diff options
Diffstat (limited to 'net/wanrouter')
-rw-r--r-- | net/wanrouter/Makefile | 7 | ||||
-rw-r--r-- | net/wanrouter/af_wanpipe.c | 2611 | ||||
-rw-r--r-- | net/wanrouter/patchlevel | 1 | ||||
-rw-r--r-- | net/wanrouter/wanmain.c | 888 | ||||
-rw-r--r-- | net/wanrouter/wanproc.c | 381 |
5 files changed, 3888 insertions, 0 deletions
diff --git a/net/wanrouter/Makefile b/net/wanrouter/Makefile new file mode 100644 index 0000000..9f188ab --- /dev/null +++ b/net/wanrouter/Makefile @@ -0,0 +1,7 @@ +# +# Makefile for the Linux WAN router layer. +# + +obj-$(CONFIG_WAN_ROUTER) += wanrouter.o + +wanrouter-objs := wanproc.o wanmain.o diff --git a/net/wanrouter/af_wanpipe.c b/net/wanrouter/af_wanpipe.c new file mode 100644 index 0000000..d93b19f --- /dev/null +++ b/net/wanrouter/af_wanpipe.c @@ -0,0 +1,2611 @@ +/***************************************************************************** +* af_wanpipe.c WANPIPE(tm) Secure Socket Layer. +* +* Author: Nenad Corbic <ncorbic@sangoma.com> +* +* Copyright: (c) 2000 Sangoma Technologies 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. +* ============================================================================ +* Due Credit: +* Wanpipe socket layer is based on Packet and +* the X25 socket layers. The above sockets were +* used for the specific use of Sangoma Technoloiges +* API programs. +* Packet socket Authors: Ross Biro, Fred N. van Kempen and +* Alan Cox. +* X25 socket Author: Jonathan Naylor. +* ============================================================================ +* Mar 15, 2002 Arnaldo C. Melo o Use wp_sk()->num, as it isnt anymore in sock +* Apr 25, 2000 Nenad Corbic o Added the ability to send zero length packets. +* Mar 13, 2000 Nenad Corbic o Added a tx buffer check via ioctl call. +* Mar 06, 2000 Nenad Corbic o Fixed the corrupt sock lcn problem. +* Server and client applicaton can run +* simultaneously without conflicts. +* Feb 29, 2000 Nenad Corbic o Added support for PVC protocols, such as +* CHDLC, Frame Relay and HDLC API. +* Jan 17, 2000 Nenad Corbic o Initial version, based on AF_PACKET socket. +* X25API support only. +* +******************************************************************************/ + +#include <linux/config.h> +#include <linux/types.h> +#include <linux/sched.h> +#include <linux/mm.h> +#include <linux/fcntl.h> +#include <linux/socket.h> +#include <linux/in.h> +#include <linux/inet.h> +#include <linux/netdevice.h> +#include <linux/poll.h> +#include <linux/wireless.h> +#include <linux/kmod.h> +#include <net/ip.h> +#include <net/protocol.h> +#include <linux/skbuff.h> +#include <net/sock.h> +#include <linux/errno.h> +#include <linux/timer.h> +#include <asm/system.h> +#include <asm/uaccess.h> +#include <linux/module.h> +#include <linux/init.h> +#include <linux/wanpipe.h> +#include <linux/if_wanpipe.h> +#include <linux/pkt_sched.h> +#include <linux/tcp.h> +#include <linux/if_wanpipe_common.h> +#include <linux/sdla_x25.h> + +#ifdef CONFIG_INET +#include <net/inet_common.h> +#endif + +#define SLOW_BACKOFF 0.1*HZ +#define FAST_BACKOFF 0.01*HZ + +//#define PRINT_DEBUG +#ifdef PRINT_DEBUG + #define DBG_PRINTK(format, a...) printk(format, ## a) +#else + #define DBG_PRINTK(format, a...) +#endif + + +/* SECURE SOCKET IMPLEMENTATION + * + * TRANSMIT: + * + * When the user sends a packet via send() system call + * the wanpipe_sendmsg() function is executed. + * + * Each packet is enqueud into sk->sk_write_queue transmit + * queue. When the packet is enqueued, a delayed transmit + * timer is triggerd which acts as a Bottom Half hander. + * + * wanpipe_delay_transmit() function (BH), dequeues packets + * from the sk->sk_write_queue transmit queue and sends it + * to the deriver via dev->hard_start_xmit(skb, dev) function. + * Note, this function is actual a function pointer of if_send() + * routine in the wanpipe driver. + * + * X25API GUARANTEED DELIVERY: + * + * In order to provide 100% guaranteed packet delivery, + * an atomic 'packet_sent' counter is implemented. Counter + * is incremented for each packet enqueued + * into sk->sk_write_queue. Counter is decremented each + * time wanpipe_delayed_transmit() function successfuly + * passes the packet to the driver. Before each send(), a poll + * routine checks the sock resources The maximum value of + * packet sent counter is 1, thus if one packet is queued, the + * application will block until that packet is passed to the + * driver. + * + * RECEIVE: + * + * Wanpipe device drivers call the socket bottom half + * function, wanpipe_rcv() to queue the incoming packets + * into an AF_WANPIPE socket queue. Based on wanpipe_rcv() + * return code, the driver knows whether the packet was + * successfully queued. If the socket queue is full, + * protocol flow control is used by the driver, if any, + * to slow down the traffic until the sock queue is free. + * + * Every time a packet arrives into a socket queue the + * socket wakes up processes which are waiting to receive + * data. + * + * If the socket queue is full, the driver sets a block + * bit which signals the socket to kick the wanpipe driver + * bottom half hander when the socket queue is partialy + * empty. wanpipe_recvmsg() function performs this action. + * + * In case of x25api, packets will never be dropped, since + * flow control is available. + * + * In case of streaming protocols like CHDLC, packets will + * be dropped but the statistics will be generated. + */ + + +/* The code below is used to test memory leaks. It prints out + * a message every time kmalloc and kfree system calls get executed. + * If the calls match there is no leak :) + */ + +/***********FOR DEBUGGING PURPOSES********************************************* +#define KMEM_SAFETYZONE 8 + +static void * dbg_kmalloc(unsigned int size, int prio, int line) { + void * v = kmalloc(size,prio); + printk(KERN_INFO "line %d kmalloc(%d,%d) = %p\n",line,size,prio,v); + return v; +} +static void dbg_kfree(void * v, int line) { + printk(KERN_INFO "line %d kfree(%p)\n",line,v); + kfree(v); +} + +#define kmalloc(x,y) dbg_kmalloc(x,y,__LINE__) +#define kfree(x) dbg_kfree(x,__LINE__) +******************************************************************************/ + + +/* List of all wanpipe sockets. */ +HLIST_HEAD(wanpipe_sklist); +static DEFINE_RWLOCK(wanpipe_sklist_lock); + +atomic_t wanpipe_socks_nr; +static unsigned long wanpipe_tx_critical; + +#if 0 +/* Private wanpipe socket structures. */ +struct wanpipe_opt +{ + void *mbox; /* Mail box */ + void *card; /* Card bouded to */ + struct net_device *dev; /* Bounded device */ + unsigned short lcn; /* Binded LCN */ + unsigned char svc; /* 0=pvc, 1=svc */ + unsigned char timer; /* flag for delayed transmit*/ + struct timer_list tx_timer; + unsigned poll_cnt; + unsigned char force; /* Used to force sock release */ + atomic_t packet_sent; +}; +#endif + +static int sk_count; +extern struct proto_ops wanpipe_ops; +static unsigned long find_free_critical; + +static void wanpipe_unlink_driver(struct sock *sk); +static void wanpipe_link_driver(struct net_device *dev, struct sock *sk); +static void wanpipe_wakeup_driver(struct sock *sk); +static int execute_command(struct sock *, unsigned char, unsigned int); +static int check_dev(struct net_device *dev, sdla_t *card); +struct net_device *wanpipe_find_free_dev(sdla_t *card); +static void wanpipe_unlink_card (struct sock *); +static int wanpipe_link_card (struct sock *); +static struct sock *wanpipe_make_new(struct sock *); +static struct sock *wanpipe_alloc_socket(void); +static inline int get_atomic_device(struct net_device *dev); +static int wanpipe_exec_cmd(struct sock *, int, unsigned int); +static int get_ioctl_cmd (struct sock *, void *); +static int set_ioctl_cmd (struct sock *, void *); +static void release_device(struct net_device *dev); +static void wanpipe_kill_sock_timer (unsigned long data); +static void wanpipe_kill_sock_irq (struct sock *); +static void wanpipe_kill_sock_accept (struct sock *); +static int wanpipe_do_bind(struct sock *sk, struct net_device *dev, + int protocol); +struct sock * get_newsk_from_skb (struct sk_buff *); +static int wanpipe_debug (struct sock *, void *); +static void wanpipe_delayed_transmit (unsigned long data); +static void release_driver(struct sock *); +static void start_cleanup_timer (struct sock *); +static void check_write_queue(struct sock *); +static int check_driver_busy (struct sock *); + +/*============================================================ + * wanpipe_rcv + * + * Wanpipe socket bottom half handler. This function + * is called by the WANPIPE device drivers to queue a + * incoming packet into the socket receive queue. + * Once the packet is queued, all processes waiting to + * read are woken up. + * + * During socket bind, this function is bounded into + * WANPIPE driver private. + *===========================================================*/ + +static int wanpipe_rcv(struct sk_buff *skb, struct net_device *dev, + struct sock *sk) +{ + struct wan_sockaddr_ll *sll = (struct wan_sockaddr_ll*)skb->cb; + wanpipe_common_t *chan = dev->priv; + /* + * When we registered the protocol we saved the socket in the data + * field for just this event. + */ + + skb->dev = dev; + + sll->sll_family = AF_WANPIPE; + sll->sll_hatype = dev->type; + sll->sll_protocol = skb->protocol; + sll->sll_pkttype = skb->pkt_type; + sll->sll_ifindex = dev->ifindex; + sll->sll_halen = 0; + + if (dev->hard_header_parse) + sll->sll_halen = dev->hard_header_parse(skb, sll->sll_addr); + + /* + * WAN_PACKET_DATA : Data which should be passed up the receive queue. + * WAN_PACKET_ASYC : Asynchronous data like place call, which should + * be passed up the listening sock. + * WAN_PACKET_ERR : Asynchronous data like clear call or restart + * which should go into an error queue. + */ + switch (skb->pkt_type){ + + case WAN_PACKET_DATA: + if (sock_queue_rcv_skb(sk,skb)<0){ + return -ENOMEM; + } + break; + case WAN_PACKET_CMD: + sk->sk_state = chan->state; + /* Bug fix: update Mar6. + * Do not set the sock lcn number here, since + * cmd is not guaranteed to be executed on the + * board, thus Lcn could be wrong */ + sk->sk_data_ready(sk, skb->len); + kfree_skb(skb); + break; + case WAN_PACKET_ERR: + sk->sk_state = chan->state; + if (sock_queue_err_skb(sk,skb)<0){ + return -ENOMEM; + } + break; + default: + printk(KERN_INFO "wansock: BH Illegal Packet Type Dropping\n"); + kfree_skb(skb); + break; + } + +//?????????????????????? +// if (sk->sk_state == WANSOCK_DISCONNECTED){ +// if (sk->sk_zapped) { +// //printk(KERN_INFO "wansock: Disconnected, killing early\n"); +// wanpipe_unlink_driver(sk); +// sk->sk_bound_dev_if = 0; +// } +// } + + return 0; +} + +/*============================================================ + * wanpipe_listen_rcv + * + * Wanpipe LISTEN socket bottom half handler. This function + * is called by the WANPIPE device drivers to queue an + * incoming call into the socket listening queue. + * Once the packet is queued, the waiting accept() process + * is woken up. + * + * During socket bind, this function is bounded into + * WANPIPE driver private. + * + * IMPORTANT NOTE: + * The accept call() is waiting for an skb packet + * which contains a pointer to a device structure. + * + * When we do a bind to a device structre, we + * bind a newly created socket into "chan->sk". Thus, + * when accept receives the skb packet, it will know + * from which dev it came form, and in turn it will know + * the address of the new sock. + * + * NOTE: This function gets called from driver ISR. + *===========================================================*/ + +static int wanpipe_listen_rcv (struct sk_buff *skb, struct sock *sk) +{ + wanpipe_opt *wp = wp_sk(sk), *newwp; + struct wan_sockaddr_ll *sll = (struct wan_sockaddr_ll*)skb->cb; + struct sock *newsk; + struct net_device *dev; + sdla_t *card; + mbox_cmd_t *mbox_ptr; + wanpipe_common_t *chan; + + /* Find a free device, if none found, all svc's are busy + */ + + card = (sdla_t*)wp->card; + if (!card){ + printk(KERN_INFO "wansock: LISTEN ERROR, No Card\n"); + return -ENODEV; + } + + dev = wanpipe_find_free_dev(card); + if (!dev){ + printk(KERN_INFO "wansock: LISTEN ERROR, No Free Device\n"); + return -ENODEV; + } + + chan=dev->priv; + chan->state = WANSOCK_CONNECTING; + + /* Allocate a new sock, which accept will bind + * and pass up to the user + */ + if ((newsk = wanpipe_make_new(sk)) == NULL){ + release_device(dev); + return -ENOMEM; + } + + + /* Initialize the new sock structure + */ + newsk->sk_bound_dev_if = dev->ifindex; + newwp = wp_sk(newsk); + newwp->card = wp->card; + + /* Insert the sock into the main wanpipe + * sock list. + */ + atomic_inc(&wanpipe_socks_nr); + + /* Allocate and fill in the new Mail Box. Then + * bind the mail box to the sock. It will be + * used by the ioctl call to read call information + * and to execute commands. + */ + if ((mbox_ptr = kmalloc(sizeof(mbox_cmd_t), GFP_ATOMIC)) == NULL) { + wanpipe_kill_sock_irq (newsk); + release_device(dev); + return -ENOMEM; + } + memset(mbox_ptr, 0, sizeof(mbox_cmd_t)); + memcpy(mbox_ptr,skb->data,skb->len); + + /* Register the lcn on which incoming call came + * from. Thus, if we have to clear it, we know + * which lcn to clear + */ + + newwp->lcn = mbox_ptr->cmd.lcn; + newwp->mbox = (void *)mbox_ptr; + + DBG_PRINTK(KERN_INFO "NEWSOCK : Device %s, bind to lcn %i\n", + dev->name,mbox_ptr->cmd.lcn); + + chan->lcn = mbox_ptr->cmd.lcn; + card->u.x.svc_to_dev_map[(chan->lcn%MAX_X25_LCN)] = dev; + + sock_reset_flag(newsk, SOCK_ZAPPED); + newwp->num = htons(X25_PROT); + + if (wanpipe_do_bind(newsk, dev, newwp->num)) { + wanpipe_kill_sock_irq (newsk); + release_device(dev); + return -EINVAL; + } + newsk->sk_state = WANSOCK_CONNECTING; + + + /* Fill in the standard sock address info */ + + sll->sll_family = AF_WANPIPE; + sll->sll_hatype = dev->type; + sll->sll_protocol = skb->protocol; + sll->sll_pkttype = skb->pkt_type; + sll->sll_ifindex = dev->ifindex; + sll->sll_halen = 0; + + skb->dev = dev; + sk->sk_ack_backlog++; + + /* We must do this manually, since the sock_queue_rcv_skb() + * function sets the skb->dev to NULL. However, we use + * the dev field in the accept function.*/ + if (atomic_read(&sk->sk_rmem_alloc) + skb->truesize >= + (unsigned)sk->sk_rcvbuf) { + + wanpipe_unlink_driver(newsk); + wanpipe_kill_sock_irq (newsk); + --sk->sk_ack_backlog; + return -ENOMEM; + } + + skb_set_owner_r(skb, sk); + skb_queue_tail(&sk->sk_receive_queue, skb); + sk->sk_data_ready(sk, skb->len); + + return 0; +} + + + +/*============================================================ + * wanpipe_make_new + * + * Create a new sock, and allocate a wanpipe private + * structure to it. Also, copy the important data + * from the original sock to the new sock. + * + * This function is used by wanpipe_listen_rcv() listen + * bottom half handler. A copy of the listening sock + * is created using this function. + * + *===========================================================*/ + +static struct sock *wanpipe_make_new(struct sock *osk) +{ + struct sock *sk; + + if (osk->sk_type != SOCK_RAW) + return NULL; + + if ((sk = wanpipe_alloc_socket()) == NULL) + return NULL; + + sk->sk_type = osk->sk_type; + sk->sk_socket = osk->sk_socket; + sk->sk_priority = osk->sk_priority; + sk->sk_protocol = osk->sk_protocol; + wp_sk(sk)->num = wp_sk(osk)->num; + sk->sk_rcvbuf = osk->sk_rcvbuf; + sk->sk_sndbuf = osk->sk_sndbuf; + sk->sk_state = WANSOCK_CONNECTING; + sk->sk_sleep = osk->sk_sleep; + + if (sock_flag(osk, SOCK_DBG)) + sock_set_flag(sk, SOCK_DBG); + + return sk; +} + +/* + * FIXME: wanpipe_opt has to include a sock in its definition and stop using + * sk_protinfo, but this code is not even compilable now, so lets leave it for + * later. + */ +static struct proto wanpipe_proto = { + .name = "WANPIPE", + .owner = THIS_MODULE, + .obj_size = sizeof(struct sock), +}; + +/*============================================================ + * wanpipe_make_new + * + * Allocate memory for the a new sock, and sock + * private data. + * + * Increment the module use count. + * + * This function is used by wanpipe_create() and + * wanpipe_make_new() functions. + * + *===========================================================*/ + +static struct sock *wanpipe_alloc_socket(void) +{ + struct sock *sk; + struct wanpipe_opt *wan_opt; + + if ((sk = sk_alloc(PF_WANPIPE, GFP_ATOMIC, &wanpipe_proto, 1)) == NULL) + return NULL; + + if ((wan_opt = kmalloc(sizeof(struct wanpipe_opt), GFP_ATOMIC)) == NULL) { + sk_free(sk); + return NULL; + } + memset(wan_opt, 0x00, sizeof(struct wanpipe_opt)); + + wp_sk(sk) = wan_opt; + + /* Use timer to send data to the driver. This will act + * as a BH handler for sendmsg functions */ + init_timer(&wan_opt->tx_timer); + wan_opt->tx_timer.data = (unsigned long)sk; + wan_opt->tx_timer.function = wanpipe_delayed_transmit; + + sock_init_data(NULL, sk); + return sk; +} + + +/*============================================================ + * wanpipe_sendmsg + * + * This function implements a sendto() system call, + * for AF_WANPIPE socket family. + * During socket bind() sk->sk_bound_dev_if is initialized + * to a correct network device. This number is used + * to find a network device to which the packet should + * be passed to. + * + * Each packet is queued into sk->sk_write_queue and + * delayed transmit bottom half handler is marked for + * execution. + * + * A socket must be in WANSOCK_CONNECTED state before + * a packet is queued into sk->sk_write_queue. + *===========================================================*/ + +static int wanpipe_sendmsg(struct kiocb *iocb, struct socket *sock, + struct msghdr *msg, int len) +{ + wanpipe_opt *wp; + struct sock *sk = sock->sk; + struct wan_sockaddr_ll *saddr=(struct wan_sockaddr_ll *)msg->msg_name; + struct sk_buff *skb; + struct net_device *dev; + unsigned short proto; + unsigned char *addr; + int ifindex, err, reserve = 0; + + + if (!sock_flag(sk, SOCK_ZAPPED)) + return -ENETDOWN; + + if (sk->sk_state != WANSOCK_CONNECTED) + return -ENOTCONN; + + if (msg->msg_flags & ~(MSG_DONTWAIT|MSG_CMSG_COMPAT)) + return(-EINVAL); + + /* it was <=, now one can send + * zero length packets */ + if (len < sizeof(x25api_hdr_t)) + return -EINVAL; + + wp = wp_sk(sk); + + if (saddr == NULL) { + ifindex = sk->sk_bound_dev_if; + proto = wp->num; + addr = NULL; + + }else{ + if (msg->msg_namelen < sizeof(struct wan_sockaddr_ll)){ + return -EINVAL; + } + + ifindex = sk->sk_bound_dev_if; + proto = saddr->sll_protocol; + addr = saddr->sll_addr; + } + + dev = dev_get_by_index(ifindex); + if (dev == NULL){ + printk(KERN_INFO "wansock: Send failed, dev index: %i\n",ifindex); + return -ENXIO; + } + dev_put(dev); + + if (sock->type == SOCK_RAW) + reserve = dev->hard_header_len; + + if (len > dev->mtu+reserve){ + return -EMSGSIZE; + } + + skb = sock_alloc_send_skb(sk, len + LL_RESERVED_SPACE(dev), + msg->msg_flags & MSG_DONTWAIT, &err); + + if (skb==NULL){ + goto out_unlock; + } + + skb_reserve(skb, LL_RESERVED_SPACE(dev)); + skb->nh.raw = skb->data; + + /* Returns -EFAULT on error */ + err = memcpy_fromiovec(skb_put(skb,len), msg->msg_iov, len); + if (err){ + goto out_free; + } + + if (dev->hard_header) { + int res; + err = -EINVAL; + res = dev->hard_header(skb, dev, ntohs(proto), addr, NULL, len); + if (res<0){ + goto out_free; + } + } + + skb->protocol = proto; + skb->dev = dev; + skb->priority = sk->sk_priority; + skb->pkt_type = WAN_PACKET_DATA; + + err = -ENETDOWN; + if (!(dev->flags & IFF_UP)) + goto out_free; + + if (atomic_read(&sk->sk_wmem_alloc) + skb->truesize > + (unsigned int)sk->sk_sndbuf){ + kfree_skb(skb); + return -ENOBUFS; + } + + skb_queue_tail(&sk->sk_write_queue,skb); + atomic_inc(&wp->packet_sent); + + if (!(test_and_set_bit(0, &wp->timer))) + mod_timer(&wp->tx_timer, jiffies + 1); + + return(len); + +out_free: + kfree_skb(skb); +out_unlock: + return err; +} + +/*============================================================ + * wanpipe_delayed_tarnsmit + * + * Transmit bottom half handler. It dequeues packets + * from sk->sk_write_queue and passes them to the + * driver. If the driver is busy, the packet is + * re-enqueued. + * + * Packet Sent counter is decremented on successful + * transmission. + *===========================================================*/ + + +static void wanpipe_delayed_transmit (unsigned long data) +{ + struct sock *sk=(struct sock *)data; + struct sk_buff *skb; + wanpipe_opt *wp = wp_sk(sk); + struct net_device *dev = wp->dev; + sdla_t *card = (sdla_t*)wp->card; + + if (!card || !dev){ + clear_bit(0, &wp->timer); + DBG_PRINTK(KERN_INFO "wansock: Transmit delay, no dev or card\n"); + return; + } + + if (sk->sk_state != WANSOCK_CONNECTED || !sock_flag(sk, SOCK_ZAPPED)) { + clear_bit(0, &wp->timer); + DBG_PRINTK(KERN_INFO "wansock: Tx Timer, State not CONNECTED\n"); + return; + } + + /* If driver is executing command, we must offload + * the board by not sending data. Otherwise a + * pending command will never get a free buffer + * to execute */ + if (atomic_read(&card->u.x.command_busy)){ + wp->tx_timer.expires = jiffies + SLOW_BACKOFF; + add_timer(&wp->tx_timer); + DBG_PRINTK(KERN_INFO "wansock: Tx Timer, command bys BACKOFF\n"); + return; + } + + + if (test_and_set_bit(0,&wanpipe_tx_critical)){ + printk(KERN_INFO "WanSock: Tx timer critical %s\n",dev->name); + wp->tx_timer.expires = jiffies + SLOW_BACKOFF; + add_timer(&wp->tx_timer); + return; + } + + /* Check for a packet in the fifo and send */ + if ((skb = skb_dequeue(&sk->sk_write_queue)) != NULL){ + + if (dev->hard_start_xmit(skb, dev) != 0){ + + /* Driver failed to transmit, re-enqueue + * the packet and retry again later */ + skb_queue_head(&sk->sk_write_queue,skb); + clear_bit(0,&wanpipe_tx_critical); + return; + }else{ + + /* Packet Sent successful. Check for more packets + * if more packets, re-trigger the transmit routine + * other wise exit + */ + atomic_dec(&wp->packet_sent); + + if (skb_peek(&sk->sk_write_queue) == NULL) { + /* If there is nothing to send, kick + * the poll routine, which will trigger + * the application to send more data */ + sk->sk_data_ready(sk, 0); + clear_bit(0, &wp->timer); + }else{ + /* Reschedule as fast as possible */ + wp->tx_timer.expires = jiffies + 1; + add_timer(&wp->tx_timer); + } + } + } + clear_bit(0,&wanpipe_tx_critical); +} + +/*============================================================ + * execute_command + * + * Execute x25api commands. The atomic variable + * chan->command is used to indicate to the driver that + * command is pending for execution. The acutal command + * structure is placed into a sock mbox structure + * (wp_sk(sk)->mbox). + * + * The sock private structure, mbox is + * used as shared memory between sock and the driver. + * Driver uses the sock mbox to execute the command + * and return the result. + * + * For all command except PLACE CALL, the function + * waits for the result. PLACE CALL can be ether + * blocking or nonblocking. The user sets this option + * via ioctl call. + *===========================================================*/ + + +static int execute_command(struct sock *sk, unsigned char cmd, unsigned int flags) +{ + wanpipe_opt *wp = wp_sk(sk); + struct net_device *dev; + wanpipe_common_t *chan=NULL; + int err=0; + DECLARE_WAITQUEUE(wait, current); + + dev = dev_get_by_index(sk->sk_bound_dev_if); + if (dev == NULL){ + printk(KERN_INFO "wansock: Exec failed no dev %i\n", + sk->sk_bound_dev_if); + return -ENODEV; + } + dev_put(dev); + + if ((chan=dev->priv) == NULL){ + printk(KERN_INFO "wansock: Exec cmd failed no priv area\n"); + return -ENODEV; + } + + if (atomic_read(&chan->command)){ + printk(KERN_INFO "wansock: ERROR: Command already running %x, %s\n", + atomic_read(&chan->command),dev->name); + return -EINVAL; + } + + if (!wp->mbox) { + printk(KERN_INFO "wansock: In execute without MBOX\n"); + return -EINVAL; + } + + ((mbox_cmd_t*)wp->mbox)->cmd.command = cmd; + ((mbox_cmd_t*)wp->mbox)->cmd.lcn = wp->lcn; + ((mbox_cmd_t*)wp->mbox)->cmd.result = 0x7F; + + + if (flags & O_NONBLOCK){ + cmd |= 0x80; + atomic_set(&chan->command, cmd); + }else{ + atomic_set(&chan->command, cmd); + } + + add_wait_queue(sk->sk_sleep,&wait); + current->state = TASK_INTERRUPTIBLE; + for (;;){ + if (((mbox_cmd_t*)wp->mbox)->cmd.result != 0x7F) { + err = 0; + break; + } + if (signal_pending(current)) { + err = -ERESTARTSYS; + break; + } + schedule(); + } + current->state = TASK_RUNNING; + remove_wait_queue(sk->sk_sleep,&wait); + + return err; +} + +/*============================================================ + * wanpipe_destroy_timer + * + * Used by wanpipe_release, to delay release of + * the socket. + *===========================================================*/ + +static void wanpipe_destroy_timer(unsigned long data) +{ + struct sock *sk=(struct sock *)data; + wanpipe_opt *wp = wp_sk(sk); + + if ((!atomic_read(&sk->sk_wmem_alloc) && + !atomic_read(&sk->sk_rmem_alloc)) || + (++wp->force == 5)) { + + if (atomic_read(&sk->sk_wmem_alloc) || + atomic_read(&sk->sk_rmem_alloc)) + printk(KERN_INFO "wansock: Warning, Packet Discarded due to sock shutdown!\n"); + + kfree(wp); + wp_sk(sk) = NULL; + + if (atomic_read(&sk->sk_refcnt) != 1) { + atomic_set(&sk->sk_refcnt, 1); + DBG_PRINTK(KERN_INFO "wansock: Error, wrong reference count: %i ! :delay.\n", + atomic_read(&sk->sk_refcnt)); + } + sock_put(sk); + atomic_dec(&wanpipe_socks_nr); + return; + } + + sk->sk_timer.expires = jiffies + 5 * HZ; + add_timer(&sk->sk_timer); + printk(KERN_INFO "wansock: packet sk destroy delayed\n"); +} + +/*============================================================ + * wanpipe_unlink_driver + * + * When the socket is released, this function is + * used to remove links that bind the sock and the + * driver together. + *===========================================================*/ +static void wanpipe_unlink_driver (struct sock *sk) +{ + struct net_device *dev; + wanpipe_common_t *chan=NULL; + + sock_reset_flag(sk, SOCK_ZAPPED); + sk->sk_state = WANSOCK_DISCONNECTED; + wp_sk(sk)->dev = NULL; + + dev = dev_get_by_index(sk->sk_bound_dev_if); + if (!dev){ + printk(KERN_INFO "wansock: No dev on release\n"); + return; + } + dev_put(dev); + + if ((chan = dev->priv) == NULL){ + printk(KERN_INFO "wansock: No Priv Area on release\n"); + return; + } + + set_bit(0,&chan->common_critical); + chan->sk=NULL; + chan->func=NULL; + chan->mbox=NULL; + chan->tx_timer=NULL; + clear_bit(0,&chan->common_critical); + release_device(dev); + + return; +} + +/*============================================================ + * wanpipe_link_driver + * + * Upon successful bind(), sock is linked to a driver + * by binding in the wanpipe_rcv() bottom half handler + * to the driver function pointer, as well as sock and + * sock mailbox addresses. This way driver can pass + * data up the socket. + *===========================================================*/ + +static void wanpipe_link_driver(struct net_device *dev, struct sock *sk) +{ + wanpipe_opt *wp = wp_sk(sk); + wanpipe_common_t *chan = dev->priv; + if (!chan) + return; + set_bit(0,&chan->common_critical); + chan->sk=sk; + chan->func=wanpipe_rcv; + chan->mbox = wp->mbox; + chan->tx_timer = &wp->tx_timer; + wp->dev = dev; + sock_set_flag(sk, SOCK_ZAPPED); + clear_bit(0,&chan->common_critical); +} + + +/*============================================================ + * release_device + * + * During sock release, clear a critical bit, which + * marks the device a being taken. + *===========================================================*/ + + +static void release_device(struct net_device *dev) +{ + wanpipe_common_t *chan=dev->priv; + clear_bit(0,(void*)&chan->rw_bind); +} + +/*============================================================ + * wanpipe_release + * + * Close a PACKET socket. This is fairly simple. We + * immediately go to 'closed' state and remove our + * protocol entry in the device list. + *===========================================================*/ + +static int wanpipe_release(struct socket *sock) +{ + wanpipe_opt *wp; + struct sock *sk = sock->sk; + + if (!sk) + return 0; + + wp = wp_sk(sk); + check_write_queue(sk); + + /* Kill the tx timer, if we don't kill it now, the timer + * will run after we kill the sock. Timer code will + * try to access the sock which has been killed and cause + * kernel panic */ + + del_timer(&wp->tx_timer); + + /* + * Unhook packet receive handler. + */ + + if (wp->num == htons(X25_PROT) && + sk->sk_state != WANSOCK_DISCONNECTED && sock_flag(sk, SOCK_ZAPPED)) { + struct net_device *dev = dev_get_by_index(sk->sk_bound_dev_if); + wanpipe_common_t *chan; + if (dev){ + chan=dev->priv; + atomic_set(&chan->disconnect,1); + DBG_PRINTK(KERN_INFO "wansock: Sending Clear Indication %i\n", + sk->sk_state); + dev_put(dev); + } + } + + set_bit(1,&wanpipe_tx_critical); + write_lock(&wanpipe_sklist_lock); + sk_del_node_init(sk); + write_unlock(&wanpipe_sklist_lock); + clear_bit(1,&wanpipe_tx_critical); + + + + release_driver(sk); + + + /* + * Now the socket is dead. No more input will appear. + */ + + sk->sk_state_change(sk); /* It is useless. Just for sanity. */ + + sock->sk = NULL; + sk->sk_socket = NULL; + sock_set_flag(sk, SOCK_DEAD); + + /* Purge queues */ + skb_queue_purge(&sk->sk_receive_queue); + skb_queue_purge(&sk->sk_write_queue); + skb_queue_purge(&sk->sk_error_queue); + + if (atomic_read(&sk->sk_rmem_alloc) || + atomic_read(&sk->sk_wmem_alloc)) { + del_timer(&sk->sk_timer); + printk(KERN_INFO "wansock: Killing in Timer R %i , W %i\n", + atomic_read(&sk->sk_rmem_alloc), + atomic_read(&sk->sk_wmem_alloc)); + sk->sk_timer.data = (unsigned long)sk; + sk->sk_timer.expires = jiffies + HZ; + sk->sk_timer.function = wanpipe_destroy_timer; + add_timer(&sk->sk_timer); + return 0; + } + + kfree(wp); + wp_sk(sk) = NULL; + + if (atomic_read(&sk->sk_refcnt) != 1) { + DBG_PRINTK(KERN_INFO "wansock: Error, wrong reference count: %i !:release.\n", + atomic_read(&sk->sk_refcnt)); + atomic_set(&sk->sk_refcnt, 1); + } + sock_put(sk); + atomic_dec(&wanpipe_socks_nr); + return 0; +} + +/*============================================================ + * check_write_queue + * + * During sock shutdown, if the sock state is + * WANSOCK_CONNECTED and there is transmit data + * pending. Wait until data is released + * before proceeding. + *===========================================================*/ + +static void check_write_queue(struct sock *sk) +{ + + if (sk->sk_state != WANSOCK_CONNECTED) + return; + + if (!atomic_read(&sk->sk_wmem_alloc)) + return; + + printk(KERN_INFO "wansock: MAJOR ERROR, Data lost on sock release !!!\n"); + +} + +/*============================================================ + * release_driver + * + * This function is called during sock shutdown, to + * release any resources and links that bind the sock + * to the driver. It also changes the state of the + * sock to WANSOCK_DISCONNECTED + *===========================================================*/ + +static void release_driver(struct sock *sk) +{ + wanpipe_opt *wp; + struct sk_buff *skb=NULL; + struct sock *deadsk=NULL; + + if (sk->sk_state == WANSOCK_LISTEN || + sk->sk_state == WANSOCK_BIND_LISTEN) { + while ((skb = skb_dequeue(&sk->sk_receive_queue)) != NULL) { + if ((deadsk = get_newsk_from_skb(skb))){ + DBG_PRINTK (KERN_INFO "wansock: RELEASE: FOUND DEAD SOCK\n"); + sock_set_flag(deadsk, SOCK_DEAD); + start_cleanup_timer(deadsk); + } + kfree_skb(skb); + } + if (sock_flag(sk, SOCK_ZAPPED)) + wanpipe_unlink_card(sk); + }else{ + if (sock_flag(sk, SOCK_ZAPPED)) + wanpipe_unlink_driver(sk); + } + sk->sk_state = WANSOCK_DISCONNECTED; + sk->sk_bound_dev_if = 0; + sock_reset_flag(sk, SOCK_ZAPPED); + wp = wp_sk(sk); + + if (wp && wp->mbox) { + kfree(wp->mbox); + wp->mbox = NULL; + } +} + +/*============================================================ + * start_cleanup_timer + * + * If new incoming call's are pending but the socket + * is being released, start the timer which will + * envoke the kill routines for pending socks. + *===========================================================*/ + + +static void start_cleanup_timer (struct sock *sk) +{ + del_timer(&sk->sk_timer); + sk->sk_timer.data = (unsigned long)sk; + sk->sk_timer.expires = jiffies + HZ; + sk->sk_timer.function = wanpipe_kill_sock_timer; + add_timer(&sk->sk_timer); +} + + +/*============================================================ + * wanpipe_kill_sock + * + * This is a function which performs actual killing + * of the sock. It releases socket resources, + * and unlinks the sock from the driver. + *===========================================================*/ + +static void wanpipe_kill_sock_timer (unsigned long data) +{ + + struct sock *sk = (struct sock *)data; + struct sock **skp; + + if (!sk) + return; + + /* This function can be called from interrupt. We must use + * appropriate locks */ + + if (test_bit(1,&wanpipe_tx_critical)){ + sk->sk_timer.expires = jiffies + 10; + add_timer(&sk->sk_timer); + return; + } + + write_lock(&wanpipe_sklist_lock); + sk_del_node_init(sk); + write_unlock(&wanpipe_sklist_lock); + + + if (wp_sk(sk)->num == htons(X25_PROT) && + sk->sk_state != WANSOCK_DISCONNECTED) { + struct net_device *dev = dev_get_by_index(sk->sk_bound_dev_if); + wanpipe_common_t *chan; + if (dev){ + chan=dev->priv; + atomic_set(&chan->disconnect,1); + dev_put(dev); + } + } + + release_driver(sk); + + sk->sk_socket = NULL; + + /* Purge queues */ + skb_queue_purge(&sk->sk_receive_queue); + skb_queue_purge(&sk->sk_write_queue); + skb_queue_purge(&sk->sk_error_queue); + + if (atomic_read(&sk->sk_rmem_alloc) || + atomic_read(&sk->sk_wmem_alloc)) { + del_timer(&sk->sk_timer); + printk(KERN_INFO "wansock: Killing SOCK in Timer\n"); + sk->sk_timer.data = (unsigned long)sk; + sk->sk_timer.expires = jiffies + HZ; + sk->sk_timer.function = wanpipe_destroy_timer; + add_timer(&sk->sk_timer); + return; + } + + if (wp_sk(sk)) { + kfree(wp_sk(sk)); + wp_sk(sk) = NULL; + } + + if (atomic_read(&sk->sk_refcnt) != 1) { + atomic_set(&sk->sk_refcnt, 1); + DBG_PRINTK(KERN_INFO "wansock: Error, wrong reference count: %i ! :timer.\n", + atomic_read(&sk->sk_refcnt)); + } + sock_put(sk); + atomic_dec(&wanpipe_socks_nr); + return; +} + +static void wanpipe_kill_sock_accept (struct sock *sk) +{ + + struct sock **skp; + + if (!sk) + return; + + /* This function can be called from interrupt. We must use + * appropriate locks */ + + write_lock(&wanpipe_sklist_lock); + sk_del_node_init(sk); + write_unlock(&wanpipe_sklist_lock); + + sk->sk_socket = NULL; + + + if (wp_sk(sk)) { + kfree(wp_sk(sk)); + wp_sk(sk) = NULL; + } + + if (atomic_read(&sk->sk_refcnt) != 1) { + atomic_set(&sk->sk_refcnt, 1); + DBG_PRINTK(KERN_INFO "wansock: Error, wrong reference count: %i ! :timer.\n", + atomic_read(&sk->sk_refcnt)); + } + sock_put(sk); + atomic_dec(&wanpipe_socks_nr); + return; +} + + +static void wanpipe_kill_sock_irq (struct sock *sk) +{ + + if (!sk) + return; + + sk->sk_socket = NULL; + + if (wp_sk(sk)) { + kfree(wp_sk(sk)); + wp_sk(sk) = NULL; + } + + if (atomic_read(&sk->sk_refcnt) != 1) { + atomic_set(&sk->sk_refcnt, 1); + DBG_PRINTK(KERN_INFO "wansock: Error, wrong reference count: %i !:listen.\n", + atomic_read(&sk->sk_refcnt)); + } + sock_put(sk); + atomic_dec(&wanpipe_socks_nr); +} + + +/*============================================================ + * wanpipe_do_bind + * + * Bottom half of the binding system call. + * Once the wanpipe_bind() function checks the + * legality of the call, this function binds the + * sock to the driver. + *===========================================================*/ + +static int wanpipe_do_bind(struct sock *sk, struct net_device *dev, + int protocol) +{ + wanpipe_opt *wp = wp_sk(sk); + wanpipe_common_t *chan=NULL; + int err=0; + + if (sock_flag(sk, SOCK_ZAPPED)) { + err = -EALREADY; + goto bind_unlock_exit; + } + + wp->num = protocol; + + if (protocol == 0){ + release_device(dev); + err = -EINVAL; + goto bind_unlock_exit; + } + + if (dev) { + if (dev->flags&IFF_UP) { + chan=dev->priv; + sk->sk_state = chan->state; + + if (wp->num == htons(X25_PROT) && + sk->sk_state != WANSOCK_DISCONNECTED && + sk->sk_state != WANSOCK_CONNECTING) { + DBG_PRINTK(KERN_INFO + "wansock: Binding to Device not DISCONNECTED %i\n", + sk->sk_state); + release_device(dev); + err = -EAGAIN; + goto bind_unlock_exit; + } + + wanpipe_link_driver(dev,sk); + sk->sk_bound_dev_if = dev->ifindex; + + /* X25 Specific option */ + if (wp->num == htons(X25_PROT)) + wp_sk(sk)->svc = chan->svc; + + } else { + sk->sk_err = ENETDOWN; + sk->sk_error_report(sk); + release_device(dev); + err = -EINVAL; + } + } else { + err = -ENODEV; + } +bind_unlock_exit: + /* FIXME where is this lock */ + + return err; +} + +/*============================================================ + * wanpipe_bind + * + * BIND() System call, which is bound to the AF_WANPIPE + * operations structure. It checks for correct wanpipe + * card name, and cross references interface names with + * the card names. Thus, interface name must belong to + * the actual card. + *===========================================================*/ + + +static int wanpipe_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len) +{ + struct wan_sockaddr_ll *sll = (struct wan_sockaddr_ll*)uaddr; + struct sock *sk=sock->sk; + wanpipe_opt *wp = wp_sk(sk); + struct net_device *dev = NULL; + sdla_t *card=NULL; + char name[15]; + + /* + * Check legality + */ + + if (addr_len < sizeof(struct wan_sockaddr_ll)){ + printk(KERN_INFO "wansock: Address length error\n"); + return -EINVAL; + } + if (sll->sll_family != AF_WANPIPE){ + printk(KERN_INFO "wansock: Illegal family name specified.\n"); + return -EINVAL; + } + + card = wanpipe_find_card (sll->sll_card); + if (!card){ + printk(KERN_INFO "wansock: Wanpipe card not found: %s\n",sll->sll_card); + return -ENODEV; + }else{ + wp_sk(sk)->card = (void *)card; + } + + if (!strcmp(sll->sll_device,"svc_listen")){ + + /* Bind a sock to a card structure for listening + */ + int err=0; + + /* This is x25 specific area if protocol doesn't + * match, return error */ + if (sll->sll_protocol != htons(X25_PROT)) + return -EINVAL; + + err= wanpipe_link_card (sk); + if (err < 0) + return err; + + if (sll->sll_protocol) + wp->num = sll->sll_protocol; + sk->sk_state = WANSOCK_BIND_LISTEN; + return 0; + + }else if (!strcmp(sll->sll_device,"svc_connect")){ + + /* This is x25 specific area if protocol doesn't + * match, return error */ + if (sll->sll_protocol != htons(X25_PROT)) + return -EINVAL; + + /* Find a free device + */ + dev = wanpipe_find_free_dev(card); + if (dev == NULL){ + DBG_PRINTK(KERN_INFO "wansock: No free network devices for card %s\n", + card->devname); + return -EINVAL; + } + }else{ + /* Bind a socket to a interface name + * This is used by PVC mostly + */ + strlcpy(name,sll->sll_device,sizeof(name)); + dev = dev_get_by_name(name); + if (dev == NULL){ + printk(KERN_INFO "wansock: Failed to get Dev from name: %s,\n", + name); + return -ENODEV; + } + + dev_put(dev); + + if (check_dev(dev, card)){ + printk(KERN_INFO "wansock: Device %s, doesn't belong to card %s\n", + dev->name, card->devname); + return -EINVAL; + } + if (get_atomic_device (dev)) + return -EINVAL; + } + + return wanpipe_do_bind(sk, dev, sll->sll_protocol ? : wp->num); +} + +/*============================================================ + * get_atomic_device + * + * Sets a bit atomically which indicates that + * the interface is taken. This avoids race conditions. + *===========================================================*/ + + +static inline int get_atomic_device(struct net_device *dev) +{ + wanpipe_common_t *chan = dev->priv; + if (!test_and_set_bit(0,(void *)&chan->rw_bind)){ + return 0; + } + return 1; +} + +/*============================================================ + * check_dev + * + * Check that device name belongs to a particular card. + *===========================================================*/ + +static int check_dev(struct net_device *dev, sdla_t *card) +{ + struct net_device* tmp_dev; + + for (tmp_dev = card->wandev.dev; tmp_dev; + tmp_dev = *((struct net_device **)tmp_dev->priv)) { + if (tmp_dev->ifindex == dev->ifindex){ + return 0; + } + } + return 1; +} + +/*============================================================ + * wanpipe_find_free_dev + * + * Find a free network interface. If found set atomic + * bit indicating that the interface is taken. + * X25API Specific. + *===========================================================*/ + +struct net_device *wanpipe_find_free_dev(sdla_t *card) +{ + struct net_device* dev; + volatile wanpipe_common_t *chan; + + if (test_and_set_bit(0,&find_free_critical)){ + printk(KERN_INFO "CRITICAL in Find Free\n"); + } + + for (dev = card->wandev.dev; dev; + dev = *((struct net_device **)dev->priv)) { + chan = dev->priv; + if (!chan) + continue; + if (chan->usedby == API && chan->svc){ + if (!get_atomic_device (dev)){ + if (chan->state != WANSOCK_DISCONNECTED){ + release_device(dev); + }else{ + clear_bit(0,&find_free_critical); + return dev; + } + } + } + } + clear_bit(0,&find_free_critical); + return NULL; +} + +/*============================================================ + * wanpipe_create + * + * SOCKET() System call. It allocates a sock structure + * and adds the socket to the wanpipe_sk_list. + * Crates AF_WANPIPE socket. + *===========================================================*/ + +static int wanpipe_create(struct socket *sock, int protocol) +{ + struct sock *sk; + + //FIXME: This checks for root user, SECURITY ? + //if (!capable(CAP_NET_RAW)) + // return -EPERM; + + if (sock->type != SOCK_DGRAM && sock->type != SOCK_RAW) + return -ESOCKTNOSUPPORT; + + sock->state = SS_UNCONNECTED; + + if ((sk = wanpipe_alloc_socket()) == NULL) + return -ENOBUFS; + + sk->sk_reuse = 1; + sock->ops = &wanpipe_ops; + sock_init_data(sock,sk); + + sock_reset_flag(sk, SOCK_ZAPPED); + sk->sk_family = PF_WANPIPE; + wp_sk(sk)->num = protocol; + sk->sk_state = WANSOCK_DISCONNECTED; + sk->sk_ack_backlog = 0; + sk->sk_bound_dev_if = 0; + + atomic_inc(&wanpipe_socks_nr); + + /* We must disable interrupts because the ISR + * can also change the list */ + set_bit(1,&wanpipe_tx_critical); + write_lock(&wanpipe_sklist_lock); + sk_add_node(sk, &wanpipe_sklist); + write_unlock(&wanpipe_sklist_lock); + clear_bit(1,&wanpipe_tx_critical); + + return(0); +} + + +/*============================================================ + * wanpipe_recvmsg + * + * Pull a packet from our receive queue and hand it + * to the user. If necessary we block. + *===========================================================*/ + +static int wanpipe_recvmsg(struct kiocb *iocb, struct socket *sock, + struct msghdr *msg, int len, int flags) +{ + struct sock *sk = sock->sk; + struct sk_buff *skb; + int copied, err=-ENOBUFS; + + + /* + * If the address length field is there to be filled in, we fill + * it in now. + */ + + msg->msg_namelen = sizeof(struct wan_sockaddr_ll); + + /* + * Call the generic datagram receiver. This handles all sorts + * of horrible races and re-entrancy so we can forget about it + * in the protocol layers. + * + * Now it will return ENETDOWN, if device have just gone down, + * but then it will block. + */ + + if (flags & MSG_OOB){ + skb = skb_dequeue(&sk->sk_error_queue); + }else{ + skb=skb_recv_datagram(sk,flags,1,&err); + } + /* + * An error occurred so return it. Because skb_recv_datagram() + * handles the blocking we don't see and worry about blocking + * retries. + */ + + if(skb==NULL) + goto out; + + /* + * You lose any data beyond the buffer you gave. If it worries a + * user program they can ask the device for its MTU anyway. + */ + + copied = skb->len; + if (copied > len) + { + copied=len; + msg->msg_flags|=MSG_TRUNC; + } + + wanpipe_wakeup_driver(sk); + + /* We can't use skb_copy_datagram here */ + err = memcpy_toiovec(msg->msg_iov, skb->data, copied); + if (err) + goto out_free; + + sock_recv_timestamp(msg, sk, skb); + + if (msg->msg_name) + memcpy(msg->msg_name, skb->cb, msg->msg_namelen); + + /* + * Free or return the buffer as appropriate. Again this + * hides all the races and re-entrancy issues from us. + */ + err = (flags&MSG_TRUNC) ? skb->len : copied; + +out_free: + skb_free_datagram(sk, skb); +out: + return err; +} + + +/*============================================================ + * wanpipe_wakeup_driver + * + * If socket receive buffer is full and driver cannot + * pass data up the sock, it sets a packet_block flag. + * This function check that flag and if sock receive + * queue has room it kicks the driver BH handler. + * + * This way, driver doesn't have to poll the sock + * receive queue. + *===========================================================*/ + +static void wanpipe_wakeup_driver(struct sock *sk) +{ + struct net_device *dev = NULL; + wanpipe_common_t *chan=NULL; + + dev = dev_get_by_index(sk->sk_bound_dev_if); + if (!dev) + return; + + dev_put(dev); + + if ((chan = dev->priv) == NULL) + return; + + if (atomic_read(&chan->receive_block)){ + if (atomic_read(&sk->sk_rmem_alloc) < + ((unsigned)sk->sk_rcvbuf * 0.9)) { + printk(KERN_INFO "wansock: Queuing task for wanpipe\n"); + atomic_set(&chan->receive_block,0); + wanpipe_queue_tq(&chan->wanpipe_task); + wanpipe_mark_bh(); + } + } +} + +/*============================================================ + * wanpipe_getname + * + * I don't know what to do with this yet. + * User can use this function to get sock address + * information. Not very useful for Sangoma's purposes. + *===========================================================*/ + + +static int wanpipe_getname(struct socket *sock, struct sockaddr *uaddr, + int *uaddr_len, int peer) +{ + struct net_device *dev; + struct sock *sk = sock->sk; + struct wan_sockaddr_ll *sll = (struct wan_sockaddr_ll*)uaddr; + + sll->sll_family = AF_WANPIPE; + sll->sll_ifindex = sk->sk_bound_dev_if; + sll->sll_protocol = wp_sk(sk)->num; + dev = dev_get_by_index(sk->sk_bound_dev_if); + if (dev) { + sll->sll_hatype = dev->type; + sll->sll_halen = dev->addr_len; + memcpy(sll->sll_addr, dev->dev_addr, dev->addr_len); + } else { + sll->sll_hatype = 0; /* Bad: we have no ARPHRD_UNSPEC */ + sll->sll_halen = 0; + } + *uaddr_len = sizeof(*sll); + + dev_put(dev); + + return 0; +} + +/*============================================================ + * wanpipe_notifier + * + * If driver turns off network interface, this function + * will be envoked. Currently I treate it as a + * call disconnect. More thought should go into this + * function. + * + * FIXME: More thought should go into this function. + * + *===========================================================*/ + +static int wanpipe_notifier(struct notifier_block *this, unsigned long msg, void *data) +{ + struct sock *sk; + hlist_node *node; + struct net_device *dev = (struct net_device *)data; + + sk_for_each(sk, node, &wanpipe_sklist) { + struct wanpipe_opt *po = wp_sk(sk); + + if (!po) + continue; + if (dev == NULL) + continue; + + switch (msg) { + case NETDEV_DOWN: + case NETDEV_UNREGISTER: + if (dev->ifindex == sk->sk_bound_dev_if) { + printk(KERN_INFO "wansock: Device down %s\n",dev->name); + if (sock_flag(sk, SOCK_ZAPPED)) { + wanpipe_unlink_driver(sk); + sk->sk_err = ENETDOWN; + sk->sk_error_report(sk); + } + + if (msg == NETDEV_UNREGISTER) { + printk(KERN_INFO "wansock: Unregistering Device: %s\n", + dev->name); + wanpipe_unlink_driver(sk); + sk->sk_bound_dev_if = 0; + } + } + break; + case NETDEV_UP: + if (dev->ifindex == sk->sk_bound_dev_if && + po->num && !sock_flag(sk, SOCK_ZAPPED)) { + printk(KERN_INFO "wansock: Registering Device: %s\n", + dev->name); + wanpipe_link_driver(dev,sk); + } + break; + } + } + return NOTIFY_DONE; +} + +/*============================================================ + * wanpipe_ioctl + * + * Execute a user commands, and set socket options. + * + * FIXME: More thought should go into this function. + * + *===========================================================*/ + +static int wanpipe_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg) +{ + struct sock *sk = sock->sk; + int err; + + switch(cmd) + { + case SIOCGSTAMP: + return sock_get_timestamp(sk, (struct timeval __user *)arg); + + case SIOC_WANPIPE_CHECK_TX: + + return atomic_read(&sk->sk_wmem_alloc); + + case SIOC_WANPIPE_SOCK_STATE: + + if (sk->sk_state == WANSOCK_CONNECTED) + return 0; + + return 1; + + + case SIOC_WANPIPE_GET_CALL_DATA: + + return get_ioctl_cmd (sk,(void*)arg); + + case SIOC_WANPIPE_SET_CALL_DATA: + + return set_ioctl_cmd (sk,(void*)arg); + + case SIOC_WANPIPE_ACCEPT_CALL: + case SIOC_WANPIPE_CLEAR_CALL: + case SIOC_WANPIPE_RESET_CALL: + + if ((err=set_ioctl_cmd(sk,(void*)arg)) < 0) + return err; + + err=wanpipe_exec_cmd(sk,cmd,0); + get_ioctl_cmd(sk,(void*)arg); + return err; + + case SIOC_WANPIPE_DEBUG: + + return wanpipe_debug(sk,(void*)arg); + + case SIOC_WANPIPE_SET_NONBLOCK: + + if (sk->sk_state != WANSOCK_DISCONNECTED) + return -EINVAL; + + sock->file->f_flags |= O_NONBLOCK; + return 0; + +#ifdef CONFIG_INET + case SIOCADDRT: + case SIOCDELRT: + case SIOCDARP: + case SIOCGARP: + case SIOCSARP: + case SIOCDRARP: + case SIOCGRARP: + case SIOCSRARP: + case SIOCGIFADDR: + case SIOCSIFADDR: + case SIOCGIFBRDADDR: + case SIOCSIFBRDADDR: + case SIOCGIFNETMASK: + case SIOCSIFNETMASK: + case SIOCGIFDSTADDR: + case SIOCSIFDSTADDR: + case SIOCSIFFLAGS: + return inet_dgram_ops.ioctl(sock, cmd, arg); +#endif + + default: + return dev_ioctl(cmd,(void __user *) arg); + } + /*NOTREACHED*/ +} + +/*============================================================ + * wanpipe_debug + * + * This function will pass up information about all + * active sockets. + * + * FIXME: More thought should go into this function. + * + *===========================================================*/ + +static int wanpipe_debug (struct sock *origsk, void *arg) +{ + struct sock *sk; + struct hlist_node *node; + struct net_device *dev = NULL; + wanpipe_common_t *chan=NULL; + int cnt=0, err=0; + wan_debug_t *dbg_data = (wan_debug_t *)arg; + + sk_for_each(sk, node, &wanpipe_sklist) { + wanpipe_opt *wp = wp_sk(sk); + + if (sk == origsk){ + continue; + } + + if ((err=put_user(1, &dbg_data->debug[cnt].free))) + return err; + if ((err = put_user(sk->sk_state, + &dbg_data->debug[cnt].state_sk))) + return err; + if ((err = put_user(sk->sk_rcvbuf, + &dbg_data->debug[cnt].rcvbuf))) + return err; + if ((err = put_user(atomic_read(&sk->sk_rmem_alloc), + &dbg_data->debug[cnt].rmem))) + return err; + if ((err = put_user(atomic_read(&sk->sk_wmem_alloc), + &dbg_data->debug[cnt].wmem))) + return err; + if ((err = put_user(sk->sk_sndbuf, + &dbg_data->debug[cnt].sndbuf))) + return err; + if ((err=put_user(sk_count, &dbg_data->debug[cnt].sk_count))) + return err; + if ((err=put_user(wp->poll_cnt, &dbg_data->debug[cnt].poll_cnt))) + return err; + if ((err = put_user(sk->sk_bound_dev_if, + &dbg_data->debug[cnt].bound))) + return err; + + if (sk->sk_bound_dev_if) { + dev = dev_get_by_index(sk->sk_bound_dev_if); + if (!dev) + continue; + + chan=dev->priv; + dev_put(dev); + + if ((err=put_user(chan->state, &dbg_data->debug[cnt].d_state))) + return err; + if ((err=put_user(chan->svc, &dbg_data->debug[cnt].svc))) + return err; + + if ((err=put_user(atomic_read(&chan->command), + &dbg_data->debug[cnt].command))) + return err; + + + if (wp){ + sdla_t *card = (sdla_t*)wp->card; + + if (card){ + if ((err=put_user(atomic_read(&card->u.x.command_busy), + &dbg_data->debug[cnt].cmd_busy))) + return err; + } + + if ((err=put_user(wp->lcn, + &dbg_data->debug[cnt].lcn))) + return err; + + if (wp->mbox) { + if ((err=put_user(1, &dbg_data->debug[cnt].mbox))) + return err; + } + } + + if ((err=put_user(atomic_read(&chan->receive_block), + &dbg_data->debug[cnt].rblock))) + return err; + + if (copy_to_user(dbg_data->debug[cnt].name, dev->name, strlen(dev->name))) + return -EFAULT; + } + + if (++cnt == MAX_NUM_DEBUG) + break; + } + return 0; +} + +/*============================================================ + * get_ioctl_cmd + * + * Pass up the contents of socket MBOX to the user. + *===========================================================*/ + +static int get_ioctl_cmd (struct sock *sk, void *arg) +{ + x25api_t *usr_data = (x25api_t *)arg; + mbox_cmd_t *mbox_ptr; + int err; + + if (usr_data == NULL) + return -EINVAL; + + if (!wp_sk(sk)->mbox) { + return -EINVAL; + } + + mbox_ptr = (mbox_cmd_t *)wp_sk(sk)->mbox; + + if ((err=put_user(mbox_ptr->cmd.qdm, &usr_data->hdr.qdm))) + return err; + if ((err=put_user(mbox_ptr->cmd.cause, &usr_data->hdr.cause))) + return err; + if ((err=put_user(mbox_ptr->cmd.diagn, &usr_data->hdr.diagn))) + return err; + if ((err=put_user(mbox_ptr->cmd.length, &usr_data->hdr.length))) + return err; + if ((err=put_user(mbox_ptr->cmd.result, &usr_data->hdr.result))) + return err; + if ((err=put_user(mbox_ptr->cmd.lcn, &usr_data->hdr.lcn))) + return err; + + if (mbox_ptr->cmd.length > 0){ + if (mbox_ptr->cmd.length > X25_MAX_DATA) + return -EINVAL; + + if (copy_to_user(usr_data->data, mbox_ptr->data, mbox_ptr->cmd.length)){ + printk(KERN_INFO "wansock: Copy failed !!!\n"); + return -EFAULT; + } + } + return 0; +} + +/*============================================================ + * set_ioctl_cmd + * + * Before command can be execute, socket MBOX must + * be created, and initialized with user data. + *===========================================================*/ + +static int set_ioctl_cmd (struct sock *sk, void *arg) +{ + x25api_t *usr_data = (x25api_t *)arg; + mbox_cmd_t *mbox_ptr; + int err; + + if (!wp_sk(sk)->mbox) { + void *mbox_ptr; + struct net_device *dev = dev_get_by_index(sk->sk_bound_dev_if); + if (!dev) + return -ENODEV; + + dev_put(dev); + + if ((mbox_ptr = kmalloc(sizeof(mbox_cmd_t), GFP_ATOMIC)) == NULL) + return -ENOMEM; + + memset(mbox_ptr, 0, sizeof(mbox_cmd_t)); + wp_sk(sk)->mbox = mbox_ptr; + + wanpipe_link_driver(dev,sk); + } + + mbox_ptr = (mbox_cmd_t*)wp_sk(sk)->mbox; + memset(mbox_ptr, 0, sizeof(mbox_cmd_t)); + + if (usr_data == NULL){ + return 0; + } + if ((err=get_user(mbox_ptr->cmd.qdm, &usr_data->hdr.qdm))) + return err; + if ((err=get_user(mbox_ptr->cmd.cause, &usr_data->hdr.cause))) + return err; + if ((err=get_user(mbox_ptr->cmd.diagn, &usr_data->hdr.diagn))) + return err; + if ((err=get_user(mbox_ptr->cmd.length, &usr_data->hdr.length))) + return err; + if ((err=get_user(mbox_ptr->cmd.result, &usr_data->hdr.result))) + return err; + + if (mbox_ptr->cmd.length > 0){ + if (mbox_ptr->cmd.length > X25_MAX_DATA) + return -EINVAL; + + if (copy_from_user(mbox_ptr->data, usr_data->data, mbox_ptr->cmd.length)){ + printk(KERN_INFO "Copy failed\n"); + return -EFAULT; + } + } + return 0; +} + + +/*====================================================================== + * wanpipe_poll + * + * Datagram poll: Again totally generic. This also handles + * sequenced packet sockets providing the socket receive queue + * is only ever holding data ready to receive. + * + * Note: when you _don't_ use this routine for this protocol, + * and you use a different write policy from sock_writeable() + * then please supply your own write_space callback. + *=====================================================================*/ + +unsigned int wanpipe_poll(struct file * file, struct socket *sock, poll_table *wait) +{ + struct sock *sk = sock->sk; + unsigned int mask; + + ++wp_sk(sk)->poll_cnt; + + poll_wait(file, sk->sk_sleep, wait); + mask = 0; + + /* exceptional events? */ + if (sk->sk_err || !skb_queue_empty(&sk->sk_error_queue)) { + mask |= POLLPRI; + return mask; + } + if (sk->sk_shutdown & RCV_SHUTDOWN) + mask |= POLLHUP; + + /* readable? */ + if (!skb_queue_empty(&sk->sk_receive_queue)) { + mask |= POLLIN | POLLRDNORM; + } + + /* connection hasn't started yet */ + if (sk->sk_state == WANSOCK_CONNECTING) { + return mask; + } + + if (sk->sk_state == WANSOCK_DISCONNECTED) { + mask = POLLPRI; + return mask; + } + + /* This check blocks the user process if there is + * a packet already queued in the socket write queue. + * This option is only for X25API protocol, for other + * protocol like chdlc enable streaming mode, + * where multiple packets can be pending in the socket + * transmit queue */ + + if (wp_sk(sk)->num == htons(X25_PROT)) { + if (atomic_read(&wp_sk(sk)->packet_sent)) + return mask; + } + + /* writable? */ + if (sock_writeable(sk)){ + mask |= POLLOUT | POLLWRNORM | POLLWRBAND; + }else{ + set_bit(SOCK_ASYNC_NOSPACE, &sk->sk_socket->flags); + } + + return mask; +} + +/*====================================================================== + * wanpipe_listen + * + * X25API Specific function. Set a socket into LISTENING MODE. + *=====================================================================*/ + + +static int wanpipe_listen(struct socket *sock, int backlog) +{ + struct sock *sk = sock->sk; + + /* This is x25 specific area if protocol doesn't + * match, return error */ + if (wp_sk(sk)->num != htons(X25_PROT)) + return -EINVAL; + + if (sk->sk_state == WANSOCK_BIND_LISTEN) { + + sk->sk_max_ack_backlog = backlog; + sk->sk_state = WANSOCK_LISTEN; + return 0; + }else{ + printk(KERN_INFO "wansock: Listening sock was not binded\n"); + } + + return -EINVAL; +} + +/*====================================================================== + * wanpipe_link_card + * + * Connects the listening socket to the driver + *=====================================================================*/ + +static int wanpipe_link_card (struct sock *sk) +{ + sdla_t *card = (sdla_t*)wp_sk(sk)->card; + + if (!card) + return -ENOMEM; + + if ((card->sk != NULL) || (card->func != NULL)){ + printk(KERN_INFO "wansock: Listening queue is already established\n"); + return -EINVAL; + } + + card->sk=sk; + card->func=wanpipe_listen_rcv; + sock_set_flag(sk, SOCK_ZAPPED); + + return 0; +} + +/*====================================================================== + * wanpipe_listen + * + * X25API Specific function. Disconnect listening socket from + * the driver. + *=====================================================================*/ + +static void wanpipe_unlink_card (struct sock *sk) +{ + sdla_t *card = (sdla_t*)wp_sk(sk)->card; + + if (card){ + card->sk=NULL; + card->func=NULL; + } +} + +/*====================================================================== + * wanpipe_exec_cmd + * + * Ioctl function calls this function to execute user command. + * Connect() sytem call also calls this function to execute + * place call. This function blocks until command is executed. + *=====================================================================*/ + +static int wanpipe_exec_cmd(struct sock *sk, int cmd, unsigned int flags) +{ + int err = -EINVAL; + wanpipe_opt *wp = wp_sk(sk); + mbox_cmd_t *mbox_ptr = (mbox_cmd_t*)wp->mbox; + + if (!mbox_ptr){ + printk(KERN_INFO "NO MBOX PTR !!!!!\n"); + return -EINVAL; + } + + /* This is x25 specific area if protocol doesn't + * match, return error */ + if (wp->num != htons(X25_PROT)) + return -EINVAL; + + + switch (cmd){ + + case SIOC_WANPIPE_ACCEPT_CALL: + + if (sk->sk_state != WANSOCK_CONNECTING) { + err = -EHOSTDOWN; + break; + } + + err = execute_command(sk,X25_ACCEPT_CALL,0); + if (err < 0) + break; + + /* Update. Mar6 2000. + * Do not set the sock lcn number here, since + * it is done in wanpipe_listen_rcv(). + */ + if (sk->sk_state == WANSOCK_CONNECTED) { + wp->lcn = ((mbox_cmd_t*)wp->mbox)->cmd.lcn; + DBG_PRINTK(KERN_INFO "\nwansock: Accept OK %i\n", + wp->lcn); + err = 0; + + }else{ + DBG_PRINTK (KERN_INFO "\nwansock: Accept Failed %i\n", + wp->lcn); + wp->lcn = 0; + err = -ECONNREFUSED; + } + break; + + case SIOC_WANPIPE_CLEAR_CALL: + + if (sk->sk_state == WANSOCK_DISCONNECTED) { + err = -EINVAL; + break; + } + + + /* Check if data buffers are pending for transmission, + * if so, check whether user wants to wait until data + * is transmitted, or clear a call and drop packets */ + + if (atomic_read(&sk->sk_wmem_alloc) || + check_driver_busy(sk)) { + mbox_cmd_t *mbox = wp->mbox; + if (mbox->cmd.qdm & 0x80){ + mbox->cmd.result = 0x35; + err = -EAGAIN; + break; + } + } + + sk->sk_state = WANSOCK_DISCONNECTING; + + err = execute_command(sk,X25_CLEAR_CALL,0); + if (err < 0) + break; + + err = -ECONNREFUSED; + if (sk->sk_state == WANSOCK_DISCONNECTED) { + DBG_PRINTK(KERN_INFO "\nwansock: CLEAR OK %i\n", + wp->lcn); + wp->lcn = 0; + err = 0; + } + break; + + case SIOC_WANPIPE_RESET_CALL: + + if (sk->sk_state != WANSOCK_CONNECTED) { + err = -EINVAL; + break; + } + + + /* Check if data buffers are pending for transmission, + * if so, check whether user wants to wait until data + * is transmitted, or reset a call and drop packets */ + + if (atomic_read(&sk->sk_wmem_alloc) || + check_driver_busy(sk)) { + mbox_cmd_t *mbox = wp->mbox; + if (mbox->cmd.qdm & 0x80){ + mbox->cmd.result = 0x35; + err = -EAGAIN; + break; + } + } + + + err = execute_command(sk, X25_RESET,0); + if (err < 0) + break; + + err = mbox_ptr->cmd.result; + break; + + + case X25_PLACE_CALL: + + err=execute_command(sk,X25_PLACE_CALL,flags); + if (err < 0) + break; + + if (sk->sk_state == WANSOCK_CONNECTED) { + + wp->lcn = ((mbox_cmd_t*)wp->mbox)->cmd.lcn; + + DBG_PRINTK(KERN_INFO "\nwansock: PLACE CALL OK %i\n", + wp->lcn); + err = 0; + + } else if (sk->sk_state == WANSOCK_CONNECTING && + (flags & O_NONBLOCK)) { + wp->lcn = ((mbox_cmd_t*)wp->mbox)->cmd.lcn; + DBG_PRINTK(KERN_INFO "\nwansock: Place Call OK: Waiting %i\n", + wp->lcn); + + err = 0; + + }else{ + DBG_PRINTK(KERN_INFO "\nwansock: Place call Failed\n"); + err = -ECONNREFUSED; + } + + break; + + default: + return -EINVAL; + } + + return err; +} + +static int check_driver_busy (struct sock *sk) +{ + struct net_device *dev = dev_get_by_index(sk->sk_bound_dev_if); + wanpipe_common_t *chan; + + if (!dev) + return 0; + + dev_put(dev); + + if ((chan=dev->priv) == NULL) + return 0; + + return atomic_read(&chan->driver_busy); +} + + +/*====================================================================== + * wanpipe_accept + * + * ACCEPT() System call. X25API Specific function. + * For each incoming call, create a new socket and + * return it to the user. + *=====================================================================*/ + +static int wanpipe_accept(struct socket *sock, struct socket *newsock, int flags) +{ + struct sock *sk; + struct sock *newsk; + struct sk_buff *skb; + DECLARE_WAITQUEUE(wait, current); + int err=0; + + if (newsock->sk != NULL){ + wanpipe_kill_sock_accept(newsock->sk); + newsock->sk=NULL; + } + + if ((sk = sock->sk) == NULL) + return -EINVAL; + + if (sk->sk_type != SOCK_RAW) + return -EOPNOTSUPP; + + if (sk->sk_state != WANSOCK_LISTEN) + return -EINVAL; + + if (wp_sk(sk)->num != htons(X25_PROT)) + return -EINVAL; + + add_wait_queue(sk->sk_sleep,&wait); + current->state = TASK_INTERRUPTIBLE; + for (;;){ + skb = skb_dequeue(&sk->sk_receive_queue); + if (skb){ + err=0; + break; + } + if (signal_pending(current)) { + err = -ERESTARTSYS; + break; + } + schedule(); + } + current->state = TASK_RUNNING; + remove_wait_queue(sk->sk_sleep,&wait); + + if (err != 0) + return err; + + newsk = get_newsk_from_skb(skb); + if (!newsk){ + return -EINVAL; + } + + set_bit(1,&wanpipe_tx_critical); + write_lock(&wanpipe_sklist_lock); + sk_add_node(newsk, &wanpipe_sklist); + write_unlock(&wanpipe_sklist_lock); + clear_bit(1,&wanpipe_tx_critical); + + newsk->sk_socket = newsock; + newsk->sk_sleep = &newsock->wait; + + /* Now attach up the new socket */ + sk->sk_ack_backlog--; + newsock->sk = newsk; + + kfree_skb(skb); + + DBG_PRINTK(KERN_INFO "\nwansock: ACCEPT Got LCN %i\n", + wp_sk(newsk)->lcn); + return 0; +} + +/*====================================================================== + * get_newsk_from_skb + * + * Accept() uses this function to get the address of the new + * socket structure. + *=====================================================================*/ + +struct sock * get_newsk_from_skb (struct sk_buff *skb) +{ + struct net_device *dev = skb->dev; + wanpipe_common_t *chan; + + if (!dev){ + return NULL; + } + + if ((chan = dev->priv) == NULL){ + return NULL; + } + + if (!chan->sk){ + return NULL; + } + return (struct sock *)chan->sk; +} + +/*====================================================================== + * wanpipe_connect + * + * CONNECT() System Call. X25API specific function + * Check the state of the sock, and execute PLACE_CALL command. + * Connect can ether block or return without waiting for connection, + * if specified by user. + *=====================================================================*/ + +static int wanpipe_connect(struct socket *sock, struct sockaddr *uaddr, int addr_len, int flags) +{ + struct sock *sk = sock->sk; + struct wan_sockaddr_ll *addr = (struct wan_sockaddr_ll*)uaddr; + struct net_device *dev; + int err; + + if (wp_sk(sk)->num != htons(X25_PROT)) + return -EINVAL; + + if (sk->sk_state == WANSOCK_CONNECTED) + return -EISCONN; /* No reconnect on a seqpacket socket */ + + if (sk->sk_state != WAN_DISCONNECTED) { + printk(KERN_INFO "wansock: Trying to connect on channel NON DISCONNECT\n"); + return -ECONNREFUSED; + } + + sk->sk_state = WANSOCK_DISCONNECTED; + sock->state = SS_UNCONNECTED; + + if (addr_len != sizeof(struct wan_sockaddr_ll)) + return -EINVAL; + + if (addr->sll_family != AF_WANPIPE) + return -EINVAL; + + if ((dev = dev_get_by_index(sk->sk_bound_dev_if)) == NULL) + return -ENETUNREACH; + + dev_put(dev); + + if (!sock_flag(sk, SOCK_ZAPPED)) /* Must bind first - autobinding does not work */ + return -EINVAL; + + sock->state = SS_CONNECTING; + sk->sk_state = WANSOCK_CONNECTING; + + if (!wp_sk(sk)->mbox) { + if (wp_sk (sk)->svc) + return -EINVAL; + else { + int err; + if ((err=set_ioctl_cmd(sk,NULL)) < 0) + return err; + } + } + + if ((err=wanpipe_exec_cmd(sk, X25_PLACE_CALL,flags)) != 0){ + sock->state = SS_UNCONNECTED; + sk->sk_state = WANSOCK_CONNECTED; + return err; + } + + if (sk->sk_state != WANSOCK_CONNECTED && (flags & O_NONBLOCK)) { + return 0; + } + + if (sk->sk_state != WANSOCK_CONNECTED) { + sock->state = SS_UNCONNECTED; + return -ECONNREFUSED; + } + + sock->state = SS_CONNECTED; + return 0; +} + +struct proto_ops wanpipe_ops = { + .family = PF_WANPIPE, + .owner = THIS_MODULE, + .release = wanpipe_release, + .bind = wanpipe_bind, + .connect = wanpipe_connect, + .socketpair = sock_no_socketpair, + .accept = wanpipe_accept, + .getname = wanpipe_getname, + .poll = wanpipe_poll, + .ioctl = wanpipe_ioctl, + .listen = wanpipe_listen, + .shutdown = sock_no_shutdown, + .setsockopt = sock_no_setsockopt, + .getsockopt = sock_no_getsockopt, + .sendmsg = wanpipe_sendmsg, + .recvmsg = wanpipe_recvmsg +}; + +static struct net_proto_family wanpipe_family_ops = { + .family = PF_WANPIPE, + .create = wanpipe_create, + .owner = THIS_MODULE, +}; + +struct notifier_block wanpipe_netdev_notifier = { + .notifier_call = wanpipe_notifier, +}; + + +#ifdef MODULE +void cleanup_module(void) +{ + printk(KERN_INFO "wansock: Cleaning up \n"); + unregister_netdevice_notifier(&wanpipe_netdev_notifier); + sock_unregister(PF_WANPIPE); + proto_unregister(&wanpipe_proto); +} + +int init_module(void) +{ + int rc; + + printk(KERN_INFO "wansock: Registering Socket \n"); + + rc = proto_register(&wanpipe_proto, 0); + if (rc != 0) + goto out; + + sock_register(&wanpipe_family_ops); + register_netdevice_notifier(&wanpipe_netdev_notifier); +out: + return rc; +} +#endif +MODULE_LICENSE("GPL"); +MODULE_ALIAS_NETPROTO(PF_WANPIPE); diff --git a/net/wanrouter/patchlevel b/net/wanrouter/patchlevel new file mode 100644 index 0000000..c043eea --- /dev/null +++ b/net/wanrouter/patchlevel @@ -0,0 +1 @@ +2.2.1 diff --git a/net/wanrouter/wanmain.c b/net/wanrouter/wanmain.c new file mode 100644 index 0000000..956c17f --- /dev/null +++ b/net/wanrouter/wanmain.c @@ -0,0 +1,888 @@ +/***************************************************************************** +* wanmain.c WAN Multiprotocol Router Module. Main code. +* +* This module is completely hardware-independent and provides +* the following common services for the WAN Link Drivers: +* o WAN device managenment (registering, unregistering) +* o Network interface management +* o Physical connection management (dial-up, incoming calls) +* o Logical connection management (switched virtual circuits) +* o Protocol encapsulation/decapsulation +* +* Author: Gideon Hack +* +* Copyright: (c) 1995-1999 Sangoma Technologies 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. +* ============================================================================ +* Nov 24, 2000 Nenad Corbic Updated for 2.4.X kernels +* Nov 07, 2000 Nenad Corbic Fixed the Mulit-Port PPP for kernels 2.2.16 and +* greater. +* Aug 2, 2000 Nenad Corbic Block the Multi-Port PPP from running on +* kernels 2.2.16 or greater. The SyncPPP +* has changed. +* Jul 13, 2000 Nenad Corbic Added SyncPPP support +* Added extra debugging in device_setup(). +* Oct 01, 1999 Gideon Hack Update for s514 PCI card +* Dec 27, 1996 Gene Kozin Initial version (based on Sangoma's WANPIPE) +* Jan 16, 1997 Gene Kozin router_devlist made public +* Jan 31, 1997 Alan Cox Hacked it about a bit for 2.1 +* Jun 27, 1997 Alan Cox realigned with vendor code +* Oct 15, 1997 Farhan Thawar changed wan_encapsulate to add a pad byte of 0 +* Apr 20, 1998 Alan Cox Fixed 2.1 symbols +* May 17, 1998 K. Baranowski Fixed SNAP encapsulation in wan_encapsulate +* Dec 15, 1998 Arnaldo Melo support for firmwares of up to 128000 bytes +* check wandev->setup return value +* Dec 22, 1998 Arnaldo Melo vmalloc/vfree used in device_setup to allocate +* kernel memory and copy configuration data to +* kernel space (for big firmwares) +* Jun 02, 1999 Gideon Hack Updates for Linux 2.0.X and 2.2.X kernels. +*****************************************************************************/ + +#include <linux/config.h> +#include <linux/stddef.h> /* offsetof(), etc. */ +#include <linux/errno.h> /* return codes */ +#include <linux/kernel.h> +#include <linux/init.h> +#include <linux/module.h> /* support for loadable modules */ +#include <linux/slab.h> /* kmalloc(), kfree() */ +#include <linux/mm.h> /* verify_area(), etc. */ +#include <linux/string.h> /* inline mem*, str* functions */ + +#include <asm/byteorder.h> /* htons(), etc. */ +#include <linux/wanrouter.h> /* WAN router API definitions */ + +#include <linux/vmalloc.h> /* vmalloc, vfree */ +#include <asm/uaccess.h> /* copy_to/from_user */ +#include <linux/init.h> /* __initfunc et al. */ +#include <net/syncppp.h> + +#define KMEM_SAFETYZONE 8 + +/***********FOR DEBUGGING PURPOSES********************************************* +static void * dbg_kmalloc(unsigned int size, int prio, int line) { + int i = 0; + void * v = kmalloc(size+sizeof(unsigned int)+2*KMEM_SAFETYZONE*8,prio); + char * c1 = v; + c1 += sizeof(unsigned int); + *((unsigned int *)v) = size; + + for (i = 0; i < KMEM_SAFETYZONE; i++) { + c1[0] = 'D'; c1[1] = 'E'; c1[2] = 'A'; c1[3] = 'D'; + c1[4] = 'B'; c1[5] = 'E'; c1[6] = 'E'; c1[7] = 'F'; + c1 += 8; + } + c1 += size; + for (i = 0; i < KMEM_SAFETYZONE; i++) { + c1[0] = 'M'; c1[1] = 'U'; c1[2] = 'N'; c1[3] = 'G'; + c1[4] = 'W'; c1[5] = 'A'; c1[6] = 'L'; c1[7] = 'L'; + c1 += 8; + } + v = ((char *)v) + sizeof(unsigned int) + KMEM_SAFETYZONE*8; + printk(KERN_INFO "line %d kmalloc(%d,%d) = %p\n",line,size,prio,v); + return v; +} +static void dbg_kfree(void * v, int line) { + unsigned int * sp = (unsigned int *)(((char *)v) - (sizeof(unsigned int) + KMEM_SAFETYZONE*8)); + unsigned int size = *sp; + char * c1 = ((char *)v) - KMEM_SAFETYZONE*8; + int i = 0; + for (i = 0; i < KMEM_SAFETYZONE; i++) { + if ( c1[0] != 'D' || c1[1] != 'E' || c1[2] != 'A' || c1[3] != 'D' + || c1[4] != 'B' || c1[5] != 'E' || c1[6] != 'E' || c1[7] != 'F') { + printk(KERN_INFO "kmalloced block at %p has been corrupted (underrun)!\n",v); + printk(KERN_INFO " %4x: %2x %2x %2x %2x %2x %2x %2x %2x\n", i*8, + c1[0],c1[1],c1[2],c1[3],c1[4],c1[5],c1[6],c1[7] ); + } + c1 += 8; + } + c1 += size; + for (i = 0; i < KMEM_SAFETYZONE; i++) { + if ( c1[0] != 'M' || c1[1] != 'U' || c1[2] != 'N' || c1[3] != 'G' + || c1[4] != 'W' || c1[5] != 'A' || c1[6] != 'L' || c1[7] != 'L' + ) { + printk(KERN_INFO "kmalloced block at %p has been corrupted (overrun):\n",v); + printk(KERN_INFO " %4x: %2x %2x %2x %2x %2x %2x %2x %2x\n", i*8, + c1[0],c1[1],c1[2],c1[3],c1[4],c1[5],c1[6],c1[7] ); + } + c1 += 8; + } + printk(KERN_INFO "line %d kfree(%p)\n",line,v); + v = ((char *)v) - (sizeof(unsigned int) + KMEM_SAFETYZONE*8); + kfree(v); +} + +#define kmalloc(x,y) dbg_kmalloc(x,y,__LINE__) +#define kfree(x) dbg_kfree(x,__LINE__) +*****************************************************************************/ + +/* + * Function Prototypes + */ + +/* + * WAN device IOCTL handlers + */ + +static int wanrouter_device_setup(struct wan_device *wandev, + wandev_conf_t __user *u_conf); +static int wanrouter_device_stat(struct wan_device *wandev, + wandev_stat_t __user *u_stat); +static int wanrouter_device_shutdown(struct wan_device *wandev); +static int wanrouter_device_new_if(struct wan_device *wandev, + wanif_conf_t __user *u_conf); +static int wanrouter_device_del_if(struct wan_device *wandev, + char __user *u_name); + +/* + * Miscellaneous + */ + +static struct wan_device *wanrouter_find_device(char *name); +static int wanrouter_delete_interface(struct wan_device *wandev, char *name); +void lock_adapter_irq(spinlock_t *lock, unsigned long *smp_flags); +void unlock_adapter_irq(spinlock_t *lock, unsigned long *smp_flags); + + + +/* + * Global Data + */ + +static char wanrouter_fullname[] = "Sangoma WANPIPE Router"; +static char wanrouter_copyright[] = "(c) 1995-2000 Sangoma Technologies Inc."; +static char wanrouter_modname[] = ROUTER_NAME; /* short module name */ +struct wan_device* wanrouter_router_devlist; /* list of registered devices */ + +/* + * Organize Unique Identifiers for encapsulation/decapsulation + */ + +static unsigned char wanrouter_oui_ether[] = { 0x00, 0x00, 0x00 }; +#if 0 +static unsigned char wanrouter_oui_802_2[] = { 0x00, 0x80, 0xC2 }; +#endif + +static int __init wanrouter_init(void) +{ + int err; + + printk(KERN_INFO "%s v%u.%u %s\n", + wanrouter_fullname, ROUTER_VERSION, ROUTER_RELEASE, + wanrouter_copyright); + + err = wanrouter_proc_init(); + if (err) + printk(KERN_INFO "%s: can't create entry in proc filesystem!\n", + wanrouter_modname); + + return err; +} + +static void __exit wanrouter_cleanup (void) +{ + wanrouter_proc_cleanup(); +} + +/* + * This is just plain dumb. We should move the bugger to drivers/net/wan, + * slap it first in directory and make it module_init(). The only reason + * for subsys_initcall() here is that net goes after drivers (why, BTW?) + */ +subsys_initcall(wanrouter_init); +module_exit(wanrouter_cleanup); + +/* + * Kernel APIs + */ + +/* + * Register WAN device. + * o verify device credentials + * o create an entry for the device in the /proc/net/router directory + * o initialize internally maintained fields of the wan_device structure + * o link device data space to a singly-linked list + * o if it's the first device, then start kernel 'thread' + * o increment module use count + * + * Return: + * 0 Ok + * < 0 error. + * + * Context: process + */ + + +int register_wan_device(struct wan_device *wandev) +{ + int err, namelen; + + if ((wandev == NULL) || (wandev->magic != ROUTER_MAGIC) || + (wandev->name == NULL)) + return -EINVAL; + + namelen = strlen(wandev->name); + if (!namelen || (namelen > WAN_DRVNAME_SZ)) + return -EINVAL; + + if (wanrouter_find_device(wandev->name)) + return -EEXIST; + +#ifdef WANDEBUG + printk(KERN_INFO "%s: registering WAN device %s\n", + wanrouter_modname, wandev->name); +#endif + + /* + * Register /proc directory entry + */ + err = wanrouter_proc_add(wandev); + if (err) { + printk(KERN_INFO + "%s: can't create /proc/net/router/%s entry!\n", + wanrouter_modname, wandev->name); + return err; + } + + /* + * Initialize fields of the wan_device structure maintained by the + * router and update local data. + */ + + wandev->ndev = 0; + wandev->dev = NULL; + wandev->next = wanrouter_router_devlist; + wanrouter_router_devlist = wandev; + return 0; +} + +/* + * Unregister WAN device. + * o shut down device + * o unlink device data space from the linked list + * o delete device entry in the /proc/net/router directory + * o decrement module use count + * + * Return: 0 Ok + * <0 error. + * Context: process + */ + + +int unregister_wan_device(char *name) +{ + struct wan_device *wandev, *prev; + + if (name == NULL) + return -EINVAL; + + for (wandev = wanrouter_router_devlist, prev = NULL; + wandev && strcmp(wandev->name, name); + prev = wandev, wandev = wandev->next) + ; + if (wandev == NULL) + return -ENODEV; + +#ifdef WANDEBUG + printk(KERN_INFO "%s: unregistering WAN device %s\n", + wanrouter_modname, name); +#endif + + if (wandev->state != WAN_UNCONFIGURED) + wanrouter_device_shutdown(wandev); + + if (prev) + prev->next = wandev->next; + else + wanrouter_router_devlist = wandev->next; + + wanrouter_proc_delete(wandev); + return 0; +} + +/* + * Encapsulate packet. + * + * Return: encapsulation header size + * < 0 - unsupported Ethertype + * + * Notes: + * 1. This function may be called on interrupt context. + */ + + +int wanrouter_encapsulate(struct sk_buff *skb, struct net_device *dev, + unsigned short type) +{ + int hdr_len = 0; + + switch (type) { + case ETH_P_IP: /* IP datagram encapsulation */ + hdr_len += 1; + skb_push(skb, 1); + skb->data[0] = NLPID_IP; + break; + + case ETH_P_IPX: /* SNAP encapsulation */ + case ETH_P_ARP: + hdr_len += 7; + skb_push(skb, 7); + skb->data[0] = 0; + skb->data[1] = NLPID_SNAP; + memcpy(&skb->data[2], wanrouter_oui_ether, + sizeof(wanrouter_oui_ether)); + *((unsigned short*)&skb->data[5]) = htons(type); + break; + + default: /* Unknown packet type */ + printk(KERN_INFO + "%s: unsupported Ethertype 0x%04X on interface %s!\n", + wanrouter_modname, type, dev->name); + hdr_len = -EINVAL; + } + return hdr_len; +} + + +/* + * Decapsulate packet. + * + * Return: Ethertype (in network order) + * 0 unknown encapsulation + * + * Notes: + * 1. This function may be called on interrupt context. + */ + + +unsigned short wanrouter_type_trans(struct sk_buff *skb, struct net_device *dev) +{ + int cnt = skb->data[0] ? 0 : 1; /* there may be a pad present */ + unsigned short ethertype; + + switch (skb->data[cnt]) { + case NLPID_IP: /* IP datagramm */ + ethertype = htons(ETH_P_IP); + cnt += 1; + break; + + case NLPID_SNAP: /* SNAP encapsulation */ + if (memcmp(&skb->data[cnt + 1], wanrouter_oui_ether, + sizeof(wanrouter_oui_ether))){ + printk(KERN_INFO + "%s: unsupported SNAP OUI %02X-%02X-%02X " + "on interface %s!\n", wanrouter_modname, + skb->data[cnt+1], skb->data[cnt+2], + skb->data[cnt+3], dev->name); + return 0; + } + ethertype = *((unsigned short*)&skb->data[cnt+4]); + cnt += 6; + break; + + /* add other protocols, e.g. CLNP, ESIS, ISIS, if needed */ + + default: + printk(KERN_INFO + "%s: unsupported NLPID 0x%02X on interface %s!\n", + wanrouter_modname, skb->data[cnt], dev->name); + return 0; + } + skb->protocol = ethertype; + skb->pkt_type = PACKET_HOST; /* Physically point to point */ + skb_pull(skb, cnt); + skb->mac.raw = skb->data; + return ethertype; +} + + +/* + * WAN device IOCTL. + * o find WAN device associated with this node + * o execute requested action or pass command to the device driver + */ + +int wanrouter_ioctl(struct inode *inode, struct file *file, + unsigned int cmd, unsigned long arg) +{ + int err = 0; + struct proc_dir_entry *dent; + struct wan_device *wandev; + void __user *data = (void __user *)arg; + + if (!capable(CAP_NET_ADMIN)) + return -EPERM; + + if ((cmd >> 8) != ROUTER_IOCTL) + return -EINVAL; + + dent = PDE(inode); + if ((dent == NULL) || (dent->data == NULL)) + return -EINVAL; + + wandev = dent->data; + if (wandev->magic != ROUTER_MAGIC) + return -EINVAL; + + switch (cmd) { + case ROUTER_SETUP: + err = wanrouter_device_setup(wandev, data); + break; + + case ROUTER_DOWN: + err = wanrouter_device_shutdown(wandev); + break; + + case ROUTER_STAT: + err = wanrouter_device_stat(wandev, data); + break; + + case ROUTER_IFNEW: + err = wanrouter_device_new_if(wandev, data); + break; + + case ROUTER_IFDEL: + err = wanrouter_device_del_if(wandev, data); + break; + + case ROUTER_IFSTAT: + break; + + default: + if ((cmd >= ROUTER_USER) && + (cmd <= ROUTER_USER_MAX) && + wandev->ioctl) + err = wandev->ioctl(wandev, cmd, arg); + else err = -EINVAL; + } + return err; +} + +/* + * WAN Driver IOCTL Handlers + */ + +/* + * Setup WAN link device. + * o verify user address space + * o allocate kernel memory and copy configuration data to kernel space + * o if configuration data includes extension, copy it to kernel space too + * o call driver's setup() entry point + */ + +static int wanrouter_device_setup(struct wan_device *wandev, + wandev_conf_t __user *u_conf) +{ + void *data = NULL; + wandev_conf_t *conf; + int err = -EINVAL; + + if (wandev->setup == NULL) { /* Nothing to do ? */ + printk(KERN_INFO "%s: ERROR, No setup script: wandev->setup()\n", + wandev->name); + return 0; + } + + conf = kmalloc(sizeof(wandev_conf_t), GFP_KERNEL); + if (conf == NULL){ + printk(KERN_INFO "%s: ERROR, Failed to allocate kernel memory !\n", + wandev->name); + return -ENOBUFS; + } + + if (copy_from_user(conf, u_conf, sizeof(wandev_conf_t))) { + printk(KERN_INFO "%s: Failed to copy user config data to kernel space!\n", + wandev->name); + kfree(conf); + return -EFAULT; + } + + if (conf->magic != ROUTER_MAGIC) { + kfree(conf); + printk(KERN_INFO "%s: ERROR, Invalid MAGIC Number\n", + wandev->name); + return -EINVAL; + } + + if (conf->data_size && conf->data) { + if (conf->data_size > 128000 || conf->data_size < 0) { + printk(KERN_INFO + "%s: ERROR, Invalid firmware data size %i !\n", + wandev->name, conf->data_size); + kfree(conf); + return -EINVAL; + } + + data = vmalloc(conf->data_size); + if (!data) { + printk(KERN_INFO + "%s: ERROR, Faild allocate kernel memory !\n", + wandev->name); + kfree(conf); + return -ENOBUFS; + } + if (!copy_from_user(data, conf->data, conf->data_size)) { + conf->data = data; + err = wandev->setup(wandev, conf); + } else { + printk(KERN_INFO + "%s: ERROR, Faild to copy from user data !\n", + wandev->name); + err = -EFAULT; + } + vfree(data); + } else { + printk(KERN_INFO + "%s: ERROR, No firmware found ! Firmware size = %i !\n", + wandev->name, conf->data_size); + } + + kfree(conf); + return err; +} + +/* + * Shutdown WAN device. + * o delete all not opened logical channels for this device + * o call driver's shutdown() entry point + */ + +static int wanrouter_device_shutdown(struct wan_device *wandev) +{ + struct net_device *dev; + int err=0; + + if (wandev->state == WAN_UNCONFIGURED) + return 0; + + printk(KERN_INFO "\n%s: Shutting Down!\n",wandev->name); + + for (dev = wandev->dev; dev;) { + err = wanrouter_delete_interface(wandev, dev->name); + if (err) + return err; + /* The above function deallocates the current dev + * structure. Therefore, we cannot use dev->priv + * as the next element: wandev->dev points to the + * next element */ + dev = wandev->dev; + } + + if (wandev->ndev) + return -EBUSY; /* there are opened interfaces */ + + if (wandev->shutdown) + err=wandev->shutdown(wandev); + + return err; +} + +/* + * Get WAN device status & statistics. + */ + +static int wanrouter_device_stat(struct wan_device *wandev, + wandev_stat_t __user *u_stat) +{ + wandev_stat_t stat; + + memset(&stat, 0, sizeof(stat)); + + /* Ask device driver to update device statistics */ + if ((wandev->state != WAN_UNCONFIGURED) && wandev->update) + wandev->update(wandev); + + /* Fill out structure */ + stat.ndev = wandev->ndev; + stat.state = wandev->state; + + if (copy_to_user(u_stat, &stat, sizeof(stat))) + return -EFAULT; + + return 0; +} + +/* + * Create new WAN interface. + * o verify user address space + * o copy configuration data to kernel address space + * o allocate network interface data space + * o call driver's new_if() entry point + * o make sure there is no interface name conflict + * o register network interface + */ + +static int wanrouter_device_new_if(struct wan_device *wandev, + wanif_conf_t __user *u_conf) +{ + wanif_conf_t *cnf; + struct net_device *dev = NULL; +#ifdef CONFIG_WANPIPE_MULTPPP + struct ppp_device *pppdev=NULL; +#endif + int err; + + if ((wandev->state == WAN_UNCONFIGURED) || (wandev->new_if == NULL)) + return -ENODEV; + + cnf = kmalloc(sizeof(wanif_conf_t), GFP_KERNEL); + if (!cnf) + return -ENOBUFS; + + err = -EFAULT; + if (copy_from_user(cnf, u_conf, sizeof(wanif_conf_t))) + goto out; + + err = -EINVAL; + if (cnf->magic != ROUTER_MAGIC) + goto out; + + if (cnf->config_id == WANCONFIG_MPPP) { +#ifdef CONFIG_WANPIPE_MULTPPP + pppdev = kmalloc(sizeof(struct ppp_device), GFP_KERNEL); + err = -ENOBUFS; + if (pppdev == NULL) + goto out; + memset(pppdev, 0, sizeof(struct ppp_device)); + pppdev->dev = kmalloc(sizeof(struct net_device), GFP_KERNEL); + if (pppdev->dev == NULL) { + kfree(pppdev); + err = -ENOBUFS; + goto out; + } + memset(pppdev->dev, 0, sizeof(struct net_device)); + err = wandev->new_if(wandev, (struct net_device *)pppdev, cnf); + dev = pppdev->dev; +#else + printk(KERN_INFO "%s: Wanpipe Mulit-Port PPP support has not been compiled in!\n", + wandev->name); + err = -EPROTONOSUPPORT; + goto out; +#endif + } else { + dev = kmalloc(sizeof(struct net_device), GFP_KERNEL); + err = -ENOBUFS; + if (dev == NULL) + goto out; + memset(dev, 0, sizeof(struct net_device)); + err = wandev->new_if(wandev, dev, cnf); + } + + if (!err) { + /* Register network interface. This will invoke init() + * function supplied by the driver. If device registered + * successfully, add it to the interface list. + */ + + if (dev->name == NULL) { + err = -EINVAL; + } else { + + #ifdef WANDEBUG + printk(KERN_INFO "%s: registering interface %s...\n", + wanrouter_modname, dev->name); + #endif + + err = register_netdev(dev); + if (!err) { + struct net_device *slave = NULL; + unsigned long smp_flags=0; + + lock_adapter_irq(&wandev->lock, &smp_flags); + + if (wandev->dev == NULL) { + wandev->dev = dev; + } else { + for (slave=wandev->dev; + *((struct net_device **)slave->priv); + slave = *((struct net_device **)slave->priv)); + + *((struct net_device **)slave->priv) = dev; + } + ++wandev->ndev; + + unlock_adapter_irq(&wandev->lock, &smp_flags); + err = 0; /* done !!! */ + goto out; + } + } + if (wandev->del_if) + wandev->del_if(wandev, dev); + } + + /* This code has moved from del_if() function */ + if (dev->priv) { + kfree(dev->priv); + dev->priv = NULL; + } + +#ifdef CONFIG_WANPIPE_MULTPPP + if (cnf->config_id == WANCONFIG_MPPP) + kfree(pppdev); + else + kfree(dev); +#else + /* Sync PPP is disabled */ + if (cnf->config_id != WANCONFIG_MPPP) + kfree(dev); +#endif + +out: + kfree(cnf); + return err; +} + + +/* + * Delete WAN logical channel. + * o verify user address space + * o copy configuration data to kernel address space + */ + +static int wanrouter_device_del_if(struct wan_device *wandev, char __user *u_name) +{ + char name[WAN_IFNAME_SZ + 1]; + int err = 0; + + if (wandev->state == WAN_UNCONFIGURED) + return -ENODEV; + + memset(name, 0, sizeof(name)); + + if (copy_from_user(name, u_name, WAN_IFNAME_SZ)) + return -EFAULT; + + err = wanrouter_delete_interface(wandev, name); + if (err) + return err; + + /* If last interface being deleted, shutdown card + * This helps with administration at leaf nodes + * (You can tell if the person at the other end of the phone + * has an interface configured) and avoids DoS vulnerabilities + * in binary driver files - this fixes a problem with the current + * Sangoma driver going into strange states when all the network + * interfaces are deleted and the link irrecoverably disconnected. + */ + + if (!wandev->ndev && wandev->shutdown) + err = wandev->shutdown(wandev); + + return err; +} + +/* + * Miscellaneous Functions + */ + +/* + * Find WAN device by name. + * Return pointer to the WAN device data space or NULL if device not found. + */ + +static struct wan_device *wanrouter_find_device(char *name) +{ + struct wan_device *wandev; + + for (wandev = wanrouter_router_devlist; + wandev && strcmp(wandev->name, name); + wandev = wandev->next); + return wandev; +} + +/* + * Delete WAN logical channel identified by its name. + * o find logical channel by its name + * o call driver's del_if() entry point + * o unregister network interface + * o unlink channel data space from linked list of channels + * o release channel data space + * + * Return: 0 success + * -ENODEV channel not found. + * -EBUSY interface is open + * + * Note: If (force != 0), then device will be destroyed even if interface + * associated with it is open. It's caller's responsibility to make + * sure that opened interfaces are not removed! + */ + +static int wanrouter_delete_interface(struct wan_device *wandev, char *name) +{ + struct net_device *dev = NULL, *prev = NULL; + unsigned long smp_flags=0; + + lock_adapter_irq(&wandev->lock, &smp_flags); + dev = wandev->dev; + prev = NULL; + while (dev && strcmp(name, dev->name)) { + struct net_device **slave = dev->priv; + prev = dev; + dev = *slave; + } + unlock_adapter_irq(&wandev->lock, &smp_flags); + + if (dev == NULL) + return -ENODEV; /* interface not found */ + + if (netif_running(dev)) + return -EBUSY; /* interface in use */ + + if (wandev->del_if) + wandev->del_if(wandev, dev); + + lock_adapter_irq(&wandev->lock, &smp_flags); + if (prev) { + struct net_device **prev_slave = prev->priv; + struct net_device **slave = dev->priv; + + *prev_slave = *slave; + } else { + struct net_device **slave = dev->priv; + wandev->dev = *slave; + } + --wandev->ndev; + unlock_adapter_irq(&wandev->lock, &smp_flags); + + printk(KERN_INFO "%s: unregistering '%s'\n", wandev->name, dev->name); + + /* Due to new interface linking method using dev->priv, + * this code has moved from del_if() function.*/ + if (dev->priv){ + kfree(dev->priv); + dev->priv=NULL; + } + + unregister_netdev(dev); + + free_netdev(dev); + + return 0; +} + +void lock_adapter_irq(spinlock_t *lock, unsigned long *smp_flags) +{ + spin_lock_irqsave(lock, *smp_flags); +} + + +void unlock_adapter_irq(spinlock_t *lock, unsigned long *smp_flags) +{ + spin_unlock_irqrestore(lock, *smp_flags); +} + +EXPORT_SYMBOL(register_wan_device); +EXPORT_SYMBOL(unregister_wan_device); +EXPORT_SYMBOL(wanrouter_encapsulate); +EXPORT_SYMBOL(wanrouter_type_trans); +EXPORT_SYMBOL(lock_adapter_irq); +EXPORT_SYMBOL(unlock_adapter_irq); + +MODULE_LICENSE("GPL"); + +/* + * End + */ diff --git a/net/wanrouter/wanproc.c b/net/wanrouter/wanproc.c new file mode 100644 index 0000000..c28ba5a --- /dev/null +++ b/net/wanrouter/wanproc.c @@ -0,0 +1,381 @@ +/***************************************************************************** +* wanproc.c WAN Router Module. /proc filesystem interface. +* +* This module is completely hardware-independent and provides +* access to the router using Linux /proc filesystem. +* +* Author: Gideon Hack +* +* Copyright: (c) 1995-1999 Sangoma Technologies 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. +* ============================================================================ +* Jun 02, 1999 Gideon Hack Updates for Linux 2.2.X kernels. +* Jun 29, 1997 Alan Cox Merged with 1.0.3 vendor code +* Jan 29, 1997 Gene Kozin v1.0.1. Implemented /proc read routines +* Jan 30, 1997 Alan Cox Hacked around for 2.1 +* Dec 13, 1996 Gene Kozin Initial version (based on Sangoma's WANPIPE) +*****************************************************************************/ + +#include <linux/config.h> +#include <linux/init.h> /* __initfunc et al. */ +#include <linux/stddef.h> /* offsetof(), etc. */ +#include <linux/errno.h> /* return codes */ +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/wanrouter.h> /* WAN router API definitions */ +#include <linux/seq_file.h> +#include <linux/smp_lock.h> + +#include <asm/io.h> + +#define PROC_STATS_FORMAT "%30s: %12lu\n" + +/****** Defines and Macros **************************************************/ + +#define PROT_DECODE(prot) ((prot == WANCONFIG_FR) ? " FR" :\ + (prot == WANCONFIG_X25) ? " X25" : \ + (prot == WANCONFIG_PPP) ? " PPP" : \ + (prot == WANCONFIG_CHDLC) ? " CHDLC": \ + (prot == WANCONFIG_MPPP) ? " MPPP" : \ + " Unknown" ) + +/****** Function Prototypes *************************************************/ + +#ifdef CONFIG_PROC_FS + +/* Miscellaneous */ + +/* + * Structures for interfacing with the /proc filesystem. + * Router creates its own directory /proc/net/router with the folowing + * entries: + * config device configuration + * status global device statistics + * <device> entry for each WAN device + */ + +/* + * Generic /proc/net/router/<file> file and inode operations + */ + +/* + * /proc/net/router + */ + +static struct proc_dir_entry *proc_router; + +/* Strings */ + +/* + * Interface functions + */ + +/****** Proc filesystem entry points ****************************************/ + +/* + * Iterator + */ +static void *r_start(struct seq_file *m, loff_t *pos) +{ + struct wan_device *wandev; + loff_t l = *pos; + + lock_kernel(); + if (!l--) + return SEQ_START_TOKEN; + for (wandev = wanrouter_router_devlist; l-- && wandev; + wandev = wandev->next) + ; + return wandev; +} + +static void *r_next(struct seq_file *m, void *v, loff_t *pos) +{ + struct wan_device *wandev = v; + (*pos)++; + return (v == SEQ_START_TOKEN) ? wanrouter_router_devlist : wandev->next; +} + +static void r_stop(struct seq_file *m, void *v) +{ + unlock_kernel(); +} + +static int config_show(struct seq_file *m, void *v) +{ + struct wan_device *p = v; + if (v == SEQ_START_TOKEN) { + seq_puts(m, "Device name | port |IRQ|DMA| mem.addr |" + "mem.size|option1|option2|option3|option4\n"); + return 0; + } + if (!p->state) + return 0; + seq_printf(m, "%-15s|0x%-4X|%3u|%3u| 0x%-8lX |0x%-6X|%7u|%7u|%7u|%7u\n", + p->name, p->ioport, p->irq, p->dma, p->maddr, p->msize, + p->hw_opt[0], p->hw_opt[1], p->hw_opt[2], p->hw_opt[3]); + return 0; +} + +static int status_show(struct seq_file *m, void *v) +{ + struct wan_device *p = v; + if (v == SEQ_START_TOKEN) { + seq_puts(m, "Device name |protocol|station|interface|" + "clocking|baud rate| MTU |ndev|link state\n"); + return 0; + } + if (!p->state) + return 0; + seq_printf(m, "%-15s|%-8s| %-7s| %-9s|%-8s|%9u|%5u|%3u |", + p->name, + PROT_DECODE(p->config_id), + p->config_id == WANCONFIG_FR ? + (p->station ? "Node" : "CPE") : + (p->config_id == WANCONFIG_X25 ? + (p->station ? "DCE" : "DTE") : + ("N/A")), + p->interface ? "V.35" : "RS-232", + p->clocking ? "internal" : "external", + p->bps, + p->mtu, + p->ndev); + + switch (p->state) { + case WAN_UNCONFIGURED: + seq_printf(m, "%-12s\n", "unconfigured"); + break; + case WAN_DISCONNECTED: + seq_printf(m, "%-12s\n", "disconnected"); + break; + case WAN_CONNECTING: + seq_printf(m, "%-12s\n", "connecting"); + break; + case WAN_CONNECTED: + seq_printf(m, "%-12s\n", "connected"); + break; + default: + seq_printf(m, "%-12s\n", "invalid"); + break; + } + return 0; +} + +static struct seq_operations config_op = { + .start = r_start, + .next = r_next, + .stop = r_stop, + .show = config_show, +}; + +static struct seq_operations status_op = { + .start = r_start, + .next = r_next, + .stop = r_stop, + .show = status_show, +}; + +static int config_open(struct inode *inode, struct file *file) +{ + return seq_open(file, &config_op); +} + +static int status_open(struct inode *inode, struct file *file) +{ + return seq_open(file, &status_op); +} + +static struct file_operations config_fops = { + .owner = THIS_MODULE, + .open = config_open, + .read = seq_read, + .llseek = seq_lseek, + .release = seq_release, +}; + +static struct file_operations status_fops = { + .owner = THIS_MODULE, + .open = status_open, + .read = seq_read, + .llseek = seq_lseek, + .release = seq_release, +}; + +static int wandev_show(struct seq_file *m, void *v) +{ + struct wan_device *wandev = m->private; + + if (wandev->magic != ROUTER_MAGIC) + return 0; + + if (!wandev->state) { + seq_puts(m, "device is not configured!\n"); + return 0; + } + + /* Update device statistics */ + if (wandev->update) { + int err = wandev->update(wandev); + if (err == -EAGAIN) { + seq_puts(m, "Device is busy!\n"); + return 0; + } + if (err) { + seq_puts(m, "Device is not configured!\n"); + return 0; + } + } + + seq_printf(m, PROC_STATS_FORMAT, + "total packets received", wandev->stats.rx_packets); + seq_printf(m, PROC_STATS_FORMAT, + "total packets transmitted", wandev->stats.tx_packets); + seq_printf(m, PROC_STATS_FORMAT, + "total bytes received", wandev->stats.rx_bytes); + seq_printf(m, PROC_STATS_FORMAT, + "total bytes transmitted", wandev->stats.tx_bytes); + seq_printf(m, PROC_STATS_FORMAT, + "bad packets received", wandev->stats.rx_errors); + seq_printf(m, PROC_STATS_FORMAT, + "packet transmit problems", wandev->stats.tx_errors); + seq_printf(m, PROC_STATS_FORMAT, + "received frames dropped", wandev->stats.rx_dropped); + seq_printf(m, PROC_STATS_FORMAT, + "transmit frames dropped", wandev->stats.tx_dropped); + seq_printf(m, PROC_STATS_FORMAT, + "multicast packets received", wandev->stats.multicast); + seq_printf(m, PROC_STATS_FORMAT, + "transmit collisions", wandev->stats.collisions); + seq_printf(m, PROC_STATS_FORMAT, + "receive length errors", wandev->stats.rx_length_errors); + seq_printf(m, PROC_STATS_FORMAT, + "receiver overrun errors", wandev->stats.rx_over_errors); + seq_printf(m, PROC_STATS_FORMAT, + "CRC errors", wandev->stats.rx_crc_errors); + seq_printf(m, PROC_STATS_FORMAT, + "frame format errors (aborts)", wandev->stats.rx_frame_errors); + seq_printf(m, PROC_STATS_FORMAT, + "receiver fifo overrun", wandev->stats.rx_fifo_errors); + seq_printf(m, PROC_STATS_FORMAT, + "receiver missed packet", wandev->stats.rx_missed_errors); + seq_printf(m, PROC_STATS_FORMAT, + "aborted frames transmitted", wandev->stats.tx_aborted_errors); + return 0; +} + +static int wandev_open(struct inode *inode, struct file *file) +{ + return single_open(file, wandev_show, PDE(inode)->data); +} + +static struct file_operations wandev_fops = { + .owner = THIS_MODULE, + .open = wandev_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, + .ioctl = wanrouter_ioctl, +}; + +/* + * Initialize router proc interface. + */ + +int __init wanrouter_proc_init(void) +{ + struct proc_dir_entry *p; + proc_router = proc_mkdir(ROUTER_NAME, proc_net); + if (!proc_router) + goto fail; + + p = create_proc_entry("config", S_IRUGO, proc_router); + if (!p) + goto fail_config; + p->proc_fops = &config_fops; + p = create_proc_entry("status", S_IRUGO, proc_router); + if (!p) + goto fail_stat; + p->proc_fops = &status_fops; + return 0; +fail_stat: + remove_proc_entry("config", proc_router); +fail_config: + remove_proc_entry(ROUTER_NAME, proc_net); +fail: + return -ENOMEM; +} + +/* + * Clean up router proc interface. + */ + +void wanrouter_proc_cleanup(void) +{ + remove_proc_entry("config", proc_router); + remove_proc_entry("status", proc_router); + remove_proc_entry(ROUTER_NAME, proc_net); +} + +/* + * Add directory entry for WAN device. + */ + +int wanrouter_proc_add(struct wan_device* wandev) +{ + if (wandev->magic != ROUTER_MAGIC) + return -EINVAL; + + wandev->dent = create_proc_entry(wandev->name, S_IRUGO, proc_router); + if (!wandev->dent) + return -ENOMEM; + wandev->dent->proc_fops = &wandev_fops; + wandev->dent->data = wandev; + return 0; +} + +/* + * Delete directory entry for WAN device. + */ +int wanrouter_proc_delete(struct wan_device* wandev) +{ + if (wandev->magic != ROUTER_MAGIC) + return -EINVAL; + remove_proc_entry(wandev->name, proc_router); + return 0; +} + +#else + +/* + * No /proc - output stubs + */ + +int __init wanrouter_proc_init(void) +{ + return 0; +} + +void wanrouter_proc_cleanup(void) +{ +} + +int wanrouter_proc_add(struct wan_device *wandev) +{ + return 0; +} + +int wanrouter_proc_delete(struct wan_device *wandev) +{ + return 0; +} + +#endif + +/* + * End + */ + |