[Ksummit-2012-discuss] PCI breakout session
H. Peter Anvin
hpa at zytor.com
Mon Jun 18 21:35:56 UTC 2012
On 06/18/2012 02:04 PM, Benjamin Herrenschmidt wrote:
> I'n not saying you "can't use it". I'm saying using it (especially as
> the sole access method) for something that is essentially emulated PCI
> in this day an age is stupid.
> IO space is a piece of history and should remain as such, designing new
> stuff around IO space is silly. It's limitations such as the 64K overall
> window accross all domains on x86 are a killer for example and trying to
> increase the granularity is yet another hackish workaround akin to
> plastering a wooden leg.
It makes sense to use it for emulated PCI on x86, because in the virtual
environment I/O space references effectively become native hypercalls --
they are *very* efficient. Just like the S390 people like to use their
channel interfaces for the same purpose, using I/O space makes a ton of
sense on x86.
And no, it's not just "a hackish workaround". It's a useful repurposing
of a legacy interface.
More information about the Ksummit-2012-discuss