From mboxrd@z Thu Jan 1 00:00:00 1970 From: Benjamin Herrenschmidt Subject: Re: [RFC PATCH] kernel vfio: enabled and supported on power Date: Sun, 13 May 2012 08:29:43 +1000 Message-ID: <1336861783.31450.4.camel@pasglop> References: <4FAE11D1.9090103@ozlabs.ru> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Cc: qemu-devel@nongnu.org, Alex Williamson , Alex Graf , kvm@vger.kernel.org, David Gibson To: Alexey Kardashevskiy Return-path: In-Reply-To: <4FAE11D1.9090103@ozlabs.ru> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: qemu-devel-bounces+gceq-qemu-devel=gmane.org@nongnu.org Sender: qemu-devel-bounces+gceq-qemu-devel=gmane.org@nongnu.org List-Id: kvm.vger.kernel.org On Sat, 2012-05-12 at 17:31 +1000, Alexey Kardashevskiy wrote: > And drivers/vfio/vfio_iommu_x86.c works perfect for POWER except a DMA > window setup > which I dropped for now and simply use quite popular configuration on > power (1Gb DMA window starting > from 0x0). > > As for me, it is too complicated. We do not need either > - platform specific IOMMU chardev or > - IOMMU API at all > > What do I miss? I think we want to get rid of the "generic" iommu API and make it x86 specific at this stage. A lot of what the guest is going to do with the iommu is really platform specific at this stage, especially when we start adding the EEH functionality, which is PE-based, or when we want to implement the guest calls for creating dynamic windows. Cheers, Ben.