From: Cindy Lu <lulu@redhat.com>
To: Jason Wang <jasowang@redhat.com>
Cc: cohuck@redhat.com, Michael Tsirkin <mst@redhat.com>,
mhabets@solarflare.com, qemu-devel@nongnu.org, hanand@xilinx.com,
rob.miller@broadcom.com, saugatm@xilinx.com, armbru@redhat.com,
hch@infradead.org, Eugenio Perez Martin <eperezma@redhat.com>,
jgg@mellanox.com, shahafs@mellanox.com, kevin.tian@intel.com,
parav@mellanox.com, vmireyno@marvell.com, "Liang,
Cunming" <cunming.liang@intel.com>,
gdawar@xilinx.com, jiri@mellanox.com, xiao.w.wang@intel.com,
stefanha@redhat.com, "Wang, Zhihong" <zhihong.wang@intel.com>,
Ariel Adam <aadam@redhat.com>,
rdunlap@infradead.org, maxime.coquelin@redhat.com, "Zhu,
Lingshan" <lingshan.zhu@intel.com>
Subject: Re: [RFC v1 3/4] vhost-vdpa: implement vhost-vdpa backend
Date: Tue, 21 Apr 2020 17:12:41 +0800 [thread overview]
Message-ID: <CACLfguX+SB0eQfrv0DwEnNpnX3LjPH+vdt_QHe+dwQGunDtiRA@mail.gmail.com> (raw)
In-Reply-To: <a6310935-83e3-903b-9221-c6ead28dffac@redhat.com>
[-- Attachment #1: Type: text/plain, Size: 25253 bytes --]
On Tue, Apr 21, 2020 at 11:57 AM Jason Wang <jasowang@redhat.com> wrote:
>
> On 2020/4/20 下午5:32, Cindy Lu wrote:
> > Currently we have 2 types of vhost backends in QEMU: vhost kernel and
> > vhost-user. The above patch provides a generic device for vDPA purpose,
> > this vDPA device exposes to user space a non-vendor-specific
> configuration
> > interface for setting up a vhost HW accelerator, this patch set
> introduces
> > a third vhost backend called vhost-vdpa based on the vDPA interface.
> >
> > Vhost-vdpa usage:
> >
> > qemu-system-x86_64 -cpu host -enable-kvm \
> > ......
> > -netdev type=vhost-vdpa,vhostdev=/dev/vhost-vdpa-id,id=vhost-vdpa0 \
> > -device virtio-net-pci,netdev=vhost-vdpa0,page-per-vq=on \
>
>
> Actually, this part should belongs to patch 2.
>
> And we probably need to add a comment that vIOMMU is not supported right
> now.
>
>
> Will fix this problem
> >
> > Author: Tiwei Bie
> > Signed-off-by: Cindy Lu <lulu@redhat.com>
> > ---
> > hw/net/vhost_net.c | 43 ++++
> > hw/net/virtio-net.c | 9 +
> > hw/virtio/Makefile.objs | 2 +-
> > hw/virtio/vhost-backend.c | 3 +
> > hw/virtio/vhost-vdpa.c | 379 ++++++++++++++++++++++++++++++
> > hw/virtio/vhost.c | 5 +
> > include/hw/virtio/vhost-backend.h | 6 +-
> > include/hw/virtio/vhost-vdpa.h | 14 ++
> > 8 files changed, 459 insertions(+), 2 deletions(-)
> > create mode 100644 hw/virtio/vhost-vdpa.c
> > create mode 100644 include/hw/virtio/vhost-vdpa.h
> >
> > diff --git a/hw/net/vhost_net.c b/hw/net/vhost_net.c
> > index 4096d64aaf..0d13fda2fc 100644
> > --- a/hw/net/vhost_net.c
> > +++ b/hw/net/vhost_net.c
> > @@ -17,8 +17,10 @@
> > #include "net/net.h"
> > #include "net/tap.h"
> > #include "net/vhost-user.h"
> > +#include "net/vhost-vdpa.h"
> >
> > #include "standard-headers/linux/vhost_types.h"
> > +#include "linux-headers/linux/vhost.h"
> > #include "hw/virtio/virtio-net.h"
> > #include "net/vhost_net.h"
> > #include "qemu/error-report.h"
> > @@ -85,6 +87,29 @@ static const int user_feature_bits[] = {
> > VHOST_INVALID_FEATURE_BIT
> > };
> >
> > +static const int vdpa_feature_bits[] = {
> > + VIRTIO_F_NOTIFY_ON_EMPTY,
> > + VIRTIO_RING_F_INDIRECT_DESC,
> > + VIRTIO_RING_F_EVENT_IDX,
> > + VIRTIO_F_ANY_LAYOUT,
> > + VIRTIO_F_VERSION_1,
> > + VIRTIO_NET_F_CSUM,
> > + VIRTIO_NET_F_GUEST_CSUM,
> > + VIRTIO_NET_F_GSO,
> > + VIRTIO_NET_F_GUEST_TSO4,
> > + VIRTIO_NET_F_GUEST_TSO6,
> > + VIRTIO_NET_F_GUEST_ECN,
> > + VIRTIO_NET_F_GUEST_UFO,
> > + VIRTIO_NET_F_HOST_TSO4,
> > + VIRTIO_NET_F_HOST_TSO6,
> > + VIRTIO_NET_F_HOST_ECN,
> > + VIRTIO_NET_F_HOST_UFO,
> > + VIRTIO_NET_F_MRG_RXBUF,
> > + VIRTIO_NET_F_MTU,
> > + VIRTIO_F_IOMMU_PLATFORM,
> > + VIRTIO_NET_F_GUEST_ANNOUNCE,
> > + VHOST_INVALID_FEATURE_BIT
> > +};
> > static const int *vhost_net_get_feature_bits(struct vhost_net *net)
> > {
> > const int *feature_bits = 0;
> > @@ -96,6 +121,9 @@ static const int *vhost_net_get_feature_bits(struct
> vhost_net *net)
> > case NET_CLIENT_DRIVER_VHOST_USER:
> > feature_bits = user_feature_bits;
> > break;
> > + case NET_CLIENT_DRIVER_VHOST_VDPA:
> > + feature_bits = vdpa_feature_bits;
> > + break;
> > default:
> > error_report("Feature bits not defined for this type: %d",
> > net->nc->info->type);
> > @@ -434,6 +462,10 @@ VHostNetState *get_vhost_net(NetClientState *nc)
> > assert(vhost_net);
> > break;
> > #endif
> > + case NET_CLIENT_DRIVER_VHOST_VDPA:
> > + vhost_net = vhost_vdpa_get_vhost_net(nc);
> > + assert(vhost_net);
> > + break;
> > default:
> > break;
> > }
> > @@ -465,3 +497,14 @@ int vhost_net_set_mtu(struct vhost_net *net,
> uint16_t mtu)
> >
> > return vhost_ops->vhost_net_set_mtu(&net->dev, mtu);
> > }
> > +int vhost_set_state(NetClientState *nc, int state)
> > +{
> > + struct vhost_net *net = get_vhost_net(nc);
> > + struct vhost_dev *hdev = &net->dev;
> > + if (nc->info->type == NET_CLIENT_DRIVER_VHOST_VDPA) {
> > + if (hdev->vhost_ops->vhost_set_state) {
> > + return hdev->vhost_ops->vhost_set_state(hdev, state);
> > + }
> > + }
> > + return 0;
> > +}
> > diff --git a/hw/net/virtio-net.c b/hw/net/virtio-net.c
> > index db3d7c38e6..bbecd7ab96 100644
> > --- a/hw/net/virtio-net.c
> > +++ b/hw/net/virtio-net.c
> > @@ -206,6 +206,9 @@ static void virtio_net_vhost_status(VirtIONet *n,
> uint8_t status)
> > VirtIODevice *vdev = VIRTIO_DEVICE(n);
> > NetClientState *nc = qemu_get_queue(n->nic);
> > int queues = n->multiqueue ? n->max_queues : 1;
> > + NetClientState *peer = nc->peer;
>
>
> qemu_get_peer()?
>
> will fix it
>
> > + uint8_t status_set = vdev->status ;
> > + uint8_t vhost_started_pre = n->vhost_started;
> >
> > if (!get_vhost_net(nc->peer)) {
> > return;
> > @@ -245,6 +248,7 @@ static void virtio_net_vhost_status(VirtIONet *n,
> uint8_t status)
> > return;
> > }
> > }
> > + status_set = status_set | VIRTIO_CONFIG_S_DRIVER_OK;
> >
> > n->vhost_started = 1;
> > r = vhost_net_start(vdev, n->nic->ncs, queues);
> > @@ -252,11 +256,16 @@ static void virtio_net_vhost_status(VirtIONet *n,
> uint8_t status)
> > error_report("unable to start vhost net: %d: "
> > "falling back on userspace virtio", -r);
> > n->vhost_started = 0;
> > + status_set = status_set & ~VIRTIO_CONFIG_S_DRIVER_OK;
> > }
> > } else {
> > vhost_net_stop(vdev, n->nic->ncs, queues);
> > + status_set = status_set & ~VIRTIO_CONFIG_S_DRIVER_OK;
> > n->vhost_started = 0;
> > }
> > + if (vhost_started_pre != n->vhost_started) {
> > + vhost_set_state(peer, status_set);
> > + }
> > }
>
>
> I think this deserves an independent patch.
>
>
> will fix it
>
> >
> > static int virtio_net_set_vnet_endian_one(VirtIODevice *vdev,
> > diff --git a/hw/virtio/Makefile.objs b/hw/virtio/Makefile.objs
> > index e2f70fbb89..17361d959e 100644
> > --- a/hw/virtio/Makefile.objs
> > +++ b/hw/virtio/Makefile.objs
> > @@ -2,7 +2,7 @@ ifeq ($(CONFIG_VIRTIO),y)
> > common-obj-y += virtio-bus.o
> > obj-y += virtio.o
> >
> > -obj-$(call lor,$(CONFIG_VHOST_USER),$(CONFIG_VHOST_KERNEL)) += vhost.o
> vhost-backend.o
> > +obj-$(call lor,$(CONFIG_VHOST_USER),$(CONFIG_VHOST_KERNEL)) += vhost.o
> vhost-backend.o vhost-vdpa.o
> > common-obj-$(call lnot,$(call
> lor,$(CONFIG_VHOST_USER),$(CONFIG_VHOST_KERNEL))) += vhost-stub.o
> > obj-$(CONFIG_VHOST_USER) += vhost-user.o
> >
> > diff --git a/hw/virtio/vhost-backend.c b/hw/virtio/vhost-backend.c
> > index 48905383f8..935cd9e561 100644
> > --- a/hw/virtio/vhost-backend.c
> > +++ b/hw/virtio/vhost-backend.c
> > @@ -286,6 +286,9 @@ int vhost_set_backend_type(struct vhost_dev *dev,
> VhostBackendType backend_type)
> > dev->vhost_ops = &user_ops;
> > break;
> > #endif
> > + case VHOST_BACKEND_TYPE_VDPA:
> > + dev->vhost_ops = &vdpa_ops;
> > + break;
> > default:
> > error_report("Unknown vhost backend type");
> > r = -1;
> > diff --git a/hw/virtio/vhost-vdpa.c b/hw/virtio/vhost-vdpa.c
> > new file mode 100644
> > index 0000000000..213b327600
> > --- /dev/null
> > +++ b/hw/virtio/vhost-vdpa.c
> > @@ -0,0 +1,379 @@
> > +/*
> > + * vhost-vdpa
> > + *
> > + * Copyright(c) 2017-2018 Intel Corporation. All rights reserved.
> > + * Copyright(c) 2020 Red Hat, Inc.
> > + *
> > + * This work is licensed under the terms of the GNU GPL, version 2 or
> later.
> > + * See the COPYING file in the top-level directory.
> > + *
> > + */
> > +
> > +#include "qemu/osdep.h"
> > +#include <linux/vhost.h>
> > +#include <linux/vfio.h>
> > +#include <sys/eventfd.h>
> > +#include <sys/ioctl.h>
> > +#include "hw/virtio/vhost.h"
> > +#include "hw/virtio/vhost-backend.h"
> > +#include "hw/virtio/virtio-net.h"
> > +#include "hw/virtio/vhost-vdpa.h"
> > +#include "qemu/main-loop.h"
> > +#include <linux/kvm.h>
> > +#include "sysemu/kvm.h"
> > +
> > +
> > +static bool vhost_vdpa_listener_skipped_section(MemoryRegionSection
> *section)
> > +{
> > + return (!memory_region_is_ram(section->mr) &&
> > + !memory_region_is_iommu(section->mr)) ||
> > + /*
> > + * Sizing an enabled 64-bit BAR can cause spurious mappings
> to
> > + * addresses in the upper part of the 64-bit address space.
> These
> > + * are never accessed by the CPU and beyond the address
> width of
> > + * some IOMMU hardware. TODO: VDPA should tell us the IOMMU
> width.
> > + */
> > + section->offset_within_address_space & (1ULL << 63);
> > +}
> > +
> > +static int vhost_vdpa_dma_map(struct vhost_vdpa *v, hwaddr iova, hwaddr
> size,
> > + void *vaddr, bool readonly)
> > +{
> > + struct vhost_msg_v2 msg;
> > + int fd = v->device_fd;
> > + int ret = 0;
> > +
> > + msg.type = VHOST_IOTLB_MSG_V2;
>
>
> Since V2 of the message is used here, I believe we need a kernel patch
> to allow the querying of backend capability.
>
> Sure, I will provide another patch for kernel
>
> > + msg.iotlb.iova = iova;
> > + msg.iotlb.size = size;
> > + msg.iotlb.uaddr = (uint64_t)vaddr;
> > + msg.iotlb.perm = readonly ? VHOST_ACCESS_RO : VHOST_ACCESS_RW;
> > + msg.iotlb.type = VHOST_IOTLB_UPDATE;
> > +
> > + if (write(fd, &msg, sizeof(msg)) != sizeof(msg)) {
> > + error_report("failed to write, fd=%d, errno=%d (%s)",
> > + fd, errno, strerror(errno));
> > + return -EIO ;
> > + }
> > +
> > + return ret;
> > +}
> > +
> > +static int vhost_vdpa_dma_unmap(struct vhost_vdpa *v, hwaddr iova,
> > + hwaddr size)
> > +{
> > + struct vhost_msg_v2 msg;
> > + int fd = v->device_fd;
> > + int ret = 0;
> > +
> > + msg.type = VHOST_IOTLB_MSG_V2;
> > + msg.iotlb.iova = iova;
> > + msg.iotlb.size = size;
> > + msg.iotlb.type = VHOST_IOTLB_INVALIDATE;
> > +
> > + if (write(fd, &msg, sizeof(msg)) != sizeof(msg)) {
> > + error_report("failed to write, fd=%d, errno=%d (%s)",
> > + fd, errno, strerror(errno));
> > + return -EIO ;
> > + }
> > +
> > + return ret;
> > +}
> > +
> > +static void vhost_vdpa_listener_region_add(MemoryListener *listener,
> > + MemoryRegionSection *section)
> > +{
> > + struct vhost_vdpa *v = container_of(listener, struct vhost_vdpa,
> listener);
> > + hwaddr iova;
> > + Int128 llend, llsize;
> > + void *vaddr;
> > + int ret;
> > +
> > + if (vhost_vdpa_listener_skipped_section(section)) {
> > + return;
> > + }
> > +
> > + if (unlikely((section->offset_within_address_space &
> ~TARGET_PAGE_MASK) !=
> > + (section->offset_within_region & ~TARGET_PAGE_MASK))) {
> > + error_report("%s received unaligned region", __func__);
> > + return;
> > + }
> > +
> > + iova = TARGET_PAGE_ALIGN(section->offset_within_address_space);
> > + llend = int128_make64(section->offset_within_address_space);
> > + llend = int128_add(llend, section->size);
> > + llend = int128_and(llend, int128_exts64(TARGET_PAGE_MASK));
> > +
> > + if (int128_ge(int128_make64(iova), llend)) {
> > + return;
> > + }
> > +
> > + memory_region_ref(section->mr);
> > +
> > + /* Here we assume that memory_region_is_ram(section->mr)==true */
> > +
> > + vaddr = memory_region_get_ram_ptr(section->mr) +
> > + section->offset_within_region +
> > + (iova - section->offset_within_address_space);
> > +
> > + llsize = int128_sub(llend, int128_make64(iova));
> > +
> > + ret = vhost_vdpa_dma_map(v, iova, int128_get64(llsize),
> > + vaddr, section->readonly);
> > + if (ret) {
> > + error_report("vhost vdpa map fail!");
> > + if (memory_region_is_ram_device(section->mr)) {
> > + /* Allow unexpected mappings not to be fatal for RAM
> devices */
> > + error_report("map ram fail!");
> > + return ;
> > + }
> > + goto fail;
> > + }
> > +
> > + return;
> > +
> > +fail:
> > + if (memory_region_is_ram_device(section->mr)) {
> > + error_report("failed to vdpa_dma_map. pci p2p may not work");
> > + return;
> > +
> > + }
> > + /*
> > + * On the initfn path, store the first error in the container so we
> > + * can gracefully fail. Runtime, there's not much we can do other
> > + * than throw a hardware error.
> > + */
> > + error_report("vhost-vdpa: DMA mapping failed, unable to continue");
> > + return;
> > +
> > +}
> > +
> > +static void vhost_vdpa_listener_region_del(MemoryListener *listener,
> > + MemoryRegionSection *section)
> > +{
> > + struct vhost_vdpa *v = container_of(listener, struct vhost_vdpa,
> listener);
> > + hwaddr iova;
> > + Int128 llend, llsize;
> > + int ret;
> > + bool try_unmap = true;
> > +
> > + if (vhost_vdpa_listener_skipped_section(section)) {
> > + return;
> > + }
> > +
> > + if (unlikely((section->offset_within_address_space &
> ~TARGET_PAGE_MASK) !=
> > + (section->offset_within_region & ~TARGET_PAGE_MASK))) {
> > + error_report("%s received unaligned region", __func__);
> > + return;
> > + }
> > +
> > + iova = TARGET_PAGE_ALIGN(section->offset_within_address_space);
> > + llend = int128_make64(section->offset_within_address_space);
> > + llend = int128_add(llend, section->size);
> > + llend = int128_and(llend, int128_exts64(TARGET_PAGE_MASK));
> > +
> > + if (int128_ge(int128_make64(iova), llend)) {
> > + return;
> > + }
> > +
> > + llsize = int128_sub(llend, int128_make64(iova));
> > +
> > + if (try_unmap) {
> > + ret = vhost_vdpa_dma_unmap(v, iova, int128_get64(llsize));
> > + if (ret) {
> > + error_report("vhost_vdpa dma unmap error!");
> > + }
> > + }
> > +
> > + memory_region_unref(section->mr);
> > +}
> > +
>
>
> I think it's better to add comment to explain why vhost-vdpa use a
> different listener other than the one used by other vhost backends (e.g
> kernel or user).
>
> will fix it
>
> > +static const MemoryListener vhost_vdpa_memory_listener = {
> > + .region_add = vhost_vdpa_listener_region_add,
> > + .region_del = vhost_vdpa_listener_region_del,
> > +};
> > +
> > +
> > +static int vhost_vdpa_call(struct vhost_dev *dev, unsigned long int
> request,
> > + void *arg)
> > +{
> > + struct vhost_vdpa *v = dev->opaque;
> > + int fd = v->device_fd;
> > +
> > + assert(dev->vhost_ops->backend_type == VHOST_BACKEND_TYPE_VDPA);
> > +
> > + return ioctl(fd, request, arg);
> > +}
> > +
> > +
> > +
> > +static int vhost_vdpa_init(struct vhost_dev *dev, void *opaque)
> > +{
> > + struct vhost_vdpa *v;
> > +
> > + assert(dev->vhost_ops->backend_type == VHOST_BACKEND_TYPE_VDPA);
> > +
> > + v = opaque;
> > + dev->opaque = opaque ;
> > +
> > + v->listener = vhost_vdpa_memory_listener;
> > + memory_listener_register(&v->listener, &address_space_memory);
> > +
> > + return 0;
> > +}
> > +
> > +static int vhost_vdpa_cleanup(struct vhost_dev *dev)
> > +{
> > + struct vhost_vdpa *v;
> > + assert(dev->vhost_ops->backend_type == VHOST_BACKEND_TYPE_VDPA);
> > +
> > + v = dev->opaque;
> > + memory_listener_unregister(&v->listener);
> > +
> > + dev->opaque = NULL;
> > + return 0;
> > +}
> > +
>
>
> A comment here is need to explain why INT_MAX is used.
>
>
> will do
> > +static int vhost_vdpa_memslots_limit(struct vhost_dev *dev)
> > +{
> > + return INT_MAX;
> > +}
> > +
> > +static int vhost_vdpa_set_log_base(struct vhost_dev *dev, uint64_t base,
> > + struct vhost_log *log)
> > +{
> > + return 0;
> > +}
>
>
> I think we should fail this function since we don't support dirty page
> tracking now.
>
> And it's not guarantee to use dirty page bitmap in the future.
>
>
> > +
> > +static int vhost_vdpa_set_mem_table(struct vhost_dev *dev,
> > + struct vhost_memory *mem)
> > +{
> > +
> > + if (mem->padding) {
> > + return -1;
> > + }
> > +
> > + return 0;
>
>
> A comment is need to explain why mem table is not used. (E.g we used
> IOTLB API instead).
>
> will do
>
> > +}
> > +
> > +static int vhost_vdpa_set_vring_addr(struct vhost_dev *dev,
> > + struct vhost_vring_addr *addr)
> > +{
> > + return vhost_vdpa_call(dev, VHOST_SET_VRING_ADDR, addr);
> > +}
> > +
> > +static int vhost_vdpa_set_vring_num(struct vhost_dev *dev,
> > + struct vhost_vring_state *ring)
> > +{
> > + return vhost_vdpa_call(dev, VHOST_SET_VRING_NUM, ring);
> > +}
> > +
> > +static int vhost_vdpa_set_vring_base(struct vhost_dev *dev,
> > + struct vhost_vring_state *ring)
> > +{
> > + return vhost_vdpa_call(dev, VHOST_GET_VRING_BASE, ring);
> > +}
> > +
> > +static int vhost_vdpa_get_vring_base(struct vhost_dev *dev,
> > + struct vhost_vring_state *ring)
> > +{
> > +
> > + return vhost_vdpa_call(dev, VHOST_GET_VRING_BASE, ring);
> > +}
> > +
> > +static int vhost_vdpa_set_vring_kick(struct vhost_dev *dev,
> > + struct vhost_vring_file *file)
> > +{
> > + return vhost_vdpa_call(dev, VHOST_SET_VRING_KICK, file);
> > +}
> > +
> > +static int vhost_vdpa_set_vring_call(struct vhost_dev *dev,
> > + struct vhost_vring_file *file)
> > +{
> > + return vhost_vdpa_call(dev, VHOST_SET_VRING_CALL, file);
> > +}
> > +
> > +static int vhost_vdpa_set_features(struct vhost_dev *dev,
> > + uint64_t features)
> > +{
> > +
> > + features |= (1ULL << VIRTIO_F_IOMMU_PLATFORM);
>
>
> This seems tricky, I don't think we need this actually.
>
>
> I will double check for this problem
> > + return vhost_vdpa_call(dev, VHOST_SET_FEATURES, &features);
> > +
> > +}
> > +
> > +static int vhost_vdpa_get_features(struct vhost_dev *dev,
> > + uint64_t *features)
> > +{
> > + return vhost_vdpa_call(dev, VHOST_GET_FEATURES, features);
> > +}
> > +
> > +static int vhost_vdpa_set_owner(struct vhost_dev *dev)
> > +{
> > + return vhost_vdpa_call(dev, VHOST_SET_OWNER, NULL);
> > +}
> > +
> > +static int vhost_vdpa_reset_device(struct vhost_dev *dev)
> > +{
> > + return vhost_vdpa_call(dev, VHOST_RESET_OWNER, NULL);
> > +}
> > +
> > +static int vhost_vdpa_get_vq_index(struct vhost_dev *dev, int idx)
> > +{
> > + assert(idx >= dev->vq_index && idx < dev->vq_index + dev->nvqs);
> > +
> > + return idx - dev->vq_index;
> > +}
> > +
> > +static int vhost_vdpa_set_vring_enable(struct vhost_dev *dev, int
> enable)
> > +{
> > + int i;
> > +
> > + for (i = 0; i < dev->nvqs; ++i) {
> > + struct vhost_vring_state state = {
> > + .index = dev->vq_index + i,
> > + .num = enable,
> > + };
> > +
> > + state.num = 1;
> > +
> > + vhost_vdpa_call(dev, VHOST_VDPA_SET_VRING_ENABLE, &state);
>
>
> Please make sure patch 4 comes first then we don't need to fix this in
> patch 4.
>
> will do
>
> > + }
> > +
> > + return 0;
> > +}
> > +
> > +static int vhost_vdpa_set_state(struct vhost_dev *dev, int state)
> > +{
> > + return vhost_vdpa_call(dev, VHOST_VDPA_SET_STATUS, &state);
> > +}
> > +
> > +
> > +const VhostOps vdpa_ops = {
> > + .backend_type = VHOST_BACKEND_TYPE_VDPA,
> > + .vhost_backend_init = vhost_vdpa_init,
> > + .vhost_backend_cleanup = vhost_vdpa_cleanup,
> > + .vhost_backend_memslots_limit = vhost_vdpa_memslots_limit,
> > + .vhost_set_log_base = vhost_vdpa_set_log_base,
> > + .vhost_set_mem_table = vhost_vdpa_set_mem_table,
> > + .vhost_set_vring_addr = vhost_vdpa_set_vring_addr,
> > + .vhost_set_vring_endian = NULL,
> > + .vhost_set_vring_num = vhost_vdpa_set_vring_num,
> > + .vhost_set_vring_base = vhost_vdpa_set_vring_base,
> > + .vhost_get_vring_base = vhost_vdpa_get_vring_base,
> > + .vhost_set_vring_kick = vhost_vdpa_set_vring_kick,
> > + .vhost_set_vring_call = vhost_vdpa_set_vring_call,
> > + .vhost_set_features = vhost_vdpa_set_features,
> > + .vhost_get_features = vhost_vdpa_get_features,
> > + .vhost_set_owner = vhost_vdpa_set_owner,
> > + .vhost_reset_device = vhost_vdpa_reset_device,
> > + .vhost_get_vq_index = vhost_vdpa_get_vq_index,
> > + .vhost_set_vring_enable = vhost_vdpa_set_vring_enable,
> > + .vhost_requires_shm_log = NULL,
> > + .vhost_migration_done = NULL,
> > + .vhost_backend_can_merge = NULL,
> > + .vhost_net_set_mtu = NULL,
> > + .vhost_set_iotlb_callback = NULL,
> > + .vhost_send_device_iotlb_msg = NULL,
> > + .vhost_set_state = vhost_vdpa_set_state,
> > +};
> > diff --git a/hw/virtio/vhost.c b/hw/virtio/vhost.c
> > index 4da0d5a6c5..d1f2c4add7 100644
> > --- a/hw/virtio/vhost.c
> > +++ b/hw/virtio/vhost.c
> > @@ -746,6 +746,11 @@ static int vhost_virtqueue_set_addr(struct
> vhost_dev *dev,
> > .log_guest_addr = vq->used_phys,
> > .flags = enable_log ? (1 << VHOST_VRING_F_LOG) : 0,
> > };
> > + if (dev->vhost_ops->backend_type == VHOST_BACKEND_TYPE_VDPA) {
> > + addr.desc_user_addr = (uint64_t)(unsigned long)vq->desc_phys;
> > + addr.avail_user_addr = (uint64_t)(unsigned long)vq->avail_phys;
> > + addr.used_user_addr = (uint64_t)(unsigned long)vq->used_phys;
> > + }
>
>
> Comment is needed to explain why vDPA differs from others.
>
> Thanks
>
> will do
>
> > int r = dev->vhost_ops->vhost_set_vring_addr(dev, &addr);
> > if (r < 0) {
> > VHOST_OPS_DEBUG("vhost_set_vring_addr failed");
> > diff --git a/include/hw/virtio/vhost-backend.h
> b/include/hw/virtio/vhost-backend.h
> > index 6f6670783f..d81bd9885f 100644
> > --- a/include/hw/virtio/vhost-backend.h
> > +++ b/include/hw/virtio/vhost-backend.h
> > @@ -17,7 +17,8 @@ typedef enum VhostBackendType {
> > VHOST_BACKEND_TYPE_NONE = 0,
> > VHOST_BACKEND_TYPE_KERNEL = 1,
> > VHOST_BACKEND_TYPE_USER = 2,
> > - VHOST_BACKEND_TYPE_MAX = 3,
> > + VHOST_BACKEND_TYPE_VDPA = 3,
> > + VHOST_BACKEND_TYPE_MAX = 4,
> > } VhostBackendType;
> >
> > typedef enum VhostSetConfigType {
> > @@ -112,6 +113,7 @@ typedef int (*vhost_get_inflight_fd_op)(struct
> vhost_dev *dev,
> > typedef int (*vhost_set_inflight_fd_op)(struct vhost_dev *dev,
> > struct vhost_inflight
> *inflight);
> >
> > +typedef int (*vhost_set_state_op)(struct vhost_dev *dev, int state);
> > typedef struct VhostOps {
> > VhostBackendType backend_type;
> > vhost_backend_init vhost_backend_init;
> > @@ -152,9 +154,11 @@ typedef struct VhostOps {
> > vhost_backend_mem_section_filter_op
> vhost_backend_mem_section_filter;
> > vhost_get_inflight_fd_op vhost_get_inflight_fd;
> > vhost_set_inflight_fd_op vhost_set_inflight_fd;
> > + vhost_set_state_op vhost_set_state;
> > } VhostOps;
> >
> > extern const VhostOps user_ops;
> > +extern const VhostOps vdpa_ops;
> >
> > int vhost_set_backend_type(struct vhost_dev *dev,
> > VhostBackendType backend_type);
> > diff --git a/include/hw/virtio/vhost-vdpa.h
> b/include/hw/virtio/vhost-vdpa.h
> > new file mode 100644
> > index 0000000000..889c1a4410
> > --- /dev/null
> > +++ b/include/hw/virtio/vhost-vdpa.h
> > @@ -0,0 +1,14 @@
> > +
> > +#ifndef HW_VIRTIO_VHOST_VDPA_H
> > +#define HW_VIRTIO_VHOST_VDPA_H
> > +
> > +#include "hw/virtio/virtio.h"
> > +
> > +typedef struct vhost_vdpa {
> > + int device_fd;
> > + MemoryListener listener;
> > +} VhostVDPA;
> > +
> > +extern AddressSpace address_space_memory;
> > +
> > +#endif
>
>
[-- Attachment #2: Type: text/html, Size: 32575 bytes --]
next prev parent reply other threads:[~2020-04-21 9:14 UTC|newest]
Thread overview: 28+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-04-20 9:32 [RFC v1 0/4] vDPA support in qemu Cindy Lu
2020-04-20 9:32 ` [RFC v1 1/4] net: Introduce qemu_get_peer Cindy Lu
2020-04-21 3:23 ` Jason Wang
2020-04-21 8:10 ` Cindy Lu
2020-04-21 15:01 ` Laurent Vivier
2020-04-20 9:32 ` [RFC v1 2/4] vhost-vdpa: introduce vhost-vdpa net client Cindy Lu
2020-04-20 14:49 ` Eric Blake
2020-04-21 3:40 ` Jason Wang
2020-04-21 9:46 ` Cindy Lu
2020-04-21 15:06 ` Laurent Vivier
2020-04-21 15:47 ` Laurent Vivier
2020-04-22 9:21 ` Cindy Lu
2020-04-20 9:32 ` [RFC v1 3/4] vhost-vdpa: implement vhost-vdpa backend Cindy Lu
2020-04-20 14:51 ` Eric Blake
2020-04-21 3:56 ` Jason Wang
2020-04-21 9:12 ` Cindy Lu [this message]
2020-04-21 15:54 ` Laurent Vivier
2020-04-22 9:24 ` Cindy Lu
2020-05-07 15:12 ` Maxime Coquelin
2020-05-07 15:56 ` Cindy Lu
2020-05-07 15:30 ` Maxime Coquelin
2020-05-07 16:02 ` Cindy Lu
2020-04-20 9:32 ` [RFC v1 4/4] vhost: introduce vhost_set_vring_ready method Cindy Lu
2020-04-21 3:59 ` Jason Wang
2020-04-21 8:42 ` Cindy Lu
2020-04-21 4:03 ` [RFC v1 0/4] vDPA support in qemu Jason Wang
2020-04-21 4:05 ` Jason Wang
2020-04-21 9:47 ` Cindy Lu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CACLfguX+SB0eQfrv0DwEnNpnX3LjPH+vdt_QHe+dwQGunDtiRA@mail.gmail.com \
--to=lulu@redhat.com \
--cc=aadam@redhat.com \
--cc=armbru@redhat.com \
--cc=cohuck@redhat.com \
--cc=cunming.liang@intel.com \
--cc=eperezma@redhat.com \
--cc=gdawar@xilinx.com \
--cc=hanand@xilinx.com \
--cc=hch@infradead.org \
--cc=jasowang@redhat.com \
--cc=jgg@mellanox.com \
--cc=jiri@mellanox.com \
--cc=kevin.tian@intel.com \
--cc=lingshan.zhu@intel.com \
--cc=maxime.coquelin@redhat.com \
--cc=mhabets@solarflare.com \
--cc=mst@redhat.com \
--cc=parav@mellanox.com \
--cc=qemu-devel@nongnu.org \
--cc=rdunlap@infradead.org \
--cc=rob.miller@broadcom.com \
--cc=saugatm@xilinx.com \
--cc=shahafs@mellanox.com \
--cc=stefanha@redhat.com \
--cc=vmireyno@marvell.com \
--cc=xiao.w.wang@intel.com \
--cc=zhihong.wang@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).