CE never supported this BT chip.
I understand that. Was the CE team already aware of the original mod that you linked to?
What I meant to ask is if it might be worth testing a BT driver compiled from that original mod source, without any of the changes made in Noob404’s version?
From what I see noob404yt took this sources from coral.googlesource.com (or maybe some other place) for BT and didn’t made any changes to them.
Only wifi driver from khadas repo has changes probably because of 5.4 kernel.
1 Like
Oh ok, that makes sense. He was probably only focused on the WiFi drivers since BT works in 5.4 but WiFi doesn’t.
Even though the driver you compiled from that mod might not be functional enough to include in CE, it would still be worth making it available for individuals as an addon for the limited br/edr functionality.
You wrote wifi had some issues with connection? Is the original version without BT working better?
Is it possible to use the MT7886 WiFi drivers already included with CE rather than the modded version? Or are the WiFi/BT drivers a package set?
I did have some issues with the WiFi network GUI screen freezing once or twice, requiring a Kodi restart. But after I got my WiFi password in there I didn’t notice any other WiFi issues.
I’m still working on fixing up a DTB for my device, so I can’t rule out that any issues I might have been experiencing aren’t due to something other than the MT7886 drivers.
EDIT: the original CE mt7668 WiFi drivers work without any issues at all as far as I can tell. It was the modded WiFi drivers that might have had some issues
Need to think how to include this BT part.
1 Like
Btw: we also try with this driver but forgot what was conclusion. I would say it didn’t work at that time. And it is hard to work on such things without actual testing.
bt version in noob404yt repo
#define VERSION "v0.0.0.53_2018072701_YOCTO"
bt version in this ce mt76x8 repo - much newer
#define VERSION "v0.0.1.00_2021011401"
1 Like
Is there a compiled version of that driver ( v0.0.1.00_2021011401)? I will test it and report back how it compares to (v0.0.0.53_2018072701_YOCTO).
I guess it depends on what your threshold is for ‘functional’. Even if all it’s good for is connecting non LE peripherals like mice and keryboards, that’s better than nothing.
I forgot the status if this newer driver. Because we didn’t push it in use then something was wrong.
And this was at CE19 time.
Let me see if I can make test image with CE20 with this driver version.
1 Like
Try this with driver version v0.0.1.00_2021011401. Focus on BT first but also look on WIFI.
1 Like
Thanks @vpeter, this driver adds a lot more BT initializing text to dmesg, but when I enable BT through the GUI, I see ‘No Bluetooth adapter found’.
From bluetoothctl, when toggling BT on from the GUI, it returns ‘Agent registered’. But bluetoothctl: list doesn’t show any BT controllers.
I’ll keep looking just in case there is anything i can do from ssh
btw WiFi appears to be working without any problems
Try this if maybe BT behaves better: 205.42 MB file on MEGA
post dmesg and journal
journalctl -l | paste
1 Like
The BT behavior is the same. When enabling BT, no bluetooth adapter is found, and no controller is present in bluetoothctl: list
When I ran journalctl -l | paste it looked like the request was being processed, but no url was returned. Is there a size limit to the paste function? I copied the journal contents and uploaded the file below.
ps I still have the first CE version you compiled with the much older BT drivers on another flash drive. Let me know if you want any logs from that device for comparison.
Can you check if bt is maybe blocked?
rfkill list
If not then nothing much can be done.
I looks like it’s ok
0: bt-dev: bluetooth
Soft blocked: no
Hard blocked: no
1: phy0: wlan
Soft blocked: no
Hard blocked: no
2: phy1: wlan
Soft blocked: no
Hard blocked: no
3: phy2: wlan
Soft blocked: no
Hard blocked: no
One of the few notable differences between the newer driver, and older driver that can load and access the BT controller, are a couple differences in the control registers that are accessed. Could it be that cr 0x80000000 needs to be read to gain access to the BT controller?
driver version v0.0.1.00_2021011401
[ 9.086679@4]- btmtk_sdio_bt_memRegister_read: read cr 80000004
[ 9.100790@4]- btmtk_sdio_bt_memRegister_read: retrytime 2
[ 9.100830@4]- btmtk_sdio_bt_memRegister_read: ger cr 0x80000004 value 0x8b01
[ 9.101677@4]- btmtk_sdio_download_rom_patch Fw Version 0x8b01
[ 9.102408@4]- btmtk_sdio_bt_memRegister_read: read cr 80000008
[ 9.119827@4]- btmtk_sdio_bt_memRegister_read: retrytime 2
[ 9.119877@4]- btmtk_sdio_bt_memRegister_read: ger cr 0x80000008 value 0x7668
driver version v0.0.0.53_2018072701_YOCTO
[ 12.064426@1]- btmtk_sdio_bt_memRegister_read: read cr 80000000
[ 12.075766@2]- btmtk_sdio_bt_memRegister_read: retrytime 299
[ 12.075805@2]- btmtk_sdio_bt_memRegister_read: ger cr 0x80000000 value 0x8b11
[ 12.076673@2]- btmtk_sdio_download_rom_patch uhwversion 0x8b11
Register 0x80000000 is jut to read hw version - I doubt there would be any difference reading it.
The source code between this two versions is very, very different. And without knowing what to look is hard to spot any difference 
1 Like
Here another version of this driver:
#define VERSION "v0.0.0.47_2018012902"
“Import from FireTVStick4K-6.2.6.5-20190808”
This is even older,& may better fit to old kernel.
1 Like
Does it matter that the drivers are for an Android kernel? Amazon uses the MT7668 chipset in a number of devices, and has the source available for all their devices. I could collect a handful of versions if they can be tested with little changes.