This commit was manufactured by cvs2svn to create branch 'vserver'.
[linux-2.6.git] / arch / xtensa / platform-iss / network.c
diff --git a/arch/xtensa/platform-iss/network.c b/arch/xtensa/platform-iss/network.c
new file mode 100644 (file)
index 0000000..0dc55cc
--- /dev/null
@@ -0,0 +1,824 @@
+/*
+ *
+ * arch/xtensa/platform-iss/network.c
+ *
+ * Platform specific initialization.
+ *
+ * Authors: Chris Zankel <chris@zankel.net>
+ * Based on work form the UML team.
+ *
+ * Copyright 2005 Tensilica Inc.
+ *
+ * This program is free software; you can redistribute  it and/or modify it
+ * under  the terms of  the GNU General  Public License as published by the
+ * Free Software Foundation;  either version 2 of the  License, or (at your
+ * option) any later version.
+ *
+ */
+
+#include <linux/config.h>
+#include <linux/list.h>
+#include <linux/irq.h>
+#include <linux/spinlock.h>
+#include <linux/slab.h>
+#include <linux/timer.h>
+#include <linux/if_ether.h>
+#include <linux/inetdevice.h>
+#include <linux/init.h>
+#include <linux/if_tun.h>
+#include <linux/etherdevice.h>
+#include <linux/interrupt.h>
+#include <linux/ioctl.h>
+#include <linux/bootmem.h>
+#include <linux/ethtool.h>
+#include <linux/rtnetlink.h>
+#include <linux/timer.h>
+#include <linux/platform_device.h>
+
+#include <xtensa/simcall.h>
+
+#define DRIVER_NAME "iss-netdev"
+#define ETH_MAX_PACKET 1500
+#define ETH_HEADER_OTHER 14
+#define ISS_NET_TIMER_VALUE (2 * HZ)
+
+
+static DEFINE_SPINLOCK(opened_lock);
+static LIST_HEAD(opened);
+
+static DEFINE_SPINLOCK(devices_lock);
+static LIST_HEAD(devices);
+
+/* ------------------------------------------------------------------------- */
+
+/* We currently only support the TUNTAP transport protocol. */
+
+#define TRANSPORT_TUNTAP_NAME "tuntap"
+#define TRANSPORT_TUNTAP_MTU ETH_MAX_PACKET
+
+struct tuntap_info {
+       char dev_name[IFNAMSIZ];
+       int fixed_config;
+       unsigned char gw[ETH_ALEN];
+       int fd;
+};
+
+/* ------------------------------------------------------------------------- */
+
+
+/* This structure contains out private information for the driver. */
+
+struct iss_net_private {
+
+       struct list_head device_list;
+       struct list_head opened_list;
+
+       spinlock_t lock;
+       struct net_device *dev;
+       struct platform_device pdev;
+       struct timer_list tl;
+       struct net_device_stats stats;
+
+       struct timer_list timer;
+       unsigned int timer_val;
+
+       int index;
+       int mtu;
+
+       unsigned char mac[ETH_ALEN];
+       int have_mac;
+
+       struct {
+               union {
+                       struct tuntap_info tuntap;
+               } info;
+
+               int (*open)(struct iss_net_private *lp);
+               void (*close)(struct iss_net_private *lp);
+               int (*read)(struct iss_net_private *lp, struct sk_buff **skb);
+               int (*write)(struct iss_net_private *lp, struct sk_buff **skb);
+               unsigned short (*protocol)(struct sk_buff *skb);
+               int (*poll)(struct iss_net_private *lp);
+       } tp;
+
+};
+
+/* ======================= ISS SIMCALL INTERFACE =========================== */
+
+/* Note: __simc must _not_ be declared inline! */
+
+static int errno;
+
+static int __simc (int a, int b, int c, int d, int e, int f)
+{
+       int ret;
+       __asm__ __volatile__ ("simcall\n"
+                             "mov %0, a2\n"
+                             "mov %1, a3\n" : "=a" (ret), "=a" (errno)
+                             : : "a2", "a3");
+       return ret;
+}
+
+static int inline simc_open(char *file, int flags, int mode)
+{
+       return __simc(SYS_open, (int) file, flags, mode, 0, 0);
+}
+
+static int inline simc_close(int fd)
+{
+       return __simc(SYS_close, fd, 0, 0, 0, 0);
+}
+
+static int inline simc_ioctl(int fd, int request, void *arg)
+{
+       return __simc(SYS_ioctl, fd, request, (int) arg, 0, 0);
+}
+
+static int inline simc_read(int fd, void *buf, size_t count)
+{
+       return __simc(SYS_read, fd, (int) buf, count, 0, 0);
+}
+
+static int inline simc_write(int fd, void *buf, size_t count)
+{
+       return __simc(SYS_write, fd, (int) buf, count, 0, 0);
+}
+
+static int inline simc_poll(int fd)
+{
+       struct timeval tv = { .tv_sec = 0, .tv_usec = 0 };
+
+       return __simc(SYS_select_one, fd, XTISS_SELECT_ONE_READ, (int)&tv,0,0);
+}
+
+/* ================================ HELPERS ================================ */
+
+
+static char *split_if_spec(char *str, ...)
+{
+       char **arg, *end;
+       va_list ap;
+
+       va_start(ap, str);
+       while ((arg = va_arg(ap, char**)) != NULL) {
+               if (*str == '\0')
+                       return NULL;
+               end = strchr(str, ',');
+               if (end != str)
+                       *arg = str;
+               if (end == NULL)
+                       return NULL;
+               *end ++ = '\0';
+               str = end;
+       }
+       va_end(ap);
+       return str;
+}
+
+
+#if 0
+/* Adjust SKB. */
+
+struct sk_buff *ether_adjust_skb(struct sk_buff *skb, int extra)
+{
+       if ((skb != NULL) && (skb_tailroom(skb) < extra)) {
+               struct sk_buff *skb2;
+
+               skb2 = skb_copy_expand(skb, 0, extra, GFP_ATOMIC);
+               dev_kfree_skb(skb);
+               skb = skb2;
+       }
+       if (skb != NULL)
+               skb_put(skb, extra);
+
+       return skb;
+}
+#endif
+
+/* Return the IP address as a string for a given device. */
+
+static void dev_ip_addr(void *d, char *buf, char *bin_buf)
+{
+       struct net_device *dev = d;
+       struct in_device *ip = dev->ip_ptr;
+       struct in_ifaddr *in;
+       u32 addr;
+
+       if ((ip == NULL) || ((in = ip->ifa_list) == NULL)) {
+               printk(KERN_WARNING "Device not assigned an IP address!\n");
+               return;
+       }
+
+       addr = in->ifa_address;
+       sprintf(buf, "%d.%d.%d.%d", addr & 0xff, (addr >> 8) & 0xff,
+               (addr >> 16) & 0xff, addr >> 24);
+
+       if (bin_buf) {
+               bin_buf[0] = addr & 0xff;
+               bin_buf[1] = (addr >> 8) & 0xff;
+               bin_buf[2] = (addr >> 16) & 0xff;
+               bin_buf[3] = addr >> 24;
+       }
+}
+
+/* Set Ethernet address of the specified device. */
+
+static void inline set_ether_mac(void *d, unsigned char *addr)
+{
+       struct net_device *dev = d;
+       memcpy(dev->dev_addr, addr, ETH_ALEN);
+}
+
+
+/* ======================= TUNTAP TRANSPORT INTERFACE ====================== */
+
+static int tuntap_open(struct iss_net_private *lp)
+{
+       struct ifreq ifr;
+       char *dev_name = lp->tp.info.tuntap.dev_name;
+       int err = -EINVAL;
+       int fd;
+
+       /* We currently only support a fixed configuration. */
+
+       if (!lp->tp.info.tuntap.fixed_config)
+               return -EINVAL;
+
+       if ((fd = simc_open("/dev/net/tun", 02, 0)) < 0) {      /* O_RDWR */
+               printk("Failed to open /dev/net/tun, returned %d "
+                      "(errno = %d)\n", fd, errno);
+               return fd;
+       }
+
+       memset(&ifr, 0, sizeof ifr);
+       ifr.ifr_flags = IFF_TAP | IFF_NO_PI;
+       strlcpy(ifr.ifr_name, dev_name, sizeof ifr.ifr_name - 1);
+
+       if ((err = simc_ioctl(fd, TUNSETIFF, (void*) &ifr)) < 0) {
+               printk("Failed to set interface, returned %d "
+                      "(errno = %d)\n", err, errno);
+               simc_close(fd);
+               return err;
+       }
+
+       lp->tp.info.tuntap.fd = fd;
+       return err;
+}
+
+static void tuntap_close(struct iss_net_private *lp)
+{
+#if 0
+       if (lp->tp.info.tuntap.fixed_config)
+               iter_addresses(lp->tp.info.tuntap.dev, close_addr, lp->host.dev_name);
+#endif
+       simc_close(lp->tp.info.tuntap.fd);
+       lp->tp.info.tuntap.fd = -1;
+}
+
+static int tuntap_read (struct iss_net_private *lp, struct sk_buff **skb)
+{
+#if 0
+       *skb = ether_adjust_skb(*skb, ETH_HEADER_OTHER);
+       if (*skb == NULL)
+               return -ENOMEM;
+#endif
+
+       return simc_read(lp->tp.info.tuntap.fd,
+                       (*skb)->data, (*skb)->dev->mtu + ETH_HEADER_OTHER);
+}
+
+static int tuntap_write (struct iss_net_private *lp, struct sk_buff **skb)
+{
+       return simc_write(lp->tp.info.tuntap.fd, (*skb)->data, (*skb)->len);
+}
+
+unsigned short tuntap_protocol(struct sk_buff *skb)
+{
+       return eth_type_trans(skb, skb->dev);
+}
+
+static int tuntap_poll(struct iss_net_private *lp)
+{
+       return simc_poll(lp->tp.info.tuntap.fd);
+}
+
+/*
+ * Currently only a device name is supported.
+ * ethX=tuntap[,[mac address][,[device name]]]
+ */
+
+static int tuntap_probe(struct iss_net_private *lp, int index, char *init)
+{
+       const int len = strlen(TRANSPORT_TUNTAP_NAME);
+       char *dev_name = NULL, *mac_str = NULL, *rem = NULL;
+
+       /* Transport should be 'tuntap': ethX=tuntap,mac,dev_name */
+
+       if (strncmp(init, TRANSPORT_TUNTAP_NAME, len))
+               return 0;
+
+       if (*(init += strlen(TRANSPORT_TUNTAP_NAME)) == ',') {
+               if ((rem=split_if_spec(init+1, &mac_str, &dev_name)) != NULL) {
+                       printk("Extra garbage on specification : '%s'\n", rem);
+                       return 0;
+               }
+       } else if (*init != '\0') {
+               printk("Invalid argument: %s. Skipping device!\n", init);
+               return 0;
+       }
+
+       if (dev_name) {
+               strncpy(lp->tp.info.tuntap.dev_name, dev_name,
+                        sizeof lp->tp.info.tuntap.dev_name);
+               lp->tp.info.tuntap.fixed_config = 1;
+       } else
+               strcpy(lp->tp.info.tuntap.dev_name, TRANSPORT_TUNTAP_NAME);
+
+
+#if 0
+       if (setup_etheraddr(mac_str, lp->mac))
+               lp->have_mac = 1;
+#endif
+       lp->mtu = TRANSPORT_TUNTAP_MTU;
+
+       //lp->info.tuntap.gate_addr = gate_addr;
+
+       lp->tp.info.tuntap.fd = -1;
+
+       lp->tp.open = tuntap_open;
+       lp->tp.close = tuntap_close;
+       lp->tp.read = tuntap_read;
+       lp->tp.write = tuntap_write;
+       lp->tp.protocol = tuntap_protocol;
+       lp->tp.poll = tuntap_poll;
+
+       printk("TUN/TAP backend - ");
+#if 0
+       if (lp->host.gate_addr != NULL)
+               printk("IP = %s", lp->host.gate_addr);
+#endif
+       printk("\n");
+
+       return 1;
+}
+
+/* ================================ ISS NET ================================ */
+
+static int iss_net_rx(struct net_device *dev)
+{
+       struct iss_net_private *lp = dev->priv;
+       int pkt_len;
+       struct sk_buff *skb;
+
+       /* Check if there is any new data. */
+
+       if (lp->tp.poll(lp) == 0)
+               return 0;
+
+       /* Try to allocate memory, if it fails, try again next round. */
+
+       if ((skb = dev_alloc_skb(dev->mtu + 2 + ETH_HEADER_OTHER)) == NULL) {
+               lp->stats.rx_dropped++;
+               return 0;
+       }
+
+       skb_reserve(skb, 2);
+
+       /* Setup skb */
+
+       skb->dev = dev;
+       skb->mac.raw = skb->data;
+       pkt_len = lp->tp.read(lp, &skb);
+       skb_put(skb, pkt_len);
+
+       if (pkt_len > 0) {
+               skb_trim(skb, pkt_len);
+               skb->protocol = lp->tp.protocol(skb);
+       //      netif_rx(skb);
+               netif_rx_ni(skb);
+
+               lp->stats.rx_bytes += skb->len;
+               lp->stats.rx_packets++;
+               return pkt_len;
+       }
+       kfree_skb(skb);
+       return pkt_len;
+}
+
+static int iss_net_poll(void)
+{
+       struct list_head *ele;
+       int err, ret = 0;
+
+       spin_lock(&opened_lock);
+
+       list_for_each(ele, &opened) {
+               struct iss_net_private *lp;
+
+               lp = list_entry(ele, struct iss_net_private, opened_list);
+
+               if (!netif_running(lp->dev))
+                       break;
+
+               spin_lock(&lp->lock);
+
+               while ((err = iss_net_rx(lp->dev)) > 0)
+                       ret++;
+
+               spin_unlock(&lp->lock);
+
+               if (err < 0) {
+                       printk(KERN_ERR "Device '%s' read returned %d, "
+                              "shutting it down\n", lp->dev->name, err);
+                       dev_close(lp->dev);
+               } else {
+                       // FIXME reactivate_fd(lp->fd, ISS_ETH_IRQ);
+               }
+       }
+
+       spin_unlock(&opened_lock);
+       return ret;
+}
+
+
+static void iss_net_timer(unsigned long priv)
+{
+       struct iss_net_private* lp = (struct iss_net_private*) priv;
+
+       spin_lock(&lp->lock);
+
+       iss_net_poll();
+
+       mod_timer(&lp->timer, jiffies + lp->timer_val);
+
+       spin_unlock(&lp->lock);
+}
+
+
+static int iss_net_open(struct net_device *dev)
+{
+       struct iss_net_private *lp = dev->priv;
+       char addr[sizeof "255.255.255.255\0"];
+       int err;
+
+       spin_lock(&lp->lock);
+
+       if ((err = lp->tp.open(lp)) < 0)
+               goto out;
+
+       if (!lp->have_mac) {
+               dev_ip_addr(dev, addr, &lp->mac[2]);
+               set_ether_mac(dev, lp->mac);
+       }
+
+       netif_start_queue(dev);
+
+       /* clear buffer - it can happen that the host side of the interface
+        * is full when we gethere. In this case, new data is never queued,
+        * SIGIOs never arrive, and the net never works.
+        */
+       while ((err = iss_net_rx(dev)) > 0)
+               ;
+
+       spin_lock(&opened_lock);
+       list_add(&lp->opened_list, &opened);
+       spin_unlock(&opened_lock);
+
+       init_timer(&lp->timer);
+       lp->timer_val = ISS_NET_TIMER_VALUE;
+       lp->timer.data = (unsigned long) lp;
+       lp->timer.function = iss_net_timer;
+       mod_timer(&lp->timer, jiffies + lp->timer_val);
+
+out:
+       spin_unlock(&lp->lock);
+       return err;
+}
+
+static int iss_net_close(struct net_device *dev)
+{
+       struct iss_net_private *lp = dev->priv;
+printk("iss_net_close!\n");
+       netif_stop_queue(dev);
+       spin_lock(&lp->lock);
+
+       spin_lock(&opened_lock);
+       list_del(&opened);
+       spin_unlock(&opened_lock);
+
+       del_timer_sync(&lp->timer);
+
+       lp->tp.close(lp);
+
+       spin_unlock(&lp->lock);
+       return 0;
+}
+
+static int iss_net_start_xmit(struct sk_buff *skb, struct net_device *dev)
+{
+       struct iss_net_private *lp = dev->priv;
+       unsigned long flags;
+       int len;
+
+       netif_stop_queue(dev);
+       spin_lock_irqsave(&lp->lock, flags);
+
+       len = lp->tp.write(lp, &skb);
+
+       if (len == skb->len) {
+               lp->stats.tx_packets++;
+               lp->stats.tx_bytes += skb->len;
+               dev->trans_start = jiffies;
+               netif_start_queue(dev);
+
+               /* this is normally done in the interrupt when tx finishes */
+               netif_wake_queue(dev);
+
+       } else if (len == 0) {
+               netif_start_queue(dev);
+               lp->stats.tx_dropped++;
+
+       } else {
+               netif_start_queue(dev);
+               printk(KERN_ERR "iss_net_start_xmit: failed(%d)\n", len);
+       }
+
+       spin_unlock_irqrestore(&lp->lock, flags);
+
+       dev_kfree_skb(skb);
+       return 0;
+}
+
+
+static struct net_device_stats *iss_net_get_stats(struct net_device *dev)
+{
+       struct iss_net_private *lp = dev->priv;
+       return &lp->stats;
+}
+
+static void iss_net_set_multicast_list(struct net_device *dev)
+{
+#if 0
+       if (dev->flags & IFF_PROMISC)
+               return;
+       else if (dev->mc_count)
+               dev->flags |= IFF_ALLMULTI;
+       else
+               dev->flags &= ~IFF_ALLMULTI;
+#endif
+}
+
+static void iss_net_tx_timeout(struct net_device *dev)
+{
+#if 0
+       dev->trans_start = jiffies;
+       netif_wake_queue(dev);
+#endif
+}
+
+static int iss_net_set_mac(struct net_device *dev, void *addr)
+{
+#if 0
+       struct iss_net_private *lp = dev->priv;
+       struct sockaddr *hwaddr = addr;
+
+       spin_lock(&lp->lock);
+       memcpy(dev->dev_addr, hwaddr->sa_data, ETH_ALEN);
+       spin_unlock(&lp->lock);
+#endif
+
+       return 0;
+}
+
+static int iss_net_change_mtu(struct net_device *dev, int new_mtu)
+{
+#if 0
+       struct iss_net_private *lp = dev->priv;
+       int err = 0;
+
+       spin_lock(&lp->lock);
+
+       // FIXME not needed new_mtu = transport_set_mtu(new_mtu, &lp->user);
+
+       if (new_mtu < 0)
+               err = new_mtu;
+       else
+               dev->mtu = new_mtu;
+
+       spin_unlock(&lp->lock);
+       return err;
+#endif
+       return -EINVAL;
+}
+
+void iss_net_user_timer_expire(unsigned long _conn)
+{
+}
+
+
+static struct platform_driver iss_net_driver = {
+       .driver = {
+               .name  = DRIVER_NAME,
+       },
+};
+
+static int driver_registered;
+
+static int iss_net_configure(int index, char *init)
+{
+       struct net_device *dev;
+       struct iss_net_private *lp;
+       int err;
+
+       if ((dev = alloc_etherdev(sizeof *lp)) == NULL) {
+               printk(KERN_ERR "eth_configure: failed to allocate device\n");
+               return 1;
+       }
+
+       /* Initialize private element. */
+
+       lp = dev->priv;
+       *lp = ((struct iss_net_private) {
+               .device_list            = LIST_HEAD_INIT(lp->device_list),
+               .opened_list            = LIST_HEAD_INIT(lp->opened_list),
+               .lock                   = SPIN_LOCK_UNLOCKED,
+               .dev                    = dev,
+               .index                  = index,
+               //.fd                   = -1,
+               .mac                    = { 0xfe, 0xfd, 0x0, 0x0, 0x0, 0x0 },
+               .have_mac               = 0,
+               });
+
+       /*
+        * Try all transport protocols.
+        * Note: more protocols can be added by adding '&& !X_init(lp, eth)'.
+        */
+
+       if (!tuntap_probe(lp, index, init)) {
+               printk("Invalid arguments. Skipping device!\n");
+               goto errout;
+       }
+
+       printk(KERN_INFO "Netdevice %d ", index);
+       if (lp->have_mac)
+               printk("(%02x:%02x:%02x:%02x:%02x:%02x) ",
+                               lp->mac[0], lp->mac[1],
+                               lp->mac[2], lp->mac[3],
+                               lp->mac[4], lp->mac[5]);
+       printk(": ");
+
+       /* sysfs register */
+
+       if (!driver_registered) {
+               platform_driver_register(&iss_net_driver);
+               driver_registered = 1;
+       }
+
+       spin_lock(&devices_lock);
+       list_add(&lp->device_list, &devices);
+       spin_unlock(&devices_lock);
+
+       lp->pdev.id = index;
+       lp->pdev.name = DRIVER_NAME;
+       platform_device_register(&lp->pdev);
+       SET_NETDEV_DEV(dev,&lp->pdev.dev);
+
+       /*
+        * If this name ends up conflicting with an existing registered
+        * netdevice, that is OK, register_netdev{,ice}() will notice this
+        * and fail.
+        */
+       snprintf(dev->name, sizeof dev->name, "eth%d", index);
+
+       dev->mtu = lp->mtu;
+       dev->open = iss_net_open;
+       dev->hard_start_xmit = iss_net_start_xmit;
+       dev->stop = iss_net_close;
+       dev->get_stats = iss_net_get_stats;
+       dev->set_multicast_list = iss_net_set_multicast_list;
+       dev->tx_timeout = iss_net_tx_timeout;
+       dev->set_mac_address = iss_net_set_mac;
+       dev->change_mtu = iss_net_change_mtu;
+       dev->watchdog_timeo = (HZ >> 1);
+       dev->irq = -1;
+
+       rtnl_lock();
+       err = register_netdevice(dev);
+       rtnl_unlock();
+
+       if (err) {
+               printk("Error registering net device!\n");
+               /* XXX: should we call ->remove() here? */
+               free_netdev(dev);
+               return 1;
+       }
+
+       init_timer(&lp->tl);
+       lp->tl.function = iss_net_user_timer_expire;
+
+#if 0
+       if (lp->have_mac)
+               set_ether_mac(dev, lp->mac);
+#endif
+       return 0;
+
+errout:
+       // FIXME: unregister; free, etc..
+       return -EIO;
+
+}
+
+/* ------------------------------------------------------------------------- */
+
+/* Filled in during early boot */
+
+struct list_head eth_cmd_line = LIST_HEAD_INIT(eth_cmd_line);
+
+struct iss_net_init {
+       struct list_head list;
+       char *init;             /* init string */
+       int index;
+};
+
+/*
+ * Parse the command line and look for 'ethX=...' fields, and register all
+ * those fields. They will be later initialized in iss_net_init.
+ */
+
+#define ERR KERN_ERR "iss_net_setup: "
+
+static int iss_net_setup(char *str)
+{
+       struct iss_net_private *device = NULL;
+       struct iss_net_init *new;
+       struct list_head *ele;
+       char *end;
+       int n;
+
+       n = simple_strtoul(str, &end, 0);
+       if (end == str) {
+               printk(ERR "Failed to parse '%s'\n", str);
+               return 1;
+       }
+       if (n < 0) {
+               printk(ERR "Device %d is negative\n", n);
+               return 1;
+       }
+       if (*(str = end) != '=') {
+               printk(ERR "Expected '=' after device number\n");
+               return 1;
+       }
+
+       spin_lock(&devices_lock);
+
+       list_for_each(ele, &devices) {
+               device = list_entry(ele, struct iss_net_private, device_list);
+               if (device->index == n)
+                       break;
+       }
+
+       spin_unlock(&devices_lock);
+
+       if (device && device->index == n) {
+               printk(ERR "Device %d already configured\n", n);
+               return 1;
+       }
+
+       if ((new = alloc_bootmem(sizeof new)) == NULL) {
+               printk("Alloc_bootmem failed\n");
+               return 1;
+       }
+
+       INIT_LIST_HEAD(&new->list);
+       new->index = n;
+       new->init = str + 1;
+
+       list_add_tail(&new->list, &eth_cmd_line);
+       return 1;
+}
+
+#undef ERR
+
+__setup("eth", iss_net_setup);
+
+/*
+ * Initialize all ISS Ethernet devices previously registered in iss_net_setup.
+ */
+
+static int iss_net_init(void)
+{
+       struct list_head *ele, *next;
+
+       /* Walk through all Ethernet devices specified in the command line. */
+
+       list_for_each_safe(ele, next, &eth_cmd_line) {
+               struct iss_net_init *eth;
+               eth = list_entry(ele, struct iss_net_init, list);
+               iss_net_configure(eth->index, eth->init);
+       }
+
+       return 1;
+}
+
+module_init(iss_net_init);
+