On Wed, 27 Jan 1999 09:07:45 -0500 
 "Jason L. Wright" <jason@thought.net> wrote:

 > The 'qec' dma engine appears to be much like the 'hme' engine, so it may be
 > able to do full duplex.  The problem is the phy on the 'qec+be' doesn't
 > claim to do full duplex, and no amount of coaxing can convince it to do it
 > (nsphy when found on the hme requires a bit of convincing about full duplex
 > operation).  My guess is that if an external phy capable of full duplex is
 > hooked up to a 'qec+be', then the card might be able to do full duplex, but
 > the onboard phy won't do it.

Full-duplex requires some support in the MAC, as well.  Many MACs determine
it automatically based on electrical signals from the PHY... but some require
frobbing of a control register directly, even when you put the PHY into
full-duplex mode.

        -- Jason R. Thorpe <thorpej@nas.nasa.gov>