vhost-user-scsi: Introduce a vhost-user-scsi sample application

This commit introduces a vhost-user-scsi backend sample application. It
must be linked with libiscsi and libvhost-user.

To use it, compile with:
  $ make vhost-user-scsi

And run as follows:
  $ ./vhost-user-scsi -u vus.sock -i iscsi://uri_to_target/
  $ qemu-system-x86_64 --enable-kvm -m 512 \
      -object memory-backend-file,id=mem,size=512m,share=on,mem-path=guestmem \
      -numa node,memdev=mem \
      -chardev socket,id=vhost-user-scsi,path=vus.sock \
      -device vhost-user-scsi-pci,chardev=vhost-user-scsi \

The application is currently limited at one LUN only and it processes
requests synchronously (therefore only achieving QD1). The purpose of
the code is to show how a backend can be implemented and to test the
vhost-user-scsi Qemu implementation.

If a different instance of this vhost-user-scsi application is executed
at a remote host, a VM can be live migrated to such a host.

Signed-off-by: Felipe Franciosi <felipe@nutanix.com>
Message-Id: <1488479153-21203-5-git-send-email-felipe@nutanix.com>

Makefile

@@ -269,6 +269,7 @@ dummy := $(call unnest-vars,, \
                 ivshmem-client-obj-y \
                 ivshmem-server-obj-y \
                 libvhost-user-obj-y \
+                vhost-user-scsi-obj-y \
                 qga-vss-dll-obj-y \
                 block-obj-y \
                 block-obj-m \
@@ -473,6 +474,8 @@ ivshmem-client$(EXESUF): $(ivshmem-client-obj-y) $(COMMON_LDADDS)
 	$(call LINK, $^)
 ivshmem-server$(EXESUF): $(ivshmem-server-obj-y) $(COMMON_LDADDS)
 	$(call LINK, $^)
+vhost-user-scsi$(EXESUF): $(vhost-user-scsi-obj-y)
+	$(call LINK, $^)
 
 module_block.h: $(SRC_PATH)/scripts/modules/module_block.py config-host.mak
 	$(call quiet-command,$(PYTHON) $< $@ \

Makefile.objs

@@ -110,6 +110,10 @@ qga-vss-dll-obj-y = qga/
 ivshmem-client-obj-y = contrib/ivshmem-client/
 ivshmem-server-obj-y = contrib/ivshmem-server/
 libvhost-user-obj-y = contrib/libvhost-user/
+vhost-user-scsi.o-cflags := $(LIBISCSI_CFLAGS)
+vhost-user-scsi.o-libs := $(LIBISCSI_LIBS)
+vhost-user-scsi-obj-y = contrib/vhost-user-scsi/
+vhost-user-scsi-obj-y += contrib/libvhost-user/libvhost-user.o
 
 ######################################################################
 trace-events-subdirs =

contrib/libvhost-user/libvhost-user.h

@@ -17,6 +17,7 @@
 #include <stdint.h>
 #include <stdbool.h>
 #include <stddef.h>
+#include <sys/poll.h>
 #include <linux/vhost.h>
 #include "standard-headers/linux/virtio_ring.h"
 
@@ -192,11 +193,11 @@ typedef struct VuVirtq {
 } VuVirtq;
 
 enum VuWatchCondtion {
-    VU_WATCH_IN = 1 << 0,
-    VU_WATCH_OUT = 1 << 1,
-    VU_WATCH_PRI = 1 << 2,
-    VU_WATCH_ERR = 1 << 3,
-    VU_WATCH_HUP = 1 << 4,
+    VU_WATCH_IN = POLLIN,
+    VU_WATCH_OUT = POLLOUT,
+    VU_WATCH_PRI = POLLPRI,
+    VU_WATCH_ERR = POLLERR,
+    VU_WATCH_HUP = POLLHUP,
 };
 
 typedef void (*vu_panic_cb) (VuDev *dev, const char *err);

contrib/vhost-user-scsi/Makefile.objs

@@ -0,0 +1 @@
+vhost-user-scsi-obj-y = vhost-user-scsi.o

contrib/vhost-user-scsi/vhost-user-scsi.c

@@ -0,0 +1,886 @@
+/*
+ * vhost-user-scsi sample application
+ *
+ * Copyright (c) 2016 Nutanix Inc. All rights reserved.
+ *
+ * Author:
+ *  Felipe Franciosi <felipe@nutanix.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 only.
+ * See the COPYING file in the top-level directory.
+ */
+
+#include "qemu/osdep.h"
+#include "contrib/libvhost-user/libvhost-user.h"
+#include "hw/virtio/virtio-scsi.h"
+#include "iscsi/iscsi.h"
+
+#include <glib.h>
+
+/* Small compat shim from glib 2.32 */
+#ifndef G_SOURCE_CONTINUE
+#define G_SOURCE_CONTINUE TRUE
+#endif
+#ifndef G_SOURCE_REMOVE
+#define G_SOURCE_REMOVE FALSE
+#endif
+
+/* #define VUS_DEBUG 1 */
+
+/** Log helpers **/
+
+#define PPRE                                                          \
+    struct timespec ts;                                               \
+    char   timebuf[64];                                               \
+    struct tm tm;                                                     \
+    (void)clock_gettime(CLOCK_REALTIME, &ts);                         \
+    (void)strftime(timebuf, 64, "%Y%m%d %T", gmtime_r(&ts.tv_sec, &tm))
+
+#define PEXT(lvl, msg, ...) do {                                      \
+    PPRE;                                                             \
+    fprintf(stderr, "%s.%06ld " lvl ": %s:%s():%d: " msg "\n",        \
+            timebuf, ts.tv_nsec / 1000,                               \
+            __FILE__, __func__, __LINE__, ## __VA_ARGS__);            \
+} while (0)
+
+#define PNOR(lvl, msg, ...) do {                                      \
+    PPRE;                                                             \
+    fprintf(stderr, "%s.%06ld " lvl ": " msg "\n",                    \
+            timebuf, ts.tv_nsec / 1000, ## __VA_ARGS__);              \
+} while (0)
+
+#ifdef VUS_DEBUG
+#define PDBG(msg, ...) PEXT("DBG", msg, ## __VA_ARGS__)
+#define PERR(msg, ...) PEXT("ERR", msg, ## __VA_ARGS__)
+#define PLOG(msg, ...) PEXT("LOG", msg, ## __VA_ARGS__)
+#else
+#define PDBG(msg, ...) { }
+#define PERR(msg, ...) PNOR("ERR", msg, ## __VA_ARGS__)
+#define PLOG(msg, ...) PNOR("LOG", msg, ## __VA_ARGS__)
+#endif
+
+/** vhost-user-scsi specific definitions **/
+
+ /* Only 1 LUN and device supported today */
+#define VUS_MAX_LUNS 1
+#define VUS_MAX_DEVS 1
+
+#define VUS_ISCSI_INITIATOR "iqn.2016-11.com.nutanix:vhost-user-scsi"
+
+typedef struct iscsi_lun {
+    struct iscsi_context *iscsi_ctx;
+    int iscsi_lun;
+} iscsi_lun_t;
+
+typedef struct vhost_scsi_dev {
+    VuDev vu_dev;
+    int server_sock;
+    GMainLoop *loop;
+    GTree *fdmap;   /* fd -> gsource context id */
+    iscsi_lun_t luns[VUS_MAX_LUNS];
+} vhost_scsi_dev_t;
+
+static vhost_scsi_dev_t *vhost_scsi_devs[VUS_MAX_DEVS];
+
+/** glib event loop integration for libvhost-user and misc callbacks **/
+
+QEMU_BUILD_BUG_ON((int)G_IO_IN != (int)VU_WATCH_IN);
+QEMU_BUILD_BUG_ON((int)G_IO_OUT != (int)VU_WATCH_OUT);
+QEMU_BUILD_BUG_ON((int)G_IO_PRI != (int)VU_WATCH_PRI);
+QEMU_BUILD_BUG_ON((int)G_IO_ERR != (int)VU_WATCH_ERR);
+QEMU_BUILD_BUG_ON((int)G_IO_HUP != (int)VU_WATCH_HUP);
+
+typedef struct vus_gsrc {
+    GSource parent;
+    vhost_scsi_dev_t *vdev_scsi;
+    GPollFD gfd;
+    vu_watch_cb vu_cb;
+} vus_gsrc_t;
+
+static gint vus_fdmap_compare(gconstpointer a, gconstpointer b)
+{
+    return (b > a) - (b < a);
+}
+
+static gboolean vus_gsrc_prepare(GSource *src, gint *timeout)
+{
+    assert(timeout);
+
+    *timeout = -1;
+    return FALSE;
+}
+
+static gboolean vus_gsrc_check(GSource *src)
+{
+    vus_gsrc_t *vus_src = (vus_gsrc_t *)src;
+
+    assert(vus_src);
+
+    return vus_src->gfd.revents & vus_src->gfd.events;
+}
+
+static gboolean vus_gsrc_dispatch(GSource *src, GSourceFunc cb, gpointer data)
+{
+    vhost_scsi_dev_t *vdev_scsi;
+    vus_gsrc_t *vus_src = (vus_gsrc_t *)src;
+
+    assert(vus_src);
+    assert(!(vus_src->vu_cb && cb));
+
+    vdev_scsi = vus_src->vdev_scsi;
+
+    assert(vdev_scsi);
+
+    if (cb) {
+        return cb(data);
+    }
+    if (vus_src->vu_cb) {
+        vus_src->vu_cb(&vdev_scsi->vu_dev, vus_src->gfd.revents, data);
+    }
+    return G_SOURCE_CONTINUE;
+}
+
+static GSourceFuncs vus_gsrc_funcs = {
+    vus_gsrc_prepare,
+    vus_gsrc_check,
+    vus_gsrc_dispatch,
+    NULL
+};
+
+static int vus_gsrc_new(vhost_scsi_dev_t *vdev_scsi, int fd, GIOCondition cond,
+                        vu_watch_cb vu_cb, GSourceFunc gsrc_cb, gpointer data)
+{
+    GSource *vus_gsrc;
+    vus_gsrc_t *vus_src;
+    guint id;
+
+    assert(vdev_scsi);
+    assert(fd >= 0);
+    assert(vu_cb || gsrc_cb);
+    assert(!(vu_cb && gsrc_cb));
+
+    vus_gsrc = g_source_new(&vus_gsrc_funcs, sizeof(vus_gsrc_t));
+    if (!vus_gsrc) {
+        PERR("Error creating GSource for new watch");
+        return -1;
+    }
+    vus_src = (vus_gsrc_t *)vus_gsrc;
+
+    vus_src->vdev_scsi = vdev_scsi;
+    vus_src->gfd.fd = fd;
+    vus_src->gfd.events = cond;
+    vus_src->vu_cb = vu_cb;
+
+    g_source_add_poll(vus_gsrc, &vus_src->gfd);
+    g_source_set_callback(vus_gsrc, gsrc_cb, data, NULL);
+    id = g_source_attach(vus_gsrc, NULL);
+    assert(id);
+    g_source_unref(vus_gsrc);
+
+    g_tree_insert(vdev_scsi->fdmap, (gpointer)(uintptr_t)fd,
+                                    (gpointer)(uintptr_t)id);
+
+    return 0;
+}
+
+/* from libiscsi's scsi-lowlevel.h **
+ *
+ * nb. We can't directly include scsi-lowlevel.h due to a namespace conflict:
+ *     QEMU's scsi.h also defines "SCSI_XFER_NONE".
+ */
+
+#define SCSI_CDB_MAX_SIZE           16
+
+struct scsi_iovector {
+    struct scsi_iovec *iov;
+    int niov;
+    int nalloc;
+    size_t offset;
+    int consumed;
+};
+
+struct scsi_allocated_memory {
+    struct scsi_allocated_memory *next;
+    char buf[0];
+};
+
+struct scsi_data {
+    int            size;
+    unsigned char *data;
+};
+
+enum scsi_sense_key {
+    SCSI_SENSE_NO_SENSE            = 0x00,
+    SCSI_SENSE_RECOVERED_ERROR     = 0x01,
+    SCSI_SENSE_NOT_READY           = 0x02,
+    SCSI_SENSE_MEDIUM_ERROR        = 0x03,
+    SCSI_SENSE_HARDWARE_ERROR      = 0x04,
+    SCSI_SENSE_ILLEGAL_REQUEST     = 0x05,
+    SCSI_SENSE_UNIT_ATTENTION      = 0x06,
+    SCSI_SENSE_DATA_PROTECTION     = 0x07,
+    SCSI_SENSE_BLANK_CHECK         = 0x08,
+    SCSI_SENSE_VENDOR_SPECIFIC     = 0x09,
+    SCSI_SENSE_COPY_ABORTED        = 0x0a,
+    SCSI_SENSE_COMMAND_ABORTED     = 0x0b,
+    SCSI_SENSE_OBSOLETE_ERROR_CODE = 0x0c,
+    SCSI_SENSE_OVERFLOW_COMMAND    = 0x0d,
+    SCSI_SENSE_MISCOMPARE          = 0x0e
+};
+
+struct scsi_sense {
+    unsigned char       error_type;
+    enum scsi_sense_key key;
+    int                 ascq;
+    unsigned            sense_specific:1;
+    unsigned            ill_param_in_cdb:1;
+    unsigned            bit_pointer_valid:1;
+    unsigned char       bit_pointer;
+    uint16_t            field_pointer;
+};
+
+enum scsi_residual {
+    SCSI_RESIDUAL_NO_RESIDUAL = 0,
+    SCSI_RESIDUAL_UNDERFLOW,
+    SCSI_RESIDUAL_OVERFLOW
+};
+
+struct scsi_task {
+    int status;
+    int cdb_size;
+    int xfer_dir;
+    int expxferlen;
+    unsigned char cdb[SCSI_CDB_MAX_SIZE];
+    enum scsi_residual residual_status;
+    size_t residual;
+    struct scsi_sense sense;
+    struct scsi_data datain;
+    struct scsi_allocated_memory *mem;
+    void *ptr;
+
+    uint32_t itt;
+    uint32_t cmdsn;
+    uint32_t lun;
+
+    struct scsi_iovector iovector_in;
+    struct scsi_iovector iovector_out;
+};
+
+/** libiscsi integration **/
+
+static int iscsi_add_lun(iscsi_lun_t *lun, char *iscsi_uri)
+{
+    struct iscsi_url *iscsi_url;
+    struct iscsi_context *iscsi_ctx;
+    int ret = 0;
+
+    assert(lun);
+    assert(iscsi_uri);
+
+    iscsi_ctx = iscsi_create_context(VUS_ISCSI_INITIATOR);
+    if (!iscsi_ctx) {
+        PERR("Unable to create iSCSI context");
+        return -1;
+    }
+
+    iscsi_url = iscsi_parse_full_url(iscsi_ctx, iscsi_uri);
+    if (!iscsi_url) {
+        PERR("Unable to parse iSCSI URL: %s", iscsi_get_error(iscsi_ctx));
+        goto fail;
+    }
+
+    iscsi_set_session_type(iscsi_ctx, ISCSI_SESSION_NORMAL);
+    iscsi_set_header_digest(iscsi_ctx, ISCSI_HEADER_DIGEST_NONE_CRC32C);
+    if (iscsi_full_connect_sync(iscsi_ctx, iscsi_url->portal, iscsi_url->lun)) {
+        PERR("Unable to login to iSCSI portal: %s", iscsi_get_error(iscsi_ctx));
+        goto fail;
+    }
+
+    lun->iscsi_ctx = iscsi_ctx;
+    lun->iscsi_lun = iscsi_url->lun;
+
+    PDBG("Context %p created for lun 0: %s", iscsi_ctx, iscsi_uri);
+
+out:
+    if (iscsi_url) {
+        iscsi_destroy_url(iscsi_url);
+    }
+    return ret;
+
+fail:
+    (void)iscsi_destroy_context(iscsi_ctx);
+    ret = -1;
+    goto out;
+}
+
+static struct scsi_task *scsi_task_new(int cdb_len, uint8_t *cdb, int dir,
+                                       int xfer_len) {
+    struct scsi_task *task;
+
+    assert(cdb_len > 0);
+    assert(cdb);
+
+    task = calloc(1, sizeof(struct scsi_task));
+    if (!task) {
+        PERR("Error allocating task: %s", strerror(errno));
+        return NULL;
+    }
+
+    memcpy(task->cdb, cdb, cdb_len);
+    task->cdb_size = cdb_len;
+    task->xfer_dir = dir;
+    task->expxferlen = xfer_len;
+
+    return task;
+}
+
+static int get_cdb_len(uint8_t *cdb)
+{
+    assert(cdb);
+
+    switch (cdb[0] >> 5) {
+    case 0: return 6;
+    case 1: /* fall through */
+    case 2: return 10;
+    case 4: return 16;
+    case 5: return 12;
+    }
+    PERR("Unable to determine cdb len (0x%02hhX)", cdb[0] >> 5);
+    return -1;
+}
+
+static int handle_cmd_sync(struct iscsi_context *ctx,
+                           VirtIOSCSICmdReq *req,
+                           struct iovec *out, unsigned int out_len,
+                           VirtIOSCSICmdResp *rsp,
+                           struct iovec *in, unsigned int in_len) {
+    struct scsi_task *task;
+    uint32_t dir;
+    uint32_t len;
+    int cdb_len;
+    int i;
+
+    assert(ctx);
+    assert(req);
+    assert(rsp);
+
+    if (!(!req->lun[1] && req->lun[2] == 0x40 && !req->lun[3])) {
+        /* Ignore anything different than target=0, lun=0 */
+        PDBG("Ignoring unconnected lun (0x%hhX, 0x%hhX)",
+             req->lun[1], req->lun[3]);
+        rsp->status = SCSI_STATUS_CHECK_CONDITION;
+        memset(rsp->sense, 0, sizeof(rsp->sense));
+        rsp->sense_len = 18;
+        rsp->sense[0] = 0x70;
+        rsp->sense[2] = SCSI_SENSE_ILLEGAL_REQUEST;
+        rsp->sense[7] = 10;
+        rsp->sense[12] = 0x24;
+
+        return 0;
+    }
+
+    cdb_len = get_cdb_len(req->cdb);
+    if (cdb_len == -1) {
+        return -1;
+    }
+
+    len = 0;
+    if (!out_len && !in_len) {
+        dir = SCSI_XFER_NONE;
+    } else if (out_len) {
+        dir = SCSI_XFER_TO_DEV;
+        for (i = 0; i < out_len; i++) {
+            len += out[i].iov_len;
+        }
+    } else {
+        dir = SCSI_XFER_FROM_DEV;
+        for (i = 0; i < in_len; i++) {
+            len += in[i].iov_len;
+        }
+    }
+
+    task = scsi_task_new(cdb_len, req->cdb, dir, len);
+    if (!task) {
+        PERR("Unable to create iscsi task");
+        return -1;
+    }
+
+    if (dir == SCSI_XFER_TO_DEV) {
+        task->iovector_out.iov = (struct scsi_iovec *)out;
+        task->iovector_out.niov = out_len;
+    } else if (dir == SCSI_XFER_FROM_DEV) {
+        task->iovector_in.iov = (struct scsi_iovec *)in;
+        task->iovector_in.niov = in_len;
+    }
+
+    PDBG("Sending iscsi cmd (cdb_len=%d, dir=%d, task=%p)",
+         cdb_len, dir, task);
+    if (!iscsi_scsi_command_sync(ctx, 0, task, NULL)) {
+        PERR("Error serving SCSI command");
+        free(task);
+        return -1;
+    }
+
+    memset(rsp, 0, sizeof(*rsp));
+
+    rsp->status = task->status;
+    rsp->resid  = task->residual;
+
+    if (task->status == SCSI_STATUS_CHECK_CONDITION) {
+        rsp->response = VIRTIO_SCSI_S_FAILURE;
+        rsp->sense_len = task->datain.size - 2;
+        memcpy(rsp->sense, &task->datain.data[2], rsp->sense_len);
+    }
+
+    free(task);
+
+    PDBG("Filled in rsp: status=%hhX, resid=%u, response=%hhX, sense_len=%u",
+         rsp->status, rsp->resid, rsp->response, rsp->sense_len);
+
+    return 0;
+}
+
+/** libvhost-user callbacks **/
+
+static vhost_scsi_dev_t *vdev_scsi_find_by_vu(VuDev *vu_dev);
+
+static void vus_panic_cb(VuDev *vu_dev, const char *buf)
+{
+    vhost_scsi_dev_t *vdev_scsi;
+
+    assert(vu_dev);
+
+    vdev_scsi = vdev_scsi_find_by_vu(vu_dev);
+
+    if (buf) {
+        PERR("vu_panic: %s", buf);
+    }
+
+    if (vdev_scsi) {
+        assert(vdev_scsi->loop);
+        g_main_loop_quit(vdev_scsi->loop);
+    }
+}
+
+static void vus_add_watch_cb(VuDev *vu_dev, int fd, int vu_evt, vu_watch_cb cb,
+                             void *pvt) {
+    vhost_scsi_dev_t *vdev_scsi;
+    guint id;
+
+    assert(vu_dev);
+    assert(fd >= 0);
+    assert(cb);
+
+    vdev_scsi = vdev_scsi_find_by_vu(vu_dev);
+    if (!vdev_scsi) {
+        vus_panic_cb(vu_dev, NULL);
+        return;
+    }
+
+    id = (guint)(uintptr_t)g_tree_lookup(vdev_scsi->fdmap,
+                                         (gpointer)(uintptr_t)fd);
+    if (id) {
+        GSource *vus_src = g_main_context_find_source_by_id(NULL, id);
+        assert(vus_src);
+        g_source_destroy(vus_src);
+        (void)g_tree_remove(vdev_scsi->fdmap, (gpointer)(uintptr_t)fd);
+    }
+
+    if (vus_gsrc_new(vdev_scsi, fd, vu_evt, cb, NULL, pvt)) {
+        vus_panic_cb(vu_dev, NULL);
+    }
+}
+
+static void vus_del_watch_cb(VuDev *vu_dev, int fd)
+{
+    vhost_scsi_dev_t *vdev_scsi;
+    guint id;
+
+    assert(vu_dev);
+    assert(fd >= 0);
+
+    vdev_scsi = vdev_scsi_find_by_vu(vu_dev);
+    if (!vdev_scsi) {
+        vus_panic_cb(vu_dev, NULL);
+        return;
+    }
+
+    id = (guint)(uintptr_t)g_tree_lookup(vdev_scsi->fdmap,
+                                         (gpointer)(uintptr_t)fd);
+    if (id) {
+        GSource *vus_src = g_main_context_find_source_by_id(NULL, id);
+        assert(vus_src);
+        g_source_destroy(vus_src);
+        (void)g_tree_remove(vdev_scsi->fdmap, (gpointer)(uintptr_t)fd);
+    }
+}
+
+static void vus_proc_ctl(VuDev *vu_dev, int idx)
+{
+    /* Control VQ not implemented */
+}
+
+static void vus_proc_evt(VuDev *vu_dev, int idx)
+{
+    /* Event VQ not implemented */
+}
+
+static void vus_proc_req(VuDev *vu_dev, int idx)
+{
+    vhost_scsi_dev_t *vdev_scsi;
+    VuVirtq *vq;
+
+    assert(vu_dev);
+
+    vdev_scsi = vdev_scsi_find_by_vu(vu_dev);
+    if (!vdev_scsi) {
+        vus_panic_cb(vu_dev, NULL);
+        return;
+    }
+
+    if ((idx < 0) || (idx >= VHOST_MAX_NR_VIRTQUEUE)) {
+        PERR("VQ Index out of range: %d", idx);
+        vus_panic_cb(vu_dev, NULL);
+        return;
+    }
+
+    vq = vu_get_queue(vu_dev, idx);
+    if (!vq) {
+        PERR("Error fetching VQ (dev=%p, idx=%d)", vu_dev, idx);
+        vus_panic_cb(vu_dev, NULL);
+        return;
+    }
+
+    PDBG("Got kicked on vq[%d]@%p", idx, vq);
+
+    while (1) {
+        VuVirtqElement *elem;
+        VirtIOSCSICmdReq *req;
+        VirtIOSCSICmdResp *rsp;
+
+        elem = vu_queue_pop(vu_dev, vq, sizeof(VuVirtqElement));
+        if (!elem) {
+            PDBG("No more elements pending on vq[%d]@%p", idx, vq);
+            break;
+        }
+        PDBG("Popped elem@%p", elem);
+
+        assert(!((elem->out_num > 1) && (elem->in_num > 1)));
+        assert((elem->out_num > 0) && (elem->in_num > 0));
+
+        if (elem->out_sg[0].iov_len < sizeof(VirtIOSCSICmdReq)) {
+            PERR("Invalid virtio-scsi req header");
+            vus_panic_cb(vu_dev, NULL);
+            break;
+        }
+        req = (VirtIOSCSICmdReq *)elem->out_sg[0].iov_base;
+
+        if (elem->in_sg[0].iov_len < sizeof(VirtIOSCSICmdResp)) {
+            PERR("Invalid virtio-scsi rsp header");
+            vus_panic_cb(vu_dev, NULL);
+            break;
+        }
+        rsp = (VirtIOSCSICmdResp *)elem->in_sg[0].iov_base;
+
+        if (handle_cmd_sync(vdev_scsi->luns[0].iscsi_ctx,
+                            req, &elem->out_sg[1], elem->out_num - 1,
+                            rsp, &elem->in_sg[1], elem->in_num - 1) != 0) {
+            vus_panic_cb(vu_dev, NULL);
+            break;
+        }
+
+        vu_queue_push(vu_dev, vq, elem, 0);
+        vu_queue_notify(vu_dev, vq);
+
+        free(elem);
+    }
+}
+
+static void vus_queue_set_started(VuDev *vu_dev, int idx, bool started)
+{
+    VuVirtq *vq;
+
+    assert(vu_dev);
+
+    if ((idx < 0) || (idx >= VHOST_MAX_NR_VIRTQUEUE)) {
+        PERR("VQ Index out of range: %d", idx);
+        vus_panic_cb(vu_dev, NULL);
+        return;
+    }
+
+    vq = vu_get_queue(vu_dev, idx);
+
+    switch (idx) {
+    case 0:
+        vu_set_queue_handler(vu_dev, vq, started ? vus_proc_ctl : NULL);
+        break;
+    case 1:
+        vu_set_queue_handler(vu_dev, vq, started ? vus_proc_evt : NULL);
+        break;
+    default:
+        vu_set_queue_handler(vu_dev, vq, started ? vus_proc_req : NULL);
+    }
+}
+
+static const VuDevIface vus_iface = {
+    .queue_set_started = vus_queue_set_started,
+};
+
+static gboolean vus_vhost_cb(gpointer data)
+{
+    VuDev *vu_dev = (VuDev *)data;
+
+    assert(vu_dev);
+
+    if (!vu_dispatch(vu_dev) != 0) {
+        PERR("Error processing vhost message");
+        vus_panic_cb(vu_dev, NULL);
+        return G_SOURCE_REMOVE;
+    }
+
+    return G_SOURCE_CONTINUE;
+}
+
+/** misc helpers **/
+
+static int unix_sock_new(char *unix_fn)
+{
+    int sock;
+    struct sockaddr_un un;
+    size_t len;
+
+    assert(unix_fn);
+
+    sock = socket(AF_UNIX, SOCK_STREAM, 0);
+    if (sock <= 0) {
+        perror("socket");
+        return -1;
+    }
+
+    un.sun_family = AF_UNIX;
+    (void)snprintf(un.sun_path, sizeof(un.sun_path), "%s", unix_fn);
+    len = sizeof(un.sun_family) + strlen(un.sun_path);
+
+    (void)unlink(unix_fn);
+    if (bind(sock, (struct sockaddr *)&un, len) < 0) {
+        perror("bind");
+        goto fail;
+    }
+
+    if (listen(sock, 1) < 0) {
+        perror("listen");
+        goto fail;
+    }
+
+    return sock;
+
+fail:
+    (void)close(sock);
+
+    return -1;
+}
+
+/** vhost-user-scsi **/
+
+static vhost_scsi_dev_t *vdev_scsi_find_by_vu(VuDev *vu_dev)
+{
+    int i;
+
+    assert(vu_dev);
+
+    for (i = 0; i < VUS_MAX_DEVS; i++) {
+        if (&vhost_scsi_devs[i]->vu_dev == vu_dev) {
+            return vhost_scsi_devs[i];
+        }
+    }
+
+    PERR("Unknown VuDev %p", vu_dev);
+    return NULL;
+}
+
+static void vdev_scsi_deinit(vhost_scsi_dev_t *vdev_scsi)
+{
+    if (!vdev_scsi) {
+        return;
+    }
+
+    if (vdev_scsi->server_sock >= 0) {
+        struct sockaddr_storage ss;
+        socklen_t sslen = sizeof(ss);
+
+        if (getsockname(vdev_scsi->server_sock, (struct sockaddr *)&ss,
+                        &sslen) == 0) {
+            struct sockaddr_un *su = (struct sockaddr_un *)&ss;
+            (void)unlink(su->sun_path);
+        }
+
+        (void)close(vdev_scsi->server_sock);
+        vdev_scsi->server_sock = -1;
+    }
+
+    if (vdev_scsi->loop) {
+        g_main_loop_unref(vdev_scsi->loop);
+        vdev_scsi->loop = NULL;
+    }
+}
+
+static vhost_scsi_dev_t *vdev_scsi_new(char *unix_fn)
+{
+    vhost_scsi_dev_t *vdev_scsi = NULL;
+
+    assert(unix_fn);
+
+    vdev_scsi = calloc(1, sizeof(vhost_scsi_dev_t));
+    if (!vdev_scsi) {
+        PERR("calloc: %s", strerror(errno));
+        return NULL;
+    }
+
+    vdev_scsi->server_sock = unix_sock_new(unix_fn);
+    if (vdev_scsi->server_sock < 0) {
+        goto err;
+    }
+
+    vdev_scsi->loop = g_main_loop_new(NULL, FALSE);
+    if (!vdev_scsi->loop) {
+        PERR("Error creating glib event loop");
+        goto err;
+    }
+
+    vdev_scsi->fdmap = g_tree_new(vus_fdmap_compare);
+    if (!vdev_scsi->fdmap) {
+        PERR("Error creating glib tree for fdmap");
+        goto err;
+    }
+
+    return vdev_scsi;
+
+err:
+    vdev_scsi_deinit(vdev_scsi);
+    free(vdev_scsi);
+
+    return NULL;
+}
+
+static int vdev_scsi_add_iscsi_lun(vhost_scsi_dev_t *vdev_scsi,
+                                   char *iscsi_uri, uint32_t lun) {
+    assert(vdev_scsi);
+    assert(iscsi_uri);
+    assert(lun < VUS_MAX_LUNS);
+
+    if (vdev_scsi->luns[lun].iscsi_ctx) {
+        PERR("Lun %d already configured", lun);
+        return -1;
+    }
+
+    if (iscsi_add_lun(&vdev_scsi->luns[lun], iscsi_uri) != 0) {
+        return -1;
+    }
+
+    return 0;
+}
+
+static int vdev_scsi_run(vhost_scsi_dev_t *vdev_scsi)
+{
+    int cli_sock;
+    int ret = 0;
+
+    assert(vdev_scsi);
+    assert(vdev_scsi->server_sock >= 0);
+    assert(vdev_scsi->loop);
+
+    cli_sock = accept(vdev_scsi->server_sock, (void *)0, (void *)0);
+    if (cli_sock < 0) {
+        perror("accept");
+        return -1;
+    }
+
+    vu_init(&vdev_scsi->vu_dev,
+            cli_sock,
+            vus_panic_cb,
+            vus_add_watch_cb,
+            vus_del_watch_cb,
+            &vus_iface);
+
+    if (vus_gsrc_new(vdev_scsi, cli_sock, G_IO_IN, NULL, vus_vhost_cb,
+                     &vdev_scsi->vu_dev)) {
+        goto fail;
+    }
+
+    g_main_loop_run(vdev_scsi->loop);
+
+out:
+    vu_deinit(&vdev_scsi->vu_dev);
+
+    return ret;
+
+fail:
+    ret = -1;
+    goto out;
+}
+
+int main(int argc, char **argv)
+{
+    vhost_scsi_dev_t *vdev_scsi = NULL;
+    char *unix_fn = NULL;
+    char *iscsi_uri = NULL;
+    int opt, err = EXIT_SUCCESS;
+
+    while ((opt = getopt(argc, argv, "u:i:")) != -1) {
+        switch (opt) {
+        case 'h':
+            goto help;
+        case 'u':
+            unix_fn = strdup(optarg);
+            break;
+        case 'i':
+            iscsi_uri = strdup(optarg);
+            break;
+        default:
+            goto help;
+        }
+    }
+    if (!unix_fn || !iscsi_uri) {
+        goto help;
+    }
+
+    vdev_scsi = vdev_scsi_new(unix_fn);
+    if (!vdev_scsi) {
+        goto err;
+    }
+    vhost_scsi_devs[0] = vdev_scsi;
+
+    if (vdev_scsi_add_iscsi_lun(vdev_scsi, iscsi_uri, 0) != 0) {
+        goto err;
+    }
+
+    if (vdev_scsi_run(vdev_scsi) != 0) {
+        goto err;
+    }
+
+out:
+    if (vdev_scsi) {
+        vdev_scsi_deinit(vdev_scsi);
+        free(vdev_scsi);
+    }
+    if (unix_fn) {
+        free(unix_fn);
+    }
+    if (iscsi_uri) {
+        free(iscsi_uri);
+    }
+
+    return err;
+
+err:
+    err = EXIT_FAILURE;
+    goto out;
+
+help:
+    fprintf(stderr, "Usage: %s [ -u unix_sock_path -i iscsi_uri ] | [ -h ]\n",
+            argv[0]);
+    fprintf(stderr, "          -u path to unix socket\n");
+    fprintf(stderr, "          -i iscsi uri for lun 0\n");
+    fprintf(stderr, "          -h print help and quit\n");
+
+    goto err;
+}