Thor said:

>> It's bogus for the thing to claim to be a pcnet if it in fact is not
>> quite a pcnet, and since the number of supported s/g elements is a
>> documented in the pcnet and ilacc databooks... sigh.

Actually, as I recall, it's the number of total descriptors that is  
documented in the data books, not the number of descriptors allowed  
per-packet.

In any case, it should be trivial to deal with this using a quirk in  
the pcn driver -- we can look for the OUI for VMware in the MAC  
address, and simply drop the number of Tx segments that bus_dma will  
create for a packet to 15 (from 16).  I think it's pretty rare that we  
bump up against the limit anyway (the driver can even gather these  
statistics for you).

-- thorpej