From mboxrd@z Thu Jan 1 00:00:00 1970 From: Joerg Roedel Subject: Re: [PATCH 1/1] iommu: Bind process address spaces to devices Date: Tue, 26 Feb 2019 14:02:04 +0100 Message-ID: <20190226130204.GA23259@8bytes.org> References: <20190220142759.33308-1-jean-philippe.brucker@arm.com> <20190220142759.33308-2-jean-philippe.brucker@arm.com> <20190226111743.GK20740@8bytes.org> <559a7ef6-8591-3936-b208-3bbae555a15a@arm.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <559a7ef6-8591-3936-b208-3bbae555a15a@arm.com> Sender: linux-kernel-owner@vger.kernel.org To: Jean-Philippe Brucker Cc: kevin.tian@intel.com, alex.williamson@redhat.com, ashok.raj@intel.com, linux-kernel@vger.kernel.org, iommu@lists.linux-foundation.org, christian.koenig@amd.com List-Id: iommu@lists.linux-foundation.org On Tue, Feb 26, 2019 at 12:49:15PM +0000, Jean-Philippe Brucker wrote: > On 26/02/2019 11:17, Joerg Roedel wrote: > > int iommu_sva_get_pasid(struct iommu_sva *handle); > > void iommu_sva_set_exit_handler(struct iommu_sva *handle, > > iommu_mm_exit_handler_t mm_exit); > > Ok sounds good. It doesn't look like this interface requires a lot of > changes on my side (iommu_sva corresponds to the iommu_bond structure > I've been using internally) but I might find problems while implementing it. Great! > Device drivers will also want to have some private data to easily > identify the faulting or exiting context. How about: > > struct iommu_sva_ops { > void (*mm_exit)(struct iommu_sva *handle, void *drvdata); > }; > int iommu_sva_set_ops(struct iommu_sva *handle, > const struct iommu_sva_ops *ops, > void *drvdata); Okay, we can also do it this way. But then please pass the drvdata via the bind() call and not via set_ops(). Set_ops() should then really only pass the call-backs, as the name implies. > I now think that device driver should always call unbind() to release > the iommu_sva handle, even if they got notified by mm_exit. Yes, this should be required. I think it also makes driver implementation easier because it doesn't need to care too much about this special case. Regards, Joerg