September-21st-2018, 11:16 AM
I investigated the problem and finally I found a solution: resetting the BCM chip physically before initializing the firmware is the key point
This is enabled by a tool called vcmailbox coming from Raspbian userland space.
The usage is the following:
$ /opt/vc/bin/vcmailbox 0x38041 8 8 128 0
$ sleep 1
$ /opt/vc/bin/vcmailbox 0x38041 8 8 128 1
$ sleep 1
after that the standard command
$ hciattach /dev/ttyAMA0 bcm43xx 921600 noflow
can be called as usual and will succeed in all cases seen so far.
The vcmailbox binary can be received from here. It needs access to the device /dev/vcio. So make sure you expose it when you start your updated container.
Our bluetooth reference implementation container has been updated already.
Thx
Armin
This is enabled by a tool called vcmailbox coming from Raspbian userland space.
The usage is the following:
$ /opt/vc/bin/vcmailbox 0x38041 8 8 128 0
$ sleep 1
$ /opt/vc/bin/vcmailbox 0x38041 8 8 128 1
$ sleep 1
after that the standard command
$ hciattach /dev/ttyAMA0 bcm43xx 921600 noflow
can be called as usual and will succeed in all cases seen so far.
The vcmailbox binary can be received from here. It needs access to the device /dev/vcio. So make sure you expose it when you start your updated container.
Our bluetooth reference implementation container has been updated already.
Thx
Armin
„You never fail until you stop trying.“, Albert Einstein (1879 - 1955)
