I found two places in the BPI repository where the driver (or suspected driver) might exist:
Examining the first link, I noticed that the file distribution within eth is very similar to the location of some files in the openwrt mainline:
After copying and pasting (the driver doesn't adapt to all API changes):
After applying it in my branch (https://github.com/libiunc/openwrt-dev/tree/siflower-hwnat/package/kernel/siflower), using make menuconfig:
> Kernel modules > Cryptographic API modules
> Kernel modules > Network Devices
This is what I've done so far. I think that besides changing the driver API, the makefiles also need to be modified and integrated to make it more "aesthetically pleasing".
In addition, the siflower target might need to be pushed to the Linux mainline, but I don't know how to do it 