Doesn't work with 3.10 kernel on raspbery pi

[lazka]

Since the kernel update bleno doesn't get any data. I've enabled debug (DEBUG='*') output and I see nothing when trying to connect.

[lazka]

This one works: raspberrypi-bootloader_1.20130902-1 (kenrel 3.6)
This one doesn't: raspberrypi-bootloader_1.20131209-1 (kernel 3.10)

http://archive.raspberrypi.org/debian/pool/main/r/raspberrypi-firmware/

[zdennis]

The 3.6.11+ kernel works fine.

Do you need to re-build libbluetooth-dev or bluez at all?

[lazka]

I didn't rebuild anything between kernel changes if that's what you mean (I also don't have bluez installed)
I can also confirm on my debian sid dev machine that 3.6 works but 3.12 doesn't.

May I ask what you use to connect to bleno? I'm using android 4.3/4.4

[zdennis]

@lazka, I'm connecting with an iOS7 device using CoreBluetooth. I haven't with Android yet.

[sandeepmistry]

Which version of BlueZ are you using? I'm using the stock 4.99 on mine, see uname output below:

pi@raspberrypi ~ $ uname -a
Linux raspberrypi 3.6.11+ #474 PREEMPT Thu Jun 13 17:14:42 BST 2013 armv6l GNU/Linux

I can also connect from my Nexus 7 running 4.4. I've heard about issues with BLE and Android - see issue #6.

Is there an easy way to switch kernel versions?

[pdominique]

Lazka > I have an issue that might be similar: when I run the test-ibeacon.js application I can see the beacon on my Android app (iBeacon Locate on Android 4.4) or iPhone app (iLocate) only for a few seconds and then it disappears from the list. I just got my Pi and I installed Raspbian via NOOBS 1.3.4. My kernel is:

Linux raspberrypi 3.10.27+ #630 PREEMPT Fri Jan 17 19:44:36 GMT 2014 armv6l GNU/Linux

[pdominique]

Just to confirm that my issue is also gone now that I use kernel 3.6.11+ instead of 3.10.27+.

[sandeepmistry]

Thanks @pdominique I'll get a another SD setup for my RPi to try out the 3.10 kernel.

[sandeepmistry]

Changing the following in hci-ble.c seems to makes things work on kernel 3.10.x

// set advertisement and scan data
hci_le_set_advertising_data(hciSocket, (uint8_t*)&advertisementDataBuf, advertisementDataLen, 1000); 
hci_le_set_scan_response_data(hciSocket, (uint8_t*)&scanDataBuf, scanDataLen, 1000);

// start advertising
hci_le_set_advertise_enable(hciSocket, 1, 1000);

to

// set scan data
hci_le_set_scan_response_data(hciSocket, (uint8_t*)&scanDataBuf, scanDataLen, 1000);

// start advertising
hci_le_set_advertise_enable(hciSocket, 1, 1000);

// set advertisement data
hci_le_set_advertising_data(hciSocket, (uint8_t*)&advertisementDataBuf, advertisementDataLen, 1000); 

followed by a node-gyp rebuild.

Need to verify this doesn't break anything in kernel 3.6.x before pushing to master.

[lazka]

Thanks for looking into this, I'll give it a try.

btw I did use hcitool to set up advertising and disabled it in bleno, but maybe it's the same problem there.

[lazka]

I can confirm that the above mentioned change makes it work here with kernel 3.12.8 and libbluetooth 4.101

Thanks!

[joelind]

Has anybody tried this with Bluez 5? I am not able to connect with Bluez 5
/ Kernel 3.12, but the same code runs fine on Bluez 4.101 and Kernel 3.9.

On Mon, Feb 3, 2014 at 8:00 AM, Christoph Reiter
notifications@github.comwrote:

I can confirm that the above mentioned change makes it work here with
kernel 3.12.8 and libbluetooth 4.101

Thanks!

Reply to this email directly or view it on GitHubhttps://github.com//issues/18#issuecomment-33952114
.

[lazka]

@joelind I've just checked with libbluetooth 5.5 (what's in debian experimental.. 5.14 would be the newest) and it also works.

Make sure to disable/enable bluetooth on Android before testing.

[sandeepmistry]

Thanks @lazka, have you had a chance to try kernel 3.6 with the change? I still haven't had a chance to try it out, hopefully I get a chance by the end of the weekend.

@joelind someone emailed me with the same issue running BlueZ 5.14 on Fedora 20. Can you try stopping bluetoothd? I'm thinking it might be listening for ATT L2CAP connections ...

[joelind]

@sandeepmistry that did the trick!

[sandeepmistry]

@joelind thanks for trying it out, now to figure out what the user space strategy (if there is any) for the newer BlueZ. There doesn't seem to be a way to turn off this behaviour in bluetoothd.

I've created #24 to track this.