[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH] vfio/pci: Don't remove irqchip notifier if not registered
From: |
Peter Xu |
Subject: |
Re: [PATCH] vfio/pci: Don't remove irqchip notifier if not registered |
Date: |
Tue, 7 Jan 2020 08:33:32 -0500 |
On Mon, Jan 06, 2020 at 03:13:24PM -0700, Alex Williamson wrote:
> On Tue, 31 Dec 2019 08:39:15 -0500
> Peter Xu <address@hidden> wrote:
>
> > The kvm irqchip notifier is only registered if the device supports
> > INTx, however it's unconditionally removed. If the assigned device
> > does not support INTx, this will cause QEMU to crash when unplugging
> > the device from the system. Change it to conditionally remove the
> > notifier only if the notify hook is setup.
> >
> > CC: Eduardo Habkost <address@hidden>
> > CC: David Gibson <address@hidden>
> > CC: Alex Williamson <address@hidden>
> > Reported-by: address@hidden
> > Fixes: c5478fea27 ("vfio/pci: Respond to KVM irqchip change notifier",
> > 2019-11-26)
> > Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1782678
> > Signed-off-by: Peter Xu <address@hidden>
> > ---
> > hw/vfio/pci.c | 4 +++-
> > 1 file changed, 3 insertions(+), 1 deletion(-)
> >
> > diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
> > index 2d40b396f2..337a173ce7 100644
> > --- a/hw/vfio/pci.c
> > +++ b/hw/vfio/pci.c
> > @@ -3076,7 +3076,9 @@ static void vfio_exitfn(PCIDevice *pdev)
> > vfio_unregister_req_notifier(vdev);
> > vfio_unregister_err_notifier(vdev);
> > pci_device_set_intx_routing_notifier(&vdev->pdev, NULL);
> > - kvm_irqchip_remove_change_notifier(&vdev->irqchip_change_notifier);
> > + if (vdev->irqchip_change_notifier.notify) {
> > + kvm_irqchip_remove_change_notifier(&vdev->irqchip_change_notifier);
> > + }
> > vfio_disable_interrupts(vdev);
> > if (vdev->intx.mmap_timer) {
> > timer_free(vdev->intx.mmap_timer);
>
>
> Thanks, Peter! Sent a pull request with David and Greg's R-b, stable
> tag, and debug credit to Eduardo as I think he spotted the issue first.
> Thanks,
Yes thanks! I wished Eduardo had even updated the bz when he debugged
so I won't be needinfo-ed and did it twice without notice :)
--
Peter Xu