Re: [PULL v3 12/55] virtio-pci: add support for configure interrupt

[Volker Rümelin]

Am 09.01.22 um 17:11 schrieb Michael S. Tsirkin:
> On Sun, Jan 09, 2022 at 07:17:30AM +0100, Volker Rümelin wrote:
> > Hi,
> >
> > > From: Cindy Lu <lulu@redhat.com>
> > >
> > > Add support for configure interrupt, The process is used kvm_irqfd_assign
> > > to set the gsi to kernel. When the configure notifier was signal by
> > > host, qemu will inject a msix interrupt to guest
> > >
> > > Signed-off-by: Cindy Lu <lulu@redhat.com>
> > > Message-Id: <20211104164827.21911-11-lulu@redhat.com>
> > > Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
> > > Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> > > ---
> > >    hw/virtio/virtio-pci.h |  4 +-
> > >    hw/virtio/virtio-pci.c | 92 ++++++++++++++++++++++++++++++++++++------
> > >    2 files changed, 83 insertions(+), 13 deletions(-)
> > Since this commit I see the following warnings.
> >
> > With -drive
> > if=virtio,id=disk1,file=/srv/cdimg/Linux/images/opensuse.qcow2,discard=unmap
> >
> > qemu-system-x86_64: virtio-blk failed to set guest notifier (-16), ensure
> > -accel kvm is set.
> > qemu-system-x86_64: virtio_bus_start_ioeventfd: failed. Fallback to
> > userspace (slower).
> >
> > With libvirt
> >      <controller type='pci' index='1' model='pcie-root-port'>
> >        <address type='pci' domain='0x0000' bus='0x00' slot='0x1c'
> > function='0'
> >         multifunction='on'/>
> >      </controller>
> >      <controller type='pci' index='2' model='pcie-root-port'>
> >        <address type='pci' domain='0x0000' bus='0x00' slot='0x1c'
> > function='1'/>
> >      </controller>
> >      <controller type='scsi' index='0' model='virtio-scsi'>
> >        <driver queues='4'/>
> >        <address type='pci' domain='0x0000' bus='0x01' slot='0x00'
> > function='0'/>
> >      </controller>
> >      <disk type='block' device='disk'>
> >        <driver name='qemu' type='raw' cache='none' discard='unmap'
> > io='io_uring'/>
> >        <source dev='/dev/vgtmp/lnxpowerm1'/>
> >        <target dev='sda' bus='scsi'/>
> >        <address type='drive' controller='0' bus='0' unit='0'/>
> >        <boot order='1'/>
> >      </disk>
> >
> > 2022-01-08T17:45:26.911491Z qemu-system-x86_64: virtio-scsi: Failed to set
> > guest notifiers (-16), ensure -accel kvm is set.
> > 2022-01-08T17:45:26.911505Z qemu-system-x86_64: virtio_bus_start_ioeventfd:
> > failed. Fallback to userspace (slower).
> >
> > The messages appear around the time the Linux guest initializes the drivers.
> >
> > With best regards,
> > Volker
> I guess it's a host that has an oldish kernel?

It's an openSUSE 5.3.18 frankenstein kernel.

> Does the following help?

No.

> diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
> index 98fb5493ae..b77cd69f97 100644
> --- a/hw/virtio/virtio-pci.c
> +++ b/hw/virtio/virtio-pci.c
> @@ -1130,15 +1130,15 @@ static int virtio_pci_set_guest_notifiers(DeviceState 
> *d, int nvqs, bool assign)
>               proxy->vector_irqfd =
>                   g_malloc0(sizeof(*proxy->vector_irqfd) *
>                             msix_nr_vectors_allocated(&proxy->pci_dev));
> +            r = kvm_virtio_pci_vector_config_use(proxy);
> +            if (r < 0) {
> +                goto config_error;
> +            }
>               r = kvm_virtio_pci_vector_use(proxy, nvqs);
>               if (r < 0) {
>                   goto config_assign_error;
>               }
>           }
> -        r = kvm_virtio_pci_vector_config_use(proxy);
> -        if (r < 0) {
> -            goto config_error;
> -        }
>           r = msix_set_vector_notifiers(&proxy->pci_dev, 
> virtio_pci_vector_unmask,
>                                         virtio_pci_vector_mask,
>                                         virtio_pci_vector_poll);

With and without this patch msix_set_vector_notifiers() returns -16.

> @@ -1155,7 +1155,9 @@ notifiers_error:
>           kvm_virtio_pci_vector_release(proxy, nvqs);
>       }
>   config_error:
> -    kvm_virtio_pci_vector_config_release(proxy);
> +    if (with_irqfd) {
> +        kvm_virtio_pci_vector_config_release(proxy);
> +    }
>   config_assign_error:
>       virtio_pci_set_guest_notifier(d, VIRTIO_CONFIG_IRQ_IDX, !assign,
>                                     with_irqfd);