diff options
Diffstat (limited to 'net-wireless/bluez/files/btattach-bcm-service.sh')
| -rw-r--r-- | net-wireless/bluez/files/btattach-bcm-service.sh | 30 |
1 files changed, 30 insertions, 0 deletions
diff --git a/net-wireless/bluez/files/btattach-bcm-service.sh b/net-wireless/bluez/files/btattach-bcm-service.sh new file mode 100644 index 000000000000..c7f8b1784679 --- /dev/null +++ b/net-wireless/bluez/files/btattach-bcm-service.sh @@ -0,0 +1,30 @@ +#!/bin/bash + +# Simple shell script to wait for the tty for an uart using BT HCI to show up +# and then invoke btattach with the right parameters, this is intended to be +# invoked from a hardware-activated systemd service +# +# For now this only suports ACPI enumerated Broadcom BT HCIs. +# This has been tested on Bay and Cherry Trail devices with both ACPI and +# PCI enumerated UARTs. +# +# Note the kernel bt developers are working on solving this entirely in the +# kernel, so it is not worth the trouble to write something better then this. + +BT_DEV="/sys/bus/platform/devices/$1" +BT_DEV="$(readlink -f $BT_DEV)" +UART_DEV="$(dirname $BT_DEV)" + +# Stupid GPD-pocket has USB BT with id 0000:0000, but still claims to have +# an uart attached bt +if [ "$1" = "BCM2E7E:00" ] && lsusb | grep -q "ID 0000:0000"; then + exit 0 +fi + +while [ ! -d "$UART_DEV/tty" ]; do + sleep .2 +done + +TTY="$(ls $UART_DEV/tty)" + +exec btattach --bredr "/dev/$TTY" -P bcm |