Comment 1 Johan Hedberg 2010-01-19 09:14:58 UTC

We'll need the bluetoothd debug logs and an hcidump to investigate this
further.

(In reply to comment #1)
> We'll need the bluetoothd debug logs and an hcidump to investigate this
> further.
> 

I would like to do that. Can you show me how to pull these logs from my device?

Comment 3 Johan Hedberg 2010-01-20 10:29:04 UTC

Some general info about the packages is available here:
http://wiki.maemo.org/Documentation/devtools/maemo5/syslog

The packages themselves are available from repository.maemo.org:
http://repository.maemo.org/pool/maemo5.0/free/

syslogd & klogd:
http://repository.maemo.org/pool/maemo5.0/free/s/sysklogd/

bluez-hcidump:
http://repository.maemo.org/pool/maemo5.0/free/b/bluez-hcidump/

You'll need to download and install (from the command line as root with dpkg -i
*.deb) the armel .deb files from those URL's.

In case you don't know how to get root privileges consult this page:
http://wiki.maemo.org/Root_access

Once you've installed the packages and rebooted your device, enable bluetoothd
debug logs with "killall -USR2 bluetoothd" (as root), perform the test and
gather any produced logs with "grep bluetoothd /var/log/syslog >
bluetothd-logs.txt". For HCI logs run "hcidump -XV" (also as root) while
performing the trace. If the HCI trace gets very long you can also create a
binary trace to a file with "hcidump -w hcidump.log -B"

How can I download these two files?
I opened the provided links from the browser but an error occurred while
processing the request.

Comment 5 Andre Klapper 2010-01-23 22:22:01 UTC

That's better answered in talk.maemo.org - please post your exact problems
there.

Created an attachment (id=2123) [details]
Logs that show the problem

It's happening for me also in the bluetooth that comes with the 330cd.

I'm attaching a tar.gz with the following information:

hci.txt: tried to pair devices -> FAILED
hci2.txt: second attempt to pair devices -> FAILED
hci3.txt: third attempt to pair devices -> WORKED but...

when tried to read the contacts it was not able to do that. So in my case most
of times I cannot connect, and when I am able to connect the addressbook cannot
be read from N900.

I'm also attaching a log from bluetoothd and a file called "time.txt" that
specifies the time when each test was run.

I have the same problem with my BMW 520D 2006 (dec).

Bluetooth connection works fine in headset mode. So I can call and the car can
initiate calls (by using buttons on steering wheel e.g. voice controlled
choosing of numbers stored in the car memory and by manually entering numbers
using the BMW information centre). However the contacts list from the N900 is
not available.

Some of the comments posted here are referring to this bug. Maybe a duplicate?
https://bugs.maemo.org/show_bug.cgi?id=6368

(In reply to comment #8)
> Some of the comments posted here are referring to this bug. Maybe a duplicate?
> https://bugs.maemo.org/show_bug.cgi?id=6368
> 

It's not a duplicate because the other bug is exclusively about contacts
synchronization. What we are stressing in this bug is that we cannot even
*connect* the device to the bluetooth set which is even worst.

Comment 10 Andre Klapper 2010-03-01 19:52:31 UTC

Making this Sergio's report as original reporter does not answer.
Changing summary.

(In reply to comment #10)
> Making this Sergio's report as original reporter does not answer.
> Changing summary.
> 

Do you know if is there anyone taking care of this Andre?

Comment 12 ext-daniel.orstadius 2010-03-03 12:07:11 UTC

From the logs it looks like N900 is already paired with the carkit, and on the
failed connection attempts the HFP service record seems to be hidden.

The first thing happening on the logs is that the carkit tries to create a
connection. Is a connection established if you don't manually try to connect
from N900?

*** This bug has been confirmed by popular vote. ***

(In reply to comment #12)
> From the logs it looks like N900 is already paired with the carkit, and on the
> failed connection attempts the HFP service record seems to be hidden.
> 
> The first thing happening on the logs is that the carkit tries to create a
> connection. Is a connection established if you don't manually try to connect
> from N900?

Not normally. I just observed it a couple of times during the past months

Created an attachment (id=2426) [details]
More logs showing the problem on another car

I have a similar problem on a 2004 BMW 300ci.

The vehicle and phone are paired fine, autoconnect very rerely works (maybe 1
time in 20).

After autoconnect fails and I then try to manually connect I receive a message
to say that connection has failed.

If I then switch the bluetooth off and on again on the phone while leaving the
carrunning and try to connect manually again the connection is successful and
the headphone icon appears at the top of the screen.

Log attached showning all 3 scenarios for both hcidump and bluetoothd:
log1: Vehicle attempts auto connect, fails.
Log2: Vehicle attempts auto connect, fails, user attempts manual connect,
fails.
Log3: Immediately after turning bluetooth off and on again on phone, vehicle
attempts auto connect, fails, user attempts manual connect, succedes

Thanks

I forgot to say I am using phone firmware version 2.2009.51-1.205.1

Comment 17 Andre Klapper 2010-03-12 15:39:40 UTC

Looks like N900 doesn't answer the SDP connection request from the carkit.

(In reply to comment #15)
I have exactly the same behaviour on my BMW X5 E53 2005, which has BMW option
Z6A Comms pack including 633 Bluetooth phone prep + telematics. I can pair and
use the N900 (10.2010.19-1.203.1 UK) once, then it will not re-connect once the
connection has been broken.
Happy to get ksyslogd logs if this helps pursue the matter, I can also get logs
from other BMWs too if that is useful - just let me know.

Comment 19 Andre Klapper 2010-06-02 13:40:20 UTC

Stuart: Logs are always welcome, but please file a separate report as this
might be different software used by the different car. (Always better to later
mark as duplicate instead of having several issues in one report.)

(In reply to comment #17)
> Looks like N900 doesn't answer the SDP connection request from the carkit.
> 

Ok, this comment was made in march. There is something obviously wrong with the
bluetooth software stack on the N900. Bug #7796 is basically the same issue. So
the N900 isn't answering the SDP connection request? WHO IS GOING TO FIX IT?

This is getting to be a joke, two bugs both full of he logs that have been
requested. Confirmed by many people and still no developer assigned and no one
bothering to comment on any progress being made.

Will this be fix be released in the meego that we wont be getting? Or will we
need to wait yet more months to make a phone call from our cars?

Again, this is basic functionality expected from any bluetooth enabled phone
that DOES NOT WORK on the N900. When can we expect it to be resolved because so
far this has just been a joke. At least assign it to someone...

If I took this long to resolve a bug I'd be out of a job, let alone to assign
the task to be looked at. Im sure Nokia can spare some BMWs to the lucky
developer who takes a look into this. I hear you can rent them for a fair
price.

Comment 21 Johan Hedberg 2010-06-13 11:45:21 UTC

(In reply to comment #20)
> (In reply to comment #17)
> > Looks like N900 doesn't answer the SDP connection request from the carkit.
> > 
> 
> Ok, this comment was made in march.

Yes, by our bugmaster Andre who doesn't necessarily have enough knowledge to
make this conclusion. In the first log of the second attachment this is the
case, however in all other logs (the 4 or 5 other ones of the second attachment
and all in the first attachment) the issue seems different.

>There is something obviously wrong with the
> bluetooth software stack on the N900.

Actually it looks like more of a hardware or bluetooth chip firmware issue on
the N900 side, or then a bug at the car kit side in sniff mode handling. The
following is seen in almost all logs:

< HCI Command: Sniff Mode (0x02|0x0003) plen 10
    handle 11 max 800 min 80 attempt 4 timeout 1
> HCI Event: Command Status (0x0f) plen 4
    Sniff Mode (0x02|0x0003) status 0x00 ncmd 1
> HCI Event: Max Slots Change (0x1b) plen 3
    handle 11 slots 1
> HCI Event: Max Slots Change (0x1b) plen 3
    handle 11 slots 5
> HCI Event: Mode Change (0x14) plen 6
    status 0x24 handle 11 mode 0x00 interval 0
    Error: LMP PDU Not Allowed

I.e. the N900 is trying to enter sniff mode while waiting for a response from
the car kits SDP server. At the Bluetooth UnplugFest from which I just returned
we saw a few similar issues with sniff mode (not with N900 HW though). One way
to confirm whether this is a sniff mode issue would be to disable it from the
N900 side. It should be possible to do it with the following command (as root):
echo 0 > /sys/class/bluetooth/hci0/idle_timeout

If that doesn't help (you'll know that it has succeeded when you don't see the
N900 sending the sniff mode command anymore) the following might also work
(again as root):
hciconfig hci0 lp RSWITCH

> Bug #7796 is basically the same issue. So
> the N900 isn't answering the SDP connection request?

No, in all except one of the logs the issue is the car kit not responding to
the N900's SDP request (most likely because the N900 requests sniff mode before
the car kit responds).

Comment 22 Andre Klapper 2010-06-13 19:11:34 UTC

Internal comment says that there are two issues here. For the N900, "L2CAP
packet is received before we got Connection Complete with connection handle".
For BMW, "BMW carkit replies with malformed SDP response."