[v2,2/4] vhost-user-blk: introduce a new vhost-user-blk host device

This commit introduces a new vhost-user device for block, it uses a
chardev to connect with the backend, same with Qemu virito-blk device,
Guest OS still uses the virtio-blk frontend driver.

To use it, start Qemu with command line like this:

qemu-system-x86_64 \
    -chardev socket,id=char0,path=/path/vhost.socket \
    -device vhost-user-blk-pci,chardev=char0,num_queues=...

Different with exist Qemu virtio-blk host device, it makes more easy
for users to implement their own I/O processing logic, such as all
user space I/O stack against hardware block device. It uses the new
vhost messages(VHOST_USER_GET_CONFIG) to get block virtio config
information from backend process.

Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
---
 configure                          |  11 ++
 hw/block/Makefile.objs             |   3 +
 hw/block/vhost-user-blk.c          | 360 +++++++++++++++++++++++++++++++++++++
 hw/virtio/virtio-pci.c             |  55 ++++++
 hw/virtio/virtio-pci.h             |  18 ++
 include/hw/virtio/vhost-user-blk.h |  40 +++++
 6 files changed, 487 insertions(+)
 create mode 100644 hw/block/vhost-user-blk.c
 create mode 100644 include/hw/virtio/vhost-user-blk.h

Hi

----- Original Message -----
> This commit introduces a new vhost-user device for block, it uses a
> chardev to connect with the backend, same with Qemu virito-blk device,
> Guest OS still uses the virtio-blk frontend driver.
> 
> To use it, start Qemu with command line like this:
> 
> qemu-system-x86_64 \
>     -chardev socket,id=char0,path=/path/vhost.socket \
>     -device vhost-user-blk-pci,chardev=char0,num_queues=...
> 
> Different with exist Qemu virtio-blk host device, it makes more easy
> for users to implement their own I/O processing logic, such as all
> user space I/O stack against hardware block device. It uses the new
> vhost messages(VHOST_USER_GET_CONFIG) to get block virtio config
> information from backend process.
> 
> Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
> ---
>  configure                          |  11 ++
>  hw/block/Makefile.objs             |   3 +
>  hw/block/vhost-user-blk.c          | 360
>  +++++++++++++++++++++++++++++++++++++
>  hw/virtio/virtio-pci.c             |  55 ++++++
>  hw/virtio/virtio-pci.h             |  18 ++
>  include/hw/virtio/vhost-user-blk.h |  40 +++++
>  6 files changed, 487 insertions(+)
>  create mode 100644 hw/block/vhost-user-blk.c
>  create mode 100644 include/hw/virtio/vhost-user-blk.h
> 
> diff --git a/configure b/configure
> index dd73cce..1452c66 100755
> --- a/configure
> +++ b/configure
> @@ -305,6 +305,7 @@ tcg="yes"
>  
>  vhost_net="no"
>  vhost_scsi="no"
> +vhost_user_blk="no"
>  vhost_vsock="no"
>  vhost_user=""
>  kvm="no"
> @@ -779,6 +780,7 @@ Linux)
>    kvm="yes"
>    vhost_net="yes"
>    vhost_scsi="yes"
> +  vhost_user_blk="yes"
>    vhost_vsock="yes"
>    QEMU_INCLUDES="-I\$(SRC_PATH)/linux-headers -I$(pwd)/linux-headers
>    $QEMU_INCLUDES"
>    supported_os="yes"
> @@ -1136,6 +1138,10 @@ for opt do
>    ;;
>    --enable-vhost-scsi) vhost_scsi="yes"
>    ;;
> +  --disable-vhost-user-blk) vhost_user_blk="no"
> +  ;;
> +  --enable-vhost-user-blk) vhost_user_blk="yes"
> +  ;;

I suggest we don't add yet another configure option, but reuse the recently introduced --enable-vhost-user (that should cover all vhost-user devices for now, but may learn to enable specific devices if needed in the future).

>    --disable-vhost-vsock) vhost_vsock="no"
>    ;;
>    --enable-vhost-vsock) vhost_vsock="yes"
> @@ -1506,6 +1512,7 @@ disabled with --disable-FEATURE, default is enabled if
> available:
>    cap-ng          libcap-ng support
>    attr            attr and xattr support
>    vhost-net       vhost-net acceleration support
> +  vhost-user-blk  VM virtio-blk acceleration in user space
>    spice           spice
>    rbd             rados block device (rbd)
>    libiscsi        iscsi support
> @@ -5365,6 +5372,7 @@ echo "posix_madvise     $posix_madvise"
>  echo "libcap-ng support $cap_ng"
>  echo "vhost-net support $vhost_net"
>  echo "vhost-scsi support $vhost_scsi"
> +echo "vhost-user-blk support $vhost_user_blk"
>  echo "vhost-vsock support $vhost_vsock"
>  echo "vhost-user support $vhost_user"
>  echo "Trace backends    $trace_backends"
> @@ -5776,6 +5784,9 @@ fi
>  if test "$vhost_scsi" = "yes" ; then
>    echo "CONFIG_VHOST_SCSI=y" >> $config_host_mak
>  fi
> +if test "$vhost_user_blk" = "yes" ; then
> +  echo "CONFIG_VHOST_USER_BLK=y" >> $config_host_mak
> +fi
>  if test "$vhost_net" = "yes" -a "$vhost_user" = "yes"; then
>    echo "CONFIG_VHOST_NET_USED=y" >> $config_host_mak
>  fi
> diff --git a/hw/block/Makefile.objs b/hw/block/Makefile.objs
> index e0ed980..4c19a58 100644
> --- a/hw/block/Makefile.objs
> +++ b/hw/block/Makefile.objs
> @@ -13,3 +13,6 @@ obj-$(CONFIG_SH4) += tc58128.o
>  
>  obj-$(CONFIG_VIRTIO) += virtio-blk.o
>  obj-$(CONFIG_VIRTIO) += dataplane/
> +ifeq ($(CONFIG_VIRTIO),y)
> +obj-$(CONFIG_VHOST_USER_BLK) += vhost-user-blk.o
> +endif
> diff --git a/hw/block/vhost-user-blk.c b/hw/block/vhost-user-blk.c
> new file mode 100644
> index 0000000..8aa9fa9
> --- /dev/null
> +++ b/hw/block/vhost-user-blk.c
> @@ -0,0 +1,360 @@
> +/*
> + * vhost-user-blk host device
> + *
> + * Copyright IBM, Corp. 2011
> + * Copyright(C) 2017 Intel Corporation.
> + *
> + * Authors:
> + *  Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
> + *  Changpeng Liu <changpeng.liu@intel.com>
> + *
> + * This work is licensed under the terms of the GNU LGPL, version 2 or
> later.
> + * See the COPYING.LIB file in the top-level directory.
> + *
> + */
> +
> +#include "qemu/osdep.h"
> +#include "qapi/error.h"
> +#include "qemu/error-report.h"
> +#include "qemu/typedefs.h"
> +#include "qemu/cutils.h"
> +#include "qom/object.h"
> +#include "hw/qdev-core.h"
> +#include "hw/virtio/vhost.h"
> +#include "hw/virtio/vhost-user-blk.h"
> +#include "hw/virtio/virtio.h"
> +#include "hw/virtio/virtio-bus.h"
> +#include "hw/virtio/virtio-access.h"
> +
> +static const int user_feature_bits[] = {
> +    VIRTIO_BLK_F_SIZE_MAX,
> +    VIRTIO_BLK_F_SEG_MAX,
> +    VIRTIO_BLK_F_GEOMETRY,
> +    VIRTIO_BLK_F_BLK_SIZE,
> +    VIRTIO_BLK_F_TOPOLOGY,
> +    VIRTIO_BLK_F_SCSI,
> +    VIRTIO_BLK_F_MQ,
> +    VIRTIO_BLK_F_RO,
> +    VIRTIO_BLK_F_FLUSH,
> +    VIRTIO_BLK_F_BARRIER,
> +    VIRTIO_BLK_F_WCE,
> +    VIRTIO_F_VERSION_1,
> +    VIRTIO_RING_F_INDIRECT_DESC,
> +    VIRTIO_RING_F_EVENT_IDX,
> +    VIRTIO_F_NOTIFY_ON_EMPTY,
> +    VHOST_INVALID_FEATURE_BIT
> +};
> +
> +static void vhost_user_blk_update_config(VirtIODevice *vdev, uint8_t
> *config)
> +{
> +    VHostUserBlk *s = VHOST_USER_BLK(vdev);
> +
> +    memcpy(config, &s->blkcfg, sizeof(struct virtio_blk_config));
> +}
> +
> +static void vhost_user_blk_set_config(VirtIODevice *vdev, const uint8_t
> *config)
> +{
> +    VHostUserBlk *s = VHOST_USER_BLK(vdev);
> +    struct virtio_blk_config *blkcfg = (struct virtio_blk_config *)config;
> +    int ret;
> +
> +    if (blkcfg->wce == s->blkcfg.wce) {
> +        return;

Is write-cache the only config change the slave is interested in?

> +    }
> +
> +    ret = vhost_dev_set_config(&s->dev, config,
> +                              sizeof(struct virtio_blk_config));
> +    if (ret) {
> +        error_report("set device config space failed");
> +        return;
> +    }
> +
> +    s->blkcfg.wce = blkcfg->wce;
> +}
> +
> +static void vhost_user_blk_handle_config_change(struct vhost_dev *dev)
> +{
> +    int ret;
> +    struct virtio_blk_config blkcfg;
> +    VHostUserBlk *s = VHOST_USER_BLK(dev->vdev);
> +
> +    ret = vhost_dev_get_config(dev, (uint8_t *)&blkcfg,
> +                               sizeof(struct virtio_blk_config));
> +    if (ret < 0) {
> +        error_report("get config space failed");
> +        return;
> +    }
> +
> +    memcpy(&s->blkcfg, &blkcfg, sizeof(struct virtio_blk_config));
> +    memcpy(dev->vdev->config, &blkcfg, sizeof(struct virtio_blk_config));

Why do you need to have s->blkcfg if you can use dev->vdev->config ?

> +    virtio_notify_config(dev->vdev);
> +}
> +
> +const VhostDevConfigOps blk_ops = {
> +    .vhost_dev_config_notifier = vhost_user_blk_handle_config_change,
> +};
> +
> +static void vhost_user_blk_start(VirtIODevice *vdev)
> +{
> +    VHostUserBlk *s = VHOST_USER_BLK(vdev);
> +    BusState *qbus = BUS(qdev_get_parent_bus(DEVICE(vdev)));
> +    VirtioBusClass *k = VIRTIO_BUS_GET_CLASS(qbus);
> +    int i, ret;
> +
> +    if (!k->set_guest_notifiers) {
> +        error_report("binding does not support guest notifiers");
> +        return;
> +    }
> +
> +    ret = vhost_dev_enable_notifiers(&s->dev, vdev);
> +    if (ret < 0) {
> +        error_report("Error enabling host notifiers: %d", -ret);
> +        return;
> +    }
> +
> +    ret = k->set_guest_notifiers(qbus->parent, s->dev.nvqs, true);
> +    if (ret < 0) {
> +        error_report("Error binding guest notifier: %d", -ret);
> +        goto err_host_notifiers;
> +    }
> +
> +    s->dev.acked_features = vdev->guest_features;
> +    ret = vhost_dev_start(&s->dev, vdev);
> +    if (ret < 0) {
> +        error_report("Error starting vhost: %d", -ret);
> +        goto err_guest_notifiers;
> +    }
> +
> +    /* guest_notifier_mask/pending not used yet, so just unmask
> +     * everything here. virtio-pci will do the right thing by
> +     * enabling/disabling irqfd.
> +     */
> +    for (i = 0; i < s->dev.nvqs; i++) {
> +        vhost_virtqueue_mask(&s->dev, vdev, i, false);
> +    }
> +
> +    return;
> +
> +err_guest_notifiers:
> +    k->set_guest_notifiers(qbus->parent, s->dev.nvqs, false);
> +err_host_notifiers:
> +    vhost_dev_disable_notifiers(&s->dev, vdev);
> +}
> +
> +static void vhost_user_blk_stop(VirtIODevice *vdev)
> +{
> +    VHostUserBlk *s = VHOST_USER_BLK(vdev);
> +    BusState *qbus = BUS(qdev_get_parent_bus(DEVICE(vdev)));
> +    VirtioBusClass *k = VIRTIO_BUS_GET_CLASS(qbus);
> +    int ret;
> +
> +    if (!k->set_guest_notifiers) {
> +        return;
> +    }
> +
> +    vhost_dev_stop(&s->dev, vdev);
> +
> +    ret = k->set_guest_notifiers(qbus->parent, s->dev.nvqs, false);
> +    if (ret < 0) {
> +        error_report("vhost guest notifier cleanup failed: %d", ret);
> +        return;
> +    }
> +
> +    vhost_dev_disable_notifiers(&s->dev, vdev);
> +}
> +
> +static void vhost_user_blk_set_status(VirtIODevice *vdev, uint8_t status)
> +{
> +    VHostUserBlk *s = VHOST_USER_BLK(vdev);
> +    bool should_start = status & VIRTIO_CONFIG_S_DRIVER_OK;
> +
> +    if (!vdev->vm_running) {
> +        should_start = false;
> +    }
> +
> +    if (s->dev.started == should_start) {
> +        return;
> +    }
> +
> +    if (should_start) {
> +        vhost_user_blk_start(vdev);
> +    } else {
> +        vhost_user_blk_stop(vdev);
> +    }
> +
> +}
> +
> +static uint64_t vhost_user_blk_get_features(VirtIODevice *vdev,
> +                                            uint64_t features,
> +                                            Error **errp)
> +{
> +    VHostUserBlk *s = VHOST_USER_BLK(vdev);
> +    uint64_t get_features;
> +
> +    /* Turn on pre-defined features */
> +    features |= s->host_features;
> +
> +    get_features = vhost_get_features(&s->dev, user_feature_bits, features);
> +
> +    return get_features;
> +}
> +
> +static void vhost_user_blk_handle_output(VirtIODevice *vdev, VirtQueue *vq)
> +{
> +
> +}
> +
> +static void vhost_user_blk_device_realize(DeviceState *dev, Error **errp)
> +{
> +    VirtIODevice *vdev = VIRTIO_DEVICE(dev);
> +    VHostUserBlk *s = VHOST_USER_BLK(vdev);
> +    int i, ret;
> +
> +    if (!s->chardev.chr) {
> +        error_setg(errp, "vhost-user-blk: chardev is mandatory");
> +        return;
> +    }
> +
> +    if (!s->num_queues || s->num_queues > VIRTIO_QUEUE_MAX) {
> +        error_setg(errp, "vhost-user-blk: invalid number of IO queues");
> +        return;
> +    }
> +
> +    if (!s->queue_size) {
> +        error_setg(errp, "vhost-user-blk: queue size must be non-zero");
> +        return;
> +    }
> +
> +    virtio_init(vdev, "virtio-blk", VIRTIO_ID_BLOCK,
> +                sizeof(struct virtio_blk_config));
> +
> +    for (i = 0; i < s->num_queues; i++) {
> +        virtio_add_queue(vdev, s->queue_size,
> +                         vhost_user_blk_handle_output);
> +    }
> +
> +    s->dev.nvqs = s->num_queues;
> +    s->dev.vqs = g_new(struct vhost_virtqueue, s->dev.nvqs);
> +    s->dev.vq_index = 0;
> +    s->dev.backend_features = 0;
> +
> +    ret = vhost_dev_init(&s->dev, &s->chardev, VHOST_BACKEND_TYPE_USER, 0);
> +    if (ret < 0) {
> +        error_setg(errp, "vhost-user-blk: vhost initialization failed: %s",
> +                   strerror(-ret));
> +        goto virtio_err;
> +    }
> +
> +    ret = vhost_dev_get_config(&s->dev, (uint8_t *)&s->blkcfg,
> +                              sizeof(struct virtio_blk_config));
> +    if (ret < 0) {
> +        error_setg(errp, "vhost-user-blk: get block config failed");
> +        goto vhost_err;
> +    }
> +
> +    if (s->blkcfg.num_queues != s->num_queues) {
> +        s->blkcfg.num_queues = s->num_queues;
> +    }
> +
> +    vhost_dev_set_config_notifier(&s->dev, &blk_ops);
> +
> +    return;
> +
> +vhost_err:
> +    vhost_dev_cleanup(&s->dev);
> +virtio_err:
> +    g_free(s->dev.vqs);
> +    virtio_cleanup(vdev);
> +}
> +
> +static void vhost_user_blk_device_unrealize(DeviceState *dev, Error **errp)
> +{
> +    VirtIODevice *vdev = VIRTIO_DEVICE(dev);
> +    VHostUserBlk *s = VHOST_USER_BLK(dev);
> +
> +    vhost_user_blk_set_status(vdev, 0);
> +    vhost_dev_cleanup(&s->dev);
> +    g_free(s->dev.vqs);
> +    virtio_cleanup(vdev);
> +}
> +
> +static void vhost_user_blk_instance_init(Object *obj)
> +{
> +    VHostUserBlk *s = VHOST_USER_BLK(obj);
> +
> +    device_add_bootindex_property(obj, &s->bootindex, "bootindex",
> +                                  "/disk@0,0", DEVICE(obj), NULL);
> +}
> +
> +static const VMStateDescription vmstate_vhost_user_blk = {
> +    .name = "vhost-user-blk",
> +    .minimum_version_id = 1,
> +    .version_id = 1,
> +    .fields = (VMStateField[]) {
> +        VMSTATE_VIRTIO_DEVICE,
> +        VMSTATE_END_OF_LIST()
> +    },
> +};
> +
> +static Property vhost_user_blk_properties[] = {
> +    DEFINE_PROP_CHR("chardev", VHostUserBlk, chardev),
> +    DEFINE_PROP_UINT16("num_queues", VHostUserBlk, num_queues, 1),
> +    DEFINE_PROP_UINT32("queue_size", VHostUserBlk, queue_size, 128),
> +    DEFINE_PROP_BIT64("f_size_max", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_SIZE_MAX, true),
> +    DEFINE_PROP_BIT64("f_sizemax", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_SIZE_MAX, true),
> +    DEFINE_PROP_BIT64("f_segmax", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_SEG_MAX, true),
> +    DEFINE_PROP_BIT64("f_geometry", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_GEOMETRY, true),
> +    DEFINE_PROP_BIT64("f_readonly", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_RO, false),
> +    DEFINE_PROP_BIT64("f_blocksize", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_BLK_SIZE, true),
> +    DEFINE_PROP_BIT64("f_topology", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_TOPOLOGY, true),
> +    DEFINE_PROP_BIT64("f_multiqueue", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_MQ, true),
> +    DEFINE_PROP_BIT64("f_flush", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_FLUSH, true),
> +    DEFINE_PROP_BIT64("f_barrier", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_BARRIER, false),
> +    DEFINE_PROP_BIT64("f_scsi", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_SCSI, false),
> +    DEFINE_PROP_BIT64("f_wce", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_WCE, false),
> +    DEFINE_PROP_END_OF_LIST(),
> +};
> +
> +static void vhost_user_blk_class_init(ObjectClass *klass, void *data)
> +{
> +    DeviceClass *dc = DEVICE_CLASS(klass);
> +    VirtioDeviceClass *vdc = VIRTIO_DEVICE_CLASS(klass);
> +
> +    dc->props = vhost_user_blk_properties;
> +    dc->vmsd = &vmstate_vhost_user_blk;
> +    set_bit(DEVICE_CATEGORY_STORAGE, dc->categories);
> +    vdc->realize = vhost_user_blk_device_realize;
> +    vdc->unrealize = vhost_user_blk_device_unrealize;
> +    vdc->get_config = vhost_user_blk_update_config;
> +    vdc->set_config = vhost_user_blk_set_config;
> +    vdc->get_features = vhost_user_blk_get_features;
> +    vdc->set_status = vhost_user_blk_set_status;
> +}
> +
> +static const TypeInfo vhost_user_blk_info = {
> +    .name = TYPE_VHOST_USER_BLK,
> +    .parent = TYPE_VIRTIO_DEVICE,
> +    .instance_size = sizeof(VHostUserBlk),
> +    .instance_init = vhost_user_blk_instance_init,
> +    .class_init = vhost_user_blk_class_init,
> +};
> +
> +static void virtio_register_types(void)
> +{
> +    type_register_static(&vhost_user_blk_info);
> +}
> +
> +type_init(virtio_register_types)
> diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
> index 8b0d6b6..be9a992 100644
> --- a/hw/virtio/virtio-pci.c
> +++ b/hw/virtio/virtio-pci.c
> @@ -2012,6 +2012,58 @@ static const TypeInfo virtio_blk_pci_info = {
>      .class_init    = virtio_blk_pci_class_init,
>  };
>  
> +#ifdef CONFIG_VHOST_USER_BLK
> +/* vhost-user-blk */
> +
> +static Property vhost_user_blk_pci_properties[] = {
> +    DEFINE_PROP_UINT32("class", VirtIOPCIProxy, class_code, 0),
> +    DEFINE_PROP_UINT32("vectors", VirtIOPCIProxy, nvectors, 2),
> +    DEFINE_PROP_END_OF_LIST(),
> +};
> +
> +static void vhost_user_blk_pci_realize(VirtIOPCIProxy *vpci_dev, Error
> **errp)
> +{
> +    VHostUserBlkPCI *dev = VHOST_USER_BLK_PCI(vpci_dev);
> +    DeviceState *vdev = DEVICE(&dev->vdev);
> +
> +    qdev_set_parent_bus(vdev, BUS(&vpci_dev->bus));
> +    object_property_set_bool(OBJECT(vdev), true, "realized", errp);
> +}
> +
> +static void vhost_user_blk_pci_class_init(ObjectClass *klass, void *data)
> +{
> +    DeviceClass *dc = DEVICE_CLASS(klass);
> +    VirtioPCIClass *k = VIRTIO_PCI_CLASS(klass);
> +    PCIDeviceClass *pcidev_k = PCI_DEVICE_CLASS(klass);
> +
> +    set_bit(DEVICE_CATEGORY_STORAGE, dc->categories);
> +    dc->props = vhost_user_blk_pci_properties;
> +    k->realize = vhost_user_blk_pci_realize;
> +    pcidev_k->vendor_id = PCI_VENDOR_ID_REDHAT_QUMRANET;
> +    pcidev_k->device_id = PCI_DEVICE_ID_VIRTIO_BLOCK;
> +    pcidev_k->revision = VIRTIO_PCI_ABI_VERSION;
> +    pcidev_k->class_id = PCI_CLASS_STORAGE_SCSI;
> +}
> +
> +static void vhost_user_blk_pci_instance_init(Object *obj)
> +{
> +    VHostUserBlkPCI *dev = VHOST_USER_BLK_PCI(obj);
> +
> +    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
> +                                TYPE_VHOST_USER_BLK);
> +    object_property_add_alias(obj, "bootindex", OBJECT(&dev->vdev),
> +                              "bootindex", &error_abort);
> +}
> +
> +static const TypeInfo vhost_user_blk_pci_info = {
> +    .name           = TYPE_VHOST_USER_BLK_PCI,
> +    .parent         = TYPE_VIRTIO_PCI,
> +    .instance_size  = sizeof(VHostUserBlkPCI),
> +    .instance_init  = vhost_user_blk_pci_instance_init,
> +    .class_init     = vhost_user_blk_pci_class_init,
> +};
> +#endif
> +
>  /* virtio-scsi-pci */
>  
>  static Property virtio_scsi_pci_properties[] = {
> @@ -2658,6 +2710,9 @@ static void virtio_pci_register_types(void)
>      type_register_static(&virtio_9p_pci_info);
>  #endif
>      type_register_static(&virtio_blk_pci_info);
> +#ifdef CONFIG_VHOST_USER_BLK
> +    type_register_static(&vhost_user_blk_pci_info);
> +#endif
>      type_register_static(&virtio_scsi_pci_info);
>      type_register_static(&virtio_balloon_pci_info);
>      type_register_static(&virtio_serial_pci_info);
> diff --git a/hw/virtio/virtio-pci.h b/hw/virtio/virtio-pci.h
> index 69f5959..19a0d01 100644
> --- a/hw/virtio/virtio-pci.h
> +++ b/hw/virtio/virtio-pci.h
> @@ -27,6 +27,9 @@
>  #include "hw/virtio/virtio-gpu.h"
>  #include "hw/virtio/virtio-crypto.h"
>  #include "hw/virtio/vhost-user-scsi.h"
> +#ifdef CONFIG_VHOST_USER_BLK
> +#include "hw/virtio/vhost-user-blk.h"
> +#endif
>  
>  #ifdef CONFIG_VIRTFS
>  #include "hw/9pfs/virtio-9p.h"
> @@ -46,6 +49,7 @@ typedef struct VirtIOSerialPCI VirtIOSerialPCI;
>  typedef struct VirtIONetPCI VirtIONetPCI;
>  typedef struct VHostSCSIPCI VHostSCSIPCI;
>  typedef struct VHostUserSCSIPCI VHostUserSCSIPCI;
> +typedef struct VHostUserBlkPCI VHostUserBlkPCI;
>  typedef struct VirtIORngPCI VirtIORngPCI;
>  typedef struct VirtIOInputPCI VirtIOInputPCI;
>  typedef struct VirtIOInputHIDPCI VirtIOInputHIDPCI;
> @@ -241,6 +245,20 @@ struct VHostUserSCSIPCI {
>      VHostUserSCSI vdev;
>  };
>  
> +#ifdef CONFIG_VHOST_USER_BLK
> +/*
> + * vhost-user-blk-pci: This extends VirtioPCIProxy.
> + */
> +#define TYPE_VHOST_USER_BLK_PCI "vhost-user-blk-pci"
> +#define VHOST_USER_BLK_PCI(obj) \
> +        OBJECT_CHECK(VHostUserBlkPCI, (obj), TYPE_VHOST_USER_BLK_PCI)
> +
> +struct VHostUserBlkPCI {
> +    VirtIOPCIProxy parent_obj;
> +    VHostUserBlk vdev;
> +};
> +#endif
> +
>  /*
>   * virtio-blk-pci: This extends VirtioPCIProxy.
>   */
> diff --git a/include/hw/virtio/vhost-user-blk.h
> b/include/hw/virtio/vhost-user-blk.h
> new file mode 100644
> index 0000000..77d20f0
> --- /dev/null
> +++ b/include/hw/virtio/vhost-user-blk.h
> @@ -0,0 +1,40 @@
> +/*
> + * vhost-user-blk host device
> + * Copyright IBM, Corp. 2011
> + * Copyright(C) 2017 Intel Corporation.
> + *
> + * Authors:
> + *  Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
> + *  Changpeng Liu <changpeng.liu@intel.com>
> + *
> + * This work is licensed under the terms of the GNU LGPL, version 2 or
> later.
> + * See the COPYING.LIB file in the top-level directory.
> + *
> + */
> +
> +#ifndef VHOST_USER_BLK_H
> +#define VHOST_USER_BLK_H
> +
> +#include "standard-headers/linux/virtio_blk.h"
> +#include "qemu-common.h"
> +#include "hw/qdev.h"
> +#include "hw/block/block.h"
> +#include "chardev/char-fe.h"
> +#include "hw/virtio/vhost.h"
> +
> +#define TYPE_VHOST_USER_BLK "vhost-user-blk"
> +#define VHOST_USER_BLK(obj) \
> +        OBJECT_CHECK(VHostUserBlk, (obj), TYPE_VHOST_USER_BLK)
> +
> +typedef struct VHostUserBlk {
> +    VirtIODevice parent_obj;
> +    CharBackend chardev;
> +    int32_t bootindex;
> +    uint64_t host_features;
> +    struct virtio_blk_config blkcfg;
> +    uint16_t num_queues;
> +    uint32_t queue_size;
> +    struct vhost_dev dev;
> +} VHostUserBlk;
> +
> +#endif
> --
> 1.9.3
> 
>

I only had time for a quick look. More review when
you repost after release.


On Thu, Aug 10, 2017 at 06:12:29PM +0800, Changpeng Liu wrote:
> This commit introduces a new vhost-user device for block, it uses a
> chardev to connect with the backend, same with Qemu virito-blk device,
> Guest OS still uses the virtio-blk frontend driver.
> 
> To use it, start Qemu with command line like this:
> 
> qemu-system-x86_64 \
>     -chardev socket,id=char0,path=/path/vhost.socket \
>     -device vhost-user-blk-pci,chardev=char0,num_queues=...
> 
> Different with exist Qemu virtio-blk host device, it makes more easy
> for users to implement their own I/O processing logic, such as all
> user space I/O stack against hardware block device. It uses the new
> vhost messages(VHOST_USER_GET_CONFIG) to get block virtio config
> information from backend process.

I took a quick look. I think I would prefer a more direct approach
where qemu is more of a driver. So user specifies properties and
they get sent to backend at init time. Only handle geometry changes
specially.

> 
> Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
> ---
>  configure                          |  11 ++
>  hw/block/Makefile.objs             |   3 +
>  hw/block/vhost-user-blk.c          | 360 +++++++++++++++++++++++++++++++++++++
>  hw/virtio/virtio-pci.c             |  55 ++++++
>  hw/virtio/virtio-pci.h             |  18 ++
>  include/hw/virtio/vhost-user-blk.h |  40 +++++
>  6 files changed, 487 insertions(+)
>  create mode 100644 hw/block/vhost-user-blk.c
>  create mode 100644 include/hw/virtio/vhost-user-blk.h
> 
> diff --git a/configure b/configure
> index dd73cce..1452c66 100755
> --- a/configure
> +++ b/configure
> @@ -305,6 +305,7 @@ tcg="yes"
>  
>  vhost_net="no"
>  vhost_scsi="no"
> +vhost_user_blk="no"
>  vhost_vsock="no"
>  vhost_user=""
>  kvm="no"
> @@ -779,6 +780,7 @@ Linux)
>    kvm="yes"
>    vhost_net="yes"
>    vhost_scsi="yes"
> +  vhost_user_blk="yes"
>    vhost_vsock="yes"
>    QEMU_INCLUDES="-I\$(SRC_PATH)/linux-headers -I$(pwd)/linux-headers $QEMU_INCLUDES"
>    supported_os="yes"
> @@ -1136,6 +1138,10 @@ for opt do
>    ;;
>    --enable-vhost-scsi) vhost_scsi="yes"
>    ;;
> +  --disable-vhost-user-blk) vhost_user_blk="no"
> +  ;;
> +  --enable-vhost-user-blk) vhost_user_blk="yes"
> +  ;;
>    --disable-vhost-vsock) vhost_vsock="no"
>    ;;
>    --enable-vhost-vsock) vhost_vsock="yes"
> @@ -1506,6 +1512,7 @@ disabled with --disable-FEATURE, default is enabled if available:
>    cap-ng          libcap-ng support
>    attr            attr and xattr support
>    vhost-net       vhost-net acceleration support
> +  vhost-user-blk  VM virtio-blk acceleration in user space
>    spice           spice
>    rbd             rados block device (rbd)
>    libiscsi        iscsi support
> @@ -5365,6 +5372,7 @@ echo "posix_madvise     $posix_madvise"
>  echo "libcap-ng support $cap_ng"
>  echo "vhost-net support $vhost_net"
>  echo "vhost-scsi support $vhost_scsi"
> +echo "vhost-user-blk support $vhost_user_blk"
>  echo "vhost-vsock support $vhost_vsock"
>  echo "vhost-user support $vhost_user"
>  echo "Trace backends    $trace_backends"
> @@ -5776,6 +5784,9 @@ fi
>  if test "$vhost_scsi" = "yes" ; then
>    echo "CONFIG_VHOST_SCSI=y" >> $config_host_mak
>  fi
> +if test "$vhost_user_blk" = "yes" ; then
> +  echo "CONFIG_VHOST_USER_BLK=y" >> $config_host_mak
> +fi
>  if test "$vhost_net" = "yes" -a "$vhost_user" = "yes"; then
>    echo "CONFIG_VHOST_NET_USED=y" >> $config_host_mak
>  fi
> diff --git a/hw/block/Makefile.objs b/hw/block/Makefile.objs
> index e0ed980..4c19a58 100644
> --- a/hw/block/Makefile.objs
> +++ b/hw/block/Makefile.objs
> @@ -13,3 +13,6 @@ obj-$(CONFIG_SH4) += tc58128.o
>  
>  obj-$(CONFIG_VIRTIO) += virtio-blk.o
>  obj-$(CONFIG_VIRTIO) += dataplane/
> +ifeq ($(CONFIG_VIRTIO),y)
> +obj-$(CONFIG_VHOST_USER_BLK) += vhost-user-blk.o
> +endif
> diff --git a/hw/block/vhost-user-blk.c b/hw/block/vhost-user-blk.c
> new file mode 100644
> index 0000000..8aa9fa9
> --- /dev/null
> +++ b/hw/block/vhost-user-blk.c
> @@ -0,0 +1,360 @@
> +/*
> + * vhost-user-blk host device
> + *
> + * Copyright IBM, Corp. 2011
> + * Copyright(C) 2017 Intel Corporation.
> + *
> + * Authors:
> + *  Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
> + *  Changpeng Liu <changpeng.liu@intel.com>
> + *
> + * This work is licensed under the terms of the GNU LGPL, version 2 or later.
> + * See the COPYING.LIB file in the top-level directory.
> + *
> + */
> +
> +#include "qemu/osdep.h"
> +#include "qapi/error.h"
> +#include "qemu/error-report.h"
> +#include "qemu/typedefs.h"
> +#include "qemu/cutils.h"
> +#include "qom/object.h"
> +#include "hw/qdev-core.h"
> +#include "hw/virtio/vhost.h"
> +#include "hw/virtio/vhost-user-blk.h"
> +#include "hw/virtio/virtio.h"
> +#include "hw/virtio/virtio-bus.h"
> +#include "hw/virtio/virtio-access.h"
> +
> +static const int user_feature_bits[] = {
> +    VIRTIO_BLK_F_SIZE_MAX,
> +    VIRTIO_BLK_F_SEG_MAX,
> +    VIRTIO_BLK_F_GEOMETRY,
> +    VIRTIO_BLK_F_BLK_SIZE,
> +    VIRTIO_BLK_F_TOPOLOGY,
> +    VIRTIO_BLK_F_SCSI,

I don't think we want to support this.


> +    VIRTIO_BLK_F_MQ,
> +    VIRTIO_BLK_F_RO,
> +    VIRTIO_BLK_F_FLUSH,
> +    VIRTIO_BLK_F_BARRIER,
> +    VIRTIO_BLK_F_WCE,
> +    VIRTIO_F_VERSION_1,

How about forcing all remotes to implement this instead?

> +    VIRTIO_RING_F_INDIRECT_DESC,
> +    VIRTIO_RING_F_EVENT_IDX,
> +    VIRTIO_F_NOTIFY_ON_EMPTY,
> +    VHOST_INVALID_FEATURE_BIT

No reason to let remote play with that.

> +};

I think a more reasonable set of features is what Linux uses:

        VIRTIO_BLK_F_SEG_MAX, VIRTIO_BLK_F_SIZE_MAX, VIRTIO_BLK_F_GEOMETRY,
        VIRTIO_BLK_F_RO, VIRTIO_BLK_F_BLK_SIZE,
        VIRTIO_BLK_F_FLUSH, VIRTIO_BLK_F_TOPOLOGY, VIRTIO_BLK_F_CONFIG_WCE,
        VIRTIO_BLK_F_MQ,

and maybe
>     VIRTIO_RING_F_INDIRECT_DESC,
>     VIRTIO_RING_F_EVENT_IDX,

others should be forced by qemu.


> +
> +static void vhost_user_blk_update_config(VirtIODevice *vdev, uint8_t *config)
> +{
> +    VHostUserBlk *s = VHOST_USER_BLK(vdev);
> +
> +    memcpy(config, &s->blkcfg, sizeof(struct virtio_blk_config));
> +}
> +
> +static void vhost_user_blk_set_config(VirtIODevice *vdev, const uint8_t *config)
> +{
> +    VHostUserBlk *s = VHOST_USER_BLK(vdev);
> +    struct virtio_blk_config *blkcfg = (struct virtio_blk_config *)config;
> +    int ret;
> +
> +    if (blkcfg->wce == s->blkcfg.wce) {
> +        return;
> +    }
> +
> +    ret = vhost_dev_set_config(&s->dev, config,
> +                              sizeof(struct virtio_blk_config));
> +    if (ret) {
> +        error_report("set device config space failed");
> +        return;
> +    }
> +
> +    s->blkcfg.wce = blkcfg->wce;
> +}
> +
> +static void vhost_user_blk_handle_config_change(struct vhost_dev *dev)
> +{
> +    int ret;
> +    struct virtio_blk_config blkcfg;
> +    VHostUserBlk *s = VHOST_USER_BLK(dev->vdev);
> +
> +    ret = vhost_dev_get_config(dev, (uint8_t *)&blkcfg,
> +                               sizeof(struct virtio_blk_config));
> +    if (ret < 0) {
> +        error_report("get config space failed");
> +        return;
> +    }
> +
> +    memcpy(&s->blkcfg, &blkcfg, sizeof(struct virtio_blk_config));
> +    memcpy(dev->vdev->config, &blkcfg, sizeof(struct virtio_blk_config));

Will break if virtio_blk_config becomes larger than 256
bytes. Better add a build time assertion.

> +
> +    virtio_notify_config(dev->vdev);
> +}
> +
> +const VhostDevConfigOps blk_ops = {
> +    .vhost_dev_config_notifier = vhost_user_blk_handle_config_change,
> +};
> +
> +static void vhost_user_blk_start(VirtIODevice *vdev)
> +{
> +    VHostUserBlk *s = VHOST_USER_BLK(vdev);
> +    BusState *qbus = BUS(qdev_get_parent_bus(DEVICE(vdev)));
> +    VirtioBusClass *k = VIRTIO_BUS_GET_CLASS(qbus);
> +    int i, ret;
> +
> +    if (!k->set_guest_notifiers) {
> +        error_report("binding does not support guest notifiers");
> +        return;
> +    }
> +
> +    ret = vhost_dev_enable_notifiers(&s->dev, vdev);
> +    if (ret < 0) {
> +        error_report("Error enabling host notifiers: %d", -ret);
> +        return;
> +    }
> +
> +    ret = k->set_guest_notifiers(qbus->parent, s->dev.nvqs, true);
> +    if (ret < 0) {
> +        error_report("Error binding guest notifier: %d", -ret);
> +        goto err_host_notifiers;
> +    }
> +
> +    s->dev.acked_features = vdev->guest_features;
> +    ret = vhost_dev_start(&s->dev, vdev);
> +    if (ret < 0) {
> +        error_report("Error starting vhost: %d", -ret);
> +        goto err_guest_notifiers;
> +    }
> +
> +    /* guest_notifier_mask/pending not used yet, so just unmask
> +     * everything here. virtio-pci will do the right thing by
> +     * enabling/disabling irqfd.
> +     */
> +    for (i = 0; i < s->dev.nvqs; i++) {
> +        vhost_virtqueue_mask(&s->dev, vdev, i, false);
> +    }
> +
> +    return;
> +
> +err_guest_notifiers:
> +    k->set_guest_notifiers(qbus->parent, s->dev.nvqs, false);
> +err_host_notifiers:
> +    vhost_dev_disable_notifiers(&s->dev, vdev);
> +}
> +
> +static void vhost_user_blk_stop(VirtIODevice *vdev)
> +{
> +    VHostUserBlk *s = VHOST_USER_BLK(vdev);
> +    BusState *qbus = BUS(qdev_get_parent_bus(DEVICE(vdev)));
> +    VirtioBusClass *k = VIRTIO_BUS_GET_CLASS(qbus);
> +    int ret;
> +
> +    if (!k->set_guest_notifiers) {
> +        return;
> +    }
> +
> +    vhost_dev_stop(&s->dev, vdev);
> +
> +    ret = k->set_guest_notifiers(qbus->parent, s->dev.nvqs, false);
> +    if (ret < 0) {
> +        error_report("vhost guest notifier cleanup failed: %d", ret);
> +        return;
> +    }
> +
> +    vhost_dev_disable_notifiers(&s->dev, vdev);
> +}
> +
> +static void vhost_user_blk_set_status(VirtIODevice *vdev, uint8_t status)
> +{
> +    VHostUserBlk *s = VHOST_USER_BLK(vdev);
> +    bool should_start = status & VIRTIO_CONFIG_S_DRIVER_OK;
> +
> +    if (!vdev->vm_running) {
> +        should_start = false;
> +    }
> +
> +    if (s->dev.started == should_start) {
> +        return;
> +    }
> +
> +    if (should_start) {
> +        vhost_user_blk_start(vdev);
> +    } else {
> +        vhost_user_blk_stop(vdev);
> +    }
> +
> +}
> +
> +static uint64_t vhost_user_blk_get_features(VirtIODevice *vdev,
> +                                            uint64_t features,
> +                                            Error **errp)
> +{
> +    VHostUserBlk *s = VHOST_USER_BLK(vdev);
> +    uint64_t get_features;
> +
> +    /* Turn on pre-defined features */
> +    features |= s->host_features;
> +
> +    get_features = vhost_get_features(&s->dev, user_feature_bits, features);
> +
> +    return get_features;
> +}
> +
> +static void vhost_user_blk_handle_output(VirtIODevice *vdev, VirtQueue *vq)
> +{

ever called? assert here?

> +
> +}
> +
> +static void vhost_user_blk_device_realize(DeviceState *dev, Error **errp)
> +{
> +    VirtIODevice *vdev = VIRTIO_DEVICE(dev);
> +    VHostUserBlk *s = VHOST_USER_BLK(vdev);
> +    int i, ret;
> +
> +    if (!s->chardev.chr) {
> +        error_setg(errp, "vhost-user-blk: chardev is mandatory");
> +        return;
> +    }
> +
> +    if (!s->num_queues || s->num_queues > VIRTIO_QUEUE_MAX) {
> +        error_setg(errp, "vhost-user-blk: invalid number of IO queues");
> +        return;
> +    }
> +
> +    if (!s->queue_size) {
> +        error_setg(errp, "vhost-user-blk: queue size must be non-zero");
> +        return;
> +    }
> +
> +    virtio_init(vdev, "virtio-blk", VIRTIO_ID_BLOCK,
> +                sizeof(struct virtio_blk_config));
> +
> +    for (i = 0; i < s->num_queues; i++) {
> +        virtio_add_queue(vdev, s->queue_size,
> +                         vhost_user_blk_handle_output);
> +    }
> +
> +    s->dev.nvqs = s->num_queues;
> +    s->dev.vqs = g_new(struct vhost_virtqueue, s->dev.nvqs);
> +    s->dev.vq_index = 0;
> +    s->dev.backend_features = 0;
> +
> +    ret = vhost_dev_init(&s->dev, &s->chardev, VHOST_BACKEND_TYPE_USER, 0);
> +    if (ret < 0) {
> +        error_setg(errp, "vhost-user-blk: vhost initialization failed: %s",
> +                   strerror(-ret));
> +        goto virtio_err;
> +    }
> +
> +    ret = vhost_dev_get_config(&s->dev, (uint8_t *)&s->blkcfg,
> +                              sizeof(struct virtio_blk_config));
> +    if (ret < 0) {
> +        error_setg(errp, "vhost-user-blk: get block config failed");
> +        goto vhost_err;
> +    }
> +
> +    if (s->blkcfg.num_queues != s->num_queues) {
> +        s->blkcfg.num_queues = s->num_queues;
> +    }
> +
> +    vhost_dev_set_config_notifier(&s->dev, &blk_ops);
> +
> +    return;
> +
> +vhost_err:
> +    vhost_dev_cleanup(&s->dev);
> +virtio_err:
> +    g_free(s->dev.vqs);
> +    virtio_cleanup(vdev);
> +}
> +
> +static void vhost_user_blk_device_unrealize(DeviceState *dev, Error **errp)
> +{
> +    VirtIODevice *vdev = VIRTIO_DEVICE(dev);
> +    VHostUserBlk *s = VHOST_USER_BLK(dev);
> +
> +    vhost_user_blk_set_status(vdev, 0);
> +    vhost_dev_cleanup(&s->dev);
> +    g_free(s->dev.vqs);
> +    virtio_cleanup(vdev);
> +}
> +
> +static void vhost_user_blk_instance_init(Object *obj)
> +{
> +    VHostUserBlk *s = VHOST_USER_BLK(obj);
> +
> +    device_add_bootindex_property(obj, &s->bootindex, "bootindex",
> +                                  "/disk@0,0", DEVICE(obj), NULL);
> +}
> +
> +static const VMStateDescription vmstate_vhost_user_blk = {
> +    .name = "vhost-user-blk",
> +    .minimum_version_id = 1,
> +    .version_id = 1,
> +    .fields = (VMStateField[]) {
> +        VMSTATE_VIRTIO_DEVICE,
> +        VMSTATE_END_OF_LIST()
> +    },
> +};
> +
> +static Property vhost_user_blk_properties[] = {
> +    DEFINE_PROP_CHR("chardev", VHostUserBlk, chardev),
> +    DEFINE_PROP_UINT16("num_queues", VHostUserBlk, num_queues, 1),
> +    DEFINE_PROP_UINT32("queue_size", VHostUserBlk, queue_size, 128),
> +    DEFINE_PROP_BIT64("f_size_max", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_SIZE_MAX, true),
> +    DEFINE_PROP_BIT64("f_sizemax", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_SIZE_MAX, true),
> +    DEFINE_PROP_BIT64("f_segmax", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_SEG_MAX, true),
> +    DEFINE_PROP_BIT64("f_geometry", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_GEOMETRY, true),
> +    DEFINE_PROP_BIT64("f_readonly", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_RO, false),
> +    DEFINE_PROP_BIT64("f_blocksize", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_BLK_SIZE, true),
> +    DEFINE_PROP_BIT64("f_topology", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_TOPOLOGY, true),
> +    DEFINE_PROP_BIT64("f_multiqueue", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_MQ, true),
> +    DEFINE_PROP_BIT64("f_flush", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_FLUSH, true),
> +    DEFINE_PROP_BIT64("f_barrier", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_BARRIER, false),
> +    DEFINE_PROP_BIT64("f_scsi", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_SCSI, false),
> +    DEFINE_PROP_BIT64("f_wce", VHostUserBlk, host_features,
> +                      VIRTIO_BLK_F_WCE, false),
> +    DEFINE_PROP_END_OF_LIST(),
> +};
> +

So if you let users specify these, why do you need to query
them from the backend with GET_CONFIG?

> +static void vhost_user_blk_class_init(ObjectClass *klass, void *data)
> +{
> +    DeviceClass *dc = DEVICE_CLASS(klass);
> +    VirtioDeviceClass *vdc = VIRTIO_DEVICE_CLASS(klass);
> +
> +    dc->props = vhost_user_blk_properties;
> +    dc->vmsd = &vmstate_vhost_user_blk;
> +    set_bit(DEVICE_CATEGORY_STORAGE, dc->categories);
> +    vdc->realize = vhost_user_blk_device_realize;
> +    vdc->unrealize = vhost_user_blk_device_unrealize;
> +    vdc->get_config = vhost_user_blk_update_config;
> +    vdc->set_config = vhost_user_blk_set_config;
> +    vdc->get_features = vhost_user_blk_get_features;
> +    vdc->set_status = vhost_user_blk_set_status;
> +}
> +

Looks like this will pass config accesses directly to backend.
I am not sure it's a good approach.

> +static const TypeInfo vhost_user_blk_info = {
> +    .name = TYPE_VHOST_USER_BLK,
> +    .parent = TYPE_VIRTIO_DEVICE,
> +    .instance_size = sizeof(VHostUserBlk),
> +    .instance_init = vhost_user_blk_instance_init,
> +    .class_init = vhost_user_blk_class_init,
> +};
> +
> +static void virtio_register_types(void)
> +{
> +    type_register_static(&vhost_user_blk_info);
> +}
> +
> +type_init(virtio_register_types)
> diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
> index 8b0d6b6..be9a992 100644
> --- a/hw/virtio/virtio-pci.c
> +++ b/hw/virtio/virtio-pci.c
> @@ -2012,6 +2012,58 @@ static const TypeInfo virtio_blk_pci_info = {
>      .class_init    = virtio_blk_pci_class_init,
>  };
>  
> +#ifdef CONFIG_VHOST_USER_BLK
> +/* vhost-user-blk */
> +
> +static Property vhost_user_blk_pci_properties[] = {
> +    DEFINE_PROP_UINT32("class", VirtIOPCIProxy, class_code, 0),
> +    DEFINE_PROP_UINT32("vectors", VirtIOPCIProxy, nvectors, 2),
> +    DEFINE_PROP_END_OF_LIST(),
> +};
> +
> +static void vhost_user_blk_pci_realize(VirtIOPCIProxy *vpci_dev, Error **errp)
> +{
> +    VHostUserBlkPCI *dev = VHOST_USER_BLK_PCI(vpci_dev);
> +    DeviceState *vdev = DEVICE(&dev->vdev);
> +
> +    qdev_set_parent_bus(vdev, BUS(&vpci_dev->bus));
> +    object_property_set_bool(OBJECT(vdev), true, "realized", errp);
> +}
> +
> +static void vhost_user_blk_pci_class_init(ObjectClass *klass, void *data)
> +{
> +    DeviceClass *dc = DEVICE_CLASS(klass);
> +    VirtioPCIClass *k = VIRTIO_PCI_CLASS(klass);
> +    PCIDeviceClass *pcidev_k = PCI_DEVICE_CLASS(klass);
> +
> +    set_bit(DEVICE_CATEGORY_STORAGE, dc->categories);
> +    dc->props = vhost_user_blk_pci_properties;
> +    k->realize = vhost_user_blk_pci_realize;
> +    pcidev_k->vendor_id = PCI_VENDOR_ID_REDHAT_QUMRANET;
> +    pcidev_k->device_id = PCI_DEVICE_ID_VIRTIO_BLOCK;
> +    pcidev_k->revision = VIRTIO_PCI_ABI_VERSION;
> +    pcidev_k->class_id = PCI_CLASS_STORAGE_SCSI;
> +}
> +
> +static void vhost_user_blk_pci_instance_init(Object *obj)
> +{
> +    VHostUserBlkPCI *dev = VHOST_USER_BLK_PCI(obj);
> +
> +    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
> +                                TYPE_VHOST_USER_BLK);
> +    object_property_add_alias(obj, "bootindex", OBJECT(&dev->vdev),
> +                              "bootindex", &error_abort);
> +}
> +
> +static const TypeInfo vhost_user_blk_pci_info = {
> +    .name           = TYPE_VHOST_USER_BLK_PCI,
> +    .parent         = TYPE_VIRTIO_PCI,
> +    .instance_size  = sizeof(VHostUserBlkPCI),
> +    .instance_init  = vhost_user_blk_pci_instance_init,
> +    .class_init     = vhost_user_blk_pci_class_init,
> +};
> +#endif
> +
>  /* virtio-scsi-pci */
>  
>  static Property virtio_scsi_pci_properties[] = {
> @@ -2658,6 +2710,9 @@ static void virtio_pci_register_types(void)
>      type_register_static(&virtio_9p_pci_info);
>  #endif
>      type_register_static(&virtio_blk_pci_info);
> +#ifdef CONFIG_VHOST_USER_BLK
> +    type_register_static(&vhost_user_blk_pci_info);
> +#endif
>      type_register_static(&virtio_scsi_pci_info);
>      type_register_static(&virtio_balloon_pci_info);
>      type_register_static(&virtio_serial_pci_info);
> diff --git a/hw/virtio/virtio-pci.h b/hw/virtio/virtio-pci.h
> index 69f5959..19a0d01 100644
> --- a/hw/virtio/virtio-pci.h
> +++ b/hw/virtio/virtio-pci.h
> @@ -27,6 +27,9 @@
>  #include "hw/virtio/virtio-gpu.h"
>  #include "hw/virtio/virtio-crypto.h"
>  #include "hw/virtio/vhost-user-scsi.h"
> +#ifdef CONFIG_VHOST_USER_BLK
> +#include "hw/virtio/vhost-user-blk.h"
> +#endif
>  
>  #ifdef CONFIG_VIRTFS
>  #include "hw/9pfs/virtio-9p.h"
> @@ -46,6 +49,7 @@ typedef struct VirtIOSerialPCI VirtIOSerialPCI;
>  typedef struct VirtIONetPCI VirtIONetPCI;
>  typedef struct VHostSCSIPCI VHostSCSIPCI;
>  typedef struct VHostUserSCSIPCI VHostUserSCSIPCI;
> +typedef struct VHostUserBlkPCI VHostUserBlkPCI;
>  typedef struct VirtIORngPCI VirtIORngPCI;
>  typedef struct VirtIOInputPCI VirtIOInputPCI;
>  typedef struct VirtIOInputHIDPCI VirtIOInputHIDPCI;
> @@ -241,6 +245,20 @@ struct VHostUserSCSIPCI {
>      VHostUserSCSI vdev;
>  };
>  
> +#ifdef CONFIG_VHOST_USER_BLK
> +/*
> + * vhost-user-blk-pci: This extends VirtioPCIProxy.
> + */
> +#define TYPE_VHOST_USER_BLK_PCI "vhost-user-blk-pci"
> +#define VHOST_USER_BLK_PCI(obj) \
> +        OBJECT_CHECK(VHostUserBlkPCI, (obj), TYPE_VHOST_USER_BLK_PCI)
> +
> +struct VHostUserBlkPCI {
> +    VirtIOPCIProxy parent_obj;
> +    VHostUserBlk vdev;
> +};
> +#endif
> +
>  /*
>   * virtio-blk-pci: This extends VirtioPCIProxy.
>   */
> diff --git a/include/hw/virtio/vhost-user-blk.h b/include/hw/virtio/vhost-user-blk.h
> new file mode 100644
> index 0000000..77d20f0
> --- /dev/null
> +++ b/include/hw/virtio/vhost-user-blk.h
> @@ -0,0 +1,40 @@
> +/*
> + * vhost-user-blk host device
> + * Copyright IBM, Corp. 2011
> + * Copyright(C) 2017 Intel Corporation.
> + *
> + * Authors:
> + *  Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
> + *  Changpeng Liu <changpeng.liu@intel.com>
> + *
> + * This work is licensed under the terms of the GNU LGPL, version 2 or later.
> + * See the COPYING.LIB file in the top-level directory.
> + *
> + */
> +
> +#ifndef VHOST_USER_BLK_H
> +#define VHOST_USER_BLK_H
> +
> +#include "standard-headers/linux/virtio_blk.h"
> +#include "qemu-common.h"
> +#include "hw/qdev.h"
> +#include "hw/block/block.h"
> +#include "chardev/char-fe.h"
> +#include "hw/virtio/vhost.h"
> +
> +#define TYPE_VHOST_USER_BLK "vhost-user-blk"
> +#define VHOST_USER_BLK(obj) \
> +        OBJECT_CHECK(VHostUserBlk, (obj), TYPE_VHOST_USER_BLK)
> +
> +typedef struct VHostUserBlk {
> +    VirtIODevice parent_obj;
> +    CharBackend chardev;
> +    int32_t bootindex;
> +    uint64_t host_features;
> +    struct virtio_blk_config blkcfg;
> +    uint16_t num_queues;
> +    uint32_t queue_size;
> +    struct vhost_dev dev;
> +} VHostUserBlk;
> +
> +#endif
> -- 
> 1.9.3

> -----Original Message-----

> From: Marc-André Lureau [mailto:marcandre.lureau@redhat.com]

> Sent: Wednesday, August 9, 2017 11:39 PM

> To: Liu, Changpeng <changpeng.liu@intel.com>

> Cc: qemu-devel@nongnu.org; stefanha@gmail.com; pbonzini@redhat.com;

> mst@redhat.com; felipe@nutanix.com; Harris, James R

> <james.r.harris@intel.com>

> Subject: Re: [PATCH v2 2/4] vhost-user-blk: introduce a new vhost-user-blk host

> device

> 

> Hi

> 

> ----- Original Message -----

> > This commit introduces a new vhost-user device for block, it uses a

> > chardev to connect with the backend, same with Qemu virito-blk device,

> > Guest OS still uses the virtio-blk frontend driver.

> >

> > To use it, start Qemu with command line like this:

> >

> > qemu-system-x86_64 \

> >     -chardev socket,id=char0,path=/path/vhost.socket \

> >     -device vhost-user-blk-pci,chardev=char0,num_queues=...

> >

> > Different with exist Qemu virtio-blk host device, it makes more easy

> > for users to implement their own I/O processing logic, such as all

> > user space I/O stack against hardware block device. It uses the new

> > vhost messages(VHOST_USER_GET_CONFIG) to get block virtio config

> > information from backend process.

> >

> > Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>

> > ---

> >  configure                          |  11 ++

> >  hw/block/Makefile.objs             |   3 +

> >  hw/block/vhost-user-blk.c          | 360

> >  +++++++++++++++++++++++++++++++++++++

> >  hw/virtio/virtio-pci.c             |  55 ++++++

> >  hw/virtio/virtio-pci.h             |  18 ++

> >  include/hw/virtio/vhost-user-blk.h |  40 +++++

> >  6 files changed, 487 insertions(+)

> >  create mode 100644 hw/block/vhost-user-blk.c

> >  create mode 100644 include/hw/virtio/vhost-user-blk.h

> >

> > diff --git a/configure b/configure

> > index dd73cce..1452c66 100755

> > --- a/configure

> > +++ b/configure

> > @@ -305,6 +305,7 @@ tcg="yes"

> >

> >  vhost_net="no"

> >  vhost_scsi="no"

> > +vhost_user_blk="no"

> >  vhost_vsock="no"

> >  vhost_user=""

> >  kvm="no"

> > @@ -779,6 +780,7 @@ Linux)

> >    kvm="yes"

> >    vhost_net="yes"

> >    vhost_scsi="yes"

> > +  vhost_user_blk="yes"

> >    vhost_vsock="yes"

> >    QEMU_INCLUDES="-I\$(SRC_PATH)/linux-headers -I$(pwd)/linux-headers

> >    $QEMU_INCLUDES"

> >    supported_os="yes"

> > @@ -1136,6 +1138,10 @@ for opt do

> >    ;;

> >    --enable-vhost-scsi) vhost_scsi="yes"

> >    ;;

> > +  --disable-vhost-user-blk) vhost_user_blk="no"

> > +  ;;

> > +  --enable-vhost-user-blk) vhost_user_blk="yes"

> > +  ;;

> 

> I suggest we don't add yet another configure option, but reuse the recently

> introduced --enable-vhost-user (that should cover all vhost-user devices for now,

> but may learn to enable specific devices if needed in the future).

Yes, I noticed there is a new vhost-user configuration, sounds good to me if other devices
such as vhost-net and vhost-scsi also use the same configuration option.
> 

> >    --disable-vhost-vsock) vhost_vsock="no"

> >    ;;

> >    --enable-vhost-vsock) vhost_vsock="yes"

> > @@ -1506,6 +1512,7 @@ disabled with --disable-FEATURE, default is enabled if

> > available:

> >    cap-ng          libcap-ng support

> >    attr            attr and xattr support

> >    vhost-net       vhost-net acceleration support

> > +  vhost-user-blk  VM virtio-blk acceleration in user space

> >    spice           spice

> >    rbd             rados block device (rbd)

> >    libiscsi        iscsi support

> > @@ -5365,6 +5372,7 @@ echo "posix_madvise     $posix_madvise"

> >  echo "libcap-ng support $cap_ng"

> >  echo "vhost-net support $vhost_net"

> >  echo "vhost-scsi support $vhost_scsi"

> > +echo "vhost-user-blk support $vhost_user_blk"

> >  echo "vhost-vsock support $vhost_vsock"

> >  echo "vhost-user support $vhost_user"

> >  echo "Trace backends    $trace_backends"

> > @@ -5776,6 +5784,9 @@ fi

> >  if test "$vhost_scsi" = "yes" ; then

> >    echo "CONFIG_VHOST_SCSI=y" >> $config_host_mak

> >  fi

> > +if test "$vhost_user_blk" = "yes" ; then

> > +  echo "CONFIG_VHOST_USER_BLK=y" >> $config_host_mak

> > +fi

> >  if test "$vhost_net" = "yes" -a "$vhost_user" = "yes"; then

> >    echo "CONFIG_VHOST_NET_USED=y" >> $config_host_mak

> >  fi

> > diff --git a/hw/block/Makefile.objs b/hw/block/Makefile.objs

> > index e0ed980..4c19a58 100644

> > --- a/hw/block/Makefile.objs

> > +++ b/hw/block/Makefile.objs

> > @@ -13,3 +13,6 @@ obj-$(CONFIG_SH4) += tc58128.o

> >

> >  obj-$(CONFIG_VIRTIO) += virtio-blk.o

> >  obj-$(CONFIG_VIRTIO) += dataplane/

> > +ifeq ($(CONFIG_VIRTIO),y)

> > +obj-$(CONFIG_VHOST_USER_BLK) += vhost-user-blk.o

> > +endif

> > diff --git a/hw/block/vhost-user-blk.c b/hw/block/vhost-user-blk.c

> > new file mode 100644

> > index 0000000..8aa9fa9

> > --- /dev/null

> > +++ b/hw/block/vhost-user-blk.c

> > @@ -0,0 +1,360 @@

> > +/*

> > + * vhost-user-blk host device

> > + *

> > + * Copyright IBM, Corp. 2011

> > + * Copyright(C) 2017 Intel Corporation.

> > + *

> > + * Authors:

> > + *  Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>

> > + *  Changpeng Liu <changpeng.liu@intel.com>

> > + *

> > + * This work is licensed under the terms of the GNU LGPL, version 2 or

> > later.

> > + * See the COPYING.LIB file in the top-level directory.

> > + *

> > + */

> > +

> > +#include "qemu/osdep.h"

> > +#include "qapi/error.h"

> > +#include "qemu/error-report.h"

> > +#include "qemu/typedefs.h"

> > +#include "qemu/cutils.h"

> > +#include "qom/object.h"

> > +#include "hw/qdev-core.h"

> > +#include "hw/virtio/vhost.h"

> > +#include "hw/virtio/vhost-user-blk.h"

> > +#include "hw/virtio/virtio.h"

> > +#include "hw/virtio/virtio-bus.h"

> > +#include "hw/virtio/virtio-access.h"

> > +

> > +static const int user_feature_bits[] = {

> > +    VIRTIO_BLK_F_SIZE_MAX,

> > +    VIRTIO_BLK_F_SEG_MAX,

> > +    VIRTIO_BLK_F_GEOMETRY,

> > +    VIRTIO_BLK_F_BLK_SIZE,

> > +    VIRTIO_BLK_F_TOPOLOGY,

> > +    VIRTIO_BLK_F_SCSI,

> > +    VIRTIO_BLK_F_MQ,

> > +    VIRTIO_BLK_F_RO,

> > +    VIRTIO_BLK_F_FLUSH,

> > +    VIRTIO_BLK_F_BARRIER,

> > +    VIRTIO_BLK_F_WCE,

> > +    VIRTIO_F_VERSION_1,

> > +    VIRTIO_RING_F_INDIRECT_DESC,

> > +    VIRTIO_RING_F_EVENT_IDX,

> > +    VIRTIO_F_NOTIFY_ON_EMPTY,

> > +    VHOST_INVALID_FEATURE_BIT

> > +};

> > +

> > +static void vhost_user_blk_update_config(VirtIODevice *vdev, uint8_t

> > *config)

> > +{

> > +    VHostUserBlk *s = VHOST_USER_BLK(vdev);

> > +

> > +    memcpy(config, &s->blkcfg, sizeof(struct virtio_blk_config));

> > +}

> > +

> > +static void vhost_user_blk_set_config(VirtIODevice *vdev, const uint8_t

> > *config)

> > +{

> > +    VHostUserBlk *s = VHOST_USER_BLK(vdev);

> > +    struct virtio_blk_config *blkcfg = (struct virtio_blk_config *)config;

> > +    int ret;

> > +

> > +    if (blkcfg->wce == s->blkcfg.wce) {

> > +        return;

> 

> Is write-cache the only config change the slave is interested in?

> 

> > +    }

> > +

> > +    ret = vhost_dev_set_config(&s->dev, config,

> > +                              sizeof(struct virtio_blk_config));

> > +    if (ret) {

> > +        error_report("set device config space failed");

> > +        return;

> > +    }

> > +

> > +    s->blkcfg.wce = blkcfg->wce;

> > +}

> > +

> > +static void vhost_user_blk_handle_config_change(struct vhost_dev *dev)

> > +{

> > +    int ret;

> > +    struct virtio_blk_config blkcfg;

> > +    VHostUserBlk *s = VHOST_USER_BLK(dev->vdev);

> > +

> > +    ret = vhost_dev_get_config(dev, (uint8_t *)&blkcfg,

> > +                               sizeof(struct virtio_blk_config));

> > +    if (ret < 0) {

> > +        error_report("get config space failed");

> > +        return;

> > +    }

> > +

> > +    memcpy(&s->blkcfg, &blkcfg, sizeof(struct virtio_blk_config));

> > +    memcpy(dev->vdev->config, &blkcfg, sizeof(struct virtio_blk_config));

> 

> Why do you need to have s->blkcfg if you can use dev->vdev->config ?

Save a copy to local to avoid frequently get config from slave via socket message.
> 

> > +    virtio_notify_config(dev->vdev);

> > +}

> > +

> > +const VhostDevConfigOps blk_ops = {

> > +    .vhost_dev_config_notifier = vhost_user_blk_handle_config_change,

> > +};

> > +

> > +static void vhost_user_blk_start(VirtIODevice *vdev)

> > +{

> > +    VHostUserBlk *s = VHOST_USER_BLK(vdev);

> > +    BusState *qbus = BUS(qdev_get_parent_bus(DEVICE(vdev)));

> > +    VirtioBusClass *k = VIRTIO_BUS_GET_CLASS(qbus);

> > +    int i, ret;

> > +

> > +    if (!k->set_guest_notifiers) {

> > +        error_report("binding does not support guest notifiers");

> > +        return;

> > +    }

> > +

> > +    ret = vhost_dev_enable_notifiers(&s->dev, vdev);

> > +    if (ret < 0) {

> > +        error_report("Error enabling host notifiers: %d", -ret);

> > +        return;

> > +    }

> > +

> > +    ret = k->set_guest_notifiers(qbus->parent, s->dev.nvqs, true);

> > +    if (ret < 0) {

> > +        error_report("Error binding guest notifier: %d", -ret);

> > +        goto err_host_notifiers;

> > +    }

> > +

> > +    s->dev.acked_features = vdev->guest_features;

> > +    ret = vhost_dev_start(&s->dev, vdev);

> > +    if (ret < 0) {

> > +        error_report("Error starting vhost: %d", -ret);

> > +        goto err_guest_notifiers;

> > +    }

> > +

> > +    /* guest_notifier_mask/pending not used yet, so just unmask

> > +     * everything here. virtio-pci will do the right thing by

> > +     * enabling/disabling irqfd.

> > +     */

> > +    for (i = 0; i < s->dev.nvqs; i++) {

> > +        vhost_virtqueue_mask(&s->dev, vdev, i, false);

> > +    }

> > +

> > +    return;

> > +

> > +err_guest_notifiers:

> > +    k->set_guest_notifiers(qbus->parent, s->dev.nvqs, false);

> > +err_host_notifiers:

> > +    vhost_dev_disable_notifiers(&s->dev, vdev);

> > +}

> > +

> > +static void vhost_user_blk_stop(VirtIODevice *vdev)

> > +{

> > +    VHostUserBlk *s = VHOST_USER_BLK(vdev);

> > +    BusState *qbus = BUS(qdev_get_parent_bus(DEVICE(vdev)));

> > +    VirtioBusClass *k = VIRTIO_BUS_GET_CLASS(qbus);

> > +    int ret;

> > +

> > +    if (!k->set_guest_notifiers) {

> > +        return;

> > +    }

> > +

> > +    vhost_dev_stop(&s->dev, vdev);

> > +

> > +    ret = k->set_guest_notifiers(qbus->parent, s->dev.nvqs, false);

> > +    if (ret < 0) {

> > +        error_report("vhost guest notifier cleanup failed: %d", ret);

> > +        return;

> > +    }

> > +

> > +    vhost_dev_disable_notifiers(&s->dev, vdev);

> > +}

> > +

> > +static void vhost_user_blk_set_status(VirtIODevice *vdev, uint8_t status)

> > +{

> > +    VHostUserBlk *s = VHOST_USER_BLK(vdev);

> > +    bool should_start = status & VIRTIO_CONFIG_S_DRIVER_OK;

> > +

> > +    if (!vdev->vm_running) {

> > +        should_start = false;

> > +    }

> > +

> > +    if (s->dev.started == should_start) {

> > +        return;

> > +    }

> > +

> > +    if (should_start) {

> > +        vhost_user_blk_start(vdev);

> > +    } else {

> > +        vhost_user_blk_stop(vdev);

> > +    }

> > +

> > +}

> > +

> > +static uint64_t vhost_user_blk_get_features(VirtIODevice *vdev,

> > +                                            uint64_t features,

> > +                                            Error **errp)

> > +{

> > +    VHostUserBlk *s = VHOST_USER_BLK(vdev);

> > +    uint64_t get_features;

> > +

> > +    /* Turn on pre-defined features */

> > +    features |= s->host_features;

> > +

> > +    get_features = vhost_get_features(&s->dev, user_feature_bits, features);

> > +

> > +    return get_features;

> > +}

> > +

> > +static void vhost_user_blk_handle_output(VirtIODevice *vdev, VirtQueue *vq)

> > +{

> > +

> > +}

> > +

> > +static void vhost_user_blk_device_realize(DeviceState *dev, Error **errp)

> > +{

> > +    VirtIODevice *vdev = VIRTIO_DEVICE(dev);

> > +    VHostUserBlk *s = VHOST_USER_BLK(vdev);

> > +    int i, ret;

> > +

> > +    if (!s->chardev.chr) {

> > +        error_setg(errp, "vhost-user-blk: chardev is mandatory");

> > +        return;

> > +    }

> > +

> > +    if (!s->num_queues || s->num_queues > VIRTIO_QUEUE_MAX) {

> > +        error_setg(errp, "vhost-user-blk: invalid number of IO queues");

> > +        return;

> > +    }

> > +

> > +    if (!s->queue_size) {

> > +        error_setg(errp, "vhost-user-blk: queue size must be non-zero");

> > +        return;

> > +    }

> > +

> > +    virtio_init(vdev, "virtio-blk", VIRTIO_ID_BLOCK,

> > +                sizeof(struct virtio_blk_config));

> > +

> > +    for (i = 0; i < s->num_queues; i++) {

> > +        virtio_add_queue(vdev, s->queue_size,

> > +                         vhost_user_blk_handle_output);

> > +    }

> > +

> > +    s->dev.nvqs = s->num_queues;

> > +    s->dev.vqs = g_new(struct vhost_virtqueue, s->dev.nvqs);

> > +    s->dev.vq_index = 0;

> > +    s->dev.backend_features = 0;

> > +

> > +    ret = vhost_dev_init(&s->dev, &s->chardev, VHOST_BACKEND_TYPE_USER,

> 0);

> > +    if (ret < 0) {

> > +        error_setg(errp, "vhost-user-blk: vhost initialization failed: %s",

> > +                   strerror(-ret));

> > +        goto virtio_err;

> > +    }

> > +

> > +    ret = vhost_dev_get_config(&s->dev, (uint8_t *)&s->blkcfg,

> > +                              sizeof(struct virtio_blk_config));

> > +    if (ret < 0) {

> > +        error_setg(errp, "vhost-user-blk: get block config failed");

> > +        goto vhost_err;

> > +    }

> > +

> > +    if (s->blkcfg.num_queues != s->num_queues) {

> > +        s->blkcfg.num_queues = s->num_queues;

> > +    }

> > +

> > +    vhost_dev_set_config_notifier(&s->dev, &blk_ops);

> > +

> > +    return;

> > +

> > +vhost_err:

> > +    vhost_dev_cleanup(&s->dev);

> > +virtio_err:

> > +    g_free(s->dev.vqs);

> > +    virtio_cleanup(vdev);

> > +}

> > +

> > +static void vhost_user_blk_device_unrealize(DeviceState *dev, Error **errp)

> > +{

> > +    VirtIODevice *vdev = VIRTIO_DEVICE(dev);

> > +    VHostUserBlk *s = VHOST_USER_BLK(dev);

> > +

> > +    vhost_user_blk_set_status(vdev, 0);

> > +    vhost_dev_cleanup(&s->dev);

> > +    g_free(s->dev.vqs);

> > +    virtio_cleanup(vdev);

> > +}

> > +

> > +static void vhost_user_blk_instance_init(Object *obj)

> > +{

> > +    VHostUserBlk *s = VHOST_USER_BLK(obj);

> > +

> > +    device_add_bootindex_property(obj, &s->bootindex, "bootindex",

> > +                                  "/disk@0,0", DEVICE(obj), NULL);

> > +}

> > +

> > +static const VMStateDescription vmstate_vhost_user_blk = {

> > +    .name = "vhost-user-blk",

> > +    .minimum_version_id = 1,

> > +    .version_id = 1,

> > +    .fields = (VMStateField[]) {

> > +        VMSTATE_VIRTIO_DEVICE,

> > +        VMSTATE_END_OF_LIST()

> > +    },

> > +};

> > +

> > +static Property vhost_user_blk_properties[] = {

> > +    DEFINE_PROP_CHR("chardev", VHostUserBlk, chardev),

> > +    DEFINE_PROP_UINT16("num_queues", VHostUserBlk, num_queues, 1),

> > +    DEFINE_PROP_UINT32("queue_size", VHostUserBlk, queue_size, 128),

> > +    DEFINE_PROP_BIT64("f_size_max", VHostUserBlk, host_features,

> > +                      VIRTIO_BLK_F_SIZE_MAX, true),

> > +    DEFINE_PROP_BIT64("f_sizemax", VHostUserBlk, host_features,

> > +                      VIRTIO_BLK_F_SIZE_MAX, true),

> > +    DEFINE_PROP_BIT64("f_segmax", VHostUserBlk, host_features,

> > +                      VIRTIO_BLK_F_SEG_MAX, true),

> > +    DEFINE_PROP_BIT64("f_geometry", VHostUserBlk, host_features,

> > +                      VIRTIO_BLK_F_GEOMETRY, true),

> > +    DEFINE_PROP_BIT64("f_readonly", VHostUserBlk, host_features,

> > +                      VIRTIO_BLK_F_RO, false),

> > +    DEFINE_PROP_BIT64("f_blocksize", VHostUserBlk, host_features,

> > +                      VIRTIO_BLK_F_BLK_SIZE, true),

> > +    DEFINE_PROP_BIT64("f_topology", VHostUserBlk, host_features,

> > +                      VIRTIO_BLK_F_TOPOLOGY, true),

> > +    DEFINE_PROP_BIT64("f_multiqueue", VHostUserBlk, host_features,

> > +                      VIRTIO_BLK_F_MQ, true),

> > +    DEFINE_PROP_BIT64("f_flush", VHostUserBlk, host_features,

> > +                      VIRTIO_BLK_F_FLUSH, true),

> > +    DEFINE_PROP_BIT64("f_barrier", VHostUserBlk, host_features,

> > +                      VIRTIO_BLK_F_BARRIER, false),

> > +    DEFINE_PROP_BIT64("f_scsi", VHostUserBlk, host_features,

> > +                      VIRTIO_BLK_F_SCSI, false),

> > +    DEFINE_PROP_BIT64("f_wce", VHostUserBlk, host_features,

> > +                      VIRTIO_BLK_F_WCE, false),

> > +    DEFINE_PROP_END_OF_LIST(),

> > +};

> > +

> > +static void vhost_user_blk_class_init(ObjectClass *klass, void *data)

> > +{

> > +    DeviceClass *dc = DEVICE_CLASS(klass);

> > +    VirtioDeviceClass *vdc = VIRTIO_DEVICE_CLASS(klass);

> > +

> > +    dc->props = vhost_user_blk_properties;

> > +    dc->vmsd = &vmstate_vhost_user_blk;

> > +    set_bit(DEVICE_CATEGORY_STORAGE, dc->categories);

> > +    vdc->realize = vhost_user_blk_device_realize;

> > +    vdc->unrealize = vhost_user_blk_device_unrealize;

> > +    vdc->get_config = vhost_user_blk_update_config;

> > +    vdc->set_config = vhost_user_blk_set_config;

> > +    vdc->get_features = vhost_user_blk_get_features;

> > +    vdc->set_status = vhost_user_blk_set_status;

> > +}

> > +

> > +static const TypeInfo vhost_user_blk_info = {

> > +    .name = TYPE_VHOST_USER_BLK,

> > +    .parent = TYPE_VIRTIO_DEVICE,

> > +    .instance_size = sizeof(VHostUserBlk),

> > +    .instance_init = vhost_user_blk_instance_init,

> > +    .class_init = vhost_user_blk_class_init,

> > +};

> > +

> > +static void virtio_register_types(void)

> > +{

> > +    type_register_static(&vhost_user_blk_info);

> > +}

> > +

> > +type_init(virtio_register_types)

> > diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c

> > index 8b0d6b6..be9a992 100644

> > --- a/hw/virtio/virtio-pci.c

> > +++ b/hw/virtio/virtio-pci.c

> > @@ -2012,6 +2012,58 @@ static const TypeInfo virtio_blk_pci_info = {

> >      .class_init    = virtio_blk_pci_class_init,

> >  };

> >

> > +#ifdef CONFIG_VHOST_USER_BLK

> > +/* vhost-user-blk */

> > +

> > +static Property vhost_user_blk_pci_properties[] = {

> > +    DEFINE_PROP_UINT32("class", VirtIOPCIProxy, class_code, 0),

> > +    DEFINE_PROP_UINT32("vectors", VirtIOPCIProxy, nvectors, 2),

> > +    DEFINE_PROP_END_OF_LIST(),

> > +};

> > +

> > +static void vhost_user_blk_pci_realize(VirtIOPCIProxy *vpci_dev, Error

> > **errp)

> > +{

> > +    VHostUserBlkPCI *dev = VHOST_USER_BLK_PCI(vpci_dev);

> > +    DeviceState *vdev = DEVICE(&dev->vdev);

> > +

> > +    qdev_set_parent_bus(vdev, BUS(&vpci_dev->bus));

> > +    object_property_set_bool(OBJECT(vdev), true, "realized", errp);

> > +}

> > +

> > +static void vhost_user_blk_pci_class_init(ObjectClass *klass, void *data)

> > +{

> > +    DeviceClass *dc = DEVICE_CLASS(klass);

> > +    VirtioPCIClass *k = VIRTIO_PCI_CLASS(klass);

> > +    PCIDeviceClass *pcidev_k = PCI_DEVICE_CLASS(klass);

> > +

> > +    set_bit(DEVICE_CATEGORY_STORAGE, dc->categories);

> > +    dc->props = vhost_user_blk_pci_properties;

> > +    k->realize = vhost_user_blk_pci_realize;

> > +    pcidev_k->vendor_id = PCI_VENDOR_ID_REDHAT_QUMRANET;

> > +    pcidev_k->device_id = PCI_DEVICE_ID_VIRTIO_BLOCK;

> > +    pcidev_k->revision = VIRTIO_PCI_ABI_VERSION;

> > +    pcidev_k->class_id = PCI_CLASS_STORAGE_SCSI;

> > +}

> > +

> > +static void vhost_user_blk_pci_instance_init(Object *obj)

> > +{

> > +    VHostUserBlkPCI *dev = VHOST_USER_BLK_PCI(obj);

> > +

> > +    virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),

> > +                                TYPE_VHOST_USER_BLK);

> > +    object_property_add_alias(obj, "bootindex", OBJECT(&dev->vdev),

> > +                              "bootindex", &error_abort);

> > +}

> > +

> > +static const TypeInfo vhost_user_blk_pci_info = {

> > +    .name           = TYPE_VHOST_USER_BLK_PCI,

> > +    .parent         = TYPE_VIRTIO_PCI,

> > +    .instance_size  = sizeof(VHostUserBlkPCI),

> > +    .instance_init  = vhost_user_blk_pci_instance_init,

> > +    .class_init     = vhost_user_blk_pci_class_init,

> > +};

> > +#endif

> > +

> >  /* virtio-scsi-pci */

> >

> >  static Property virtio_scsi_pci_properties[] = {

> > @@ -2658,6 +2710,9 @@ static void virtio_pci_register_types(void)

> >      type_register_static(&virtio_9p_pci_info);

> >  #endif

> >      type_register_static(&virtio_blk_pci_info);

> > +#ifdef CONFIG_VHOST_USER_BLK

> > +    type_register_static(&vhost_user_blk_pci_info);

> > +#endif

> >      type_register_static(&virtio_scsi_pci_info);

> >      type_register_static(&virtio_balloon_pci_info);

> >      type_register_static(&virtio_serial_pci_info);

> > diff --git a/hw/virtio/virtio-pci.h b/hw/virtio/virtio-pci.h

> > index 69f5959..19a0d01 100644

> > --- a/hw/virtio/virtio-pci.h

> > +++ b/hw/virtio/virtio-pci.h

> > @@ -27,6 +27,9 @@

> >  #include "hw/virtio/virtio-gpu.h"

> >  #include "hw/virtio/virtio-crypto.h"

> >  #include "hw/virtio/vhost-user-scsi.h"

> > +#ifdef CONFIG_VHOST_USER_BLK

> > +#include "hw/virtio/vhost-user-blk.h"

> > +#endif

> >

> >  #ifdef CONFIG_VIRTFS

> >  #include "hw/9pfs/virtio-9p.h"

> > @@ -46,6 +49,7 @@ typedef struct VirtIOSerialPCI VirtIOSerialPCI;

> >  typedef struct VirtIONetPCI VirtIONetPCI;

> >  typedef struct VHostSCSIPCI VHostSCSIPCI;

> >  typedef struct VHostUserSCSIPCI VHostUserSCSIPCI;

> > +typedef struct VHostUserBlkPCI VHostUserBlkPCI;

> >  typedef struct VirtIORngPCI VirtIORngPCI;

> >  typedef struct VirtIOInputPCI VirtIOInputPCI;

> >  typedef struct VirtIOInputHIDPCI VirtIOInputHIDPCI;

> > @@ -241,6 +245,20 @@ struct VHostUserSCSIPCI {

> >      VHostUserSCSI vdev;

> >  };

> >

> > +#ifdef CONFIG_VHOST_USER_BLK

> > +/*

> > + * vhost-user-blk-pci: This extends VirtioPCIProxy.

> > + */

> > +#define TYPE_VHOST_USER_BLK_PCI "vhost-user-blk-pci"

> > +#define VHOST_USER_BLK_PCI(obj) \

> > +        OBJECT_CHECK(VHostUserBlkPCI, (obj), TYPE_VHOST_USER_BLK_PCI)

> > +

> > +struct VHostUserBlkPCI {

> > +    VirtIOPCIProxy parent_obj;

> > +    VHostUserBlk vdev;

> > +};

> > +#endif

> > +

> >  /*

> >   * virtio-blk-pci: This extends VirtioPCIProxy.

> >   */

> > diff --git a/include/hw/virtio/vhost-user-blk.h

> > b/include/hw/virtio/vhost-user-blk.h

> > new file mode 100644

> > index 0000000..77d20f0

> > --- /dev/null

> > +++ b/include/hw/virtio/vhost-user-blk.h

> > @@ -0,0 +1,40 @@

> > +/*

> > + * vhost-user-blk host device

> > + * Copyright IBM, Corp. 2011

> > + * Copyright(C) 2017 Intel Corporation.

> > + *

> > + * Authors:

> > + *  Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>

> > + *  Changpeng Liu <changpeng.liu@intel.com>

> > + *

> > + * This work is licensed under the terms of the GNU LGPL, version 2 or

> > later.

> > + * See the COPYING.LIB file in the top-level directory.

> > + *

> > + */

> > +

> > +#ifndef VHOST_USER_BLK_H

> > +#define VHOST_USER_BLK_H

> > +

> > +#include "standard-headers/linux/virtio_blk.h"

> > +#include "qemu-common.h"

> > +#include "hw/qdev.h"

> > +#include "hw/block/block.h"

> > +#include "chardev/char-fe.h"

> > +#include "hw/virtio/vhost.h"

> > +

> > +#define TYPE_VHOST_USER_BLK "vhost-user-blk"

> > +#define VHOST_USER_BLK(obj) \

> > +        OBJECT_CHECK(VHostUserBlk, (obj), TYPE_VHOST_USER_BLK)

> > +

> > +typedef struct VHostUserBlk {

> > +    VirtIODevice parent_obj;

> > +    CharBackend chardev;

> > +    int32_t bootindex;

> > +    uint64_t host_features;

> > +    struct virtio_blk_config blkcfg;

> > +    uint16_t num_queues;

> > +    uint32_t queue_size;

> > +    struct vhost_dev dev;

> > +} VHostUserBlk;

> > +

> > +#endif

> > --

> > 1.9.3

> >

> >

On 09/08/2017 19:10, Michael S. Tsirkin wrote:
> So user specifies properties and
> they get sent to backend at init time. Only handle geometry changes
> specially.

So QEMU would get the configuration, set these properties, and send the
result to the backend via SET_CONFIG?

vhost-user-blk-pci.cyls=uint32
vhost-user-blk-pci.secs=uint32
vhost-user-blk-pci.heads=uint32
vhost-user-blk-pci.serial=str
vhost-user-blk-pci.min_io_size=uint16
vhost-user-blk-pci.opt_io_size=uint32
vhost-user-blk-pci.logical_block_size=uint16
vhost-user-blk-pci.physical_block_size=uint16

If the properties are incompatible (e.g. too small logical block size)
SET_CONFIG fails and QEMU would fail to realize the device.  This makes
sense, I think.

Thanks,

Paolo