We used ampak ap6255 (cyw43455) on our custom arm/linux board, running mainline kernel (5.1.5).
Our hardware guy forgot to connect the BT_DEV_WAKE pin to any pin on host processor (rockchip rk3328). The pin is merely pulled-up via a resistor.
We tried to load the driver and it works. The driver complains when failing to resolve the device-wakeup-gpios option in device tree blob. But no further problems found so far.
[ 3.547897] of_get_named_gpiod_flags: can't parse 'device-wakeup-gpios' property of node '/serial@ff110000/bluetooth'
[ 3.547908] of_get_named_gpiod_flags: can't parse 'device-wakeup-gpio' property of node '/serial@ff110000/bluetooth'
AFAIK, if BT_DEV_WAKE is kept asserted, the bluetooth device won't sleep. This is not a problem for our design since the board works as a server and the power consumption is a trivial issue.
My hope is that you can confirm such a configuration is OK, both for bluetooth device and for kernel driver. Then we don't need to modify and re-manufacture PCBs, saving us some time and money.
Or may there be any potential risk in functionality or stability we don't know, please tell us. Then we have to modify the design and allocate a host pin for suspending/resuming the bluetooth device.