lkml.org 
[lkml]   [2008]   [Nov]   [20]   [last100]   RSS Feed
Views: [more markup]  [less markup]  [headers]  [forward] 
 
Messages in this thread
/
DateThu, 20 Nov 2008 08:51:39 +0100
FromIngo Molnar <>
SubjectRe: [GIT PULL] AMD IOMMU updates for 2.6.28-rc5
* FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> wrote:

> > > It had been the default option for AMD IOMMU until you hit the 
> > > bugs. IIRC, VT-d also shares a protection domain by default. It 
> > > would be nice to avoid surprising users if the two 
> > > virtualization IOMMUs works in the similar way.
> > 
> > Calgary has a per-bus protection domain, both on x86 and PPC.
> 
> I see. Then it might be better to change VT-d to use a separate 
> protection domain by default.

yes, agreed, and that should be the sane default for any IOMMU driver 
- unless the performance impact is prohibitive.

Note that this widens the positive impact of the IOMMU code: not only 
does it enable transparent support of DMA to/from devices that have a 
limited DMA range, not only does it help isolation in virtualization - 
it also acts as a daily debug helper for _native_ drivers.

Note that people will prefer to run with an IOMMU enabled even if all 
devices support the full memory range - just due to the DMA protection 
features. Just like people prefer to run an OS with paging protections 
enabled ;-)

It also puts pressure on the hw design side to treat IOMMUs not just 
as some fringe feature for compatibility with older transports or 
virtualization, but also as a prime-time native IO feature.

	Ingo


\
 
 \ /
  Last update: 2008-11-20 08:55    [from the cache]
©2003-2008