> I'd like to be wrong, but I don't think so (for Tx mbufs). I had to
> m_copy() as well in a driver for a card that can't handle byte-enable
> signals -- also, it was necessary to check all bus_dma_segments for
> proper alignment to prevent data corruption. *Uff*.

I didn't think so.  :-(

I also had to do this, but I couldn't ever get the stack to send me an
mbuf chain that consisted of more than one bus_dma_segment (packet either
in 1 mbuf or 1 cluster).  Any idea of how to provoke a multi-mbuf packet? 
I'd really like to test that code path in my driver.

//Jesse Off