On Thu, Aug 4, 2022 at 6:44 AM Jason Wang <jasowang@xxxxxxxxxx> wrote:
>
>
> 在 2022/8/4 01:18, Eugenio Pérez 写道:
> > Isolate control virtqueue in its own group, allowing to intercept control
> > commands but letting dataplane run totally passthrough to the guest.
> >
> > Signed-off-by: Eugenio Pérez <eperezma@xxxxxxxxxx>
> > ---
> > v3:
> > * Make asid related queries print a warning instead of returning an
> > error and stop the start of qemu.
> > ---
> > hw/virtio/vhost-vdpa.c | 3 +-
> > net/vhost-vdpa.c | 144 +++++++++++++++++++++++++++++++++++++++--
> > 2 files changed, 142 insertions(+), 5 deletions(-)
> >
> > diff --git a/hw/virtio/vhost-vdpa.c b/hw/virtio/vhost-vdpa.c
> > index 131100841c..a4cb68862b 100644
> > --- a/hw/virtio/vhost-vdpa.c
> > +++ b/hw/virtio/vhost-vdpa.c
> > @@ -674,7 +674,8 @@ static int vhost_vdpa_set_backend_cap(struct vhost_dev *dev)
> > {
> > uint64_t features;
> > uint64_t f = 0x1ULL << VHOST_BACKEND_F_IOTLB_MSG_V2 |
> > - 0x1ULL << VHOST_BACKEND_F_IOTLB_BATCH;
> > + 0x1ULL << VHOST_BACKEND_F_IOTLB_BATCH |
> > + 0x1ULL << VHOST_BACKEND_F_IOTLB_ASID;
> > int r;
> >
> > if (vhost_vdpa_call(dev, VHOST_GET_BACKEND_FEATURES, &features)) {
> > diff --git a/net/vhost-vdpa.c b/net/vhost-vdpa.c
> > index e3b65ed546..5f39f0edb5 100644
> > --- a/net/vhost-vdpa.c
> > +++ b/net/vhost-vdpa.c
> > @@ -37,6 +37,9 @@ typedef struct VhostVDPAState {
> > /* Control commands shadow buffers */
> > void *cvq_cmd_out_buffer, *cvq_cmd_in_buffer;
> >
> > + /* Number of address spaces supported by the device */
> > + unsigned address_space_num;
> > +
> > /* The device always have SVQ enabled */
> > bool always_svq;
> > bool started;
> > @@ -100,6 +103,8 @@ static const uint64_t vdpa_svq_device_features =
> > BIT_ULL(VIRTIO_NET_F_RSC_EXT) |
> > BIT_ULL(VIRTIO_NET_F_STANDBY);
> >
> > +#define VHOST_VDPA_NET_CVQ_ASID 1
> > +
> > VHostNetState *vhost_vdpa_get_vhost_net(NetClientState *nc)
> > {
> > VhostVDPAState *s = DO_UPCAST(VhostVDPAState, nc, nc);
> > @@ -224,6 +229,101 @@ static NetClientInfo net_vhost_vdpa_info = {
> > .check_peer_type = vhost_vdpa_check_peer_type,
> > };
> >
> > +static void vhost_vdpa_get_vring_group(int device_fd,
> > + struct vhost_vring_state *state)
> > +{
> > + int r = ioctl(device_fd, VHOST_VDPA_GET_VRING_GROUP, state);
> > + if (unlikely(r < 0)) {
> > + /*
> > + * Assume all groups are 0, the consequences are the same and we will
> > + * not abort device creation
> > + */
> > + state->num = 0;
> > + }
> > +}
> > +
> > +/**
> > + * Check if all the virtqueues of the virtio device are in a different vq than
> > + * the last vq. VQ group of last group passed in cvq_group.
> > + */
> > +static bool vhost_vdpa_cvq_group_is_independent(struct vhost_vdpa *v,
> > + struct vhost_vring_state cvq_group)
> > +{
> > + struct vhost_dev *dev = v->dev;
> > +
> > + for (int i = 0; i < (dev->vq_index_end - 1); ++i) {
> > + struct vhost_vring_state vq_group = {
> > + .index = i,
> > + };
> > +
> > + vhost_vdpa_get_vring_group(v->device_fd, &vq_group);
> > + if (unlikely(vq_group.num == cvq_group.num)) {
> > + warn_report("CVQ %u group is the same as VQ %u one (%u)",
> > + cvq_group.index, vq_group.index, cvq_group.num);
>
>
> I don't get why we need warn here.
>
Is not mandatory, but vhost will set a migration blocker and there is
no way to trace it back to the cause without it.
>
> > + return false;
> > + }
> > + }
> > +
> > + return true;
> > +}
> > +
> > +static int vhost_vdpa_set_address_space_id(struct vhost_vdpa *v,
> > + unsigned vq_group,
> > + unsigned asid_num)
> > +{
> > + struct vhost_vring_state asid = {
> > + .index = vq_group,
> > + .num = asid_num,
> > + };
> > + int ret;
> > +
> > + ret = ioctl(v->device_fd, VHOST_VDPA_SET_GROUP_ASID, &asid);
> > + if (unlikely(ret < 0)) {
> > + warn_report("Can't set vq group %u asid %u, errno=%d (%s)",
> > + asid.index, asid.num, errno, g_strerror(errno));
> > + }
> > + return ret;
> > +}
> > +
> > +static void vhost_vdpa_net_prepare(NetClientState *nc)
> > +{
> > + VhostVDPAState *s = DO_UPCAST(VhostVDPAState, nc, nc);
> > + struct vhost_vdpa *v = &s->vhost_vdpa;
> > + struct vhost_dev *dev = v->dev;
> > + struct vhost_vring_state cvq_group = {
> > + .index = v->dev->vq_index_end - 1,
> > + };
> > + int r;
> > +
> > + assert(nc->info->type == NET_CLIENT_DRIVER_VHOST_VDPA);
> > +
> > + if (dev->nvqs != 1 || dev->vq_index + dev->nvqs != dev->vq_index_end) {
> > + /* Only interested in CVQ */
> > + return;
> > + }
> > +
> > + if (s->always_svq) {
> > + /* SVQ is already enabled */
> > + return;
> > + }
> > +
> > + if (s->address_space_num < 2) {
> > + v->shadow_vqs_enabled = false;
> > + return;
> > + }
> > +
> > + vhost_vdpa_get_vring_group(v->device_fd, &cvq_group);
> > + if (!vhost_vdpa_cvq_group_is_independent(v, cvq_group)) {
>
>
> If there's no other caller of vhost_vdpa_cvq_group_is_independent(), I'd
> suggest to unitfy them into a single helper.
>
Sure, that can be done.
Thanks!
> (Btw, the name of the function is kind of too long).
>
> Thanks
>
>
> > + v->shadow_vqs_enabled = false;
> > + return;
> > + }
> > +
> > + r = vhost_vdpa_set_address_space_id(v, cvq_group.num,
> > + VHOST_VDPA_NET_CVQ_ASID);
> > + v->shadow_vqs_enabled = r == 0;
> > + s->vhost_vdpa.address_space_id = r == 0 ? 1 : 0;
> > +}
> > +
> > static void vhost_vdpa_cvq_unmap_buf(struct vhost_vdpa *v, void *addr)
> > {
> > VhostIOVATree *tree = v->iova_tree;
> > @@ -431,6 +531,7 @@ static NetClientInfo net_vhost_vdpa_cvq_info = {
> > .type = NET_CLIENT_DRIVER_VHOST_VDPA,
> > .size = sizeof(VhostVDPAState),
> > .receive = vhost_vdpa_receive,
> > + .prepare = vhost_vdpa_net_prepare,
> > .load = vhost_vdpa_net_load,
> > .cleanup = vhost_vdpa_cleanup,
> > .has_vnet_hdr = vhost_vdpa_has_vnet_hdr,
> > @@ -541,12 +642,38 @@ static const VhostShadowVirtqueueOps vhost_vdpa_net_svq_ops = {
> > .avail_handler = vhost_vdpa_net_handle_ctrl_avail,
> > };
> >
> > +static uint32_t vhost_vdpa_get_as_num(int vdpa_device_fd)
> > +{
> > + uint64_t features;
> > + unsigned num_as;
> > + int r;
> > +
> > + r = ioctl(vdpa_device_fd, VHOST_GET_BACKEND_FEATURES, &features);
> > + if (unlikely(r < 0)) {
> > + warn_report("Cannot get backend features");
> > + return 1;
> > + }
> > +
> > + if (!(features & BIT_ULL(VHOST_BACKEND_F_IOTLB_ASID))) {
> > + return 1;
> > + }
> > +
> > + r = ioctl(vdpa_device_fd, VHOST_VDPA_GET_AS_NUM, &num_as);
> > + if (unlikely(r < 0)) {
> > + warn_report("Cannot retrieve number of supported ASs");
> > + return 1;
> > + }
> > +
> > + return num_as;
> > +}
> > +
> > static NetClientState *net_vhost_vdpa_init(NetClientState *peer,
> > const char *device,
> > const char *name,
> > int vdpa_device_fd,
> > int queue_pair_index,
> > int nvqs,
> > + unsigned nas,
> > bool is_datapath,
> > bool svq,
> > VhostIOVATree *iova_tree)
> > @@ -565,6 +692,7 @@ static NetClientState *net_vhost_vdpa_init(NetClientState *peer,
> > snprintf(nc->info_str, sizeof(nc->info_str), TYPE_VHOST_VDPA);
> > s = DO_UPCAST(VhostVDPAState, nc, nc);
> >
> > + s->address_space_num = nas;
> > s->vhost_vdpa.device_fd = vdpa_device_fd;
> > s->vhost_vdpa.index = queue_pair_index;
> > s->always_svq = svq;
> > @@ -650,6 +778,8 @@ int net_init_vhost_vdpa(const Netdev *netdev, const char *name,
> > g_autoptr(VhostIOVATree) iova_tree = NULL;
> > NetClientState *nc;
> > int queue_pairs, r, i = 0, has_cvq = 0;
> > + unsigned num_as = 1;
> > + bool svq_cvq;
> >
> > assert(netdev->type == NET_CLIENT_DRIVER_VHOST_VDPA);
> > opts = &netdev->u.vhost_vdpa;
> > @@ -674,7 +804,13 @@ int net_init_vhost_vdpa(const Netdev *netdev, const char *name,
> > goto err;
> > }
> >
> > - if (opts->x_svq) {
> > + svq_cvq = opts->x_svq;
> > + if (has_cvq && !opts->x_svq) {
> > + num_as = vhost_vdpa_get_as_num(vdpa_device_fd);
> > + svq_cvq = num_as > 1;
> > + }
> > +
> > + if (opts->x_svq || svq_cvq) {
> > struct vhost_vdpa_iova_range iova_range;
> >
> > uint64_t invalid_dev_features =
> > @@ -697,15 +833,15 @@ int net_init_vhost_vdpa(const Netdev *netdev, const char *name,
> >
> > for (i = 0; i < queue_pairs; i++) {
> > ncs[i] = net_vhost_vdpa_init(peer, TYPE_VHOST_VDPA, name,
> > - vdpa_device_fd, i, 2, true, opts->x_svq,
> > - iova_tree);
> > + vdpa_device_fd, i, 2, num_as, true,
> > + opts->x_svq, iova_tree);
> > if (!ncs[i])
> > goto err;
> > }
> >
> > if (has_cvq) {
> > nc = net_vhost_vdpa_init(peer, TYPE_VHOST_VDPA, name,
> > - vdpa_device_fd, i, 1, false,
> > + vdpa_device_fd, i, 1, num_as, false,
> > opts->x_svq, iova_tree);
> > if (!nc)
> > goto err;
>
