[PATCH v2 00/10] vhost-user: VHOST_USER_PROTOCOL_F_GPA_ADDRESSES

[PATCH v2 00/10] vhost-user: VHOST_USER_PROTOCOL_F_GPA_ADDRESSES

[Vladimir Sementsov-Ogievskiy]

Hi all!

Here is a new feature to pass GPA addresses instead of UVA
to vhost-user server from QEMU. The details are in last patch
10/10 the others are preparation and refactoring.

v2:
06: reword commit message

Based-on: <20260420200339.708640-1-vsementsov@yandex-team.ru>
   "[PATCH v6 00/22] vhost refactoring and fixes"

Vladimir Sementsov-Ogievskiy (10):
  vhost: move IOTLB functions from vhost-backend.c to vhost.c
  vhost: use consistent naming for backend handlers
  vhost: rename vhost-backend.c to vhost-kernel.c
  vhost: replace .vhost_vq_get_addr() with .vhost_phys_vring_addr()
  vhost: simplify vhost_memory_map() and vhost_memory_unmap()
  qapi: remove user addresses from x-query-virtio-vhost-queue-status
  vhost: stop reusing vq->desc (and friends) to store physical address
  vhost: rename vring pointer fields to reflect user addresses
  vhost: add .vhost_phys_iotlb_msg() handler
  vhost-user: add VHOST_USER_PROTOCOL_F_GPA_ADDRESSES

 docs/interop/vhost-user.rst                   |  21 +-
 hw/virtio/meson.build                         |   2 +-
 hw/virtio/{vhost-backend.c => vhost-kernel.c} |  90 +-------
 hw/virtio/vhost-user.c                        |  32 ++-
 hw/virtio/vhost-vdpa.c                        |  19 +-
 hw/virtio/vhost.c                             | 193 +++++++++++++-----
 hw/virtio/virtio-hmp-cmds.c                   |   3 -
 hw/virtio/virtio-qmp.c                        |   3 -
 include/hw/virtio/vhost-backend.h             |  37 ++--
 include/hw/virtio/vhost-user.h                |   1 +
 include/hw/virtio/vhost.h                     |   8 +-
 qapi/virtio.json                              |  15 --
 12 files changed, 210 insertions(+), 214 deletions(-)
 rename hw/virtio/{vhost-backend.c => vhost-kernel.c} (83%)

-- 
2.52.0

[PATCH v2 01/10] vhost: move IOTLB functions from vhost-backend.c to vhost.c

[Vladimir Sementsov-Ogievskiy]

Move and rename vhost_backend_update_device_iotlb(),
vhost_backend_invalidate_device_iotlb(), and
vhost_backend_handle_iotlb_msg() from vhost-backend.c to vhost.c.
vhost-backend.c is actually about vhost-kernel backend. But these
functions are shared with vhost-user, so let's move them into
generic place. Moreover, two of three functions becomes static as
they are used only in vhost.c.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
---
 hw/virtio/vhost-backend.c         | 82 +----------------------------
 hw/virtio/vhost-user.c            |  2 +-
 hw/virtio/vhost.c                 | 85 +++++++++++++++++++++++++++++--
 include/hw/virtio/vhost-backend.h | 11 ----
 include/hw/virtio/vhost.h         |  2 +
 5 files changed, 86 insertions(+), 96 deletions(-)

diff --git a/hw/virtio/vhost-backend.c b/hw/virtio/vhost-backend.c
index 4367db0d951..fea88afc39a 100644
--- a/hw/virtio/vhost-backend.c
+++ b/hw/virtio/vhost-backend.c
@@ -298,7 +298,7 @@ static void vhost_kernel_iotlb_read(void *opaque)
                 break;
             }
 
-            vhost_backend_handle_iotlb_msg(dev, &msg.iotlb);
+            vhost_handle_iotlb_msg(dev, &msg.iotlb);
         }
     } else {
         struct vhost_msg msg;
@@ -313,7 +313,7 @@ static void vhost_kernel_iotlb_read(void *opaque)
                 break;
             }
 
-            vhost_backend_handle_iotlb_msg(dev, &msg.iotlb);
+            vhost_handle_iotlb_msg(dev, &msg.iotlb);
         }
     }
 }
@@ -392,81 +392,3 @@ const VhostOps kernel_ops = {
         .vhost_send_device_iotlb_msg = vhost_kernel_send_device_iotlb_msg,
 };
 #endif
-
-int vhost_backend_update_device_iotlb(struct vhost_dev *dev,
-                                             uint64_t iova, uint64_t uaddr,
-                                             uint64_t len,
-                                             IOMMUAccessFlags perm)
-{
-    struct vhost_iotlb_msg imsg;
-
-    imsg.iova =  iova;
-    imsg.uaddr = uaddr;
-    imsg.size = len;
-    imsg.type = VHOST_IOTLB_UPDATE;
-
-    switch (perm) {
-    case IOMMU_RO:
-        imsg.perm = VHOST_ACCESS_RO;
-        break;
-    case IOMMU_WO:
-        imsg.perm = VHOST_ACCESS_WO;
-        break;
-    case IOMMU_RW:
-        imsg.perm = VHOST_ACCESS_RW;
-        break;
-    default:
-        return -EINVAL;
-    }
-
-    if (dev->vhost_ops && dev->vhost_ops->vhost_send_device_iotlb_msg)
-        return dev->vhost_ops->vhost_send_device_iotlb_msg(dev, &imsg);
-
-    return -ENODEV;
-}
-
-int vhost_backend_invalidate_device_iotlb(struct vhost_dev *dev,
-                                                 uint64_t iova, uint64_t len)
-{
-    struct vhost_iotlb_msg imsg;
-
-    imsg.iova = iova;
-    imsg.size = len;
-    imsg.type = VHOST_IOTLB_INVALIDATE;
-
-    if (dev->vhost_ops && dev->vhost_ops->vhost_send_device_iotlb_msg)
-        return dev->vhost_ops->vhost_send_device_iotlb_msg(dev, &imsg);
-
-    return -ENODEV;
-}
-
-int vhost_backend_handle_iotlb_msg(struct vhost_dev *dev,
-                                          struct vhost_iotlb_msg *imsg)
-{
-    int ret = 0;
-
-    if (unlikely(!dev->vdev)) {
-        error_report("Unexpected IOTLB message when virtio device is stopped");
-        return -EINVAL;
-    }
-
-    switch (imsg->type) {
-    case VHOST_IOTLB_MISS:
-        ret = vhost_device_iotlb_miss(dev, imsg->iova,
-                                      imsg->perm != VHOST_ACCESS_RO);
-        break;
-    case VHOST_IOTLB_ACCESS_FAIL:
-        /* FIXME: report device iotlb error */
-        error_report("Access failure IOTLB message type not supported");
-        ret = -ENOTSUP;
-        break;
-    case VHOST_IOTLB_UPDATE:
-    case VHOST_IOTLB_INVALIDATE:
-    default:
-        error_report("Unexpected IOTLB message type");
-        ret = -EINVAL;
-        break;
-    }
-
-    return ret;
-}
diff --git a/hw/virtio/vhost-user.c b/hw/virtio/vhost-user.c
index 4fb136b7b37..6978d8ee947 100644
--- a/hw/virtio/vhost-user.c
+++ b/hw/virtio/vhost-user.c
@@ -1901,7 +1901,7 @@ static gboolean backend_read(QIOChannel *ioc, GIOCondition condition,
 
     switch (hdr.request) {
     case VHOST_USER_BACKEND_IOTLB_MSG:
-        ret = vhost_backend_handle_iotlb_msg(dev, &payload.iotlb);
+        ret = vhost_handle_iotlb_msg(dev, &payload.iotlb);
         break;
     case VHOST_USER_BACKEND_CONFIG_CHANGE_MSG:
         ret = vhost_user_backend_handle_config_change(dev);
diff --git a/hw/virtio/vhost.c b/hw/virtio/vhost.c
index c610007e812..4d42f8bda41 100644
--- a/hw/virtio/vhost.c
+++ b/hw/virtio/vhost.c
@@ -916,14 +916,92 @@ static void vhost_region_addnop(MemoryListener *listener,
     vhost_region_add_section(dev, section);
 }
 
+static int vhost_update_device_iotlb(struct vhost_dev *dev,
+                                     uint64_t iova, uint64_t uaddr,
+                                     uint64_t len,
+                                     IOMMUAccessFlags perm)
+{
+    struct vhost_iotlb_msg imsg;
+
+    imsg.iova =  iova;
+    imsg.uaddr = uaddr;
+    imsg.size = len;
+    imsg.type = VHOST_IOTLB_UPDATE;
+
+    switch (perm) {
+    case IOMMU_RO:
+        imsg.perm = VHOST_ACCESS_RO;
+        break;
+    case IOMMU_WO:
+        imsg.perm = VHOST_ACCESS_WO;
+        break;
+    case IOMMU_RW:
+        imsg.perm = VHOST_ACCESS_RW;
+        break;
+    default:
+        return -EINVAL;
+    }
+
+    if (dev->vhost_ops && dev->vhost_ops->vhost_send_device_iotlb_msg) {
+        return dev->vhost_ops->vhost_send_device_iotlb_msg(dev, &imsg);
+    }
+
+    return -ENODEV;
+}
+
+static int vhost_invalidate_device_iotlb(struct vhost_dev *dev,
+                                         uint64_t iova, uint64_t len)
+{
+    struct vhost_iotlb_msg imsg;
+
+    imsg.iova = iova;
+    imsg.size = len;
+    imsg.type = VHOST_IOTLB_INVALIDATE;
+
+    if (dev->vhost_ops && dev->vhost_ops->vhost_send_device_iotlb_msg) {
+        return dev->vhost_ops->vhost_send_device_iotlb_msg(dev, &imsg);
+    }
+
+    return -ENODEV;
+}
+
+int vhost_handle_iotlb_msg(struct vhost_dev *dev, struct vhost_iotlb_msg *imsg)
+{
+    int ret = 0;
+
+    if (unlikely(!dev->vdev)) {
+        error_report("Unexpected IOTLB message when virtio device is stopped");
+        return -EINVAL;
+    }
+
+    switch (imsg->type) {
+    case VHOST_IOTLB_MISS:
+        ret = vhost_device_iotlb_miss(dev, imsg->iova,
+                                      imsg->perm != VHOST_ACCESS_RO);
+        break;
+    case VHOST_IOTLB_ACCESS_FAIL:
+        /* FIXME: report device iotlb error */
+        error_report("Access failure IOTLB message type not supported");
+        ret = -ENOTSUP;
+        break;
+    case VHOST_IOTLB_UPDATE:
+    case VHOST_IOTLB_INVALIDATE:
+    default:
+        error_report("Unexpected IOTLB message type");
+        ret = -EINVAL;
+        break;
+    }
+
+    return ret;
+}
+
 static void vhost_iommu_unmap_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
 {
     struct vhost_iommu *iommu = container_of(n, struct vhost_iommu, n);
     struct vhost_dev *hdev = iommu->hdev;
     hwaddr iova = iotlb->iova + iommu->iommu_offset;
 
-    if (vhost_backend_invalidate_device_iotlb(hdev, iova,
-                                              iotlb->addr_mask + 1)) {
+    if (vhost_invalidate_device_iotlb(hdev, iova, iotlb->addr_mask + 1)) {
         error_report("Fail to invalidate device iotlb");
     }
 }
@@ -1304,8 +1382,7 @@ int vhost_device_iotlb_miss(struct vhost_dev *dev, uint64_t iova, int write)
         len = MIN(iotlb.addr_mask + 1, len);
         iova = iova & ~iotlb.addr_mask;
 
-        ret = vhost_backend_update_device_iotlb(dev, iova, uaddr,
-                                                len, iotlb.perm);
+        ret = vhost_update_device_iotlb(dev, iova, uaddr, len, iotlb.perm);
         if (ret) {
             trace_vhost_iotlb_miss(dev, 4);
             error_report("Fail to update device iotlb");
diff --git a/include/hw/virtio/vhost-backend.h b/include/hw/virtio/vhost-backend.h
index ff94fa17342..57497e197aa 100644
--- a/include/hw/virtio/vhost-backend.h
+++ b/include/hw/virtio/vhost-backend.h
@@ -222,17 +222,6 @@ typedef struct VhostOps {
     vhost_check_device_state_op vhost_check_device_state;
 } VhostOps;
 
-int vhost_backend_update_device_iotlb(struct vhost_dev *dev,
-                                             uint64_t iova, uint64_t uaddr,
-                                             uint64_t len,
-                                             IOMMUAccessFlags perm);
-
-int vhost_backend_invalidate_device_iotlb(struct vhost_dev *dev,
-                                                 uint64_t iova, uint64_t len);
-
-int vhost_backend_handle_iotlb_msg(struct vhost_dev *dev,
-                                          struct vhost_iotlb_msg *imsg);
-
 int vhost_user_gpu_set_socket(struct vhost_dev *dev, int fd);
 
 int vhost_user_get_shared_object(struct vhost_dev *dev, unsigned char *uuid,
diff --git a/include/hw/virtio/vhost.h b/include/hw/virtio/vhost.h
index 54862f9b445..b292e9f0fed 100644
--- a/include/hw/virtio/vhost.h
+++ b/include/hw/virtio/vhost.h
@@ -403,6 +403,8 @@ int vhost_dev_set_inflight(struct vhost_dev *dev,
 int vhost_dev_get_inflight(struct vhost_dev *dev, uint16_t queue_size,
                            struct vhost_inflight *inflight);
 bool vhost_dev_has_iommu(struct vhost_dev *dev);
+int vhost_handle_iotlb_msg(struct vhost_dev *dev, struct vhost_iotlb_msg *imsg);
+
 
 static inline bool vhost_dev_has_feature(struct vhost_dev *dev,
                                          uint64_t feature)
-- 
2.52.0

[PATCH v2 02/10] vhost: use consistent naming for backend handlers

[Vladimir Sementsov-Ogievskiy]

Most of handlers start from simply .vhost_, except for these
four, starting from .vhost_backend_. Let's rename them to be
consistent.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
---
 hw/virtio/vhost-backend.c         |  6 +++---
 hw/virtio/vhost-user.c            |  8 ++++----
 hw/virtio/vhost-vdpa.c            |  6 +++---
 hw/virtio/vhost.c                 | 14 +++++++-------
 include/hw/virtio/vhost-backend.h | 17 ++++++++---------
 5 files changed, 25 insertions(+), 26 deletions(-)

diff --git a/hw/virtio/vhost-backend.c b/hw/virtio/vhost-backend.c
index fea88afc39a..caeadd069b0 100644
--- a/hw/virtio/vhost-backend.c
+++ b/hw/virtio/vhost-backend.c
@@ -358,9 +358,9 @@ static void vhost_kernel_set_iotlb_callback(struct vhost_dev *dev,
 
 const VhostOps kernel_ops = {
         .backend_type = VHOST_BACKEND_TYPE_KERNEL,
-        .vhost_backend_init = vhost_kernel_init,
-        .vhost_backend_cleanup = vhost_kernel_cleanup,
-        .vhost_backend_memslots_limit = vhost_kernel_memslots_limit,
+        .vhost_init = vhost_kernel_init,
+        .vhost_cleanup = vhost_kernel_cleanup,
+        .vhost_memslots_limit = vhost_kernel_memslots_limit,
         .vhost_net_set_backend = vhost_kernel_net_set_backend,
         .vhost_scsi_set_endpoint = vhost_kernel_scsi_set_endpoint,
         .vhost_scsi_clear_endpoint = vhost_kernel_scsi_clear_endpoint,
diff --git a/hw/virtio/vhost-user.c b/hw/virtio/vhost-user.c
index 6978d8ee947..78ffb25d6b1 100644
--- a/hw/virtio/vhost-user.c
+++ b/hw/virtio/vhost-user.c
@@ -3115,10 +3115,10 @@ void vhost_user_qmp_status(struct vhost_dev *dev, VirtioStatus *status)
 
 const VhostOps user_ops = {
         .backend_type = VHOST_BACKEND_TYPE_USER,
-        .vhost_backend_init = vhost_user_backend_init,
-        .vhost_backend_cleanup = vhost_user_backend_cleanup,
-        .vhost_backend_memslots_limit = vhost_user_memslots_limit,
-        .vhost_backend_no_private_memslots = vhost_user_no_private_memslots,
+        .vhost_init = vhost_user_backend_init,
+        .vhost_cleanup = vhost_user_backend_cleanup,
+        .vhost_memslots_limit = vhost_user_memslots_limit,
+        .vhost_no_private_memslots = vhost_user_no_private_memslots,
         .vhost_set_log_base = vhost_user_set_log_base,
         .vhost_set_mem_table = vhost_user_set_mem_table,
         .vhost_set_vring_addr = vhost_user_set_vring_addr,
diff --git a/hw/virtio/vhost-vdpa.c b/hw/virtio/vhost-vdpa.c
index 9c7634e2439..93474624898 100644
--- a/hw/virtio/vhost-vdpa.c
+++ b/hw/virtio/vhost-vdpa.c
@@ -1590,8 +1590,8 @@ static bool  vhost_vdpa_force_iommu(struct vhost_dev *dev)
 
 const VhostOps vdpa_ops = {
         .backend_type = VHOST_BACKEND_TYPE_VDPA,
-        .vhost_backend_init = vhost_vdpa_init,
-        .vhost_backend_cleanup = vhost_vdpa_cleanup,
+        .vhost_init = vhost_vdpa_init,
+        .vhost_cleanup = vhost_vdpa_cleanup,
         .vhost_set_log_base = vhost_vdpa_set_log_base,
         .vhost_set_vring_addr = vhost_vdpa_set_vring_addr,
         .vhost_set_vring_num = vhost_vdpa_set_vring_num,
@@ -1602,7 +1602,7 @@ const VhostOps vdpa_ops = {
         .vhost_get_features = vhost_vdpa_get_features,
         .vhost_set_owner = vhost_vdpa_set_owner,
         .vhost_set_vring_endian = NULL,
-        .vhost_backend_memslots_limit = vhost_vdpa_memslots_limit,
+        .vhost_memslots_limit = vhost_vdpa_memslots_limit,
         .vhost_set_mem_table = vhost_vdpa_set_mem_table,
         .vhost_set_features = vhost_vdpa_set_features,
         .vhost_reset_device = vhost_vdpa_reset_device,
diff --git a/hw/virtio/vhost.c b/hw/virtio/vhost.c
index 4d42f8bda41..ffd7d404cc7 100644
--- a/hw/virtio/vhost.c
+++ b/hw/virtio/vhost.c
@@ -58,7 +58,7 @@ unsigned int vhost_get_max_memslots(void)
     struct vhost_dev *hdev;
 
     QLIST_FOREACH(hdev, &vhost_devices, entry) {
-        max = MIN(max, hdev->vhost_ops->vhost_backend_memslots_limit(hdev));
+        max = MIN(max, hdev->vhost_ops->vhost_memslots_limit(hdev));
     }
     return max;
 }
@@ -69,7 +69,7 @@ unsigned int vhost_get_free_memslots(void)
     struct vhost_dev *hdev;
 
     QLIST_FOREACH(hdev, &vhost_devices, entry) {
-        unsigned int r = hdev->vhost_ops->vhost_backend_memslots_limit(hdev);
+        unsigned int r = hdev->vhost_ops->vhost_memslots_limit(hdev);
         unsigned int cur_free = r - hdev->mem->nregions;
 
         if (unlikely(r < hdev->mem->nregions)) {
@@ -664,8 +664,8 @@ static bool vhost_section(struct vhost_dev *dev, MemoryRegionSection *section)
          */
         if ((memory_region_get_fd(section->mr) < 0 ||
             !qemu_ram_is_shared(section->mr->ram_block)) &&
-            dev->vhost_ops->vhost_backend_no_private_memslots &&
-            dev->vhost_ops->vhost_backend_no_private_memslots(dev)) {
+            dev->vhost_ops->vhost_no_private_memslots &&
+            dev->vhost_ops->vhost_no_private_memslots(dev)) {
             trace_vhost_reject_section(mr->name, 2);
             return false;
         }
@@ -1674,7 +1674,7 @@ int vhost_dev_init(struct vhost_dev *hdev, void *opaque,
     r = vhost_set_backend_type(hdev, backend_type);
     assert(r >= 0);
 
-    r = hdev->vhost_ops->vhost_backend_init(hdev, opaque, errp);
+    r = hdev->vhost_ops->vhost_init(hdev, opaque, errp);
     if (r < 0) {
         goto fail;
     }
@@ -1691,7 +1691,7 @@ int vhost_dev_init(struct vhost_dev *hdev, void *opaque,
         goto fail;
     }
 
-    limit = hdev->vhost_ops->vhost_backend_memslots_limit(hdev);
+    limit = hdev->vhost_ops->vhost_memslots_limit(hdev);
     if (limit < MEMORY_DEVICES_SAFE_MAX_MEMSLOTS &&
         memory_devices_memslot_auto_decision_active()) {
         error_setg(errp, "some memory device (like virtio-mem)"
@@ -1809,7 +1809,7 @@ void vhost_dev_cleanup(struct vhost_dev *hdev)
     g_free(hdev->mem);
     g_free(hdev->mem_sections);
     if (hdev->vhost_ops) {
-        hdev->vhost_ops->vhost_backend_cleanup(hdev);
+        hdev->vhost_ops->vhost_cleanup(hdev);
     }
     assert(!hdev->log);
 
diff --git a/include/hw/virtio/vhost-backend.h b/include/hw/virtio/vhost-backend.h
index 57497e197aa..aab6bf3ef7e 100644
--- a/include/hw/virtio/vhost-backend.h
+++ b/include/hw/virtio/vhost-backend.h
@@ -51,10 +51,9 @@ struct vhost_scsi_target;
 struct vhost_iotlb_msg;
 struct vhost_virtqueue;
 
-typedef int (*vhost_backend_init)(struct vhost_dev *dev, void *opaque,
-                                  Error **errp);
-typedef int (*vhost_backend_cleanup)(struct vhost_dev *dev);
-typedef int (*vhost_backend_memslots_limit)(struct vhost_dev *dev);
+typedef int (*vhost_init)(struct vhost_dev *dev, void *opaque, Error **errp);
+typedef int (*vhost_cleanup)(struct vhost_dev *dev);
+typedef int (*vhost_memslots_limit)(struct vhost_dev *dev);
 
 typedef int (*vhost_net_set_backend_op)(struct vhost_dev *dev,
                                 struct vhost_vring_file *file);
@@ -131,7 +130,7 @@ typedef int (*vhost_crypto_create_session_op)(struct vhost_dev *dev,
 typedef int (*vhost_crypto_close_session_op)(struct vhost_dev *dev,
                                              uint64_t session_id);
 
-typedef bool (*vhost_backend_no_private_memslots_op)(struct vhost_dev *dev);
+typedef bool (*vhost_no_private_memslots_op)(struct vhost_dev *dev);
 
 typedef int (*vhost_get_inflight_fd_op)(struct vhost_dev *dev,
                                         uint16_t queue_size,
@@ -166,10 +165,10 @@ typedef int (*vhost_check_device_state_op)(struct vhost_dev *dev, Error **errp);
 
 typedef struct VhostOps {
     VhostBackendType backend_type;
-    vhost_backend_init vhost_backend_init;
-    vhost_backend_cleanup vhost_backend_cleanup;
-    vhost_backend_memslots_limit vhost_backend_memslots_limit;
-    vhost_backend_no_private_memslots_op vhost_backend_no_private_memslots;
+    vhost_init vhost_init;
+    vhost_cleanup vhost_cleanup;
+    vhost_memslots_limit vhost_memslots_limit;
+    vhost_no_private_memslots_op vhost_no_private_memslots;
     vhost_net_set_backend_op vhost_net_set_backend;
     vhost_net_set_mtu_op vhost_net_set_mtu;
     vhost_scsi_set_endpoint_op vhost_scsi_set_endpoint;
-- 
2.52.0

[PATCH v2 03/10] vhost: rename vhost-backend.c to vhost-kernel.c

[Vladimir Sementsov-Ogievskiy]

This file is totally about vhost-kernel implementation, not generic
code. Let's finally give it corresponding name.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
---
 hw/virtio/meson.build                         | 2 +-
 hw/virtio/{vhost-backend.c => vhost-kernel.c} | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
 rename hw/virtio/{vhost-backend.c => vhost-kernel.c} (99%)

diff --git a/hw/virtio/meson.build b/hw/virtio/meson.build
index adc3a4f748a..fbc1b1ad121 100644
--- a/hw/virtio/meson.build
+++ b/hw/virtio/meson.build
@@ -14,7 +14,7 @@ system_virtio_ss.add(files('virtio-qmp.c'))
 
 if have_vhost
   system_virtio_ss.add(files('vhost.c'))
-  system_virtio_ss.add(files('vhost-backend.c', 'vhost-iova-tree.c'))
+  system_virtio_ss.add(files('vhost-kernel.c', 'vhost-iova-tree.c'))
   if have_vhost_user
     system_virtio_ss.add(files('vhost-user.c'))
     system_virtio_ss.add(files('vhost-user-base.c'))
diff --git a/hw/virtio/vhost-backend.c b/hw/virtio/vhost-kernel.c
similarity index 99%
rename from hw/virtio/vhost-backend.c
rename to hw/virtio/vhost-kernel.c
index caeadd069b0..3390b48c6f1 100644
--- a/hw/virtio/vhost-backend.c
+++ b/hw/virtio/vhost-kernel.c
@@ -1,5 +1,5 @@
 /*
- * vhost-backend
+ * vhost-kernel
  *
  * Copyright (c) 2013 Virtual Open Systems Sarl.
  *
-- 
2.52.0

[PATCH v2 04/10] vhost: replace .vhost_vq_get_addr() with .vhost_phys_vring_addr()

[Vladimir Sementsov-Ogievskiy]

Make a simper handler to consolidate the logic in one place. That
helps further changes.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
---
 hw/virtio/vhost-vdpa.c            | 13 +++----------
 hw/virtio/vhost.c                 | 12 ++++++------
 include/hw/virtio/vhost-backend.h |  6 ++----
 3 files changed, 11 insertions(+), 20 deletions(-)

diff --git a/hw/virtio/vhost-vdpa.c b/hw/virtio/vhost-vdpa.c
index 93474624898..9e1aa4860a8 100644
--- a/hw/virtio/vhost-vdpa.c
+++ b/hw/virtio/vhost-vdpa.c
@@ -1571,16 +1571,9 @@ static int vhost_vdpa_set_owner(struct vhost_dev *dev)
     return 0;
 }
 
-static int vhost_vdpa_vq_get_addr(struct vhost_dev *dev,
-                    struct vhost_vring_addr *addr, struct vhost_virtqueue *vq)
+static bool vhost_vdpa_phys_vring_addr(struct vhost_dev *dev)
 {
-    assert(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;
-    trace_vhost_vdpa_vq_get_addr(dev, vq, addr->desc_user_addr,
-                                 addr->avail_user_addr, addr->used_user_addr);
-    return 0;
+    return true;
 }
 
 static bool  vhost_vdpa_force_iommu(struct vhost_dev *dev)
@@ -1617,7 +1610,7 @@ const VhostOps vdpa_ops = {
         .vhost_send_device_iotlb_msg = NULL,
         .vhost_dev_start = vhost_vdpa_dev_start,
         .vhost_get_device_id = vhost_vdpa_get_device_id,
-        .vhost_vq_get_addr = vhost_vdpa_vq_get_addr,
+        .vhost_phys_vring_addr = vhost_vdpa_phys_vring_addr,
         .vhost_force_iommu = vhost_vdpa_force_iommu,
         .vhost_set_config_call = vhost_vdpa_set_config_call,
         .vhost_reset_status = vhost_vdpa_reset_status,
diff --git a/hw/virtio/vhost.c b/hw/virtio/vhost.c
index ffd7d404cc7..564a5d46982 100644
--- a/hw/virtio/vhost.c
+++ b/hw/virtio/vhost.c
@@ -1093,16 +1093,16 @@ static int vhost_virtqueue_set_addr(struct vhost_dev *dev,
                                     struct vhost_virtqueue *vq,
                                     unsigned idx, bool enable_log)
 {
+    bool phys = dev->vhost_ops->vhost_phys_vring_addr &&
+        dev->vhost_ops->vhost_phys_vring_addr(dev);
     struct vhost_vring_addr addr;
     int r;
     memset(&addr, 0, sizeof(struct vhost_vring_addr));
 
-    if (dev->vhost_ops->vhost_vq_get_addr) {
-        r = dev->vhost_ops->vhost_vq_get_addr(dev, &addr, vq);
-        if (r < 0) {
-            VHOST_OPS_DEBUG(r, "vhost_vq_get_addr failed");
-            return r;
-        }
+    if (phys) {
+        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;
     } else {
         addr.desc_user_addr = (uint64_t)(unsigned long)vq->desc;
         addr.avail_user_addr = (uint64_t)(unsigned long)vq->avail;
diff --git a/include/hw/virtio/vhost-backend.h b/include/hw/virtio/vhost-backend.h
index aab6bf3ef7e..7cc3bb54418 100644
--- a/include/hw/virtio/vhost-backend.h
+++ b/include/hw/virtio/vhost-backend.h
@@ -141,9 +141,7 @@ typedef int (*vhost_set_inflight_fd_op)(struct vhost_dev *dev,
 
 typedef int (*vhost_dev_start_op)(struct vhost_dev *dev, bool started);
 
-typedef int (*vhost_vq_get_addr_op)(struct vhost_dev *dev,
-                    struct vhost_vring_addr *addr,
-                    struct vhost_virtqueue *vq);
+typedef bool (*vhost_phys_vring_addr_op)(struct vhost_dev *dev);
 
 typedef int (*vhost_get_device_id_op)(struct vhost_dev *dev, uint32_t *dev_id);
 
@@ -211,7 +209,7 @@ typedef struct VhostOps {
     vhost_get_inflight_fd_op vhost_get_inflight_fd;
     vhost_set_inflight_fd_op vhost_set_inflight_fd;
     vhost_dev_start_op vhost_dev_start;
-    vhost_vq_get_addr_op  vhost_vq_get_addr;
+    vhost_phys_vring_addr_op vhost_phys_vring_addr;
     vhost_get_device_id_op vhost_get_device_id;
     vhost_force_iommu_op vhost_force_iommu;
     vhost_set_config_call_op vhost_set_config_call;
-- 
2.52.0

[PATCH v2 05/10] vhost: simplify vhost_memory_map() and vhost_memory_unmap()

[Vladimir Sementsov-Ogievskiy]

Make these functions simple wrappers around address_space_map() and
address_space_unmap(). Move IOMMU handling logic one layer up to the
callers.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
---
 hw/virtio/vhost.c | 43 ++++++++++++++++++++++++-------------------
 1 file changed, 24 insertions(+), 19 deletions(-)

diff --git a/hw/virtio/vhost.c b/hw/virtio/vhost.c
index 564a5d46982..cc083572f40 100644
--- a/hw/virtio/vhost.c
+++ b/hw/virtio/vhost.c
@@ -456,21 +456,17 @@ static inline void vhost_dev_log_resize(struct vhost_dev *dev, uint64_t size)
 static void *vhost_memory_map(struct vhost_dev *dev, hwaddr addr,
                               hwaddr len, bool is_write)
 {
-    if (!vhost_dev_has_iommu(dev)) {
-        hwaddr mapped_len = len;
-        void *res = address_space_map(dev->vdev->dma_as, addr, &mapped_len,
-                                      is_write, MEMTXATTRS_UNSPECIFIED);
-        if (!res) {
-            return NULL;
-        }
-        if (len != mapped_len) {
-            address_space_unmap(dev->vdev->dma_as, res, mapped_len, 0, 0);
-            return NULL;
-        }
-        return res;
-    } else {
-        return (void *)(uintptr_t)addr;
+    hwaddr mapped_len = len;
+    void *res = address_space_map(dev->vdev->dma_as, addr, &mapped_len,
+                                  is_write, MEMTXATTRS_UNSPECIFIED);
+    if (!res) {
+        return NULL;
+    }
+    if (len != mapped_len) {
+        address_space_unmap(dev->vdev->dma_as, res, mapped_len, 0, 0);
+        return NULL;
     }
+    return res;
 }
 
 static void vhost_memory_unmap(struct vhost_dev *dev, void **buffer,
@@ -481,17 +477,18 @@ static void vhost_memory_unmap(struct vhost_dev *dev, void **buffer,
         return;
     }
 
-    if (!vhost_dev_has_iommu(dev)) {
-        address_space_unmap(dev->vdev->dma_as, *buffer, len, is_write,
-                            access_len);
-    }
-
+    address_space_unmap(dev->vdev->dma_as, *buffer, len, is_write,
+                        access_len);
     *buffer = NULL;
 }
 
 static void vhost_vrings_unmap(struct vhost_dev *dev,
                                struct vhost_virtqueue *vq, bool touched)
 {
+    if (vhost_dev_has_iommu(dev)) {
+        return;
+    }
+
     vhost_memory_unmap(dev, &vq->used, vq->used_size, touched,
                        touched ? vq->used_size : 0);
     vhost_memory_unmap(dev, &vq->avail, vq->avail_size, 0,
@@ -519,6 +516,14 @@ static int vhost_vrings_map(struct vhost_dev *dev,
         /* Queue might not be ready for start */
         return 0;
     }
+
+    if (vhost_dev_has_iommu(dev)) {
+        vq->desc = (void *)(uintptr_t)vq->desc_phys;
+        vq->avail = (void *)(uintptr_t)vq->avail_phys;
+        vq->used = (void *)(uintptr_t)vq->used_phys;
+        return 1;
+    }
+
     vq->desc = vhost_memory_map(dev, vq->desc_phys, vq->desc_size, false);
     if (!vq->desc) {
         goto fail;
-- 
2.52.0

[PATCH v2 06/10] qapi: remove user addresses from x-query-virtio-vhost-queue-status

[Vladimir Sementsov-Ogievskiy]

Semantics of these (actually, internal) fields is not simple, they may
contain either virtual or physical addresses. We are going to change
this to simplify the logic. Keeping this logic only for unstable
info command seems too much. Changing semantics of info fields doesn't
seem to make real sense too. So, let's just drop them. We can dot it,
as command is experimental.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
---
 hw/virtio/virtio-hmp-cmds.c |  3 ---
 hw/virtio/virtio-qmp.c      |  3 ---
 qapi/virtio.json            | 15 ---------------
 3 files changed, 21 deletions(-)

diff --git a/hw/virtio/virtio-hmp-cmds.c b/hw/virtio/virtio-hmp-cmds.c
index 4bf9a3109d4..a91bb7dbeda 100644
--- a/hw/virtio/virtio-hmp-cmds.c
+++ b/hw/virtio/virtio-hmp-cmds.c
@@ -203,15 +203,12 @@ void hmp_vhost_queue_status(Monitor *mon, const QDict *qdict)
     monitor_printf(mon, "  call:                 %"PRId64"\n", s->call);
     monitor_printf(mon, "  VRing:\n");
     monitor_printf(mon, "    num:         %"PRId64"\n", s->num);
-    monitor_printf(mon, "    desc:        0x%016"PRIx64"\n", s->desc);
     monitor_printf(mon, "    desc_phys:   0x%016"PRIx64"\n",
                    s->desc_phys);
     monitor_printf(mon, "    desc_size:   %"PRId32"\n", s->desc_size);
-    monitor_printf(mon, "    avail:       0x%016"PRIx64"\n", s->avail);
     monitor_printf(mon, "    avail_phys:  0x%016"PRIx64"\n",
                    s->avail_phys);
     monitor_printf(mon, "    avail_size:  %"PRId32"\n", s->avail_size);
-    monitor_printf(mon, "    used:        0x%016"PRIx64"\n", s->used);
     monitor_printf(mon, "    used_phys:   0x%016"PRIx64"\n",
                    s->used_phys);
     monitor_printf(mon, "    used_size:   %"PRId32"\n", s->used_size);
diff --git a/hw/virtio/virtio-qmp.c b/hw/virtio/virtio-qmp.c
index 51e3a42c4c4..ae9be0a00f4 100644
--- a/hw/virtio/virtio-qmp.c
+++ b/hw/virtio/virtio-qmp.c
@@ -794,9 +794,6 @@ VirtVhostQueueStatus *qmp_x_query_virtio_vhost_queue_status(const char *path,
     status->name = g_strdup(vdev->name);
     status->kick = hdev->vqs[queue].kick;
     status->call = hdev->vqs[queue].call;
-    status->desc = (uintptr_t)hdev->vqs[queue].desc;
-    status->avail = (uintptr_t)hdev->vqs[queue].avail;
-    status->used = (uintptr_t)hdev->vqs[queue].used;
     status->num = hdev->vqs[queue].num;
     status->desc_phys = hdev->vqs[queue].desc_phys;
     status->desc_size = hdev->vqs[queue].desc_size;
diff --git a/qapi/virtio.json b/qapi/virtio.json
index 671f1ad793c..1fc4e38a44d 100644
--- a/qapi/virtio.json
+++ b/qapi/virtio.json
@@ -650,12 +650,6 @@
 #
 # @call: vhost_virtqueue call
 #
-# @desc: vhost_virtqueue desc
-#
-# @avail: vhost_virtqueue avail
-#
-# @used: vhost_virtqueue used
-#
 # @num: vhost_virtqueue num
 #
 # @desc-phys: vhost_virtqueue desc_phys (descriptor area physical
@@ -678,9 +672,6 @@
   'data': { 'name': 'str',
             'kick': 'int',
             'call': 'int',
-            'desc': 'uint64',
-            'avail': 'uint64',
-            'used': 'uint64',
             'num': 'int',
             'desc-phys': 'uint64',
             'desc-size': 'uint32',
@@ -720,12 +711,9 @@
 #              "avail-size": 2054,
 #              "desc-size": 16384,
 #              "used-size": 8198,
-#              "desc": 140141447430144,
 #              "num": 1024,
 #              "call": 0,
-#              "avail": 140141447446528,
 #              "desc-phys": 5216108544,
-#              "used": 140141447448640,
 #              "kick": 0
 #          }
 #        }
@@ -744,12 +732,9 @@
 #              "avail-size": 262,
 #              "desc-size": 2048,
 #              "used-size": 1030,
-#              "desc": 140141413580800,
 #              "num": 128,
 #              "call": 0,
-#              "avail": 140141413582848,
 #              "desc-phys": 5182259200,
-#              "used": 140141413583168,
 #              "kick": 0
 #          }
 #        }
-- 
2.52.0

[PATCH v2 07/10] vhost: stop reusing vq->desc (and friends) to store physical address

[Vladimir Sementsov-Ogievskiy]

Remove the hack of storing physical addresses in vq->desc/avail/used
pointers when IOMMU is present. Instead, handle this case directly in
vhost_virtqueue_set_addr() by checking for IOMMU presence.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
---
 hw/virtio/vhost.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

diff --git a/hw/virtio/vhost.c b/hw/virtio/vhost.c
index cc083572f40..6bb6b07e190 100644
--- a/hw/virtio/vhost.c
+++ b/hw/virtio/vhost.c
@@ -518,9 +518,6 @@ static int vhost_vrings_map(struct vhost_dev *dev,
     }
 
     if (vhost_dev_has_iommu(dev)) {
-        vq->desc = (void *)(uintptr_t)vq->desc_phys;
-        vq->avail = (void *)(uintptr_t)vq->avail_phys;
-        vq->used = (void *)(uintptr_t)vq->used_phys;
         return 1;
     }
 
@@ -1104,7 +1101,7 @@ static int vhost_virtqueue_set_addr(struct vhost_dev *dev,
     int r;
     memset(&addr, 0, sizeof(struct vhost_vring_addr));
 
-    if (phys) {
+    if (phys || vhost_dev_has_iommu(dev)) {
         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;
-- 
2.52.0

[PATCH v2 08/10] vhost: rename vring pointer fields to reflect user addresses

[Vladimir Sementsov-Ogievskiy]

Rename vq->desc/avail/used to vq->desc_user/avail_user/used_user to
clearly indicate these fields contain user space addresses, not
physical addresses.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
---
 hw/virtio/vhost.c         | 37 +++++++++++++++++++------------------
 include/hw/virtio/vhost.h |  6 +++---
 2 files changed, 22 insertions(+), 21 deletions(-)

diff --git a/hw/virtio/vhost.c b/hw/virtio/vhost.c
index 6bb6b07e190..e3d452a7637 100644
--- a/hw/virtio/vhost.c
+++ b/hw/virtio/vhost.c
@@ -489,11 +489,11 @@ static void vhost_vrings_unmap(struct vhost_dev *dev,
         return;
     }
 
-    vhost_memory_unmap(dev, &vq->used, vq->used_size, touched,
+    vhost_memory_unmap(dev, &vq->used_user, vq->used_size, touched,
                        touched ? vq->used_size : 0);
-    vhost_memory_unmap(dev, &vq->avail, vq->avail_size, 0,
+    vhost_memory_unmap(dev, &vq->avail_user, vq->avail_size, 0,
                        touched ? vq->avail_size : 0);
-    vhost_memory_unmap(dev, &vq->desc, vq->desc_size, 0,
+    vhost_memory_unmap(dev, &vq->desc_user, vq->desc_size, 0,
                        touched ? vq->desc_size : 0);
 }
 
@@ -504,13 +504,13 @@ static int vhost_vrings_map(struct vhost_dev *dev,
 {
     vq->desc_size = virtio_queue_get_desc_size(vdev, idx);
     vq->desc_phys = virtio_queue_get_desc_addr(vdev, idx);
-    vq->desc = NULL;
+    vq->desc_user = NULL;
     vq->avail_size = virtio_queue_get_avail_size(vdev, idx);
     vq->avail_phys = virtio_queue_get_avail_addr(vdev, idx);
-    vq->avail = NULL;
+    vq->avail_user = NULL;
     vq->used_size = virtio_queue_get_used_size(vdev, idx);
     vq->used_phys = virtio_queue_get_used_addr(vdev, idx);
-    vq->used = NULL;
+    vq->used_user = NULL;
 
     if (vq->desc_phys == 0) {
         /* Queue might not be ready for start */
@@ -521,16 +521,17 @@ static int vhost_vrings_map(struct vhost_dev *dev,
         return 1;
     }
 
-    vq->desc = vhost_memory_map(dev, vq->desc_phys, vq->desc_size, false);
-    if (!vq->desc) {
+    vq->desc_user = vhost_memory_map(dev, vq->desc_phys, vq->desc_size, false);
+    if (!vq->desc_user) {
         goto fail;
     }
-    vq->avail = vhost_memory_map(dev, vq->avail_phys, vq->avail_size, false);
-    if (!vq->avail) {
+    vq->avail_user = vhost_memory_map(dev, vq->avail_phys, vq->avail_size,
+                                      false);
+    if (!vq->avail_user) {
         goto fail;
     }
-    vq->used = vhost_memory_map(dev, vq->used_phys, vq->used_size, true);
-    if (!vq->used) {
+    vq->used_user = vhost_memory_map(dev, vq->used_phys, vq->used_size, true);
+    if (!vq->used_user) {
         goto fail;
     }
 
@@ -594,7 +595,7 @@ static int vhost_verify_ring_mappings(struct vhost_dev *dev,
 
         j = 0;
         r = vhost_verify_ring_part_mapping(
-                vq->desc, vq->desc_phys, vq->desc_size,
+                vq->desc_user, vq->desc_phys, vq->desc_size,
                 reg_hva, reg_gpa, reg_size);
         if (r) {
             break;
@@ -602,7 +603,7 @@ static int vhost_verify_ring_mappings(struct vhost_dev *dev,
 
         j++;
         r = vhost_verify_ring_part_mapping(
-                vq->avail, vq->avail_phys, vq->avail_size,
+                vq->avail_user, vq->avail_phys, vq->avail_size,
                 reg_hva, reg_gpa, reg_size);
         if (r) {
             break;
@@ -610,7 +611,7 @@ static int vhost_verify_ring_mappings(struct vhost_dev *dev,
 
         j++;
         r = vhost_verify_ring_part_mapping(
-                vq->used, vq->used_phys, vq->used_size,
+                vq->used_user, vq->used_phys, vq->used_size,
                 reg_hva, reg_gpa, reg_size);
         if (r) {
             break;
@@ -1106,9 +1107,9 @@ static int vhost_virtqueue_set_addr(struct vhost_dev *dev,
         addr.avail_user_addr = (uint64_t)(unsigned long)vq->avail_phys;
         addr.used_user_addr = (uint64_t)(unsigned long)vq->used_phys;
     } else {
-        addr.desc_user_addr = (uint64_t)(unsigned long)vq->desc;
-        addr.avail_user_addr = (uint64_t)(unsigned long)vq->avail;
-        addr.used_user_addr = (uint64_t)(unsigned long)vq->used;
+        addr.desc_user_addr = (uint64_t)(unsigned long)vq->desc_user;
+        addr.avail_user_addr = (uint64_t)(unsigned long)vq->avail_user;
+        addr.used_user_addr = (uint64_t)(unsigned long)vq->used_user;
     }
     addr.index = idx;
     addr.log_guest_addr = vq->used_phys;
diff --git a/include/hw/virtio/vhost.h b/include/hw/virtio/vhost.h
index b292e9f0fed..684bafcaadd 100644
--- a/include/hw/virtio/vhost.h
+++ b/include/hw/virtio/vhost.h
@@ -24,9 +24,9 @@ struct vhost_inflight {
 struct vhost_virtqueue {
     int kick;
     int call;
-    void *desc;
-    void *avail;
-    void *used;
+    void *desc_user;
+    void *avail_user;
+    void *used_user;
     int num;
     unsigned long long desc_phys;
     unsigned desc_size;
-- 
2.52.0

[PATCH v2 09/10] vhost: add .vhost_phys_iotlb_msg() handler

[Vladimir Sementsov-Ogievskiy]

Allow backends to chose, which addr to put into vhost_iotlb_msg:
user or physical. To be used in further commit.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
---
 hw/virtio/vhost.c                 | 11 ++++++++---
 include/hw/virtio/vhost-backend.h |  3 +++
 2 files changed, 11 insertions(+), 3 deletions(-)

diff --git a/hw/virtio/vhost.c b/hw/virtio/vhost.c
index e3d452a7637..af41841b529 100644
--- a/hw/virtio/vhost.c
+++ b/hw/virtio/vhost.c
@@ -920,14 +920,18 @@ static void vhost_region_addnop(MemoryListener *listener,
 }
 
 static int vhost_update_device_iotlb(struct vhost_dev *dev,
-                                     uint64_t iova, uint64_t uaddr,
+                                     uint64_t iova,
+                                     uint64_t paddr,
+                                     uint64_t uaddr,
                                      uint64_t len,
                                      IOMMUAccessFlags perm)
 {
+    bool phys = dev->vhost_ops->vhost_phys_iotlb_msg &&
+        dev->vhost_ops->vhost_phys_iotlb_msg(dev);
     struct vhost_iotlb_msg imsg;
 
     imsg.iova =  iova;
-    imsg.uaddr = uaddr;
+    imsg.uaddr = phys ? paddr : uaddr;
     imsg.size = len;
     imsg.type = VHOST_IOTLB_UPDATE;
 
@@ -1385,7 +1389,8 @@ int vhost_device_iotlb_miss(struct vhost_dev *dev, uint64_t iova, int write)
         len = MIN(iotlb.addr_mask + 1, len);
         iova = iova & ~iotlb.addr_mask;
 
-        ret = vhost_update_device_iotlb(dev, iova, uaddr, len, iotlb.perm);
+        ret = vhost_update_device_iotlb(dev, iova, iotlb.translated_addr, uaddr,
+                                        len, iotlb.perm);
         if (ret) {
             trace_vhost_iotlb_miss(dev, 4);
             error_report("Fail to update device iotlb");
diff --git a/include/hw/virtio/vhost-backend.h b/include/hw/virtio/vhost-backend.h
index 7cc3bb54418..00881a0bd44 100644
--- a/include/hw/virtio/vhost-backend.h
+++ b/include/hw/virtio/vhost-backend.h
@@ -143,6 +143,8 @@ typedef int (*vhost_dev_start_op)(struct vhost_dev *dev, bool started);
 
 typedef bool (*vhost_phys_vring_addr_op)(struct vhost_dev *dev);
 
+typedef bool (*vhost_phys_iotlb_msg_op)(struct vhost_dev *dev);
+
 typedef int (*vhost_get_device_id_op)(struct vhost_dev *dev, uint32_t *dev_id);
 
 typedef bool (*vhost_force_iommu_op)(struct vhost_dev *dev);
@@ -210,6 +212,7 @@ typedef struct VhostOps {
     vhost_set_inflight_fd_op vhost_set_inflight_fd;
     vhost_dev_start_op vhost_dev_start;
     vhost_phys_vring_addr_op vhost_phys_vring_addr;
+    vhost_phys_iotlb_msg_op vhost_phys_iotlb_msg;
     vhost_get_device_id_op vhost_get_device_id;
     vhost_force_iommu_op vhost_force_iommu;
     vhost_set_config_call_op vhost_set_config_call;
-- 
2.52.0

[PATCH v2 10/10] vhost-user: add VHOST_USER_PROTOCOL_F_GPA_ADDRESSES

[Vladimir Sementsov-Ogievskiy]

Unlike the kernel, vhost-user backend knows nothing about QEMU's
userspace addresses. We can pass GPA instead and nothing changes.
Generally, vhost-user servers need these addresses only to calculate
offsets inside memory regions. Still, some servers (QEMU's internal
is one example) may do checks for passed addresses to be "userspace
addresses", for example check for non-zero. That's why we need
additional negotiation for the feature.

The benefit: this opens the doors for further implementation of
local migration (live-update) with passing open vhost-related FDs
through UNIX domain socket. This way the connection with backend is
kept live and untouched.

Without this change, we would have to communicate with backend to
inform it about UVA address changes, but it's better to simply use
more stable GPA numbers, which don't change after migration.

Additionally, the current implementation exposes QEMU's process
address space by passing UVA, which breaks ASLR. New protocol
feature avoids that.

Note, that we do nothing with backend messages and replies.
Frontends have to work with backends userspace addresses anyway,
because they come from userfaultfd.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
---
 docs/interop/vhost-user.rst    | 21 +++++++++++++++++----
 hw/virtio/vhost-user.c         | 22 +++++++++++++++++-----
 include/hw/virtio/vhost-user.h |  1 +
 3 files changed, 35 insertions(+), 9 deletions(-)

diff --git a/docs/interop/vhost-user.rst b/docs/interop/vhost-user.rst
index 137c9f3669d..8dcb030c530 100644
--- a/docs/interop/vhost-user.rst
+++ b/docs/interop/vhost-user.rst
@@ -164,8 +164,16 @@ A vring address description
 
 :log: a 64-bit guest address for logging
 
-Note that a ring address is an IOVA if ``VIRTIO_F_IOMMU_PLATFORM`` has
-been negotiated. Otherwise it is a user address.
+.. Note::
+   When ``VIRTIO_F_IOMMU_PLATFORM`` is negotiated, ring addresses are IOVAs.
+
+   Otherwise, when ``VHOST_USER_PROTOCOL_F_GPA_ADDRESSES`` is negotiated, the
+   ring addresses are guest physical addresses for frontend messages. That
+   does not apply to backend replies.
+
+   Finally, when neither ``VIRTIO_F_IOMMU_PLATFORM`` nor
+   ``VHOST_USER_PROTOCOL_F_GPA_ADDRESSES`` features are negotiated, ring
+   addresses are user virtual addresses.
 
 .. _memory_region_description:
 
@@ -180,7 +188,9 @@ Memory region description
 
 :size: a 64-bit size
 
-:user address: a 64-bit user address
+:user address: a 64-bit user address. When ``VHOST_USER_PROTOCOL_F_GPA_ADDRESSES``
+  is negotiated, this field contain guest physical address instead and must
+  duplicate ``guest address`` field.
 
 :mmap offset: a 64-bit offset where region starts in the mapped memory
 
@@ -252,7 +262,9 @@ An IOTLB message
 
 :size: a 64-bit size
 
-:user address: a 64-bit user address
+:user address: a 64-bit user address. When ``VHOST_USER_PROTOCOL_F_GPA_ADDRESSES``
+  is negotiated, this field contain guest physical address instead, except for
+  ``VHOST_USER_BACKEND_IOTLB_MSG``, where it's user address anyway.
 
 :permissions flags: an 8-bit value:
   - 0: No access
@@ -1063,6 +1075,7 @@ Protocol features
   #define VHOST_USER_PROTOCOL_F_SHARED_OBJECT           18
   #define VHOST_USER_PROTOCOL_F_DEVICE_STATE            19
   #define VHOST_USER_PROTOCOL_F_GET_VRING_BASE_INFLIGHT 20
+  #define VHOST_USER_PROTOCOL_F_GPA_ADDRESSES           21
 
 Front-end message types
 -----------------------
diff --git a/hw/virtio/vhost-user.c b/hw/virtio/vhost-user.c
index 78ffb25d6b1..15042c0f3c2 100644
--- a/hw/virtio/vhost-user.c
+++ b/hw/virtio/vhost-user.c
@@ -563,12 +563,22 @@ static MemoryRegion *vhost_user_get_mr_data(uint64_t addr, ram_addr_t *offset,
     return mr;
 }
 
-static void vhost_user_fill_msg_region(VhostUserMemoryRegion *dst,
+static bool vhost_user_gpa_addresses(struct vhost_dev *dev)
+{
+    return vhost_user_has_protocol_feature(
+        dev, VHOST_USER_PROTOCOL_F_GPA_ADDRESSES);
+}
+
+static void vhost_user_fill_msg_region(struct vhost_dev *dev,
+                                       VhostUserMemoryRegion *dst,
                                        struct vhost_memory_region *src,
                                        uint64_t mmap_offset)
 {
+    bool use_phys = vhost_user_gpa_addresses(dev);
+
     assert(src != NULL && dst != NULL);
-    dst->userspace_addr = src->userspace_addr;
+
+    dst->userspace_addr = use_phys ? src->guest_phys_addr : src->userspace_addr;
     dst->memory_size = src->memory_size;
     dst->guest_phys_addr = src->guest_phys_addr;
     dst->mmap_offset = mmap_offset;
@@ -606,7 +616,7 @@ static int vhost_user_fill_set_mem_table_msg(struct vhost_user *u,
                 error_report("Failed preparing vhost-user memory table msg");
                 return -ENOBUFS;
             }
-            vhost_user_fill_msg_region(&region_buffer, reg, offset);
+            vhost_user_fill_msg_region(dev, &region_buffer, reg, offset);
             msg->payload.memory.regions[*fd_num] = region_buffer;
             fds[(*fd_num)++] = fd;
         } else if (track_ramblocks) {
@@ -752,7 +762,7 @@ static int send_remove_regions(struct vhost_dev *dev,
 
         if (fd > 0) {
             msg->hdr.request = VHOST_USER_REM_MEM_REG;
-            vhost_user_fill_msg_region(&region_buffer, shadow_reg, 0);
+            vhost_user_fill_msg_region(dev, &region_buffer, shadow_reg, 0);
             msg->payload.mem_reg.region = region_buffer;
 
             ret = vhost_user_write(dev, msg, NULL, 0);
@@ -813,7 +823,7 @@ static int send_add_regions(struct vhost_dev *dev,
                 u->region_rb[reg_idx] = mr->ram_block;
             }
             msg->hdr.request = VHOST_USER_ADD_MEM_REG;
-            vhost_user_fill_msg_region(&region_buffer, reg, offset);
+            vhost_user_fill_msg_region(dev, &region_buffer, reg, offset);
             msg->payload.mem_reg.region = region_buffer;
 
             ret = vhost_user_write(dev, msg, &fd, 1);
@@ -3151,4 +3161,6 @@ const VhostOps user_ops = {
         .vhost_supports_device_state = vhost_user_supports_device_state,
         .vhost_set_device_state_fd = vhost_user_set_device_state_fd,
         .vhost_check_device_state = vhost_user_check_device_state,
+        .vhost_phys_vring_addr = vhost_user_gpa_addresses,
+        .vhost_phys_iotlb_msg = vhost_user_gpa_addresses,
 };
diff --git a/include/hw/virtio/vhost-user.h b/include/hw/virtio/vhost-user.h
index 0e576fc5386..5ab70dab7e4 100644
--- a/include/hw/virtio/vhost-user.h
+++ b/include/hw/virtio/vhost-user.h
@@ -34,6 +34,7 @@ enum VhostUserProtocolFeature {
     VHOST_USER_PROTOCOL_F_SHARED_OBJECT = 18,
     VHOST_USER_PROTOCOL_F_DEVICE_STATE = 19,
     VHOST_USER_PROTOCOL_F_GET_VRING_BASE_INFLIGHT = 20,
+    VHOST_USER_PROTOCOL_F_GPA_ADDRESSES = 21,
     VHOST_USER_PROTOCOL_F_MAX
 };
 
-- 
2.52.0

Re: [PATCH v2 06/10] qapi: remove user addresses from x-query-virtio-vhost-queue-status

[Markus Armbruster]

Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru> writes:

> Semantics of these (actually, internal) fields is not simple, they may
> contain either virtual or physical addresses. We are going to change
> this to simplify the logic. Keeping this logic only for unstable
> info command seems too much. Changing semantics of info fields doesn't
> seem to make real sense too. So, let's just drop them. We can dot it,

s/dot/do/

> as command is experimental.

Nitpick: the schema has

    # @unstable: This command is meant for debugging.

Suggest "as the command is unstable".

> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>

QAPI schema
Acked-by: Markus Armbruster <armbru@redhat.com>

Re: [PATCH v2 00/10] vhost-user: VHOST_USER_PROTOCOL_F_GPA_ADDRESSES

[Vladimir Sementsov-Ogievskiy]

ping

On 20.04.26 23:20, Vladimir Sementsov-Ogievskiy wrote:
> Hi all!
> 
> Here is a new feature to pass GPA addresses instead of UVA
> to vhost-user server from QEMU. The details are in last patch
> 10/10 the others are preparation and refactoring.
> 
> v2:
> 06: reword commit message
> 
> Based-on: <20260420200339.708640-1-vsementsov@yandex-team.ru>
>     "[PATCH v6 00/22] vhost refactoring and fixes"
> 
> Vladimir Sementsov-Ogievskiy (10):
>    vhost: move IOTLB functions from vhost-backend.c to vhost.c
>    vhost: use consistent naming for backend handlers
>    vhost: rename vhost-backend.c to vhost-kernel.c
>    vhost: replace .vhost_vq_get_addr() with .vhost_phys_vring_addr()
>    vhost: simplify vhost_memory_map() and vhost_memory_unmap()
>    qapi: remove user addresses from x-query-virtio-vhost-queue-status
>    vhost: stop reusing vq->desc (and friends) to store physical address
>    vhost: rename vring pointer fields to reflect user addresses
>    vhost: add .vhost_phys_iotlb_msg() handler
>    vhost-user: add VHOST_USER_PROTOCOL_F_GPA_ADDRESSES
> 
>   docs/interop/vhost-user.rst                   |  21 +-
>   hw/virtio/meson.build                         |   2 +-
>   hw/virtio/{vhost-backend.c => vhost-kernel.c} |  90 +-------
>   hw/virtio/vhost-user.c                        |  32 ++-
>   hw/virtio/vhost-vdpa.c                        |  19 +-
>   hw/virtio/vhost.c                             | 193 +++++++++++++-----
>   hw/virtio/virtio-hmp-cmds.c                   |   3 -
>   hw/virtio/virtio-qmp.c                        |   3 -
>   include/hw/virtio/vhost-backend.h             |  37 ++--
>   include/hw/virtio/vhost-user.h                |   1 +
>   include/hw/virtio/vhost.h                     |   8 +-
>   qapi/virtio.json                              |  15 --
>   12 files changed, 210 insertions(+), 214 deletions(-)
>   rename hw/virtio/{vhost-backend.c => vhost-kernel.c} (83%)
> 


-- 
Best regards,
Vladimir