Re: [PATCH RFC] Bluetooth: Add generic support for vendor HCI frames

[Zijun Hu]

On 7/22/2026 3:49 AM, Luiz Augusto von Dentz wrote:
> For the user channel I would say it would be better to always include
> 0xff. That way, it remains agnostic and it won't suddenly match a
> different packet type if you change the vendor or something like that.

so, perhaps, you original proposal(Option A) is better and easier to implement than this
simpler one since it uses consistent way to decide vendor HCI frame.

1)
userspace <-> kernel : 0xff + vendor hci indicator.

2)
for RX, transport driver stamp with HCI_VENDOR_PKT before hci_recv_frame()
For TX, transport driver pull one more byte(indicator) for HCI_VENDOR_PKT frame
tell type of vendor hci frames.

let me also give a implementation for this 'Option A' as well.