Re: [PATCH 0/2] x86: per-device dma_mapping_ops

[FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>]

On Wed, 14 May 2008 08:49:24 +0300
Muli Ben-Yehuda <muli@il.ibm.com> wrote:

> On Tue, May 13, 2008 at 03:04:50PM +0900, FUJITA Tomonori wrote:
> 
> > This patchset adds per-device dma_mapping_ops support for
> > CONFIG_X86_64 like POWER architecture does. This change enables us
> > to cleanly fix the Calgary IOMMU issue that some devices are not
> > behind the IOMMU [1]. It also would be helpful to handle KVM PCI
> > passthrough.
> 
> Awesome! Much needed, thank you for doing this.

No problem. Well, as you know, it's just a base. We need more work to
solve the problems on the top of this.

I'd like to have a mechanism to register a hook called when a new pci
(or dma capable) device is created. It enables IOMMUs to set up an
appropriate dma_mapping_ops per device. It could also enables us to
simplify the IOMMUs code to initilize devices at startup (for exmple,
intel-iommu checks all the pci devices and creates a domain per device
if necessary).

I'll post an updated version against -mm. If people seems to be fine
with per-device dma_mapping_ops, then I'll work on further issues.