Re: [Qemu-devel] [PATCH v2] vhost: Fix aborting if KVM does not support eventfds

[Stefan Hajnoczi]

On Wed, Nov 18, 2015 at 11:25:32PM +0200, Michael S. Tsirkin wrote:
> On Tue, Nov 17, 2015 at 10:17:37AM +0300, Pavel Fedin wrote:
> > If you happen to have a stock kernel of old version, like 3.x,
> 
> On ARM
> 
> > and you
> > attempt to enable vhost by setting vhost=on, qemu aborts with error:
> > kvm_mem_ioeventfd_add: error adding ioeventfd: Function not implemented
> > 
> > This patch adds capability check, so that vhost gets disabled instead. A
> > warning is displayed, explaining the reason:
> > 
> > 2015-11-13T08:43:51.146802Z qemu-system-aarch64: KVM does not support 
> > eventfd binding
> > 2015-11-13T08:43:51.146915Z qemu-system-aarch64: unable to start vhost net: 
> > 38: falling back on userspace virtio
> > 
> > This problem can be observed with libvirt, which checks for /dev/vhost-net
> > availability and just inserts "vhost=on" automatically in this case.
> > 
> > Signed-off-by: Pavel Fedin <address@hidden>
> > ---
> > v1 => v2:
> > - Removed "MMIO" from warning message, because it applies not only to MMIO
> > - Add note about warning to the commit message
> > ---
> >  hw/virtio/vhost.c | 6 ++++++
> >  1 file changed, 6 insertions(+)
> > 
> > diff --git a/hw/virtio/vhost.c b/hw/virtio/vhost.c
> > index 1794f0d..50b8171 100644
> > --- a/hw/virtio/vhost.c
> > +++ b/hw/virtio/vhost.c
> > @@ -24,6 +24,7 @@
> >  #include "hw/virtio/virtio-bus.h"
> >  #include "hw/virtio/virtio-access.h"
> >  #include "migration/migration.h"
> > +#include "sysemu/kvm.h"
> >  
> >  static struct vhost_log *vhost_log;
> >  static struct vhost_log *vhost_log_shm;
> > @@ -1083,6 +1084,11 @@ int vhost_dev_enable_notifiers(struct vhost_dev 
> > *hdev, VirtIODevice *vdev)
> >          r = -ENOSYS;
> >          goto fail;
> >      }
> > +    if (!kvm_eventfds_enabled()) {
> > +        error_report("KVM does not support eventfd binding");
> > +        r = -ENOSYS;
> > +        goto fail;
> > +    }
> >  
> >      for (i = 0; i < hdev->nvqs; ++i) {
> >          r = k->set_host_notifier(qbus->parent, hdev->vq_index + i, true);
> 
> That's a wrong thing for vhost to do - we don't want it
> poking at kvm because it does not work with either kvm or
> eventfd.  It uses host notifiers, there should be an
> API to check for that.
> 
> Or better yet, actually implement host notifiers
> in userspace - should not be that hard to do, just
> do write into eventfd.

I sent this RFC patch a while back:
https://lists.nongnu.org/archive/html/qemu-devel/2015-07/msg04710.html

My intent was to write a qtest for a vhost device but I didn't complete
the work.  Anyway, this patch might be useful.

Stefan

signature.asc
Description: PGP signature