Current code does this:

  be16_to_cpu(buf[i * 2] << 8 | buf[(i * 2) + 1])

Which is effectively (neglecting the index):

  be16_to_cpu(be16_to_cpu(*((u16 *) buf)))

This means the int16 in the buffer is not converted at all.

Daniel Mack confirmed that the driver works on little endian
CPUs, leading to the conclusion that the device-side structure
is actually little endian.
This changes the code to use le16_to_cpu().

Caught by sparse.
Acked-by: Daniel Mack <zonque@gmail.com>
Signed-off-by: Eldad Zack <eldad@fogrefinery.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>