> BTW:  What happened with the bounce buffers?  Has that been
> implemented in current?

No, not yet.

For me to start implementing them, i need somebody else to finish
something that they were working on, but have been too busy to work on
recently.

Once that's done, i need to come up with the "right" interface -- i
have some ideas, but all of them seem to have a few drawbacks -- that
will work on the i386 and alpha, for ISA, EISA, and PCI.  In a
nutshell, the way DMA mapping is done now (i.e. with vtophys()) in
_ALL_ ISA, EISA, and PCI  drivers, even those which wouldn't require
bounce buffering on the i386, will not work on other, more complex,
systems, such as the Alpha.

Once that's done, i need to implement the interface -- hopefully on
the Alpha, rather than the i386 -- to be sure that it's 'sane.'

Then, i'll need to port it to the i386.

Then, i'll need to port more drivers to use it (and i'll not be
porting too many, since i can neither use nor test them).


In other words, there's a lot of work, e.g. all of the actual
implementation, to be done, and i'm doing this in my "free time."
It will get done, but it won't get done tomorrow.



cgd