Commit 5430a28f authored by mst@redhat.com's avatar mst@redhat.com Committed by Anthony Liguori
Browse files

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: default avatarMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: default avatarAnthony Liguori <aliguori@us.ibm.com>
parent f157ed20
......@@ -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;
......
......@@ -38,10 +38,12 @@ struct vhost_dev {
bool log_enabled;
vhost_log_chunk_t *log;
unsigned long long log_size;
bool force;
};
int vhost_dev_init(struct vhost_dev *hdev, int devfd);
int vhost_dev_init(struct vhost_dev *hdev, int devfd, bool force);
void vhost_dev_cleanup(struct vhost_dev *hdev);
bool vhost_dev_query(struct vhost_dev *hdev, VirtIODevice *vdev);
int vhost_dev_start(struct vhost_dev *hdev, VirtIODevice *vdev);
void vhost_dev_stop(struct vhost_dev *hdev, VirtIODevice *vdev);
......
......@@ -81,7 +81,8 @@ static int vhost_net_get_fd(VLANClientState *backend)
}
}
struct vhost_net *vhost_net_init(VLANClientState *backend, int devfd)
struct vhost_net *vhost_net_init(VLANClientState *backend, int devfd,
bool force)
{
int r;
struct vhost_net *net = qemu_malloc(sizeof *net);
......@@ -98,7 +99,7 @@ struct vhost_net *vhost_net_init(VLANClientState *backend, int devfd)
(1 << VHOST_NET_F_VIRTIO_NET_HDR);
net->backend = r;
r = vhost_dev_init(&net->dev, devfd);
r = vhost_dev_init(&net->dev, devfd, force);
if (r < 0) {
goto fail;
}
......@@ -121,6 +122,11 @@ fail:
return NULL;
}
bool vhost_net_query(VHostNetState *net, VirtIODevice *dev)
{
return vhost_dev_query(&net->dev, dev);
}
int vhost_net_start(struct vhost_net *net,
VirtIODevice *dev)
{
......@@ -188,15 +194,21 @@ void vhost_net_cleanup(struct vhost_net *net)
qemu_free(net);
}
#else
struct vhost_net *vhost_net_init(VLANClientState *backend, int devfd)
struct vhost_net *vhost_net_init(VLANClientState *backend, int devfd,
bool force)
{
return NULL;
}
bool vhost_net_query(VHostNetState *net, VirtIODevice *dev)
{
return NULL;
return false;
}
int vhost_net_start(struct vhost_net *net,
VirtIODevice *dev)
{
return -ENOSYS;
return -ENOSYS;
}
void vhost_net_stop(struct vhost_net *net,
VirtIODevice *dev)
......@@ -209,7 +221,7 @@ void vhost_net_cleanup(struct vhost_net *net)
unsigned vhost_net_get_features(struct vhost_net *net, unsigned features)
{
return features;
return features;
}
void vhost_net_ack_features(struct vhost_net *net, unsigned features)
{
......
......@@ -6,8 +6,9 @@
struct vhost_net;
typedef struct vhost_net VHostNetState;
VHostNetState *vhost_net_init(VLANClientState *backend, int devfd);
VHostNetState *vhost_net_init(VLANClientState *backend, int devfd, bool force);
bool vhost_net_query(VHostNetState *net, VirtIODevice *dev);
int vhost_net_start(VHostNetState *net, VirtIODevice *dev);
void vhost_net_stop(VHostNetState *net, VirtIODevice *dev);
......
......@@ -119,7 +119,11 @@ static void virtio_net_vhost_status(VirtIONet *n, uint8_t status)
return;
}
if (!n->vhost_started) {
int r = vhost_net_start(tap_get_vhost_net(n->nic->nc.peer), &n->vdev);
int r;
if (!vhost_net_query(tap_get_vhost_net(n->nic->nc.peer), &n->vdev)) {
return;
}
r = vhost_net_start(tap_get_vhost_net(n->nic->nc.peer), &n->vdev);
if (r < 0) {
error_report("unable to start vhost net: %d: "
"falling back on userspace virtio", -r);
......
......@@ -595,6 +595,12 @@ static int virtio_pci_set_guest_notifier(void *opaque, int n, bool assign)
return 0;
}
static bool virtio_pci_query_guest_notifiers(void *opaque)
{
VirtIOPCIProxy *proxy = opaque;
return msix_enabled(&proxy->pci_dev);
}
static int virtio_pci_set_guest_notifiers(void *opaque, bool assign)
{
VirtIOPCIProxy *proxy = opaque;
......@@ -658,6 +664,7 @@ static const VirtIOBindings virtio_pci_bindings = {
.save_queue = virtio_pci_save_queue,
.load_queue = virtio_pci_load_queue,
.get_features = virtio_pci_get_features,
.query_guest_notifiers = virtio_pci_query_guest_notifiers,
.set_host_notifier = virtio_pci_set_host_notifier,
.set_guest_notifiers = virtio_pci_set_guest_notifiers,
.vmstate_change = virtio_pci_vmstate_change,
......
......@@ -93,6 +93,7 @@ typedef struct {
int (*load_config)(void * opaque, QEMUFile *f);
int (*load_queue)(void * opaque, int n, QEMUFile *f);
unsigned (*get_features)(void * opaque);
bool (*query_guest_notifiers)(void * opaque);
int (*set_guest_notifiers)(void * opaque, bool assigned);
int (*set_host_notifier)(void * opaque, int n, bool assigned);
void (*vmstate_change)(void * opaque, bool running);
......
......@@ -491,8 +491,10 @@ int net_init_tap(QemuOpts *opts, Monitor *mon, const char *name, VLANState *vlan
}
}
if (qemu_opt_get_bool(opts, "vhost", !!qemu_opt_get(opts, "vhostfd"))) {
if (qemu_opt_get_bool(opts, "vhost", !!qemu_opt_get(opts, "vhostfd") ||
qemu_opt_get_bool(opts, "vhostforce", false))) {
int vhostfd, r;
bool force = qemu_opt_get_bool(opts, "vhostforce", false);
if (qemu_opt_get(opts, "vhostfd")) {
r = net_handle_fd_param(mon, qemu_opt_get(opts, "vhostfd"));
if (r == -1) {
......@@ -502,7 +504,7 @@ int net_init_tap(QemuOpts *opts, Monitor *mon, const char *name, VLANState *vlan
} else {
vhostfd = -1;
}
s->vhost_net = vhost_net_init(&s->nc, vhostfd);
s->vhost_net = vhost_net_init(&s->nc, vhostfd, force);
if (!s->vhost_net) {
error_report("vhost-net requested but could not be initialized");
return -1;
......
......@@ -1050,7 +1050,7 @@ DEF("net", HAS_ARG, QEMU_OPTION_net,
"-net tap[,vlan=n][,name=str],ifname=name\n"
" connect the host TAP network interface to VLAN 'n'\n"
#else
"-net tap[,vlan=n][,name=str][,fd=h][,ifname=name][,script=file][,downscript=dfile][,sndbuf=nbytes][,vnet_hdr=on|off][,vhost=on|off][,vhostfd=h]\n"
"-net tap[,vlan=n][,name=str][,fd=h][,ifname=name][,script=file][,downscript=dfile][,sndbuf=nbytes][,vnet_hdr=on|off][,vhost=on|off][,vhostfd=h][,vhostforce=on|off]\n"
" connect the host TAP network interface to VLAN 'n' and use the\n"
" network scripts 'file' (default=" DEFAULT_NETWORK_SCRIPT ")\n"
" and 'dfile' (default=" DEFAULT_NETWORK_DOWN_SCRIPT ")\n"
......@@ -1061,6 +1061,8 @@ DEF("net", HAS_ARG, QEMU_OPTION_net,
" use vnet_hdr=off to avoid enabling the IFF_VNET_HDR tap flag\n"
" use vnet_hdr=on to make the lack of IFF_VNET_HDR support an error condition\n"
" use vhost=on to enable experimental in kernel accelerator\n"
" (only has effect for virtio guests which use MSIX)\n"
" use vhostforce=on to force vhost on for non-MSIX virtio guests\n"
" use 'vhostfd=h' to connect to an already opened vhost net device\n"
#endif
"-net socket[,vlan=n][,name=str][,fd=h][,listen=[host]:port][,connect=host:port]\n"
......
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment