Three fixes for host device passthrough:
1. OUT transfers now report actual_length = transfer_buffer_length instead
of 0. The kernel needs to know how many bytes were consumed; returning 0
caused bulk writes (critical for mass storage) to appear as failures.
2. BCD version encoding (bcdUSB, bcdDevice) now properly reconstructs the
full 2-byte BCD value from the Version struct's nibble fields. Previously,
the minor field (a single nibble) was written as a full byte, corrupting
values like USB 1.1 (0x0110 → 0x0101).
3. USB transfer handlers (nusb) now propagate errors instead of silently
returning empty success responses. Failed control_in, bulk, and interrupt
transfers were falling through to Ok(UrbResponse::default()), making the
kernel think transfers succeeded with 0 bytes of data.
Also sets usb_version and dev_num from the actual device in build_usb_device.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Davíð Steinn Geirsson
889017be3b