CG2900: constant stream of error messages

HI,
Pease provide the TESTing enviroment used, Is the test is done in Ubuntu/Andriod?. What is the baseline used?.

We need to know Testing environment used, Is the test is done in Ubuntu/Andriod?. What is the baseline ( kernel ) used?.

The issue was observed with the 2012.03.01 Ubuntu image.

The kernel in there matches http://www.igloocommunity.org/gitweb/?p=kernel/igloo-kernel.git;a=commit;h=ceed3b0

           This problem is happening when the board is booted up for the first time after flashing, and I was able to reproduce the issue only once and rest of the times it is booting up properly (first boot after flashing, tried 4 times) and in the working case both lbsd and lbsappd are running. I other case I couldn’t make out whether it is running or not because after booting this message is repeating. [ 16.873229] cg2900-uart cg2900-uart.0: Could not find corresponding user to h4_channel 9

Attached both logs

Log with issue :

[ 2.186492] AppArmo
* Starting bluetooth [ 16.673217] cg2900-uart cg2900-uart.0: Could not find corresponding user to h4_channel 9 [ OK ]
saned disabled; edit /etc/default/saned
[ 16.873229] cg2900-uart cg2900-uart.0: Could not find corresponding user to h4_channel 9
[ 17.073272] cg2900-uart cg2900-uart.0: Could not find corresponding user to h4_channel 9
[ 17.275970] cg2900-uart cg2900-uart.0: Could not find corresponding user to h4_channel 9
[ 17.473236] cg2900-uart cg2900-uart.0: Could not find corresponding user to h4_channel 9
[ 17.673248] cg2900-uart cg2900-uart.0: Could not find corresponding user to h4_channel 9
[ 17.873931] cg2900-uart cg2900-uart.0: Could not find corresponding user to h4_channel 9

Working log :

              Checking for running unattended-upgrades:
 * Starting bluetooth [ OK ]
saned disabled; edit /etc/default/saned
Welcome to Linaro 12.03 (GNU/Linux 3.3.0-1000-ux500 armv7l)

This issue has to be looked by the kernel team.
Both the logs are attached to this ticket.

Given that this happens on the first boot only, I strongly suggest we lower the priority to low and take it out of the 2012.04 cycle. We have a lot on the plate already.

The issue is reproducible only in first boot.
We understand that the issue should be solved with change as below in file "/etc/init.d/lbsd":
service lbsd /system/bin/lbsd
    user root
    disabled
    group radio
on property:dev:bootcomplete=1
    start lbsd

But we are facing issue to test the same.
This is because this issue is reproducible only in first boot and we can modify this file only in command shell after boot.
So to test this we need help from someone who can create ubuntu image with this change.
Kalle,
Can you help us?

If you are using an SD card, you can simply mount the partitions on the card and edit the file before putting it in the board.

You can also mount the partitions inside an eMMC image (I believe) but that's more involved as you need to know the offset where it starts and so on.

We are using eMMC image and we have tried that.
Problem is we will not be able to reproduce the error like this.

This error was reproducible only in first boot after flash.

Hence, we will need an eMMC image with this change. Such that we can flash and test during first boot.

Will chat with you in detail today.

Sounds quirky if it's not triggered in SD first boot but requires flashing...

But, nothing prevents you from flashing the eMMC image, then booting from SD?

I would be hard-pressed to believe that SD booting does something magical that makes the problem (what is it exactly?) go away...

As for creating an eMMC image, there are steps to create and SD image at

  http://releases.linaro.org/12.03.1/ubuntu/leb-snowball/

which are applicable for creating eMMC image as well. You just need to use 'snowball_emmc' as the --dev target and '--image_file custom.img' instead of the --mmc argument.

The hardwarepack can be augmented to contain a different deb for lbsd with linaro-hwpack-replace from linaro-image-tools.

with further debugging, it was clear that issue cause is different.
debugging this along with other cg2900 kernel and userinit issues.

Need to be checked after kernel changes by Mridul.

Keeping it assigned to GPS team.
Need to be tested and verified only after kernel fix.

This issue needs to be checked by kernel team. We have provided our analysis. It is not related to GPS.
Please asgin to the relevant team.

[ 16.873229] cg2900-uart cg2900-uart.0: Could not find corresponding user to h4_channel 9

This is occuring because CG2900 has received data to be supplied to GPS daemon, but the daemon has either not initialised or has crashed.

If not, please check either the GPS module is not disabled properly, or their is no GPS client user to accept the data, and CG2900 is discarding the same.

kalle,
can you confirm if you are able to reproduce this in 12.05 release image?

This can't be reproduced on 2012.05 since the lbsd package was left out of the hwpack, thus it won't be there to crash on the first boot.

This should be tested against a snapshot, but I'm not opposed to resolving this as wontfix/fixed for the 2012.06 release given the hard-to-reproduce nature.

We can re-open it if it comes up in testing.

Not reproducible with 12.06 release candidate with 3.4 kernel.