[Intel-wired-lan] [PATCH] ixgbe: Remove bimodal SR-IOV disabling
Rose, Gregory V
gregory.v.rose at intel.com
Fri Jul 10 23:00:59 UTC 2015
> -----Original Message-----
> From: Alex Williamson [mailto:alex.williamson at redhat.com]
> Sent: Friday, July 10, 2015 3:44 PM
> To: Rose, Gregory V
> Cc: intel-wired-lan at lists.osuosl.org; Kirsher, Jeffrey T;
> netdev at vger.kernel.org; linux-kernel at vger.kernel.org
> Subject: Re: [PATCH] ixgbe: Remove bimodal SR-IOV disabling
>
> On Fri, 2015-07-10 at 21:36 +0000, Rose, Gregory V wrote:
> >
> > > -----Original Message-----
> > > From: Alex Williamson [mailto:alex.williamson at redhat.com]
> > > Sent: Friday, July 10, 2015 2:32 PM
> > > To: intel-wired-lan at lists.osuosl.org; Kirsher, Jeffrey T
> > > Cc: netdev at vger.kernel.org; linux-kernel at vger.kernel.org; Rose,
> > > Gregory V
> > > Subject: [PATCH] ixgbe: Remove bimodal SR-IOV disabling
> > >
> > > When unbinding an SR-IOV device with VFs configured from ixgbe, the
> > > driver behaves in one of two ways. If max_vfs was specified, the
> > > SR-IOV state is disabled, removing the VFs. The occurs regardless
> > > of whether the VF count was later modified through sysfs. If
> > > however max_vfs is zero, such as by not specifying the module
> > > parameter, the VFs persist after the PF is unbound from ixgbe. If
> > > the PF is then bound to vfio-pci to be assigned to a VM, the PF is
> non-functional.
> > >
> > > >From the comment, commit da36b64736cf ("ixgbe: Implement PCI SR-IOV
> > > sysfs callback operation") clearly intended this alternate behavior,
> > > but probably didn't realize the PF doesn't work in this mode.
> > >
> > > This bimodal behavior is confusing to users and results in a state
> > > where the PF is broken for other uses unless the user sets
> > > sriov_numvfs to zero prior to unbinding the device. Remove this
> > > behavior so that VFs are removed and the PF is functional for other
> > > uses after unbind, regardless of the way VFs are enabled.
> > >
> > > Signed-off-by: Alex Williamson <alex.williamson at redhat.com>
> > > Cc: Greg Rose <gregory.v.rose at intel.com>
> > > Cc: Jeff Kirsher <jeffrey.t.kirsher at intel.com>
> > > ---
> > >
> > > I can only think that not disabling SR-IOV was meant to enable some
> > > sort of persistence for VFs, but that's probably better accomplished
> > > with either udev rules and/or modprobe.d install scripts.
> > >
> > > drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 7 +------
> > > 1 file changed, 1 insertion(+), 6 deletions(-)
> > >
> > > diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> > > b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> > > index 5be12a0..de04e3e 100644
> > > --- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> > > +++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> > > @@ -8810,12 +8810,7 @@ static void ixgbe_remove(struct pci_dev *pdev)
> > > unregister_netdev(netdev);
> > >
> > > #ifdef CONFIG_PCI_IOV
> > > - /*
> > > - * Only disable SR-IOV on unload if the user specified the now
> > > - * deprecated max_vfs module parameter.
> > > - */
> > > - if (max_vfs)
> > > - ixgbe_disable_sriov(adapter);
> > > + ixgbe_disable_sriov(adapter);
> > > #endif
> > > ixgbe_clear_interrupt_scheme(adapter);
> > >
> >
> > Please remove max_vfs module parameter - it is deprecated and should be
> removed from upstream builds. Dave let us get away with a kernel module a
> few years ago because the other necessary infrastructure to enable SR-IOV
> virtual functions via the PCIe interface was not available. Now that it's
> there it should be removed and vendors/end users should be forced to move
> away from this.
>
> I can't really say I'm in favor of removing that option. It's probably
> going to break a lot of people because doing the udev rules right is hard.
> The sysfs sriov interface has been tossed over the wall as the right way
> to do things, but there's really no infrastructure to facilitate even the
> simple peanut butter, everybody gets the same number of VFs, interface
> that max_vfs provides. I think the existence of this bug is probably a
> good indication that the sysfs interface has not really been adopted yet.
> Thanks,
Alright, I'll go with that reasoning.
Acked-by: Greg Rose <gregory.v.rose at intel.com>
>
> Alex
More information about the Intel-wired-lan
mailing list