2006-05-30 14:04:48

by Jens Hottenroth

[permalink] [raw]
Subject: Re: [Bluez-users] Bluetooth Instability Problem extended info

Hello Marcel,

thanks for beeing interested in helping me.


I send you the requested information (and a bit more):

The bbt is running Kernel 2.4.25-mh5, gcc 3.3.3 with denx elkd version 3.1
The PC is running suse 9.3 professional with kernel
2.6.11.4-21.11-default and gcc 3.3.5 (3.3.3 when cross compiling for PPC)

I tested some more hardware, the D-Link DBT-120 is revision B4 (got two
of these), an Acer BT-600, and a Typhoon 20006. Everything works fine
between two PC's, when used in bbt it still hangs.

One of my co-workers is suspecting the USB-Host chip, a timing or data
size problem inside the bbt, so i'll try to check that out, too.


The commands I use are very simple,

l2ping BD_ADDR -s <size>

with <size> = {0, 20, 40, 100, 200, 300, 400, 500}


I have some debugging info from the kernel (a small part
while l2ping is active):

__recv_frame: hci0 type 4 data c3d714e0 count 7
__recv_frame: new packet len 7
hci_recv_frame: hci0 type 4 len 7
hci_num_comp_pkts_evt: hci0 num_hndl 1
hci_send_to_sock: hdev c0325a00 len 7
hci_event_packet: hci0 evt 0x13
hci_num_comp_pkts_evt: hci0 num_hndl 1
hci_send_to_sock: hdev c0325a00 len 7
hci_event_packet: hci0 evt 0x13
hci_num_comp_pkts_evt: hci0 num_hndl 1
hci_tx_task: hci0 acl 8 sco 8
hci_sched_acl: hci0
hci_low_sent: conn 00000000 quote 0
hci_sched_sco: hci0
hci_low_sent: conn 00000000 quote 0
hci_rx_task: hci0

Till here it looks normal to me, but what's that status -84?

hci_usb_rx_complete: hci0 urb c0301f34 type 2 status -84 count 192 flags 10
hci_usb_rx_complete: hci0 urb c0301f34 type 2 resubmit status 0
hci_usb_rx_complete: hci0 urb c0301f34 type 2 status 0 count 9 flags 10
__recv_frame: hci0 type 2 data c3fda800 count 9
__recv_frame: new packet len 25446
hci_usb_rx_complete: hci0 urb c0301f34 type 2 resubmit status 0
hci_usb_rx_complete: hci0 urb c0301f34 type 2 status 0 count 343 flags 10
__recv_frame: hci0 type 2 data c3fda800 count 343
hci_usb_rx_complete: hci0 urb c0301f34 type 2 resubmit status 0
hci_usb_rx_complete: hci0 urb c0301f34 type 2 status 0 count 73 flags 10
__recv_frame: hci0 type 2 data c3fda800 count 73
hci_usb_rx_complete: hci0 urb c0301f34 type 2 resubmit status 0
hci_usb_rx_complete: hci0 urb c0301f34 type 2 status 0 count 343 flags 10

And that's all he's doing from now on till reset. It' allways
shows that status -84, but sometimes (depending on <size>) it
startsretransmitting normal, sometimes it breaks down.



The hcidump tool made the following trace (not matching the above kernel
debug):

It's a bit bigger, so i made a link. The command was

l2ping -s 500 BD_ADDR

ho//tten//roth.info/files/hcidump.log (without that extra //)


To be a bit more precise, i can't update the Kernel of the bbt to 2.6
without problems.


Thanks for all input, comments and help!

Best regards from Germany

Jens

> Date: Mon, 29 May 2006 15:25:01 +0200
> From: Marcel Holtmann <[email protected]>
> Subject: Re: [Bluez-users] Bluetooth instability problem
> To: [email protected]
> Message-ID: <1148909101.31689.48.camel@localhost>
> Content-Type: text/plain
>
> Hi Jens,
>
>> I don't know if i'm at the right spot here, but i'll ask here first. I
>> have the following problem:
>>
>> I've 3 "PCs" (one is an embedded linux based broadband handheld tester,
>> afterwards refered to as bbt), all running linux with Kernels from 2.24
>> (bbt) to 2.26 (PCs), and identical bluetooth usb-sticks (D-Link BT120).
>>
>> Here's what i've done / what i know:
>>
>> Downloaded the latest bluez-libs and bluez-utils, compiled and
>> installed. Ran configure / make for a new kernel with only that
>> functions needed, installed it to the bbt.
>>
>> Running some tests with l2ping brought the following results. I can ping
>> the bbt, sometimes i get no response messages (max. 5 after another,
>> most times 4 in a row). This test runs for about 30 minutes, then i
>> canceled it. The bigger I define paketsize (defaults 44 Bytes) the
>> faster / more often come those no response messages and, here it comes!,
>> when more than 250 Bytes are specified, the connection hangs! Aprox. 2
>> minutes. The bigger the paketsize the faster it hangs.
>> After it once hung, only a total reset of the bbt gets it running again.
>>
>> Between those 2 PCs, I only had problems with rfcomm and dun, pan worked
>> just fine. After i updated the Systems (it's SuSe linux 9.1 and 9.3, i
>> forgot to say that) those problems where gone.
>>
>> The biggest Problem is, that i can't path or upgrade the kernel in the bbt.
>
> what kernel versions are you using. You mentioned onyl the bluez-utils
> versions. Please specify the exact commands you are using for testing
> and include the output of "hcidump -X -V".
>
> Regards
>
> Marcel




_______________________________________________
Bluez-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/bluez-users


2006-05-31 17:03:49

by Paulo Wollny

[permalink] [raw]
Subject: [Bluez-users] bluetooth ppd for dummies

Hi,

i used to connect my computer to the internet (GPRS) using bluez with
wvdial, under suse 9.0.
I upgraded to suse 10.0, and since then, wvdial is not working anymore.
does anyone have all the pppd configuration whuch could send to me as an
example, with all the steps?

Best regards,

Paulo Wollny



_______________________________________________
Bluez-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/bluez-users

2006-05-30 14:24:39

by Marcel Holtmann

[permalink] [raw]
Subject: Re: [Bluez-users] Bluetooth Instability Problem extended info

Hi Jens,

> I send you the requested information (and a bit more):
>
> The bbt is running Kernel 2.4.25-mh5, gcc 3.3.3 with denx elkd version 3.1
> The PC is running suse 9.3 professional with kernel
> 2.6.11.4-21.11-default and gcc 3.3.5 (3.3.3 when cross compiling for PPC)
>
> I tested some more hardware, the D-Link DBT-120 is revision B4 (got two
> of these), an Acer BT-600, and a Typhoon 20006. Everything works fine
> between two PC's, when used in bbt it still hangs.
>
> One of my co-workers is suspecting the USB-Host chip, a timing or data
> size problem inside the bbt, so i'll try to check that out, too.

and he might be right with that. The USB stack in a 2.4 kernel is
actually not good. You should really consider moving to a recent 2.6
kernel for your btt.

Regards

Marcel




_______________________________________________
Bluez-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/bluez-users