github/acrn-hypervisor
1
0
Fork 0
mirror of https://github.com/projectacrn/acrn-hypervisor.git synced 2026-06-08 18:14:53 +00:00
Code Issues Projects Releases Wiki Activity

dm: virtio poll mode support for RT

Browse Source 
Device trap has great impact on latency of real time (RT) tasks.
This patch provide a virtio poll mode to avoid trap.

According to the virtio spec, backend devices can declare the
notification is not needed so that frontend will never trap.
This means the backends make commitment to the frontends they have a
poll mechanism which don’t need any frontends notification.

This patch uses a periodic timer to give backends pseudo notifications
so that drive them processing data in their virtqueues. People should
choose a appropriate notification peroid interval to use this poll
mode. Too big interval may cause virtqueue processing latency while
too small interval may cause high SOS CPU usage. The suggested interval
is between 100us to 1ms.

The poll mode is not enabled by default and traditional trap
notification mode will be used. To use poll mode for RT with interval
1ms. You can add following acrn-dm parameter.

	--virtio_poll 1000000

Tracked-On: #1956
Signed-off-by: Jie Deng <jie.deng@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
This commit is contained in:
Jie Deng
2018-12-04 00:33:46 +08:00
committed by wenlingz
parent 7cc8566d37
commit b261e74dd5
15 changed files with 222 additions and 40 deletions
Download Patch File Download Diff File Expand all files Collapse all files

57
devicemodel/hw/pci/virtio/virtio_net.c
View File

@@ -555,7 +555,7 @@ virtio_net_tx_thread(void *param)
for (;;) {
/* note - tx mutex is locked here */
while (net->resetting || !vq_has_descs(vq)) {
vq->used->flags &= ~ACRN_VRING_USED_F_NO_NOTIFY;
vq_clear_used_ring_flags(&net->base, vq);
/* memory barrier */
mb();
if (!net->resetting && vq_has_descs(vq))
@@ -724,7 +724,7 @@ virtio_net_init(struct vmctx *ctx, struct pci_vdev *dev, char *opts)
char nstr[80];
char tname[MAXCOMLEN + 1];
struct virtio_net *net;
char *devname;
char *devname = NULL;
char *vtopts;
char *opt;
int mac_provided;
@@ -751,25 +751,10 @@ virtio_net_init(struct vmctx *ctx, struct pci_vdev *dev, char *opts)
DPRINTF(("virtio_net: pthread_mutex_init failed with "
"error %d!\n", rc));
virtio_linkup(&net->base, &virtio_net_ops, net, dev, net->queues);
net->base.mtx = &net->mtx;
net->base.device_caps = VIRTIO_NET_S_HOSTCAPS;
net->queues[VIRTIO_NET_RXQ].qsize = VIRTIO_NET_RINGSZ;
net->queues[VIRTIO_NET_RXQ].notify = virtio_net_ping_rxq;
net->queues[VIRTIO_NET_TXQ].qsize = VIRTIO_NET_RINGSZ;
net->queues[VIRTIO_NET_TXQ].notify = virtio_net_ping_txq;
#ifdef notyet
net->queues[VIRTIO_NET_CTLQ].qsize = VIRTIO_NET_RINGSZ;
net->queues[VIRTIO_NET_CTLQ].notify = virtio_net_ping_ctlq;
#endif
/*
* Attempt to open the tap device and read the MAC address
* if specified
* Read the MAC address if specified
*/
mac_provided = 0;
net->tapfd = -1;
net->vhost_net = NULL;
if (opts != NULL) {
int err;
@@ -795,14 +780,38 @@ virtio_net_init(struct vmctx *ctx, struct pci_vdev *dev, char *opts)
mac_provided = 1;
}
}
if (strncmp(devname, "tap", 3) == 0 ||
strncmp(devname, "vmnet", 5) == 0)
virtio_net_tap_setup(net, devname);
free(devname);
}
virtio_linkup(&net->base, &virtio_net_ops, net, dev, net->queues,
net->use_vhost ? BACKEND_VHOST : BACKEND_VBSU);
net->base.mtx = &net->mtx;
net->base.device_caps = VIRTIO_NET_S_HOSTCAPS;
net->queues[VIRTIO_NET_RXQ].qsize = VIRTIO_NET_RINGSZ;
net->queues[VIRTIO_NET_RXQ].notify = virtio_net_ping_rxq;
net->queues[VIRTIO_NET_TXQ].qsize = VIRTIO_NET_RINGSZ;
net->queues[VIRTIO_NET_TXQ].notify = virtio_net_ping_txq;
#ifdef notyet
net->queues[VIRTIO_NET_CTLQ].qsize = VIRTIO_NET_RINGSZ;
net->queues[VIRTIO_NET_CTLQ].notify = virtio_net_ping_ctlq;
#endif
/*
* Attempt to open the tap device
*/
net->tapfd = -1;
if (!devname) {
WPRINTF(("virtio_net: devname NULL\n"));
return -1;
}
if (strncmp(devname, "tap", 3) == 0 ||
strncmp(devname, "vmnet", 5) == 0)
virtio_net_tap_setup(net, devname);
free(devname);
/*
* The default MAC address is the standard NetApp OUI of 00-a0-98,
* followed by an MD5 of the PCI slot/func number and dev name