From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx1.redhat.com ([209.132.183.28]:30492 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1945973Ab2JaSTH (ORCPT ); Wed, 31 Oct 2012 14:19:07 -0400 Message-ID: <50916B93.9060807@redhat.com> Date: Wed, 31 Oct 2012 14:18:59 -0400 From: Don Dutile MIME-Version: 1.0 To: Ben Hutchings CC: "Rose, Gregory V" , "linux-pci@vger.kernel.org" , "bhelgaas@google.com" , "yuvalmin@broadcom.com" , "yinghai@kernel.org" , "davem@davemloft.net" Subject: Re: [PATCH 3/8] PCI: sysfs per device SRIOV control and status References: <1351190310-5043-1-git-send-email-ddutile@redhat.com> <1351190310-5043-4-git-send-email-ddutile@redhat.com> <1351196234.2662.37.camel@bwh-desktop.uk.solarflarecom.com> <508AA719.2060705@redhat.com> <1351705018.2706.9.camel@bwh-desktop.uk.solarflarecom.com> In-Reply-To: <1351705018.2706.9.camel@bwh-desktop.uk.solarflarecom.com> Content-Type: text/plain; charset=UTF-8; format=flowed Sender: linux-pci-owner@vger.kernel.org List-ID: On 10/31/2012 01:36 PM, Ben Hutchings wrote: > On Wed, 2012-10-31 at 17:01 +0000, Rose, Gregory V wrote: >>> -----Original Message----- >>> From: Don Dutile [mailto:ddutile@redhat.com] >>> Sent: Friday, October 26, 2012 8:07 AM >>> To: Ben Hutchings >>> Cc: linux-pci@vger.kernel.org; bhelgaas@google.com; yuvalmin@broadcom.com; >>> Rose, Gregory V; yinghai@kernel.org; davem@davemloft.net >>> Subject: Re: [PATCH 3/8] PCI: sysfs per device SRIOV control and status >>> >> >>> >>> Ok, my turn: >>> Any feedback on having the sysfs configure call >>> pci_sriov_[enable/disable](), as well as do the don't-disable if VFs are >>> assigned to guests? >>> >> >> Don, >> >> As I've mentioned before I still prefer to have the sysfs interface >> you've written up make the calls to pci_sriov_enable/disable() > > I think that would work for sfc, assuming the driver function is to be > called before either of those. I don't know whether it would work for > any of the other drivers with SR-IOV back-ends, though. > >> and have the checking for whether the VFs are assigned to guests done >> there also, > > I agree that this is should be centralised, though I think that could be > done as a later step without too much pain. > >> but really it isn't anything worth going to the mats about. As it >> stands I think if you address the issues brought up by Ben then I'm >> fine with what you've worked up so far. Since no one else seems to >> have an opinion about it (as demonstrated by a lack of response over >> the last 5 days) then I'd suggest we go forward with the current >> implementation. I'd really like to see this in 3.8 if possible. >> >> Thanks for all your work. > > Agreed, thanks Don. > > Ben. > Greg & Ben, Thanks for your feedback on the core doing the sriov_enable/disable. I agree that let's get existing the core stuff into 3.8 & do the enable/disable as a follow up. I'm just polishing the patch set now based on Ben's feedback.... I was delayed when my test machine was (improperly) taken from me (bad IT, bad! :-( ). Just got the machine back today to test the various (error) cases, and make checkpatch.pl happy. I expect to post a PATCH (non-RFC) later today. - Don ps -- of course, if Greg wants to give me a set of igb patches like the ones he did for ixgbe, it'd get me out of the test machine battles -- I have private test machine w/igb's. :) hint, hint...ok, more like 'beg, beg' ... again... ;)