[RFC,v1,3/4] vhost-vdpa: implement vhost-vdpa backend

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;
+    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);
+    }
 }
 
 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;
+    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);
+}
+
+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;
+}
+
+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;
+}
+
+static int vhost_vdpa_set_mem_table(struct vhost_dev *dev,
+                                    struct vhost_memory *mem)
+{
+
+    if (mem->padding) {
+        return -1;
+    }
+
+    return 0;
+}
+
+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);
+    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);
+    }
+
+    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;
+    }
     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