From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eli Cohen Subject: Re: [PATCH RFC] pci: Control whether VFs are probed on pci_enable_sriov Date: Mon, 8 Dec 2014 21:10:32 +0200 Message-ID: <20141208191032.GC24790@mtldesk30> References: <1417957693-24979-1-git-send-email-eli@mellanox.com> <20141207184242.GA24790@mtldesk30> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Eli Cohen , "bhelgaas@google.com" , David Miller , linux-pci , netdev , "ogerlitz@mellanox.com" , "yevgenyp@mellanox.com" , Donald Dutile To: Yuval Mintz Return-path: Content-Disposition: inline In-Reply-To: Sender: linux-pci-owner@vger.kernel.org List-Id: netdev.vger.kernel.org On Sun, Dec 07, 2014 at 08:08:47PM +0000, Yuval Mintz wrote: > >>>Use a parameter to pci_enable_sriov to control that policy, and modify > >>>all current callers such that they retain the same functionality. > >> > >> What's the end-game here? How eventually would this be controlled? > > >You can probe any VF at the hypervisor through sysfs files > >(bind/unbind). You can also pass them through to a VM. Nothing > >changes. > > If you're not planning on adding a logic to set this, why do we need to > add a parameter to pci_enable_sriov() - given that all callers use the > exact same logic? > [And I don't really think we'd want different devices to behave differently > by default; That would be confusing for users.] > Currently the kerenl will call probe for any device for which there is a supporting driver and I did not want to change that. > >>>Use a one shot flag on struct pci_device which is cleared after the > >>>first probe is ignored so subsequent attempts go through. > >> > >> Does a one-shot flag suffice? E.g., consider assigning a VF to VM and > >> than shutting down the VM. Assuming this feature is disabled, > >> the VF didn't appear on the hypervisor prior to the assignment but > >> will appear after its shutdown. > > >Sorry, I don't follow you here. Please clarify. > > >To be clear, the functionality proposed here is really one shot. It > >just prevents calling probe once; besides that nothing changes. > > What I meant is that device is unbinded after initial probe, > But in the scenario I've stated above, the VF will become binded once > it's returned to the hypervisor. > Now, I understand that what you're trying to achieve - but my question > is whether what you're REALLY trying to achieve is the ability to have VFs > which would only be binded to VMs and never to hypervisor [by default]? No this was not my intention. I guess any driver that wishes to achieve the same functionality can simply ignore the N first probes after the call to pci_enable_sriov() where N is the number of VFs it created.