Very Computer

(Or which models don't, whichever is easier!)

Also, does anyone out there have experience with the use of the DDI/DKI
function:

ddi_dma_set_sbus64()

Would I call this after each time I get a DMA handle?  Can this be invoked
on a machine with a 32-bit Sbus (and will it reliably fail in this case?)
Or do I look at the return value of ddi_dma_burstsizes after calling
ddi_dma_set_sbus64(), and compare the resulting burstsize with what was
requested in ddi_dma_set_sbus64()?

To say that this area is a little underdocumented is putting it mildly!
There is no mention of changes required for 64-bit SBus support in the
downloadable DDK.

Also, is the IEEE Sbus spec available on-line anywhere, or a precis/summary
of same?
--
Steve Dyer

Sun does. Or so one may conclude running cscope.

Quote:>Would I call this after each time I get a DMA handle?

Not necesserily. Allocate any handle (IOCB if applicable),
then call ddi_dma_burstsizes() and ddi_dma_set_sbus64().
If the latter returns success you may reconfigure your
device to use 64bit requests.

Quote:>  Can this be invoked
>on a machine with a 32-bit Sbus (and will it reliably fail in this case?)

Yes.

Quote:>Or do I look at the return value of ddi_dma_burstsizes after calling
>ddi_dma_set_sbus64(), and compare the resulting burstsize with what was
>requested in ddi_dma_set_sbus64()?

Like that or other way around, as you like.

Quote:>Also, is the IEEE Sbus spec available on-line anywhere, or a precis/summary
>of same?

No. IEEE makes money off this. Sun cooperates removing these materials,
if we may judge by the fate of IEEE-1275 on playground.sun.com.

And remember, you apply my suggestions on your own risk!
--Pete

P.S. I am curious, who may develop an Sbus device in 1989, and why?