Hi Jason, > -----Original Message----- > From: Jason Gunthorpe <j...@mellanox.com> > Sent: Wednesday, August 5, 2020 5:19 PM > To: Dey, Megha <megha....@intel.com> > Cc: Marc Zyngier <m...@kernel.org>; Jiang, Dave <dave.ji...@intel.com>; > vk...@kernel.org; bhelg...@google.com; raf...@kernel.org; > gre...@linuxfoundation.org; t...@linutronix.de; h...@zytor.com; > alex.william...@redhat.com; Pan, Jacob jun <jacob.jun....@intel.com>; Raj, > Ashok <ashok....@intel.com>; Liu, Yi L <yi.l....@intel.com>; Lu, Baolu > <baolu...@intel.com>; Tian, Kevin <kevin.t...@intel.com>; Kumar, Sanjay K > <sanjay.k.ku...@intel.com>; Luck, Tony <tony.l...@intel.com>; Lin, Jing > <jing....@intel.com>; Williams, Dan J <dan.j.willi...@intel.com>; > kwankh...@nvidia.com; eric.au...@redhat.com; pa...@mellanox.com; > Hansen, Dave <dave.han...@intel.com>; netan...@mellanox.com; > shah...@mellanox.com; yan.y.z...@linux.intel.com; pbonz...@redhat.com; > Ortiz, Samuel <samuel.or...@intel.com>; Hossain, Mona > <mona.hoss...@intel.com>; dmaeng...@vger.kernel.org; linux- > ker...@vger.kernel.org; x...@kernel.org; linux-...@vger.kernel.org; > k...@vger.kernel.org > Subject: Re: [PATCH RFC v2 02/18] irq/dev-msi: Add support for a new DEV_MSI > irq domain > > On Thu, Aug 06, 2020 at 12:13:24AM +0000, Dey, Megha wrote: > > > Well, I had suggested to pass in the parent struct device, but it > > > could certainly use an irq_domain instead: > > > > > > platform_msi_assign_domain(dev, > > > device_to_iommu(p_dev)->ir_domain); > > > > > > Or > > > > > > platform_msi_assign_domain(dev, pdev->msi_domain) > > > > > > ? > > > > > > Any maybe the natural expression is to add a version of > > > platform_msi_create_device_domain() that accepts a parent > > > irq_domain() and if the device doesn't already have a msi_domain > > > then it creates one. Might be too tricky to manage lifetime of the new > irq_domain though.. > > > > > > It feels cleaner to me if everything related to this is contained in > > > the platform_msi and the driver using it. Not sure it makes sense to > > > involve the iommu? > > > > Well yeah something like this can be done, but what is the missing > > piece is where the IRQ domain actually gets created, i.e where this > > new version of platform_msi_create_device_domain() is called. That is > > the only piece that is currently done in the IOMMU driver only for DSA > > mdev. Not that all devices need to do it this way.. do you have > > suggestions as to where you want to call this function? > > Oops, I was thinking of platform_msi_domain_alloc_irqs() not > create_device_domain() > > ie call it in the device driver that wishes to consume the extra MSIs. > > Is there a harm if each device driver creates a new irq_domain for its use?
Well, the only harm is if we want to reuse the irq domain. As of today, we only have DSA mdev which uses the dev-msi domain. In the IRQ domain hierarchy, We will have this: Vector-> intel-ir->dev-msi So tmrw if we have a new device, which would also want to have the intel-ir as the parent and use the same domain ops, we will simply be creating a copy of this IRQ domain, which may not be very fruitful. But apart from that, I don't think there are any issues.. What do you think is the best approach here? > > Jason