virtio: support specification 1.0
authorYuanhan Liu <yuanhan.liu@linux.intel.com>
Tue, 2 Feb 2016 13:48:19 +0000 (21:48 +0800)
committerThomas Monjalon <thomas.monjalon@6wind.com>
Wed, 3 Feb 2016 15:07:50 +0000 (16:07 +0100)
Modern (v1.0) virtio pci device defines several pci capabilities.
Each cap has a configure structure corresponding to it, and the
cap.bar and cap.offset fields tell us where to find it.

Firstly, we map the pci resources by rte_eal_pci_map_device().
We then could easily locate a cfg structure by:

    cfg_addr = dev->mem_resources[cap.bar].addr + cap.offset;

Therefore, the entrance of enabling modern (v1.0) pci device support
is to iterate the pci capability lists, and to locate some configs
we care; and they are:

- common cfg

  For generic virtio and virtqueue configuration, such as setting/getting
  features, enabling a specific queue, and so on.

- nofity cfg

  Combining with `queue_notify_off' from common cfg, we could use it to
  notify a specific virt queue.

- device cfg

  Where virtio_net_config structure is located.

- isr cfg

  Where to read isr (interrupt status).

If any of above cap is not found, we fallback to the legacy virtio
handling.

If succeed, hw->vtpci_ops is assigned to modern_ops, where all
operations are implemented by reading/writing a (or few) specific
configuration space from above 4 cfg structures. And that's basically
how this patch works.

Besides those changes, virtio 1.0 introduces a new status field:
FEATURES_OK, which is set after features negotiation is done.

Last, set the VIRTIO_F_VERSION_1 feature flag.

Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Tested-by: Qian Xu <qian.q.xu@intel.com>
Reviewed-by: Tetsuya Mukawa <mukawa@igel.co.jp>
Tested-by: Tetsuya Mukawa <mukawa@igel.co.jp>
Acked-by: Huawei Xie <huawei.xie@intel.com>
doc/guides/rel_notes/release_2_3.rst
drivers/net/virtio/virtio_ethdev.c
drivers/net/virtio/virtio_ethdev.h
drivers/net/virtio/virtio_pci.c
drivers/net/virtio/virtio_pci.h
drivers/net/virtio/virtqueue.h

index 53d048d..7945694 100644 (file)
@@ -35,6 +35,10 @@ This section should contain new features added in this release. Sample format:
 
   Refer to the previous release notes for examples.
 
+* **Virtio 1.0.**
+
+  Enabled virtio 1.0 support for virtio pmd driver.
+
 
 Resolved Issues
 ---------------
index 4f84757..755503d 100644 (file)
@@ -927,7 +927,7 @@ virtio_vlan_filter_set(struct rte_eth_dev *dev, uint16_t vlan_id, int on)
        return virtio_send_command(hw->cvq, &ctrl, &len, 1);
 }
 
-static void
+static int
 virtio_negotiate_features(struct virtio_hw *hw)
 {
        uint64_t host_features;
@@ -949,6 +949,22 @@ virtio_negotiate_features(struct virtio_hw *hw)
        hw->guest_features = vtpci_negotiate_features(hw, host_features);
        PMD_INIT_LOG(DEBUG, "features after negotiate = %" PRIx64,
                hw->guest_features);
+
+       if (hw->modern) {
+               if (!vtpci_with_feature(hw, VIRTIO_F_VERSION_1)) {
+                       PMD_INIT_LOG(ERR,
+                               "VIRTIO_F_VERSION_1 features is not enabled.");
+                       return -1;
+               }
+               vtpci_set_status(hw, VIRTIO_CONFIG_STATUS_FEATURES_OK);
+               if (!(vtpci_get_status(hw) & VIRTIO_CONFIG_STATUS_FEATURES_OK)) {
+                       PMD_INIT_LOG(ERR,
+                               "failed to set FEATURES_OK status!");
+                       return -1;
+               }
+       }
+
+       return 0;
 }
 
 /*
@@ -1032,7 +1048,8 @@ eth_virtio_dev_init(struct rte_eth_dev *eth_dev)
 
        /* Tell the host we've known how to drive the device. */
        vtpci_set_status(hw, VIRTIO_CONFIG_STATUS_DRIVER);
-       virtio_negotiate_features(hw);
+       if (virtio_negotiate_features(hw) < 0)
+               return -1;
 
        /* If host does not support status then disable LSC */
        if (!vtpci_with_feature(hw, VIRTIO_NET_F_STATUS))
@@ -1043,7 +1060,8 @@ eth_virtio_dev_init(struct rte_eth_dev *eth_dev)
        rx_func_get(eth_dev);
 
        /* Setting up rx_header size for the device */
-       if (vtpci_with_feature(hw, VIRTIO_NET_F_MRG_RXBUF))
+       if (vtpci_with_feature(hw, VIRTIO_NET_F_MRG_RXBUF) ||
+           vtpci_with_feature(hw, VIRTIO_F_VERSION_1))
                hw->vtnet_hdr_size = sizeof(struct virtio_net_hdr_mrg_rxbuf);
        else
                hw->vtnet_hdr_size = sizeof(struct virtio_net_hdr);
@@ -1159,6 +1177,7 @@ eth_virtio_dev_uninit(struct rte_eth_dev *eth_dev)
                rte_intr_callback_unregister(&pci_dev->intr_handle,
                                                virtio_interrupt_handler,
                                                eth_dev);
+       rte_eal_pci_unmap_device(pci_dev);
 
        PMD_INIT_LOG(DEBUG, "dev_uninit completed");
 
index ae2d47d..fed9571 100644 (file)
@@ -64,7 +64,8 @@
         1u << VIRTIO_NET_F_CTRL_VQ       |     \
         1u << VIRTIO_NET_F_CTRL_RX       |     \
         1u << VIRTIO_NET_F_CTRL_VLAN     |     \
-        1u << VIRTIO_NET_F_MRG_RXBUF)
+        1u << VIRTIO_NET_F_MRG_RXBUF     |     \
+        1ULL << VIRTIO_F_VERSION_1)
 
 /*
  * CQ function prototype
index 3961077..7c7621a 100644 (file)
 #include "virtio_logs.h"
 #include "virtqueue.h"
 
+/*
+ * Following macros are derived from linux/pci_regs.h, however,
+ * we can't simply include that header here, as there is no such
+ * file for non-Linux platform.
+ */
+#define PCI_CAPABILITY_LIST    0x34
+#define PCI_CAP_ID_VNDR                0x09
+
 static void
 legacy_read_dev_config(struct virtio_hw *hw, size_t offset,
                       void *dst, int length)
@@ -444,6 +452,217 @@ static const struct virtio_pci_ops legacy_ops = {
 };
 
 
+static inline uint8_t
+io_read8(uint8_t *addr)
+{
+       return *(volatile uint8_t *)addr;
+}
+
+static inline void
+io_write8(uint8_t val, uint8_t *addr)
+{
+       *(volatile uint8_t *)addr = val;
+}
+
+static inline uint16_t
+io_read16(uint16_t *addr)
+{
+       return *(volatile uint16_t *)addr;
+}
+
+static inline void
+io_write16(uint16_t val, uint16_t *addr)
+{
+       *(volatile uint16_t *)addr = val;
+}
+
+static inline uint32_t
+io_read32(uint32_t *addr)
+{
+       return *(volatile uint32_t *)addr;
+}
+
+static inline void
+io_write32(uint32_t val, uint32_t *addr)
+{
+       *(volatile uint32_t *)addr = val;
+}
+
+static inline void
+io_write64_twopart(uint64_t val, uint32_t *lo, uint32_t *hi)
+{
+       io_write32(val & ((1ULL << 32) - 1), lo);
+       io_write32(val >> 32,                hi);
+}
+
+static void
+modern_read_dev_config(struct virtio_hw *hw, size_t offset,
+                      void *dst, int length)
+{
+       int i;
+       uint8_t *p;
+       uint8_t old_gen, new_gen;
+
+       do {
+               old_gen = io_read8(&hw->common_cfg->config_generation);
+
+               p = dst;
+               for (i = 0;  i < length; i++)
+                       *p++ = io_read8((uint8_t *)hw->dev_cfg + offset + i);
+
+               new_gen = io_read8(&hw->common_cfg->config_generation);
+       } while (old_gen != new_gen);
+}
+
+static void
+modern_write_dev_config(struct virtio_hw *hw, size_t offset,
+                       const void *src, int length)
+{
+       int i;
+       const uint8_t *p = src;
+
+       for (i = 0;  i < length; i++)
+               io_write8(*p++, (uint8_t *)hw->dev_cfg + offset + i);
+}
+
+static uint64_t
+modern_get_features(struct virtio_hw *hw)
+{
+       uint32_t features_lo, features_hi;
+
+       io_write32(0, &hw->common_cfg->device_feature_select);
+       features_lo = io_read32(&hw->common_cfg->device_feature);
+
+       io_write32(1, &hw->common_cfg->device_feature_select);
+       features_hi = io_read32(&hw->common_cfg->device_feature);
+
+       return ((uint64_t)features_hi << 32) | features_lo;
+}
+
+static void
+modern_set_features(struct virtio_hw *hw, uint64_t features)
+{
+       io_write32(0, &hw->common_cfg->guest_feature_select);
+       io_write32(features & ((1ULL << 32) - 1),
+               &hw->common_cfg->guest_feature);
+
+       io_write32(1, &hw->common_cfg->guest_feature_select);
+       io_write32(features >> 32,
+               &hw->common_cfg->guest_feature);
+}
+
+static uint8_t
+modern_get_status(struct virtio_hw *hw)
+{
+       return io_read8(&hw->common_cfg->device_status);
+}
+
+static void
+modern_set_status(struct virtio_hw *hw, uint8_t status)
+{
+       io_write8(status, &hw->common_cfg->device_status);
+}
+
+static void
+modern_reset(struct virtio_hw *hw)
+{
+       modern_set_status(hw, VIRTIO_CONFIG_STATUS_RESET);
+       modern_get_status(hw);
+}
+
+static uint8_t
+modern_get_isr(struct virtio_hw *hw)
+{
+       return io_read8(hw->isr);
+}
+
+static uint16_t
+modern_set_config_irq(struct virtio_hw *hw, uint16_t vec)
+{
+       io_write16(vec, &hw->common_cfg->msix_config);
+       return io_read16(&hw->common_cfg->msix_config);
+}
+
+static uint16_t
+modern_get_queue_num(struct virtio_hw *hw, uint16_t queue_id)
+{
+       io_write16(queue_id, &hw->common_cfg->queue_select);
+       return io_read16(&hw->common_cfg->queue_size);
+}
+
+static void
+modern_setup_queue(struct virtio_hw *hw, struct virtqueue *vq)
+{
+       uint64_t desc_addr, avail_addr, used_addr;
+       uint16_t notify_off;
+
+       desc_addr = vq->mz->phys_addr;
+       avail_addr = desc_addr + vq->vq_nentries * sizeof(struct vring_desc);
+       used_addr = RTE_ALIGN_CEIL(avail_addr + offsetof(struct vring_avail,
+                                                        ring[vq->vq_nentries]),
+                                  VIRTIO_PCI_VRING_ALIGN);
+
+       io_write16(vq->vq_queue_index, &hw->common_cfg->queue_select);
+
+       io_write64_twopart(desc_addr, &hw->common_cfg->queue_desc_lo,
+                                     &hw->common_cfg->queue_desc_hi);
+       io_write64_twopart(avail_addr, &hw->common_cfg->queue_avail_lo,
+                                      &hw->common_cfg->queue_avail_hi);
+       io_write64_twopart(used_addr, &hw->common_cfg->queue_used_lo,
+                                     &hw->common_cfg->queue_used_hi);
+
+       notify_off = io_read16(&hw->common_cfg->queue_notify_off);
+       vq->notify_addr = (void *)((uint8_t *)hw->notify_base +
+                               notify_off * hw->notify_off_multiplier);
+
+       io_write16(1, &hw->common_cfg->queue_enable);
+
+       PMD_INIT_LOG(DEBUG, "queue %u addresses:", vq->vq_queue_index);
+       PMD_INIT_LOG(DEBUG, "\t desc_addr: %" PRIx64, desc_addr);
+       PMD_INIT_LOG(DEBUG, "\t aval_addr: %" PRIx64, avail_addr);
+       PMD_INIT_LOG(DEBUG, "\t used_addr: %" PRIx64, used_addr);
+       PMD_INIT_LOG(DEBUG, "\t notify addr: %p (notify offset: %u)",
+               vq->notify_addr, notify_off);
+}
+
+static void
+modern_del_queue(struct virtio_hw *hw, struct virtqueue *vq)
+{
+       io_write16(vq->vq_queue_index, &hw->common_cfg->queue_select);
+
+       io_write64_twopart(0, &hw->common_cfg->queue_desc_lo,
+                                 &hw->common_cfg->queue_desc_hi);
+       io_write64_twopart(0, &hw->common_cfg->queue_avail_lo,
+                                 &hw->common_cfg->queue_avail_hi);
+       io_write64_twopart(0, &hw->common_cfg->queue_used_lo,
+                                 &hw->common_cfg->queue_used_hi);
+
+       io_write16(0, &hw->common_cfg->queue_enable);
+}
+
+static void
+modern_notify_queue(struct virtio_hw *hw __rte_unused, struct virtqueue *vq)
+{
+       io_write16(1, vq->notify_addr);
+}
+
+static const struct virtio_pci_ops modern_ops = {
+       .read_dev_cfg   = modern_read_dev_config,
+       .write_dev_cfg  = modern_write_dev_config,
+       .reset          = modern_reset,
+       .get_status     = modern_get_status,
+       .set_status     = modern_set_status,
+       .get_features   = modern_get_features,
+       .set_features   = modern_set_features,
+       .get_isr        = modern_get_isr,
+       .set_config_irq = modern_set_config_irq,
+       .get_queue_num  = modern_get_queue_num,
+       .setup_queue    = modern_setup_queue,
+       .del_queue      = modern_del_queue,
+       .notify_queue   = modern_notify_queue,
+};
+
+
 void
 vtpci_read_dev_config(struct virtio_hw *hw, size_t offset,
                      void *dst, int length)
@@ -496,6 +715,12 @@ vtpci_set_status(struct virtio_hw *hw, uint8_t status)
        hw->vtpci_ops->set_status(hw, status);
 }
 
+uint8_t
+vtpci_get_status(struct virtio_hw *hw)
+{
+       return hw->vtpci_ops->get_status(hw);
+}
+
 uint8_t
 vtpci_isr(struct virtio_hw *hw)
 {
@@ -510,15 +735,142 @@ vtpci_irq_config(struct virtio_hw *hw, uint16_t vec)
        return hw->vtpci_ops->set_config_irq(hw, vec);
 }
 
+static void *
+get_cfg_addr(struct rte_pci_device *dev, struct virtio_pci_cap *cap)
+{
+       uint8_t  bar    = cap->bar;
+       uint32_t length = cap->length;
+       uint32_t offset = cap->offset;
+       uint8_t *base;
+
+       if (bar > 5) {
+               PMD_INIT_LOG(ERR, "invalid bar: %u", bar);
+               return NULL;
+       }
+
+       if (offset + length < offset) {
+               PMD_INIT_LOG(ERR, "offset(%u) + length(%u) overflows",
+                       offset, length);
+               return NULL;
+       }
+
+       if (offset + length > dev->mem_resource[bar].len) {
+               PMD_INIT_LOG(ERR,
+                       "invalid cap: overflows bar space: %u > %" PRIu64,
+                       offset + length, dev->mem_resource[bar].len);
+               return NULL;
+       }
+
+       base = dev->mem_resource[bar].addr;
+       if (base == NULL) {
+               PMD_INIT_LOG(ERR, "bar %u base addr is NULL", bar);
+               return NULL;
+       }
+
+       return base + offset;
+}
+
+static int
+virtio_read_caps(struct rte_pci_device *dev, struct virtio_hw *hw)
+{
+       uint8_t pos;
+       struct virtio_pci_cap cap;
+       int ret;
+
+       if (rte_eal_pci_map_device(dev) < 0) {
+               PMD_INIT_LOG(DEBUG, "failed to map pci device!");
+               return -1;
+       }
+
+       ret = rte_eal_pci_read_config(dev, &pos, 1, PCI_CAPABILITY_LIST);
+       if (ret < 0) {
+               PMD_INIT_LOG(DEBUG, "failed to read pci capability list");
+               return -1;
+       }
+
+       while (pos) {
+               ret = rte_eal_pci_read_config(dev, &cap, sizeof(cap), pos);
+               if (ret < 0) {
+                       PMD_INIT_LOG(ERR,
+                               "failed to read pci cap at pos: %x", pos);
+                       break;
+               }
+
+               if (cap.cap_vndr != PCI_CAP_ID_VNDR) {
+                       PMD_INIT_LOG(DEBUG,
+                               "[%2x] skipping non VNDR cap id: %02x",
+                               pos, cap.cap_vndr);
+                       goto next;
+               }
+
+               PMD_INIT_LOG(DEBUG,
+                       "[%2x] cfg type: %u, bar: %u, offset: %04x, len: %u",
+                       pos, cap.cfg_type, cap.bar, cap.offset, cap.length);
+
+               switch (cap.cfg_type) {
+               case VIRTIO_PCI_CAP_COMMON_CFG:
+                       hw->common_cfg = get_cfg_addr(dev, &cap);
+                       break;
+               case VIRTIO_PCI_CAP_NOTIFY_CFG:
+                       rte_eal_pci_read_config(dev, &hw->notify_off_multiplier,
+                                               4, pos + sizeof(cap));
+                       hw->notify_base = get_cfg_addr(dev, &cap);
+                       break;
+               case VIRTIO_PCI_CAP_DEVICE_CFG:
+                       hw->dev_cfg = get_cfg_addr(dev, &cap);
+                       break;
+               case VIRTIO_PCI_CAP_ISR_CFG:
+                       hw->isr = get_cfg_addr(dev, &cap);
+                       break;
+               }
+
+next:
+               pos = cap.cap_next;
+       }
+
+       if (hw->common_cfg == NULL || hw->notify_base == NULL ||
+           hw->dev_cfg == NULL    || hw->isr == NULL) {
+               PMD_INIT_LOG(INFO, "no modern virtio pci device found.");
+               return -1;
+       }
+
+       PMD_INIT_LOG(INFO, "found modern virtio pci device.");
+
+       PMD_INIT_LOG(DEBUG, "common cfg mapped at: %p", hw->common_cfg);
+       PMD_INIT_LOG(DEBUG, "device cfg mapped at: %p", hw->dev_cfg);
+       PMD_INIT_LOG(DEBUG, "isr cfg mapped at: %p", hw->isr);
+       PMD_INIT_LOG(DEBUG, "notify base: %p, notify off multiplier: %u",
+               hw->notify_base, hw->notify_off_multiplier);
+
+       return 0;
+}
+
 int
 vtpci_init(struct rte_pci_device *dev, struct virtio_hw *hw)
 {
-       hw->vtpci_ops = &legacy_ops;
+       hw->dev = dev;
 
+       /*
+        * Try if we can succeed reading virtio pci caps, which exists
+        * only on modern pci device. If failed, we fallback to legacy
+        * virtio handling.
+        */
+       if (virtio_read_caps(dev, hw) == 0) {
+               PMD_INIT_LOG(INFO, "modern virtio pci detected.");
+               hw->vtpci_ops = &modern_ops;
+               hw->modern    = 1;
+               dev->driver->drv_flags |= RTE_PCI_DRV_INTR_LSC;
+               return 0;
+       }
+
+       PMD_INIT_LOG(INFO, "trying with legacy virtio pci.");
        if (legacy_virtio_resource_init(dev) < 0)
                return -1;
+
+       hw->vtpci_ops = &legacy_ops;
        hw->use_msix = legacy_virtio_has_msix(&dev->addr);
        hw->io_base  = (uint32_t)(uintptr_t)dev->mem_resource[0].addr;
+       hw->modern   = 0;
 
        return 0;
 }
index d7bc6bb..fcac660 100644 (file)
@@ -93,6 +93,7 @@ struct virtqueue;
 #define VIRTIO_CONFIG_STATUS_ACK       0x01
 #define VIRTIO_CONFIG_STATUS_DRIVER    0x02
 #define VIRTIO_CONFIG_STATUS_DRIVER_OK 0x04
+#define VIRTIO_CONFIG_STATUS_FEATURES_OK 0x08
 #define VIRTIO_CONFIG_STATUS_FAILED    0x80
 
 /*
@@ -141,6 +142,8 @@ struct virtqueue;
 /* We support indirect buffer descriptors */
 #define VIRTIO_RING_F_INDIRECT_DESC    28
 
+#define VIRTIO_F_VERSION_1             32
+
 /*
  * Some VirtIO feature bits (currently bits 28 through 31) are
  * reserved for the transport being used (eg. virtio_ring), the
@@ -163,6 +166,60 @@ struct virtqueue;
  */
 #define VIRTIO_MAX_VIRTQUEUES 8
 
+/* Common configuration */
+#define VIRTIO_PCI_CAP_COMMON_CFG      1
+/* Notifications */
+#define VIRTIO_PCI_CAP_NOTIFY_CFG      2
+/* ISR Status */
+#define VIRTIO_PCI_CAP_ISR_CFG         3
+/* Device specific configuration */
+#define VIRTIO_PCI_CAP_DEVICE_CFG      4
+/* PCI configuration access */
+#define VIRTIO_PCI_CAP_PCI_CFG         5
+
+/* This is the PCI capability header: */
+struct virtio_pci_cap {
+       uint8_t cap_vndr;               /* Generic PCI field: PCI_CAP_ID_VNDR */
+       uint8_t cap_next;               /* Generic PCI field: next ptr. */
+       uint8_t cap_len;                /* Generic PCI field: capability length */
+       uint8_t cfg_type;               /* Identifies the structure. */
+       uint8_t bar;                    /* Where to find it. */
+       uint8_t padding[3];             /* Pad to full dword. */
+       uint32_t offset;                /* Offset within bar. */
+       uint32_t length;                /* Length of the structure, in bytes. */
+};
+
+struct virtio_pci_notify_cap {
+       struct virtio_pci_cap cap;
+       uint32_t notify_off_multiplier; /* Multiplier for queue_notify_off. */
+};
+
+/* Fields in VIRTIO_PCI_CAP_COMMON_CFG: */
+struct virtio_pci_common_cfg {
+       /* About the whole device. */
+       uint32_t device_feature_select; /* read-write */
+       uint32_t device_feature;        /* read-only */
+       uint32_t guest_feature_select;  /* read-write */
+       uint32_t guest_feature;         /* read-write */
+       uint16_t msix_config;           /* read-write */
+       uint16_t num_queues;            /* read-only */
+       uint8_t device_status;          /* read-write */
+       uint8_t config_generation;      /* read-only */
+
+       /* About a specific virtqueue. */
+       uint16_t queue_select;          /* read-write */
+       uint16_t queue_size;            /* read-write, power of 2. */
+       uint16_t queue_msix_vector;     /* read-write */
+       uint16_t queue_enable;          /* read-write */
+       uint16_t queue_notify_off;      /* read-only */
+       uint32_t queue_desc_lo;         /* read-write */
+       uint32_t queue_desc_hi;         /* read-write */
+       uint32_t queue_avail_lo;        /* read-write */
+       uint32_t queue_avail_hi;        /* read-write */
+       uint32_t queue_used_lo;         /* read-write */
+       uint32_t queue_used_hi;         /* read-write */
+};
+
 struct virtio_hw;
 
 struct virtio_pci_ops {
@@ -188,6 +245,8 @@ struct virtio_pci_ops {
        void (*notify_queue)(struct virtio_hw *hw, struct virtqueue *vq);
 };
 
+struct virtio_net_config;
+
 struct virtio_hw {
        struct virtqueue *cvq;
        uint32_t    io_base;
@@ -198,7 +257,14 @@ struct virtio_hw {
        uint8_t     vlan_strip;
        uint8_t     use_msix;
        uint8_t     started;
+       uint8_t     modern;
        uint8_t     mac_addr[ETHER_ADDR_LEN];
+       uint32_t    notify_off_multiplier;
+       uint8_t     *isr;
+       uint16_t    *notify_base;
+       struct rte_pci_device *dev;
+       struct virtio_pci_common_cfg *common_cfg;
+       struct virtio_net_config *dev_cfg;
        const struct virtio_pci_ops *vtpci_ops;
 };
 
@@ -284,6 +350,7 @@ void vtpci_reset(struct virtio_hw *);
 
 void vtpci_reinit_complete(struct virtio_hw *);
 
+uint8_t vtpci_get_status(struct virtio_hw *);
 void vtpci_set_status(struct virtio_hw *, uint8_t);
 
 uint64_t vtpci_negotiate_features(struct virtio_hw *, uint64_t);
index d7fb450..99d4fa9 100644 (file)
@@ -202,6 +202,8 @@ struct virtqueue {
        /* Size bins in array as RFC 2819, undersized [0], 64 [1], etc */
        uint64_t        size_bins[8];
 
+       uint16_t        *notify_addr;
+
        struct vq_desc_extra {
                void              *cookie;
                uint16_t          ndescs;