The problem is that the packets have an FCS at the end; there's no way,
currently, for the driver to indicate that the packets have an FCS
(other than switching from the Prism radio header to the radiotap radio
header, and indicating it in the radio header), so there's no way for
Ethereal to know that the packets have an FCS at the end without the
user telling it that they do.
The way to do that is to turn on the "Assume packets have FCS"
preference for the "IEEE 802.11" protocol. If that option is off, the
802.11 dissector assumes that the last 4 bytes of the packet are packet
data, and those 4 bytes look like an incomplete tagged parameter, and
that incomplete tagged parameter, because it *is* incomplete, causes the
frame to be marked as malformed. If that option is on, the last 4 bytes
in that capture are reported as a correct CRC.
