Palacios Public Git Repository

To checkout Palacios execute

  git clone http://v3vee.org/palacios/palacios.web/palacios.git
This will give you the master branch. You probably want the devel branch or one of the release branches. To switch to the devel branch, simply execute
  cd palacios
  git checkout --track -b devel origin/devel
The other branches are similar.


Update on the VNET Bridge virtio device
[palacios.git] / palacios / src / devices / lnx_virtio_vnet.c
index 56400ba..fea13c6 100644 (file)
 #include <palacios/vmm_dev_mgr.h>
 #include <palacios/vm_guest_mem.h>
 #include <devices/lnx_virtio_pci.h>
-
+#include <palacios/vmm_vnet.h>
+#include <palacios/vmm_sprintf.h>
 #include <devices/pci.h>
 
 
+#ifndef CONFIG_DEBUG_VNET_BRG
+#undef PrintDebug
+#define PrintDebug(fmt, args...)
+#endif
+
+
 #define QUEUE_SIZE 128
 #define NUM_QUEUES 3
 
 struct vnet_config {
-    uint32_t num_links;
+    uint32_t num_devs;
     uint32_t num_routes;
 } __attribute__((packed));
 
 
+#define CTRL_QUEUE 0
+#define RECV_QUEUE 2
+#define XMIT_QUEUE 1
 
 struct virtio_vnet_state {
+    struct v3_vm_info *vm;
     struct vnet_config vnet_cfg;
     struct virtio_config virtio_cfg;
 
-
     struct vm_device * pci_bus;
     struct pci_device * pci_dev;
-
-
-#define CTRL_QUEUE 0
-#define RECV_QUEUE 1
-#define XMIT_QUEUE 2
+       
     struct virtio_queue queue[NUM_QUEUES];
 
     struct virtio_queue * cur_queue;
 
-
     int io_range_size;
+    v3_lock_t lock;
 };
 
 #define VNET_GET_ROUTES 10
@@ -63,25 +69,42 @@ struct virtio_vnet_state {
 #define VNET_ADD_LINK 21
 #define VNET_DEL_LINK 22
 
-// structure of the symmod notifier ring structures
+// structure of the vnet command header
 struct vnet_ctrl_hdr {
     uint8_t cmd_type;
     uint32_t num_cmds;
 } __attribute__((packed));
 
+#define VIRTIO_NET_S_LINK_UP   1       /* Link is up */
+#define VIRTIO_NET_MAX_BUFSIZE (sizeof(struct virtio_net_hdr) + (64 << 10))
+
+struct virtio_net_hdr {
+       uint8_t flags;
 
-static int virtio_reset(struct virtio_vnet_state * virtio) {
+#define VIRTIO_NET_HDR_GSO_NONE                0       /* Not a GSO frame */
+       uint8_t gso_type;
+       uint16_t hdr_len;               /* Ethernet + IP + tcp/udp hdrs */
+       uint16_t gso_size;              /* Bytes to append to hdr_len per frame */
+       uint16_t csum_start;    /* Position to start checksumming from */
+       uint16_t csum_offset;   /* Offset after that to place checksum */
+}__attribute__((packed));
 
-    memset(virtio->queue, 0, sizeof(struct virtio_queue) * 2);
 
-    virtio->cur_queue = &(virtio->queue[0]);
+static int virtio_reset(struct virtio_vnet_state * vnet_brg) {
 
-    virtio->virtio_cfg.status = 0;
-    virtio->virtio_cfg.pci_isr = 0;
+    memset(vnet_brg->queue, 0, sizeof(struct virtio_queue) * 2);
 
-    virtio->queue[0].queue_size = QUEUE_SIZE;
+    vnet_brg->cur_queue = &(vnet_brg->queue[0]);
 
-    memset(&(virtio->vnet_cfg), 0, sizeof(struct vnet_config));
+    vnet_brg->virtio_cfg.status = 0;
+    vnet_brg->virtio_cfg.pci_isr = 0;
+
+    vnet_brg->queue[0].queue_size = QUEUE_SIZE;
+    vnet_brg->queue[1].queue_size = QUEUE_SIZE;
+    vnet_brg->queue[2].queue_size = QUEUE_SIZE;
+
+    memset(&(vnet_brg->vnet_cfg), 0, sizeof(struct vnet_config));
+    v3_lock_init(&(vnet_brg->lock));
 
     return 0;
 }
@@ -104,9 +127,9 @@ static int get_desc_count(struct virtio_queue * q, int index) {
 
 
 static int handle_cmd_kick(struct guest_info * core, struct virtio_vnet_state * vnet_state) {
-    struct virtio_queue * q = vnet_state->cur_queue;
+    struct virtio_queue * q = &(vnet_state->queue[0]);
     
-    PrintDebug("VirtioVNET: Virtio Kick on command  queue\n");
+    PrintDebug("VNET Bridge: Handling command  queue\n");
 
     while (q->cur_avail_idx < q->avail->index) {
        struct vring_desc * hdr_desc = NULL;
@@ -121,10 +144,10 @@ static int handle_cmd_kick(struct guest_info * core, struct virtio_vnet_state *
        uint8_t status = 0;
 
 
-       PrintDebug("Descriptor Count=%d, index=%d\n", desc_cnt, q->cur_avail_idx % QUEUE_SIZE);
+       PrintDebug("Descriptor Count=%d, index=%d, desc_idx=%d\n", desc_cnt, q->cur_avail_idx % QUEUE_SIZE, desc_idx);
 
        if (desc_cnt < 3) {
-           PrintError("Symmod loads must include at least 3 descriptors (cnt=%d)\n", desc_cnt);
+           PrintError("VNET Bridge cmd must include at least 3 descriptors (cnt=%d)\n", desc_cnt);
            return -1;
        }
        
@@ -141,20 +164,24 @@ static int handle_cmd_kick(struct guest_info * core, struct virtio_vnet_state *
            
            for (i = 0; i < hdr->num_cmds; i++) {
                uint8_t tmp_status = 0;
-               uint8_t * buf = NULL;
+               struct v3_vnet_route * route = NULL;
                
                buf_desc = &(q->desc[desc_idx]);
 
-               if (guest_pa_to_host_va(core, buf_desc->addr_gpa, (addr_t *)&(buf)) == -1) {
-                   PrintError("Could not translate buffer address\n");
+               if (guest_pa_to_host_va(core, buf_desc->addr_gpa, (addr_t *)&(route)) == -1) {
+                   PrintError("Could not translate route address\n");
                    return -1;
                }
 
                // add route
                PrintDebug("Adding VNET Route\n");
 
+               tmp_status = v3_vnet_add_route(*route);
+
+               PrintDebug("VNET Route Added\n");
+
                if (tmp_status != 0) {
-                   PrintError("Error loading module segment\n");
+                   PrintError("Error adding VNET ROUTE\n");
                    status = tmp_status;
                }
 
@@ -162,17 +189,14 @@ static int handle_cmd_kick(struct guest_info * core, struct virtio_vnet_state *
                desc_idx = buf_desc->next;
            }
 
-       } else if (hdr->cmd_type == VNET_ADD_LINK) {
-
-
-       }
+       } 
 
 
 
        status_desc = &(q->desc[desc_idx]);
 
        if (guest_pa_to_host_va(core, status_desc->addr_gpa, (addr_t *)&status_ptr) == -1) {
-           PrintError("SYMMOD Error could not translate status address\n");
+           PrintError("VirtioVNET Error could not translate status address\n");
            return -1;
        }
 
@@ -198,15 +222,199 @@ static int handle_cmd_kick(struct guest_info * core, struct virtio_vnet_state *
     return 0;
 }
 
+static int copy_data_to_desc(struct guest_info *core, 
+                                       struct vring_desc * desc, 
+                                       uchar_t * buf, 
+                                       uint_t buf_len) 
+{
+    uint32_t len;
+    uint8_t * desc_buf = NULL;
+
+    if (guest_pa_to_host_va(core, desc->addr_gpa, (addr_t *)&(desc_buf)) == -1) {
+       PrintError("Could not translate buffer address\n");
+       return -1;
+    }
+    len = (desc->length < buf_len)?desc->length:buf_len;
+    memcpy(desc_buf, buf, len);
+
+    return len;
+}
+
+
+static int vnet_brg_rx(struct v3_vnet_pkt *pkt, void * private_data) {
+    struct virtio_vnet_state * vnet_brg = (struct virtio_vnet_state *)private_data;
+    struct virtio_queue * q = &(vnet_brg->queue[RECV_QUEUE]);
+    char *buf = (char *)pkt;
+    int ret_val;
+    int pkt_head_len = ((addr_t)pkt->data) - ((addr_t)pkt); 
+    uint32_t data_len = sizeof(struct v3_vnet_pkt); //TODO: should not transfer the whole struct, only the data size 
+    unsigned long flags;
+
+    flags = v3_lock_irqsave(vnet_brg->lock);
+       
+    PrintDebug("VNET Bridge: RX: pkt sent to guest size: %d\n, pkt_header_len: %d\n", data_len, pkt_head_len);
+    v3_hexdump(buf, 100, NULL, 0);
+
+    if (q->ring_avail_addr == 0) {
+       PrintError("Queue is not set\n");
+       ret_val = -1;
+       goto exit;
+    }
+
+    if (q->last_avail_idx > q->avail->index)
+       q->idx_overflow = true;
+    q->last_avail_idx = q->avail->index;
+
+    if (q->cur_avail_idx < q->avail->index || (q->idx_overflow && q->cur_avail_idx < q->avail->index+65536)){
+       uint16_t buf_idx = q->avail->ring[q->cur_avail_idx % q->queue_size];
+       uint32_t len = 0;
+       uint32_t offset = 0;
+       struct vring_desc * buf_desc = NULL;
+
+       buf_desc = &(q->desc[buf_idx]);
+       PrintDebug("VNET Bridge RX: buffer desc len: %d\n", buf_desc->length);
+
+       len = copy_data_to_desc(&(vnet_brg->vm->cores[0]), buf_desc, buf + offset, data_len - offset);
+       if (len == -1) {
+           ret_val = -1;
+           goto exit;
+       }
+       offset += len;
+       buf_desc->length = len;
+       
+       if (offset >= data_len) {
+           buf_desc->flags &= ~VIRTIO_NEXT_FLAG;
+       }else {
+           PrintDebug("VNET Bridge: RX: Pkt not fit into one desc buf\n");
+       }
+
+       q->used->ring[q->used->index % q->queue_size].id = q->avail->ring[q->cur_avail_idx % q->queue_size];
+       q->used->ring[q->used->index % q->queue_size].length = data_len; // This should be the total length of data sent to guest (header+pkt_data)
+       q->used->index++;
+
+       int last_idx = q->cur_avail_idx;
+       q->cur_avail_idx++;
+       if (q->cur_avail_idx < last_idx)
+           q->idx_overflow = false;
+    } else {
+       PrintDebug("VNET Bridge: Handle RX: Fails to send, no available buffer: current_idx:%d, availIdx: %d\n", q->cur_avail_idx, q->avail->index);
+    }
+
+    if (!(q->avail->flags & VIRTIO_NO_IRQ_FLAG)) {
+       v3_pci_raise_irq(vnet_brg->pci_bus, 0, vnet_brg->pci_dev);
+       vnet_brg->virtio_cfg.pci_isr = 0x1;
+       PrintDebug("Raising IRQ %d\n",  vnet_brg->pci_dev->config_header.intr_line);
+    }
+
+    ret_val = 0;
+
+exit:
+
+    v3_unlock_irqrestore(vnet_brg->lock, flags);
+    return ret_val;
+}
+
+static int vnet_send(struct v3_vnet_pkt *pkt, int pkt_len, void * private_data){
+    struct guest_info *core  = (struct guest_info *)private_data; 
+
+#ifdef CONFIG_DEBUG_VNET_BRG
+    {
+       PrintDebug("VNET Bridge: send pkt size: %d\n", pkt->size);
+       v3_hexdump(pkt->data,pkt->size, NULL, 0);
+    }
+#endif
+
+#ifdef CONFIG_VNET_PROFILE
+    uint64_t start, end;
+    rdtscll(start);
+    core->vnet_times.time_copy_from_guest = start - core->vnet_times.virtio_handle_start;
+#endif
+
+    pkt->src_type = LINK_EDGE;
+
+    return v3_vnet_send_pkt(pkt, (void *)core);
+}
+
+static int pkt_tx(struct guest_info *core, struct virtio_vnet_state * vnet_brg, struct vring_desc * buf_desc) 
+{
+    uint8_t * buf = NULL;
+    uint32_t len = buf_desc->length;
+    struct v3_vnet_pkt *pkt;
+
+    if (guest_pa_to_host_va(core, buf_desc->addr_gpa, (addr_t *)&(buf)) == -1) {
+       PrintError("Could not translate buffer address\n");
+       return -1;
+    }
+    pkt = (struct v3_vnet_pkt *)buf;
+    if (vnet_send(pkt, len, (void *)core) == -1) {
+       return -1;
+    }
+
+    return 0;
+}
+
+static int handle_pkt_tx(struct guest_info *core, struct virtio_vnet_state * vnet_brg) 
+{
+    struct virtio_queue * q = &(vnet_brg->queue[XMIT_QUEUE]);
+
+    if (q->avail->index < q->last_avail_idx) {
+       q->idx_overflow = true;
+    }
+
+    q->last_avail_idx = q->avail->index;
+
+    PrintDebug("VNET Bridge Device: Handle TX\n");
+
+    while (q->cur_avail_idx < q->avail->index || 
+          (q->idx_overflow && q->cur_avail_idx < (q->avail->index + 65536))) {
+       struct vring_desc * buf_desc = NULL;
+       uint16_t desc_idx = q->avail->ring[q->cur_avail_idx % q->queue_size];
+
+       buf_desc = &(q->desc[desc_idx]);
+
+       PrintDebug("VNET Bridge: Handle TX buf_len: %d\n", buf_desc->length);
+
+       if (pkt_tx(core, vnet_brg, buf_desc) == -1) {
+               PrintError("Error handling nic operation\n");
+               return -1;
+       }
+        
+       q->used->ring[q->used->index % q->queue_size].id = q->avail->ring[q->cur_avail_idx % q->queue_size];
+       q->used->ring[q->used->index % q->queue_size].length = buf_desc->length; // What do we set this to????
+       q->used->index++;
+
+       int last_idx = q->cur_avail_idx;
+       q->cur_avail_idx ++;
+       if (q->cur_avail_idx < last_idx) {
+           q->idx_overflow = false;
+       }
+    }
+
+    if (!(q->avail->flags & VIRTIO_NO_IRQ_FLAG)) {
+       v3_pci_raise_irq(vnet_brg->pci_bus, 0, vnet_brg->pci_dev);
+       vnet_brg->virtio_cfg.pci_isr = 0x1;
+    }
+
+#ifdef CONFIG_VNET_PROFILE
+    uint64_t time;
+    rdtscll(time);
+    core->vnet_times.total_handle_time = time - core->vnet_times.virtio_handle_start;
+    core->vnet_times.print = true;
+#endif
+
+    return 0;
+}
 
 static int virtio_io_write(struct guest_info * core, uint16_t port, void * src, uint_t length, void * private_data) {
-    struct virtio_vnet_state * vnet_state = (struct virtio_vnet_state *)private_data;
-    int port_idx = port % vnet_state->io_range_size;
+    struct virtio_vnet_state * vnet_brg = (struct virtio_vnet_state *)private_data;
+    int port_idx = port % vnet_brg->io_range_size;
 
 
-    PrintDebug("VirtioVNET: VIRTIO VNET Write for port %d len=%d, value=%x\n", 
+    PrintDebug("VNET Bridge: VIRTIO VNET Write for port %d len=%d, value=%x\n", 
               port, length, *(uint32_t *)src);
-    PrintDebug("VirtioVNET: port idx=%d\n", port_idx);
+    PrintDebug("VNET Bridge: port idx=%d\n", port_idx);
 
 
     switch (port_idx) {
@@ -214,9 +422,8 @@ static int virtio_io_write(struct guest_info * core, uint16_t port, void * src,
            if (length != 4) {
                PrintError("Illegal write length for guest features\n");
                return -1;
-           }
-           
-           vnet_state->virtio_cfg.guest_features = *(uint32_t *)src;
+           }    
+           vnet_brg->virtio_cfg.guest_features = *(uint32_t *)src;
 
            break;
        case VRING_PG_NUM_PORT:
@@ -224,41 +431,39 @@ static int virtio_io_write(struct guest_info * core, uint16_t port, void * src,
                addr_t pfn = *(uint32_t *)src;
                addr_t page_addr = (pfn << VIRTIO_PAGE_SHIFT);
 
-               vnet_state->cur_queue->pfn = pfn;
+               vnet_brg->cur_queue->pfn = pfn;
                
-               vnet_state->cur_queue->ring_desc_addr = page_addr ;
-               vnet_state->cur_queue->ring_avail_addr = page_addr + (QUEUE_SIZE * sizeof(struct vring_desc));
-               vnet_state->cur_queue->ring_used_addr = ( vnet_state->cur_queue->ring_avail_addr + \
+               vnet_brg->cur_queue->ring_desc_addr = page_addr ;
+               vnet_brg->cur_queue->ring_avail_addr = page_addr + (QUEUE_SIZE * sizeof(struct vring_desc));
+               vnet_brg->cur_queue->ring_used_addr = ( vnet_brg->cur_queue->ring_avail_addr + \
                                                 sizeof(struct vring_avail)    + \
                                                 (QUEUE_SIZE * sizeof(uint16_t)));
                
                // round up to next page boundary.
-               vnet_state->cur_queue->ring_used_addr = (vnet_state->cur_queue->ring_used_addr + 0xfff) & ~0xfff;
+               vnet_brg->cur_queue->ring_used_addr = (vnet_brg->cur_queue->ring_used_addr + 0xfff) & ~0xfff;
 
-               if (guest_pa_to_host_va(core, vnet_state->cur_queue->ring_desc_addr, (addr_t *)&(vnet_state->cur_queue->desc)) == -1) {
+               if (guest_pa_to_host_va(core, vnet_brg->cur_queue->ring_desc_addr, (addr_t *)&(vnet_brg->cur_queue->desc)) == -1) {
                    PrintError("Could not translate ring descriptor address\n");
                    return -1;
                }
 
-
-               if (guest_pa_to_host_va(core, vnet_state->cur_queue->ring_avail_addr, (addr_t *)&(vnet_state->cur_queue->avail)) == -1) {
+               if (guest_pa_to_host_va(core, vnet_brg->cur_queue->ring_avail_addr, (addr_t *)&(vnet_brg->cur_queue->avail)) == -1) {
                    PrintError("Could not translate ring available address\n");
                    return -1;
                }
 
-
-               if (guest_pa_to_host_va(core, vnet_state->cur_queue->ring_used_addr, (addr_t *)&(vnet_state->cur_queue->used)) == -1) {
+               if (guest_pa_to_host_va(core, vnet_brg->cur_queue->ring_used_addr, (addr_t *)&(vnet_brg->cur_queue->used)) == -1) {
                    PrintError("Could not translate ring used address\n");
                    return -1;
                }
 
-               PrintDebug("VirtioVNET: RingDesc_addr=%p, Avail_addr=%p, Used_addr=%p\n",
-                          (void *)(vnet_state->cur_queue->ring_desc_addr),
-                          (void *)(vnet_state->cur_queue->ring_avail_addr),
-                          (void *)(vnet_state->cur_queue->ring_used_addr));
+               PrintDebug("VNET Bridge: RingDesc_addr=%p, Avail_addr=%p, Used_addr=%p\n",
+                          (void *)(vnet_brg->cur_queue->ring_desc_addr),
+                          (void *)(vnet_brg->cur_queue->ring_avail_addr),
+                          (void *)(vnet_brg->cur_queue->ring_used_addr));
 
-               PrintDebug("VirtioVNET: RingDesc=%p, Avail=%p, Used=%p\n", 
-                          vnet_state->cur_queue->desc, vnet_state->cur_queue->avail, vnet_state->cur_queue->used);
+               PrintDebug("VNET Bridge: RingDesc=%p, Avail=%p, Used=%p\n", 
+                          vnet_brg->cur_queue->desc, vnet_brg->cur_queue->avail, vnet_brg->cur_queue->used);
 
            } else {
                PrintError("Illegal write length for page frame number\n");
@@ -266,46 +471,53 @@ static int virtio_io_write(struct guest_info * core, uint16_t port, void * src,
            }
            break;
        case VRING_Q_SEL_PORT:
-           vnet_state->virtio_cfg.vring_queue_selector = *(uint16_t *)src;
+           vnet_brg->virtio_cfg.vring_queue_selector = *(uint16_t *)src;
 
-           if (vnet_state->virtio_cfg.vring_queue_selector > NUM_QUEUES) {
-               PrintError("Virtio Symbiotic device has no qeueues. Selected %d\n", 
-                          vnet_state->virtio_cfg.vring_queue_selector);
+           if (vnet_brg->virtio_cfg.vring_queue_selector > NUM_QUEUES) {
+               PrintError("VNET Bridge device has no qeueues. Selected %d\n", 
+                          vnet_brg->virtio_cfg.vring_queue_selector);
                return -1;
            }
            
-           vnet_state->cur_queue = &(vnet_state->queue[vnet_state->virtio_cfg.vring_queue_selector]);
+           vnet_brg->cur_queue = &(vnet_brg->queue[vnet_brg->virtio_cfg.vring_queue_selector]);
 
            break;
        case VRING_Q_NOTIFY_PORT: {
            uint16_t queue_idx = *(uint16_t *)src;
 
-           PrintDebug("VirtioVNET: Handling Kick\n");
+           PrintDebug("VNET Bridge: Handling Kick\n");
 
            if (queue_idx == 0) {
-               if (handle_cmd_kick(core, vnet_state) == -1) {
-                   PrintError("Could not handle VNET Control command\n");
+               if (handle_cmd_kick(core, vnet_brg) == -1) {
+                   PrintError("Could not handle Virtio VNET Control command\n");
                    return -1;
                }
+           } else if (queue_idx == 1) {
+               if (handle_pkt_tx(core, vnet_brg) == -1){
+                   PrintError("Could not handle Virtio VNET TX\n");
+                   return -1;
+               }
+           } else if (queue_idx == 2) {
+               PrintDebug("VNET Bridge: receive kick on RX Queue\n");
            } else {
-               PrintError("Kick on invalid queue (%d)\n", queue_idx);
+               PrintError("VNET Bridge: Kick on invalid queue (%d)\n", queue_idx);
                return -1;
            }
 
            break;
        }
        case VIRTIO_STATUS_PORT:
-           vnet_state->virtio_cfg.status = *(uint8_t *)src;
+           vnet_brg->virtio_cfg.status = *(uint8_t *)src;
 
-           if (vnet_state->virtio_cfg.status == 0) {
-               PrintDebug("VirtioVNET: Resetting device\n");
-               virtio_reset(vnet_state);
+           if (vnet_brg->virtio_cfg.status == 0) {
+               PrintDebug("VNET Bridge: Resetting device\n");
+               virtio_reset(vnet_brg);
            }
 
            break;
 
        case VIRTIO_ISR_PORT:
-           vnet_state->virtio_cfg.pci_isr = *(uint8_t *)src;
+           vnet_brg->virtio_cfg.pci_isr = *(uint8_t *)src;
            break;
        default:
            return -1;
@@ -388,7 +600,12 @@ static int virtio_io_read(struct guest_info * core, uint16_t port, void * dst, u
     return length;
 }
 
+static int vnet_brg_input(struct v3_vm_info * vm, 
+                               struct v3_vnet_pkt * pkt, 
+                               void * private_data){
 
+    return vnet_brg_rx(pkt, private_data);
+}
 
 static struct v3_device_ops dev_ops = {
     .free = NULL,
@@ -398,23 +615,25 @@ static struct v3_device_ops dev_ops = {
 };
 
 
-static int vnet_init(struct v3_vm_info * vm, v3_cfg_tree_t * cfg) {
+static int vnet_brg_init(struct v3_vm_info * vm, v3_cfg_tree_t * cfg) {
     struct vm_device * pci_bus = v3_find_dev(vm, v3_cfg_val(cfg, "bus"));
-    struct virtio_vnet_state * virtio_state = NULL;
+    struct virtio_vnet_state * vbrg_state = NULL;
     struct pci_device * pci_dev = NULL;
     char * name = v3_cfg_val(cfg, "name");
 
-    PrintDebug("VirtioVNET: Initializing VIRTIO VNET Control device\n");
+    PrintDebug("VNET Bridge: Initializing VNET Bridge Control device: %s\n", name);
 
     if (pci_bus == NULL) {
-       PrintError("VirtIO devices require a PCI Bus");
+       PrintError("VNET Bridge device require a PCI Bus");
        return -1;
     }
     
-    virtio_state  = (struct virtio_vnet_state *)V3_Malloc(sizeof(struct virtio_vnet_state));
-    memset(virtio_state, 0, sizeof(struct virtio_vnet_state));
+    vbrg_state  = (struct virtio_vnet_state *)V3_Malloc(sizeof(struct virtio_vnet_state));
+    memset(vbrg_state, 0, sizeof(struct virtio_vnet_state));
+       
+    vbrg_state->vm = vm;
 
-    struct vm_device * dev = v3_allocate_device(name, &dev_ops, virtio_state);
+    struct vm_device * dev = v3_allocate_device(name, &dev_ops, vbrg_state);
 
     if (v3_attach_device(vm, dev) == -1) {
        PrintError("Could not attach device %s\n", name);
@@ -429,38 +648,35 @@ static int vnet_init(struct v3_vm_info * vm, v3_cfg_tree_t * cfg) {
        int tmp_ports = num_ports;
        int i;
 
-
        // This gets the number of ports, rounded up to a power of 2
-       virtio_state->io_range_size = 1; // must be a power of 2
+       vbrg_state->io_range_size = 1; // must be a power of 2
 
        while (tmp_ports > 0) {
            tmp_ports >>= 1;
-           virtio_state->io_range_size <<= 1;
+           vbrg_state->io_range_size <<= 1;
        }
        
        // this is to account for any low order bits being set in num_ports
        // if there are none, then num_ports was already a power of 2 so we shift right to reset it
-       if ((num_ports & ((virtio_state->io_range_size >> 1) - 1)) == 0) {
-           virtio_state->io_range_size >>= 1;
+       if ((num_ports & ((vbrg_state->io_range_size >> 1) - 1)) == 0) {
+           vbrg_state->io_range_size >>= 1;
        }
 
-
        for (i = 0; i < 6; i++) {
            bars[i].type = PCI_BAR_NONE;
        }
 
        bars[0].type = PCI_BAR_IO;
        bars[0].default_base_port = -1;
-       bars[0].num_ports = virtio_state->io_range_size;
-
+       bars[0].num_ports = vbrg_state->io_range_size;
        bars[0].io_read = virtio_io_read;
        bars[0].io_write = virtio_io_write;
-       bars[0].private_data = virtio_state;
+       bars[0].private_data = vbrg_state;
 
        pci_dev = v3_pci_register_device(pci_bus, PCI_STD_DEVICE, 
                                         0, PCI_AUTO_DEV_NUM, 0,
                                         "LNX_VIRTIO_VNET", bars,
-                                        NULL, NULL, NULL, virtio_state);
+                                        NULL, NULL, NULL, vbrg_state);
 
        if (!pci_dev) {
            PrintError("Could not register PCI Device\n");
@@ -469,29 +685,45 @@ static int vnet_init(struct v3_vm_info * vm, v3_cfg_tree_t * cfg) {
        
        pci_dev->config_header.vendor_id = VIRTIO_VENDOR_ID;
        pci_dev->config_header.subsystem_vendor_id = VIRTIO_SUBVENDOR_ID;
-       
-
        pci_dev->config_header.device_id = VIRTIO_VNET_DEV_ID;
        pci_dev->config_header.class = PCI_CLASS_MEMORY;
        pci_dev->config_header.subclass = PCI_MEM_SUBCLASS_RAM;
-    
        pci_dev->config_header.subsystem_id = VIRTIO_VNET_SUBDEVICE_ID;
-
-
        pci_dev->config_header.intr_pin = 1;
-
        pci_dev->config_header.max_latency = 1; // ?? (qemu does it...)
 
 
-       virtio_state->pci_dev = pci_dev;
-       virtio_state->pci_bus = pci_bus;
+       vbrg_state->pci_dev = pci_dev;
+       vbrg_state->pci_bus = pci_bus;
     }
 
-    virtio_reset(virtio_state);
+    virtio_reset(vbrg_state);
+
+    v3_vnet_add_bridge(vm, vnet_brg_input, (void *)vbrg_state);
+
+//for temporary hack
+#if 1  
+    {
+       uchar_t dstmac[6] = {0xff,0xff,0xff,0xff,0xff,0xff};
+       uchar_t zeromac[6] = {0,0,0,0,0,0};
+
+       struct v3_vnet_route route;
+       route.dst_id = 0;
+       route.dst_type = LINK_EDGE;
+       route.src_id = -1;
+       route.src_type = LINK_ANY;
+
+       memcpy(route.dst_mac, dstmac, 6);
+       route.dst_mac_qual = MAC_NONE;
+       memcpy(route.src_mac, zeromac, 6);
+       route.src_mac_qual = MAC_ANY;
+          
+       v3_vnet_add_route(route);
+    }
+#endif
 
-   
     return 0;
 }
 
 
-device_register("LNX_VIRTIO_VNET", vnet_init)
+device_register("LNX_VNET_BRG", vnet_brg_init)