Subject: Re: DMA beyond end of isa
To: Andrew Cagney <>
From: None <Chris_G_Demetriou@NIAGARA.NECTAR.CS.CMU.EDU>
List: tech-kern
Date: 01/05/1996 10:41:43
> Excerpts from mail: 4-Jan-96 Re: DMA beyond end of isa Bill
> Studenmund@loki.Sta (2776*)
> > The outline of the spec is one or more PowerPC's on a PCI bus connected
> > to a PCI-ISA bridge. A chip on the PCI side impliments most of the
> > Mac's hardware, and chips on the ISA side do most of a PC's (such as
> > soundblaster-compatable audio, COM1-COM4, LPT1, PC keyboard, etc.)
> > I/O. Most of the IO will be accessable little-endianly and
> > big-endianly (the video buffer in particular).
> Some trivia...
> The PCI-ISA bridge would more then likely be based on the 82378 which
> supports dma (on the PCI side) to the full 4gb of RAM.  Do x86 PCI
> machines already use this chip?

that would be:

product INTEL SIO       0x0484  82378IB PCI-ISA Bridge (System I/O)

yes, PCs use it, as do many alphas, however:

it may support DMA to the full 4GB of 32-bit PCI address space, but no
matter how you cut it, ISA devices won't be able to access more than
24-bits of that address space at a time; at best it's windowed, which
is still annoying (and which is unsupported in the current code, and
relatively hard to support with a reasonable interface, at all).

"and now to rant!"

I really hope they don't go the 'normal' PC route with the 82378,
where PCI interrupts are mapped through IRQs which are handled by the
PICs on the 82378.  That would be nasty.

I long for the day when _REAL_ PCI-ISA briges exist, i.e. that:
	(1) present all 15 ISA IRQs as one PCI IRQ,
	(2) do positive address decoding on accesses, then pass
	    the accesses along (and do negative, ISA, decoding
	    on them on the far side of the bridge).
I'll be so happy them...  ISA busses will be be remappable, you'll be
able to have more than one, you'll be able to tell, reliably, if
there's no port at the address you're trying to talk to, etc. etc.

What I really want is PCI-ISA adapters to exist only so that people
can talk to legacy devices.  I want this completely
backward-compatible hardware/software interface to go away, because it
causes many more problems than it solves (for me)...  8-)