diff options
| author | mst@redhat.com <mst@redhat.com> | 2011-02-01 22:13:42 +0200 |
|---|---|---|
| committer | Anthony Liguori <aliguori@us.ibm.com> | 2011-02-01 16:50:44 -0600 |
| commit | 5430a28fe452907c9e1b2097e073bc1ea4b29f39 (patch) | |
| tree | 066db9fdbac5a5ddca922b5d282161e9b6141b0e /hw/vhost.c | |
| parent | f157ed202e51dc2492b201dc34ed28e89c973fb7 (diff) | |
vhost: force vhost off for non-MSI guests
When MSI is off, each interrupt needs to be bounced through the io
thread when it's set/cleared, so vhost-net causes more context switches and
higher CPU utilization than userspace virtio which handles networking in
the same thread.
We'll need to fix this by adding level irq support in kvm irqfd,
for now disable vhost-net in these configurations.
Added a vhostforce flag to force vhost-net back on.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Diffstat (limited to 'hw/vhost.c')
| -rw-r--r-- | hw/vhost.c | 10 |
1 files changed, 9 insertions, 1 deletions
diff --git a/hw/vhost.c b/hw/vhost.c index 6082da287e..38cc3b365b 100644 --- a/hw/vhost.c +++ b/hw/vhost.c @@ -581,7 +581,7 @@ static void vhost_virtqueue_cleanup(struct vhost_dev *dev, 0, virtio_queue_get_desc_size(vdev, idx)); } -int vhost_dev_init(struct vhost_dev *hdev, int devfd) +int vhost_dev_init(struct vhost_dev *hdev, int devfd, bool force) { uint64_t features; int r; @@ -613,6 +613,7 @@ int vhost_dev_init(struct vhost_dev *hdev, int devfd) hdev->log_enabled = false; hdev->started = false; cpu_register_phys_memory_client(&hdev->client); + hdev->force = force; return 0; fail: r = -errno; @@ -627,6 +628,13 @@ void vhost_dev_cleanup(struct vhost_dev *hdev) close(hdev->control); } +bool vhost_dev_query(struct vhost_dev *hdev, VirtIODevice *vdev) +{ + return !vdev->binding->query_guest_notifiers || + vdev->binding->query_guest_notifiers(vdev->binding_opaque) || + hdev->force; +} + int vhost_dev_start(struct vhost_dev *hdev, VirtIODevice *vdev) { int i, r; |