[\]^_`
    0020: 61 62 63 64 65 66 67 68  41 42 43 44              abcdefghABCD
2008-11-25 17:32:06.610504 < ACL data: handle 1 flags 0x02 dlen 52
    L2CAP(s): Echo rsp: dlen 44
    0000: 41 42 43 44 45 46 47 48  49 4a 4b 4c 4d 4e 4f 50  ABCDEFGHIJKLMNOP
    0010: 51 52 53 54 55 56 57 58  59 5a 5b 5c 5d 5e 5f 60  QRSTUVWXYZ[\]^_`
    0020: 61 62 63 64 65 66 67 68  41 42 43 44              abcdefghABCD
2008-11-25 17:32:06.622192 > HCI Event: Number of Completed Packets
(0x13) plen 5
    handle 1 packets 1

[ 10 seconds ]

2008-11-25 17:32:16.169464 > ACL data: handle 1 flags 0x02 dlen 52
    L2CAP(s): Echo req: dlen 44
    0000: 41 42 43 44 45 46 47 48  49 4a 4b 4c 4d 4e 4f 50  ABCDEFGHIJKLMNOP
    0010: 51 52 53 54 55 56 57 58  59 5a 5b 5c 5d 5e 5f 60  QRSTUVWXYZ[\]^_`
    0020: 61 62 63 64 65 66 67 68  41 42 43 44              abcdefghABCD
2008-11-25 17:32:16.169616 < ACL data: handle 1 flags 0x02 dlen 52
    L2CAP(s): Echo rsp: dlen 44
    0000: 41 42 43 44 45 46 47 48  49 4a 4b 4c 4d 4e 4f 50  ABCDEFGHIJKLMNOP
    0010: 51 52 53 54 55 56 57 58  59 5a 5b 5c 5d 5e 5f 60  QRSTUVWXYZ[\]^_`
    0020: 61 62 63 64 65 66 67 68  41 42 43 44              abcdefghABCD
2008-11-25 17:32:16.183380 > HCI Event: Number of Completed Packets
(0x13) plen 5
    handle 1 packets 1
2008-11-25 17:32:16.576141 < HCI Command: Disconnect (0x01|0x0006) plen 3
    handle 1 reason 0x13
    Reason: Remote User Terminated Connection
2008-11-25 17:32:16.589813 > HCI Event: Command Status (0x0f) plen 4
    Disconnect (0x01|0x0006) status 0x00 ncmd 1
2008-11-25 17:32:16.589843 > HCI Event: Disconn Complete (0x05) plen 4
    status 0x00 handle 1 reason 0x16
    Reason: Connection Terminated by Local Host

Re: l2ping only works for 10 seconds with 2.6.27

[Marcel Holtmann]

Hi Nick,

> I have noticed with 2.6.27 that a bluez receiving side of l2ping will
> detch the ACL connection (preventing further l2cap echos) after 10
> seconds. HCI log below.
>
> Is this by design, or was it unintentional?

it is not fully intentional, but it result due the fact on how we had  
to implement the Simple Pairing support. Problem here is that l2ping  
can not create a remote reference count of the connection. So BlueZ  
will clean up the ACL link. And we do have to do that since otherwise  
we waste power with an unused link.

So if you do l2test -P 1 -n <bdaddr> first and then l2ping you will  
see that the ACL link stays up since the l2test hold a remote  
reference count.

Regards

Marcel

Re: l2ping only works for 10 seconds with 2.6.27

[Nick Pelly]

On Tue, Nov 25, 2008 at 9:00 PM, Marcel Holtmann <marcel@holtmann.org> wrote:
> Hi Nick,
>
>> I have noticed with 2.6.27 that a bluez receiving side of l2ping will
>> detch the ACL connection (preventing further l2cap echos) after 10
>> seconds. HCI log below.
>>
>> Is this by design, or was it unintentional?
>
> it is not fully intentional, but it result due the fact on how we had to
> implement the Simple Pairing support. Problem here is that l2ping can not
> create a remote reference count of the connection. So BlueZ will clean up
> the ACL link. And we do have to do that since otherwise we waste power with
> an unused link.

Yeah I think it is an ok policy, like you say its bad for power to
maintain to link with no local client using it.

> So if you do l2test -P 1 -n <bdaddr> first and then l2ping you will see that
> the ACL link stays up since the l2test hold a remote reference count.

Ok, that'll do for a workaround for testing.

By the way, what happened to the HCI_UART_DEBUG patch I sent a few days ago?

Thanks,

Nick