[cap-talk] DMA vs programmed I/O (was: Midori in The Register)

[Jonathan S. Shapiro]

On Wed, 2008-08-06 at 22:34 +0100, David-Sarah Hopwood wrote:
> Kenneth Hamer-Hodges wrote:
> > Adding a second DMA version of typed hardware leads to a second set of
> > mechanism at risk and question about delegation to the DMA level open up
> > another Pandora's Box. 
> > With multi core solution so easily available today this simplicity and
> > uniformity remain the best option for building a Trusted Computer Base.
> 
> Given that existing systems use DMA, and obtain a significant performance
> advantage from doing so *for any given number of cores*, and that new
> systems must be able to compete on performance, it seems unlikely that a
> new system design that was unable to use DMA could be successful.

I agree, but Kenneth's criteria was assurability, and he therefore
focused on complexity reduction rather than performance.

As a practical matter, I don't think that a typed DMA hardware interface
would add significant complexity to the implementation, and that
complexity must be weighed against the need to show in the "extra
general core" approach that the programmed I/O implementation running on
the extra core isn't doing some additional thing, and that all the
extraneous hardware gorp in that core does not alter the semantics of
the I/O operation. As a practical matter, this seems likely to be more
challenging than assuring a typed DMA interface.


shap