patch-2.4.3 linux/drivers/net/wan/hdlc.c

Next file: linux/drivers/net/wan/hostess_sv11.c
Previous file: linux/drivers/net/wan/hd6457x.c
Back to the patch index
Back to the overall index

diff -u --recursive --new-file v2.4.2/linux/drivers/net/wan/hdlc.c linux/drivers/net/wan/hdlc.c
@@ -0,0 +1,1454 @@
+/*
+ * Generic HDLC support routines for Linux
+ *
+ * Copyright (C) 1999, 2000 Krzysztof Halasa <khc@pm.waw.pl>
+ *
+ * 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.
+ *
+ * Current status:
+ *    - this is work in progress
+ *    - not heavily tested on SMP
+ *    - currently supported:
+ *	* raw IP-in-HDLC
+ *	* Cisco HDLC
+ *	* Frame Relay with ANSI or CCITT LMI (both user and network side)
+ *	* PPP (using syncppp.c)
+ *	* X.25
+ *
+ * Use sethdlc utility to set line parameters, protocol and PVCs
+ */
+
+#include <linux/config.h>
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/poll.h>
+#include <linux/sched.h>
+#include <linux/errno.h>
+#include <linux/if_arp.h>
+#include <linux/init.h>
+#include <linux/skbuff.h>
+#include <linux/pkt_sched.h>
+#include <linux/inetdevice.h>
+#include <linux/lapb.h>
+#include <linux/rtnetlink.h>
+#include <linux/hdlc.h>
+
+/* #define DEBUG_PKT */
+/* #define DEBUG_HARD_HEADER */
+/* #define DEBUG_FECN */
+/* #define DEBUG_BECN */
+
+static const char* version = "HDLC support module revision 1.02 for Linux 2.4";
+
+
+#define CISCO_MULTICAST		0x8F	/* Cisco multicast address */
+#define CISCO_UNICAST		0x0F	/* Cisco unicast address */
+#define CISCO_KEEPALIVE		0x8035	/* Cisco keepalive protocol */
+#define CISCO_SYS_INFO		0x2000	/* Cisco interface/system info */
+#define CISCO_ADDR_REQ		0	/* Cisco address request */
+#define CISCO_ADDR_REPLY	1	/* Cisco address reply */
+#define CISCO_KEEPALIVE_REQ	2	/* Cisco keepalive request */
+
+static int hdlc_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd);
+
+/********************************************************
+ *
+ * Cisco HDLC support
+ *
+ *******************************************************/
+
+static int cisco_hard_header(struct sk_buff *skb, struct net_device *dev,
+			     u16 type, void *daddr, void *saddr,
+			     unsigned int len)
+{
+	hdlc_header *data;
+#ifdef DEBUG_HARD_HEADER
+	printk(KERN_DEBUG "%s: cisco_hard_header called\n", dev->name);
+#endif
+
+	skb_push(skb, sizeof(hdlc_header));
+	data = (hdlc_header*)skb->data;
+	if (type == CISCO_KEEPALIVE)
+		data->address = CISCO_MULTICAST;
+	else
+		data->address = CISCO_UNICAST;
+	data->control = 0;
+	data->protocol = htons(type);
+
+	return sizeof(hdlc_header);
+}
+
+
+
+static void cisco_keepalive_send(hdlc_device *hdlc, u32 type,
+				 u32 par1, u32 par2)
+{
+	struct sk_buff *skb;
+	cisco_packet *data;
+
+	skb = dev_alloc_skb(sizeof(hdlc_header)+sizeof(cisco_packet));
+	if (!skb) {
+		printk(KERN_WARNING "%s: Memory squeeze on cisco_keepalive_send()\n",
+			       hdlc_to_name(hdlc));
+		return;
+	}
+	skb_reserve(skb, 4);
+	cisco_hard_header(skb, hdlc_to_dev(hdlc), CISCO_KEEPALIVE,
+			  NULL, NULL, 0);
+	data = (cisco_packet*)skb->tail;
+
+	data->type = htonl(type);
+	data->par1 = htonl(par1);
+	data->par2 = htonl(par2);
+	data->rel = 0xFFFF;
+	data->time = htonl(jiffies * 1000 / HZ);
+
+	skb_put(skb, sizeof(cisco_packet));
+	skb->priority = TC_PRIO_CONTROL;
+	skb->dev = hdlc_to_dev(hdlc);
+
+	dev_queue_xmit(skb);
+}
+
+
+
+static void cisco_netif(hdlc_device *hdlc, struct sk_buff *skb)
+{
+	hdlc_header *data = (hdlc_header*)skb->data;
+	cisco_packet *cisco_data;
+	struct in_device *in_dev;
+	u32 addr, mask;
+
+	if (skb->len<sizeof(hdlc_header))
+		goto rx_error;
+
+	if (data->address != CISCO_MULTICAST &&
+	    data->address != CISCO_UNICAST)
+		goto rx_error;
+
+	skb_pull(skb, sizeof(hdlc_header));
+
+	switch(ntohs(data->protocol)) {
+	case ETH_P_IP:
+	case ETH_P_IPX:
+	case ETH_P_IPV6:
+		skb->protocol = data->protocol;
+		skb->dev = hdlc_to_dev(hdlc);
+		netif_rx(skb);
+		return;
+
+	case CISCO_SYS_INFO:
+		/* Packet is not needed, drop it. */
+		dev_kfree_skb_any(skb);
+		return;
+
+	case CISCO_KEEPALIVE:
+		if (skb->len != CISCO_PACKET_LEN &&
+		    skb->len != CISCO_BIG_PACKET_LEN) {
+			printk(KERN_INFO "%s: Invalid length of Cisco "
+			       "control packet (%d bytes)\n",
+			       hdlc_to_name(hdlc), skb->len);
+			goto rx_error;
+		}
+
+		cisco_data = (cisco_packet*)skb->data;
+
+		switch(ntohl (cisco_data->type)) {
+		case CISCO_ADDR_REQ: /* Stolen from syncppp.c :-) */
+			in_dev = hdlc_to_dev(hdlc)->ip_ptr;
+			addr = 0;
+			mask = ~0; /* is the mask correct? */
+
+			if (in_dev != NULL) {
+				struct in_ifaddr **ifap = &in_dev->ifa_list;
+
+				while (*ifap != NULL) {
+					if (strcmp(hdlc_to_name(hdlc),
+						   (*ifap)->ifa_label) == 0) {
+						addr = (*ifap)->ifa_local;
+						mask = (*ifap)->ifa_mask;
+						break;
+					}
+					ifap = &(*ifap)->ifa_next;
+				}
+
+				cisco_keepalive_send(hdlc, CISCO_ADDR_REPLY,
+						     addr, mask);
+			}
+			dev_kfree_skb_any(skb);
+			return;
+
+		case CISCO_ADDR_REPLY:
+			printk(KERN_INFO "%s: Unexpected Cisco IP address "
+			       "reply\n", hdlc_to_name(hdlc));
+			goto rx_error;
+
+		case CISCO_KEEPALIVE_REQ:
+			hdlc->lmi.rxseq = ntohl(cisco_data->par1);
+			if (ntohl(cisco_data->par2) == hdlc->lmi.txseq) {
+				hdlc->lmi.last_poll = jiffies;
+				if (!(hdlc->lmi.state & LINK_STATE_RELIABLE)) {
+					u32 sec, min, hrs, days;
+					sec = ntohl(cisco_data->time) / 1000;
+					min = sec / 60; sec -= min * 60;
+					hrs = min / 60; min -= hrs * 60;
+					days = hrs / 24; hrs -= days * 24;
+					printk(KERN_INFO "%s: Link up (peer "
+					       "uptime %ud%uh%um%us)\n",
+					       hdlc_to_name(hdlc), days, hrs,
+					       min, sec);
+				}
+				hdlc->lmi.state |= LINK_STATE_RELIABLE;
+			}
+
+			dev_kfree_skb_any(skb);
+			return;
+		} /* switch(keepalive type) */
+	} /* switch(protocol) */
+
+	printk(KERN_INFO "%s: Unsupported protocol %x\n", hdlc_to_name(hdlc),
+	       data->protocol);
+	dev_kfree_skb_any(skb);
+	return;
+
+ rx_error:
+	hdlc->stats.rx_errors++; /* Mark error */
+	dev_kfree_skb_any(skb);
+}
+
+
+
+static void cisco_timer(unsigned long arg)
+{
+	hdlc_device *hdlc = (hdlc_device*)arg;
+
+	if ((hdlc->lmi.state & LINK_STATE_RELIABLE) &&
+	    (jiffies - hdlc->lmi.last_poll >= hdlc->lmi.T392 * HZ)) {
+		hdlc->lmi.state &= ~LINK_STATE_RELIABLE;
+		printk(KERN_INFO "%s: Link down\n", hdlc_to_name(hdlc));
+	}
+
+	cisco_keepalive_send(hdlc, CISCO_KEEPALIVE_REQ, ++hdlc->lmi.txseq,
+			     hdlc->lmi.rxseq);
+	hdlc->timer.expires = jiffies + hdlc->lmi.T391*HZ;
+
+	hdlc->timer.function = cisco_timer;
+	hdlc->timer.data = arg;
+	add_timer(&hdlc->timer);
+}
+
+
+
+/******************************************************************
+ *
+ *     generic Frame Relay routines
+ *
+ *****************************************************************/
+
+
+static int fr_hard_header(struct sk_buff *skb, struct net_device *dev,
+			  u16 type, void *daddr, void *saddr, unsigned int len)
+{
+	u16 head_len;
+
+	if (!daddr)
+		daddr = dev->broadcast;
+
+#ifdef DEBUG_HARD_HEADER
+	printk(KERN_DEBUG "%s: fr_hard_header called\n", dev->name);
+#endif
+
+	switch(type) {
+	case ETH_P_IP:
+		head_len = 4;
+		skb_push(skb, head_len);
+		skb->data[3] = NLPID_IP;
+		break;
+
+	case ETH_P_IPV6:
+		head_len = 4;
+		skb_push(skb, head_len);
+		skb->data[3] = NLPID_IPV6;
+		break;
+
+	case LMI_PROTO:
+		head_len = 4;
+		skb_push(skb, head_len);
+		skb->data[3] = LMI_PROTO;
+		break;
+
+	default:
+		head_len = 10;
+		skb_push(skb, head_len);
+		skb->data[3] = FR_PAD;
+		skb->data[4] = NLPID_SNAP;
+		skb->data[5] = FR_PAD;
+		skb->data[6] = FR_PAD;
+		skb->data[7] = FR_PAD;
+		skb->data[8] = type>>8;
+		skb->data[9] = (u8)type;
+	}
+
+	memcpy(skb->data, daddr, 2);
+	skb->data[2] = FR_UI;
+
+	return head_len;
+}
+
+
+
+static inline void fr_log_dlci_active(pvc_device *pvc)
+{
+	printk(KERN_INFO "%s: %sactive%s\n", pvc_to_name(pvc),
+	       pvc->state & PVC_STATE_ACTIVE ? "" : "in",
+	       pvc->state & PVC_STATE_NEW ? " new" : "");
+}
+
+
+
+static inline u8 fr_lmi_nextseq(u8 x)
+{
+	x++;
+	return x ? x : 1;
+}
+
+
+
+static void fr_lmi_send(hdlc_device *hdlc, int fullrep)
+{
+	struct sk_buff *skb;
+	pvc_device *pvc = hdlc->first_pvc;
+	int len = mode_is(hdlc, MODE_FR_ANSI) ? LMI_ANSI_LENGTH : LMI_LENGTH;
+	int stat_len = 3;
+	u8 *data;
+	int i = 0;
+
+	if (mode_is(hdlc, MODE_DCE) && fullrep) {
+		len += hdlc->pvc_count * (2 + stat_len);
+		if (len > HDLC_MAX_MTU) {
+			printk(KERN_WARNING "%s: Too many PVCs while sending "
+			       "LMI full report\n", hdlc_to_name(hdlc));
+			return;
+		}
+	}
+
+	skb = dev_alloc_skb(len);
+	if (!skb) {
+		printk(KERN_WARNING "%s: Memory squeeze on fr_lmi_send()\n",
+			       hdlc_to_name(hdlc));
+		return;
+	}
+	memset(skb->data, 0, len);
+	skb_reserve(skb, 4);
+	fr_hard_header(skb, hdlc_to_dev(hdlc), LMI_PROTO, NULL, NULL, 0);
+	data = skb->tail;
+	data[i++] = LMI_CALLREF;
+	data[i++] = mode_is(hdlc, MODE_DCE) ? LMI_STATUS : LMI_STATUS_ENQUIRY;
+	if (mode_is(hdlc, MODE_FR_ANSI))
+		data[i++] = LMI_ANSI_LOCKSHIFT;
+	data[i++] = mode_is(hdlc, MODE_FR_CCITT) ? LMI_CCITT_REPTYPE :
+		LMI_REPTYPE;
+	data[i++] = LMI_REPT_LEN;
+	data[i++] = fullrep ? LMI_FULLREP : LMI_INTEGRITY;
+
+	data[i++] = mode_is(hdlc, MODE_FR_CCITT) ? LMI_CCITT_ALIVE : LMI_ALIVE;
+	data[i++] = LMI_INTEG_LEN;
+	data[i++] = hdlc->lmi.txseq = fr_lmi_nextseq(hdlc->lmi.txseq);
+	data[i++] = hdlc->lmi.rxseq;
+
+	if (mode_is(hdlc, MODE_DCE) && fullrep) {
+		while (pvc) {
+			data[i++] = mode_is(hdlc, MODE_FR_CCITT) ?
+				LMI_CCITT_PVCSTAT:LMI_PVCSTAT;
+			data[i++] = stat_len;
+
+			if ((hdlc->lmi.state & LINK_STATE_RELIABLE) &&
+			    (pvc->netdev.flags & IFF_UP) &&
+			    !(pvc->state & (PVC_STATE_ACTIVE|PVC_STATE_NEW))) {
+				pvc->state |= PVC_STATE_NEW;
+				fr_log_dlci_active(pvc);
+			}
+
+			dlci_to_status(hdlc, netdev_dlci(&pvc->netdev),
+				       data+i, pvc->state);
+			i += stat_len;
+			pvc = pvc->next;
+		}
+	}
+
+	skb_put(skb, i);
+	skb->priority = TC_PRIO_CONTROL;
+	skb->dev = hdlc_to_dev(hdlc);
+
+	dev_queue_xmit(skb);
+}
+
+
+
+static void fr_timer(unsigned long arg)
+{
+	hdlc_device *hdlc = (hdlc_device*)arg;
+	int i, cnt = 0, reliable;
+	u32 list;
+
+	if (mode_is(hdlc, MODE_DCE))
+		reliable = (jiffies - hdlc->lmi.last_poll < hdlc->lmi.T392*HZ);
+	else {
+		hdlc->lmi.last_errors <<= 1; /* Shift the list */
+		if (hdlc->lmi.state & LINK_STATE_REQUEST) {
+			printk(KERN_INFO "%s: No LMI status reply received\n",
+			       hdlc_to_name(hdlc));
+			hdlc->lmi.last_errors |= 1;
+		}
+
+		for (i = 0, list = hdlc->lmi.last_errors; i < hdlc->lmi.N393;
+		     i++, list >>= 1)
+			cnt += (list & 1);	/* errors count */
+
+		reliable = (cnt < hdlc->lmi.N392);
+	}
+
+	if ((hdlc->lmi.state & LINK_STATE_RELIABLE) !=
+	    (reliable ? LINK_STATE_RELIABLE : 0)) {
+		pvc_device *pvc = hdlc->first_pvc;
+
+		while (pvc) {/* Deactivate all PVCs */
+			pvc->state &= ~(PVC_STATE_NEW | PVC_STATE_ACTIVE);
+			pvc = pvc->next;
+		}
+
+		hdlc->lmi.state ^= LINK_STATE_RELIABLE;
+		printk(KERN_INFO "%s: Link %sreliable\n", hdlc_to_name(hdlc),
+		       reliable ? "" : "un");
+
+		if (reliable) {
+			hdlc->lmi.N391cnt = 0; /* Request full status */
+			hdlc->lmi.state |= LINK_STATE_CHANGED;
+		}
+	}
+
+	if (mode_is(hdlc, MODE_DCE))
+		hdlc->timer.expires = jiffies + hdlc->lmi.T392*HZ;
+	else {
+		if (hdlc->lmi.N391cnt)
+			hdlc->lmi.N391cnt--;
+
+		fr_lmi_send(hdlc, hdlc->lmi.N391cnt == 0);
+
+		hdlc->lmi.state |= LINK_STATE_REQUEST;
+		hdlc->timer.expires = jiffies + hdlc->lmi.T391*HZ;
+	}
+
+	hdlc->timer.function = fr_timer;
+	hdlc->timer.data = arg;
+	add_timer(&hdlc->timer);
+}
+
+
+
+static int fr_lmi_recv(hdlc_device *hdlc, struct sk_buff *skb)
+{
+	int stat_len;
+	pvc_device *pvc;
+	int reptype = -1, error;
+	u8 rxseq, txseq;
+	int i;
+
+	if (skb->len < (mode_is(hdlc, MODE_FR_ANSI) ?
+			LMI_ANSI_LENGTH : LMI_LENGTH)) {
+		printk(KERN_INFO "%s: Short LMI frame\n", hdlc_to_name(hdlc));
+		return 1;
+	}
+
+	if (skb->data[5] != (!mode_is(hdlc, MODE_DCE) ?
+			     LMI_STATUS : LMI_STATUS_ENQUIRY)) {
+		printk(KERN_INFO "%s: LMI msgtype=%x, Not LMI status %s\n",
+		       hdlc_to_name(hdlc), skb->data[2],
+		       mode_is(hdlc, MODE_DCE) ? "enquiry" : "reply");
+		return 1;
+	}
+
+	i = mode_is(hdlc, MODE_FR_ANSI) ? 7 : 6;
+
+	if (skb->data[i] !=
+	    (mode_is(hdlc, MODE_FR_CCITT) ? LMI_CCITT_REPTYPE : LMI_REPTYPE)) {
+		printk(KERN_INFO "%s: Not a report type=%x\n",
+		       hdlc_to_name(hdlc), skb->data[i]);
+		return 1;
+	}
+	i++;
+
+	i++;				/* Skip length field */
+
+	reptype = skb->data[i++];
+
+	if (skb->data[i]!=
+	    (mode_is(hdlc, MODE_FR_CCITT) ? LMI_CCITT_ALIVE : LMI_ALIVE)) {
+		printk(KERN_INFO "%s: Unsupported status element=%x\n",
+		       hdlc_to_name(hdlc), skb->data[i]);
+		return 1;
+	}
+	i++;
+
+	i++;			/* Skip length field */
+
+	hdlc->lmi.rxseq = skb->data[i++]; /* TX sequence from peer */
+	rxseq = skb->data[i++];	/* Should confirm our sequence */
+
+	txseq = hdlc->lmi.txseq;
+
+	if (mode_is(hdlc, MODE_DCE)) {
+		if (reptype != LMI_FULLREP && reptype != LMI_INTEGRITY) {
+			printk(KERN_INFO "%s: Unsupported report type=%x\n",
+			       hdlc_to_name(hdlc), reptype);
+			return 1;
+		}
+	}
+
+	error = 0;
+	if (!(hdlc->lmi.state & LINK_STATE_RELIABLE))
+		error = 1;
+
+	if (rxseq == 0 || rxseq != txseq) {
+		hdlc->lmi.N391cnt = 0; /* Ask for full report next time */
+		error = 1;
+	}
+
+	if (mode_is(hdlc, MODE_DCE)) {
+		if ((hdlc->lmi.state & LINK_STATE_FULLREP_SENT) && !error) {
+/* Stop sending full report - the last one has been confirmed by DTE */
+			hdlc->lmi.state &= ~LINK_STATE_FULLREP_SENT;
+			pvc = hdlc->first_pvc;
+			while (pvc) {
+				if (pvc->state & PVC_STATE_NEW) {
+					pvc->state &= ~PVC_STATE_NEW;
+					pvc->state |= PVC_STATE_ACTIVE;
+					fr_log_dlci_active(pvc);
+
+/* Tell DTE that new PVC is now active */
+					hdlc->lmi.state |= LINK_STATE_CHANGED;
+				}
+				pvc = pvc->next;
+			}
+		}
+
+		if (hdlc->lmi.state & LINK_STATE_CHANGED) {
+			reptype = LMI_FULLREP;
+			hdlc->lmi.state |= LINK_STATE_FULLREP_SENT;
+			hdlc->lmi.state &= ~LINK_STATE_CHANGED;
+		}
+
+		fr_lmi_send(hdlc, reptype == LMI_FULLREP ? 1 : 0);
+		return 0;
+	}
+
+	/* DTE */
+
+	if (reptype != LMI_FULLREP || error)
+		return 0;
+
+	stat_len = 3;
+	pvc = hdlc->first_pvc;
+
+	while (pvc) {
+		pvc->newstate = 0;
+		pvc = pvc->next;
+	}
+
+	while (skb->len >= i + 2 + stat_len) {
+		u16 dlci;
+		u8 state = 0;
+
+		if (skb->data[i] != (mode_is(hdlc, MODE_FR_CCITT) ?
+				     LMI_CCITT_PVCSTAT : LMI_PVCSTAT)) {
+			printk(KERN_WARNING "%s: Invalid PVCSTAT ID: %x\n",
+			       hdlc_to_name(hdlc), skb->data[i]);
+			return 1;
+		}
+		i++;
+
+		if (skb->data[i] != stat_len) {
+			printk(KERN_WARNING "%s: Invalid PVCSTAT length: %x\n",
+			       hdlc_to_name(hdlc), skb->data[i]);
+			return 1;
+		}
+		i++;
+
+		dlci = status_to_dlci(hdlc, skb->data+i, &state);
+		pvc = find_pvc(hdlc, dlci);
+
+		if (pvc)
+			pvc->newstate = state;
+		else if (state == PVC_STATE_NEW)
+			printk(KERN_INFO "%s: new PVC available, DLCI=%u\n",
+			       hdlc_to_name(hdlc), dlci);
+
+		i += stat_len;
+	}
+
+	pvc = hdlc->first_pvc;
+
+	while (pvc) {
+		if (pvc->newstate == PVC_STATE_NEW)
+			pvc->newstate = PVC_STATE_ACTIVE;
+
+		pvc->newstate |= (pvc->state &
+				  ~(PVC_STATE_NEW|PVC_STATE_ACTIVE));
+		if (pvc->state != pvc->newstate) {
+			pvc->state = pvc->newstate;
+			fr_log_dlci_active(pvc);
+		}
+		pvc = pvc->next;
+	}
+
+	/* Next full report after N391 polls */
+	hdlc->lmi.N391cnt = hdlc->lmi.N391;
+
+	return 0;
+}
+
+
+
+static void fr_netif(hdlc_device *hdlc, struct sk_buff *skb)
+{
+	fr_hdr *fh = (fr_hdr*)skb->data;
+	u8 *data = skb->data;
+	u16 dlci;
+	pvc_device *pvc;
+
+	if (skb->len<4 || fh->ea1 || data[2] != FR_UI)
+		goto rx_error;
+
+	dlci = q922_to_dlci(skb->data);
+
+	if (dlci == LMI_DLCI) {
+		if (data[3] == LMI_PROTO) {
+			if (fr_lmi_recv(hdlc, skb))
+				goto rx_error;
+			else {
+				/* No request pending */
+				hdlc->lmi.state &= ~LINK_STATE_REQUEST;
+				hdlc->lmi.last_poll = jiffies;
+				dev_kfree_skb_any(skb);
+				return;
+			}
+		}
+
+		printk(KERN_INFO "%s: Received non-LMI frame with LMI DLCI\n",
+		       hdlc_to_name(hdlc));
+		goto rx_error;
+	}
+
+	pvc = find_pvc(hdlc, dlci);
+	if (!pvc) {
+#ifdef DEBUG_PKT
+		printk(KERN_INFO "%s: No PVC for received frame's DLCI %d\n",
+		       hdlc_to_name(hdlc), dlci);
+#endif
+		goto rx_error;
+	}
+
+	if ((pvc->netdev.flags & IFF_UP) == 0) {
+#ifdef DEBUG_PKT
+		printk(KERN_INFO "%s: PVC for received frame's DLCI %d is down\n",
+		       hdlc_to_name(hdlc), dlci);
+#endif
+		goto rx_error;
+	}
+
+	pvc->stats.rx_packets++; /* PVC traffic */
+	pvc->stats.rx_bytes += skb->len;
+
+	if ((pvc->state & PVC_STATE_FECN) != (fh->fecn ? PVC_STATE_FECN : 0)) {
+#ifdef DEBUG_FECN
+		printk(KERN_DEBUG "%s: FECN O%s\n", pvc_to_name(pvc),
+		       fh->fecn ? "N" : "FF");
+#endif
+		pvc->state ^= PVC_STATE_FECN;
+	}
+
+	if ((pvc->state & PVC_STATE_BECN) != (fh->becn ? PVC_STATE_BECN : 0)) {
+#ifdef DEBUG_FECN
+		printk(KERN_DEBUG "%s: BECN O%s\n", pvc_to_name(pvc),
+		       fh->becn ? "N" : "FF");
+#endif
+		pvc->state ^= PVC_STATE_BECN;
+	}
+
+	if (pvc->state & PVC_STATE_BECN)
+		pvc->stats.rx_compressed++;
+
+	if (data[3] == NLPID_IP) {
+		skb_pull(skb, 4); /* Remove 4-byte header (hdr, UI, NLPID) */
+		skb->protocol = htons(ETH_P_IP);
+		skb->dev = &pvc->netdev;
+		netif_rx(skb);
+		return;
+	}
+
+
+	if (data[3] == NLPID_IPV6) {
+		skb_pull(skb, 4); /* Remove 4-byte header (hdr, UI, NLPID) */
+		skb->protocol = htons(ETH_P_IPV6);
+		skb->dev = &pvc->netdev;
+		netif_rx(skb);
+		return;
+	}
+
+	if (data[3] == FR_PAD && data[4] == NLPID_SNAP && data[5] == FR_PAD &&
+	    data[6] == FR_PAD && data[7] == FR_PAD &&
+	    ((data[8]<<8) | data[9]) == ETH_P_ARP) {
+		skb_pull(skb, 10);
+		skb->protocol = htons(ETH_P_ARP);
+		skb->dev = &pvc->netdev;
+		netif_rx(skb);
+		return;
+	}
+
+	printk(KERN_INFO "%s: Unusupported protocol %x\n",
+	       hdlc_to_name(hdlc), data[3]);
+	dev_kfree_skb_any(skb);
+	return;
+
+ rx_error:
+	hdlc->stats.rx_errors++; /* Mark error */
+	dev_kfree_skb_any(skb);
+}
+
+
+
+static void fr_cisco_open(hdlc_device *hdlc)
+{
+	hdlc->lmi.state = LINK_STATE_CHANGED;
+	hdlc->lmi.txseq = hdlc->lmi.rxseq = 0;
+	hdlc->lmi.last_errors = 0xFFFFFFFF;
+	hdlc->lmi.N391cnt = 0;
+
+	init_timer(&hdlc->timer);
+	hdlc->timer.expires = jiffies + HZ; /* First poll after 1 second */
+	hdlc->timer.function = mode_is(hdlc, MODE_FR) ? fr_timer : cisco_timer;
+	hdlc->timer.data = (unsigned long)hdlc;
+	add_timer(&hdlc->timer);
+}
+
+
+
+static void fr_cisco_close(hdlc_device *hdlc)
+{
+	pvc_device *pvc = hdlc->first_pvc;
+
+	del_timer_sync(&hdlc->timer);
+
+	while(pvc) {		/* NULL in Cisco mode */
+		dev_close(&pvc->netdev); /* Shutdown all PVCs for this FRAD */
+		pvc = pvc->next;
+	}
+}
+
+
+
+/******************************************************************
+ *
+ *     generic HDLC routines
+ *
+ *****************************************************************/
+
+
+
+static int hdlc_change_mtu(struct net_device *dev, int new_mtu)
+{
+	if ((new_mtu < 68) || (new_mtu > HDLC_MAX_MTU))
+		return -EINVAL;
+	dev->mtu = new_mtu;
+	return 0;
+}
+
+
+
+/********************************************************
+ *
+ * PVC device routines
+ *
+ *******************************************************/
+
+static int pvc_open(struct net_device *dev)
+{
+	pvc_device *pvc = dev_to_pvc(dev);
+	int result = 0;
+
+	if ((hdlc_to_dev(pvc->master)->flags & IFF_UP) == 0)
+		return -EIO;  /* Master must be UP in order to activate PVC */
+
+	memset(&(pvc->stats), 0, sizeof(struct net_device_stats));
+	pvc->state = 0;
+
+	if (!mode_is(pvc->master, MODE_SOFT) && pvc->master->open_pvc)
+		result = pvc->master->open_pvc(pvc);
+	if (result)
+		return result;
+
+	pvc->master->lmi.state |= LINK_STATE_CHANGED;
+	return 0;
+}
+
+
+
+static int pvc_close(struct net_device *dev)
+{
+	pvc_device *pvc = dev_to_pvc(dev);
+	pvc->state = 0;
+
+	if (!mode_is(pvc->master, MODE_SOFT) && pvc->master->close_pvc)
+		pvc->master->close_pvc(pvc);
+
+	pvc->master->lmi.state |= LINK_STATE_CHANGED;
+	return 0;
+}
+
+
+
+static int pvc_xmit(struct sk_buff *skb, struct net_device *dev)
+{
+	pvc_device *pvc = dev_to_pvc(dev);
+
+	if (pvc->state & PVC_STATE_ACTIVE) {
+		skb->dev = hdlc_to_dev(pvc->master);
+		pvc->stats.tx_bytes += skb->len;
+		pvc->stats.tx_packets++;
+		if (pvc->state & PVC_STATE_FECN)
+			pvc->stats.tx_compressed++; /* TX Congestion counter */
+		dev_queue_xmit(skb);
+	} else {
+		pvc->stats.tx_dropped++;
+		dev_kfree_skb(skb);
+	}
+
+	return 0;
+}
+
+
+
+static struct net_device_stats *pvc_get_stats(struct net_device *dev)
+{
+	pvc_device *pvc = dev_to_pvc(dev);
+	return &pvc->stats;
+}
+
+
+
+static int pvc_change_mtu(struct net_device *dev, int new_mtu)
+{
+	if ((new_mtu < 68) || (new_mtu > HDLC_MAX_MTU))
+		return -EINVAL;
+	dev->mtu = new_mtu;
+	return 0;
+}
+
+
+
+static void destroy_pvc_list(hdlc_device *hdlc)
+{
+	pvc_device *pvc = hdlc->first_pvc;
+	while(pvc) {
+		pvc_device *next = pvc->next;
+		unregister_netdevice(&pvc->netdev);
+		kfree(pvc);
+		pvc = next;
+	}
+
+	hdlc->first_pvc = NULL;	/* All PVCs destroyed */
+	hdlc->pvc_count = 0;
+	hdlc->lmi.state |= LINK_STATE_CHANGED;
+}
+
+
+
+/********************************************************
+ *
+ * X.25 protocol support routines
+ *
+ *******************************************************/
+
+#ifdef CONFIG_HDLC_X25
+/* These functions are callbacks called by LAPB layer */
+
+void x25_connect_disconnect(void *token, int reason, int code)
+{
+	hdlc_device *hdlc = token;
+	struct sk_buff *skb;
+	unsigned char *ptr;
+
+	if ((skb = dev_alloc_skb(1)) == NULL) {
+		printk(KERN_ERR "%s: out of memory\n", hdlc_to_name(hdlc));
+		return;
+	}
+
+	ptr = skb_put(skb, 1);
+	*ptr = code;
+
+	skb->dev = hdlc_to_dev(hdlc);
+	skb->protocol = htons(ETH_P_X25);
+	skb->mac.raw = skb->data;
+	skb->pkt_type = PACKET_HOST;
+
+	netif_rx(skb);
+}
+
+void x25_connected(void *token, int reason)
+{
+	x25_connect_disconnect(token, reason, 1);
+}
+
+void x25_disconnected(void *token, int reason)
+{
+	x25_connect_disconnect(token, reason, 2);
+}
+
+
+int x25_data_indication(void *token, struct sk_buff *skb)
+{
+	hdlc_device *hdlc = token;
+	unsigned char *ptr;
+
+	ptr = skb_push(skb, 1);
+	*ptr = 0;
+
+	skb->dev = hdlc_to_dev(hdlc);
+	skb->protocol = htons(ETH_P_X25);
+	skb->mac.raw = skb->data;
+	skb->pkt_type = PACKET_HOST;
+
+	return netif_rx(skb);
+}
+
+
+void x25_data_transmit(void *token, struct sk_buff *skb)
+{
+	hdlc_device *hdlc = token;
+	hdlc->xmit(hdlc, skb);	/* Ignore return value :-( */
+}
+#endif /* CONFIG_HDLC_X25 */
+
+
+/********************************************************
+ *
+ * HDLC device routines
+ *
+ *******************************************************/
+
+static int hdlc_open(struct net_device *dev)
+{
+	hdlc_device *hdlc = dev_to_hdlc(dev);
+	int result;
+
+	if (hdlc->mode == MODE_NONE)
+		return -ENOSYS;
+
+	memset(&(hdlc->stats), 0, sizeof(struct net_device_stats));
+
+	if (mode_is(hdlc, MODE_FR | MODE_SOFT) ||
+	    mode_is(hdlc, MODE_CISCO | MODE_SOFT))
+		fr_cisco_open(hdlc);
+#ifdef CONFIG_HDLC_PPP
+	else if (mode_is(hdlc, MODE_PPP | MODE_SOFT)) {
+		sppp_attach(&hdlc->pppdev);
+		/* sppp_attach nukes them. We don't need syncppp's ioctl */
+		dev->do_ioctl = hdlc_ioctl;
+		hdlc->pppdev.sppp.pp_flags &= ~PP_CISCO;
+		dev->type = ARPHRD_PPP;
+		result = sppp_open(dev);
+		if (result) {
+			sppp_detach(dev);
+			return result;
+		}
+	}
+#endif
+#ifdef CONFIG_HDLC_X25
+	else if (mode_is(hdlc, MODE_X25)) {
+		struct lapb_register_struct cb;
+
+		cb.connect_confirmation = x25_connected;
+		cb.connect_indication = x25_connected;
+		cb.disconnect_confirmation = x25_disconnected;
+		cb.disconnect_indication = x25_disconnected;
+		cb.data_indication = x25_data_indication;
+		cb.data_transmit = x25_data_transmit;
+
+		result = lapb_register(hdlc, &cb);
+		if (result != LAPB_OK)
+			return result;
+	}
+#endif
+	result = hdlc->open(hdlc);
+	if (result) {
+		if (mode_is(hdlc, MODE_FR | MODE_SOFT) ||
+		    mode_is(hdlc, MODE_CISCO | MODE_SOFT))
+			fr_cisco_close(hdlc);
+#ifdef CONFIG_HDLC_PPP
+		else if (mode_is(hdlc, MODE_PPP | MODE_SOFT)) {
+			sppp_close(dev);
+			sppp_detach(dev);
+			dev->rebuild_header = NULL;
+			dev->change_mtu = hdlc_change_mtu;
+			dev->mtu = HDLC_MAX_MTU;
+			dev->hard_header_len = 16;
+		}
+#endif
+#ifdef CONFIG_HDLC_X25
+		else if (mode_is(hdlc, MODE_X25))
+			lapb_unregister(hdlc);
+#endif
+	}
+
+	return result;
+}
+
+
+
+static int hdlc_close(struct net_device *dev)
+{
+	hdlc_device *hdlc = dev_to_hdlc(dev);
+
+	hdlc->close(hdlc);
+
+	if (mode_is(hdlc, MODE_FR | MODE_SOFT) ||
+	    mode_is(hdlc, MODE_CISCO | MODE_SOFT))
+		fr_cisco_close(hdlc);
+#ifdef CONFIG_HDLC_PPP
+	else if (mode_is(hdlc, MODE_PPP | MODE_SOFT)) {
+		sppp_close(dev);
+		sppp_detach(dev);
+		dev->rebuild_header = NULL;
+		dev->change_mtu = hdlc_change_mtu;
+		dev->mtu = HDLC_MAX_MTU;
+		dev->hard_header_len = 16;
+	}
+#endif
+#ifdef CONFIG_HDLC_X25
+	else if (mode_is(hdlc, MODE_X25))
+		lapb_unregister(hdlc);
+#endif
+	return 0;
+}
+
+
+
+static int hdlc_xmit(struct sk_buff *skb, struct net_device *dev)
+{
+	hdlc_device *hdlc = dev_to_hdlc(dev);
+
+#ifdef CONFIG_HDLC_X25
+	if (mode_is(hdlc, MODE_X25 | MODE_SOFT)) {
+		int result;
+
+
+		/* X.25 to LAPB */
+		switch (skb->data[0]) {
+		case 0:		/* Data to be transmitted */
+			skb_pull(skb, 1);
+			if ((result = lapb_data_request(hdlc, skb)) != LAPB_OK)
+				dev_kfree_skb(skb);
+			return 0;
+
+		case 1:
+			if ((result = lapb_connect_request(hdlc))!= LAPB_OK) {
+				if (result == LAPB_CONNECTED) {
+				/* Send connect confirm. msg to level 3 */
+					x25_connected(hdlc, 0);
+				} else {
+					printk(KERN_ERR "%s: LAPB connect "
+					       "request failed, error code = "
+					       "%i\n", hdlc_to_name(hdlc),
+					       result);
+				}
+			}
+			break;
+
+		case 2:
+			if ((result=lapb_disconnect_request(hdlc))!=LAPB_OK) {
+				if (result == LAPB_NOTCONNECTED) {
+				/* Send disconnect confirm. msg to level 3 */
+					x25_disconnected(hdlc, 0);
+				} else {
+					printk(KERN_ERR "%s: LAPB disconnect "
+					       "request failed, error code = "
+					       "%i\n", hdlc_to_name(hdlc),
+					       result);
+				}
+			}
+			break;
+
+		default:		/* to be defined */
+		}
+
+		dev_kfree_skb(skb);
+		return 0;
+	} /* MODE_X25 */
+#endif /* CONFIG_HDLC_X25 */
+
+	return hdlc->xmit(hdlc, skb);
+}
+
+
+
+void hdlc_netif_rx(hdlc_device *hdlc, struct sk_buff *skb)
+{
+/* skb contains raw HDLC frame, in both hard- and software modes */
+	skb->mac.raw = skb->data;
+
+	switch(hdlc->mode & MODE_MASK) {
+	case MODE_HDLC:
+		skb->protocol = htons(ETH_P_IP);
+		skb->dev = hdlc_to_dev(hdlc);
+		netif_rx(skb);
+		return;
+
+	case MODE_FR:
+		fr_netif(hdlc, skb);
+		return;
+
+	case MODE_CISCO:
+		cisco_netif(hdlc, skb);
+		return;
+
+#ifdef CONFIG_HDLC_PPP
+	case MODE_PPP:
+#if 0
+		sppp_input(hdlc_to_dev(hdlc), skb);
+#else
+		skb->protocol = htons(ETH_P_WAN_PPP);
+		skb->dev = hdlc_to_dev(hdlc);
+		netif_rx(skb);
+#endif
+		return;
+#endif
+#ifdef CONFIG_HDLC_X25
+	case MODE_X25:
+		skb->dev = hdlc_to_dev(hdlc);
+		if (lapb_data_received(hdlc, skb) == LAPB_OK)
+			return;
+		break;
+#endif
+	}
+
+	hdlc->stats.rx_errors++;
+	dev_kfree_skb_any(skb);
+}
+
+
+
+static struct net_device_stats *hdlc_get_stats(struct net_device *dev)
+{
+	return &dev_to_hdlc(dev)->stats;
+}
+
+
+
+static int hdlc_set_mode(hdlc_device *hdlc, int mode)
+{
+	int result = -1;	/* Default to soft modes */
+	struct net_device *dev = hdlc_to_dev(hdlc);
+
+	if(!capable(CAP_NET_ADMIN))
+		return -EPERM;
+
+	if(dev->flags & IFF_UP)
+		return -EBUSY;
+
+	dev->addr_len = 0;
+	dev->hard_header = NULL;
+	hdlc->mode = MODE_NONE;
+
+	if (!(mode & MODE_SOFT))
+		switch(mode & MODE_MASK) {
+		case MODE_HDLC:
+			result = hdlc->set_mode ?
+				hdlc->set_mode(hdlc, MODE_HDLC) : 0;
+			break;
+
+		case MODE_CISCO: /* By card */
+#ifdef CONFIG_HDLC_PPP
+		case MODE_PPP:
+#endif
+#ifdef CONFIG_HDLC_X25
+		case MODE_X25:
+#endif
+		case MODE_FR:
+			result = hdlc->set_mode ?
+				hdlc->set_mode(hdlc, mode) : -ENOSYS;
+			break;
+
+		default:
+			return -EINVAL;
+		}
+
+	if (result) {
+		mode |= MODE_SOFT; /* Try "host software" protocol */
+
+		switch(mode & MODE_MASK) {
+		case MODE_CISCO:
+			dev->hard_header = cisco_hard_header;
+			break;
+
+#ifdef CONFIG_HDLC_PPP
+		case MODE_PPP:
+			break;
+#endif
+#ifdef CONFIG_HDLC_X25
+		case MODE_X25:
+			break;
+#endif
+
+		case MODE_FR:
+			dev->hard_header = fr_hard_header;
+			dev->addr_len = 2;
+			*(u16*)dev->dev_addr = htons(LMI_DLCI);
+			dlci_to_q922(dev->broadcast, LMI_DLCI);
+			break;
+
+		default:
+			return -EINVAL;
+		}
+
+		result = hdlc->set_mode ?
+			hdlc->set_mode(hdlc, MODE_HDLC) : 0;
+	}
+
+	if (result)
+		return result;
+
+	hdlc->mode = mode;
+	switch(mode & MODE_MASK) {
+#ifdef CONFIG_HDLC_PPP
+	case MODE_PPP:   dev->type = ARPHRD_PPP;   break;
+#endif
+#ifdef CONFIG_HDLC_X25
+	case MODE_X25:   dev->type = ARPHRD_X25;   break;
+#endif
+	case MODE_FR:    dev->type = ARPHRD_FRAD;  break;
+	case MODE_CISCO: dev->type = ARPHRD_CISCO; break;
+	default:         dev->type = ARPHRD_RAWHDLC;
+	}
+
+	memset(&(hdlc->stats), 0, sizeof(struct net_device_stats));
+	destroy_pvc_list(hdlc);
+	return 0;
+}
+
+
+
+static int hdlc_fr_pvc(hdlc_device *hdlc, int dlci)
+{
+	pvc_device **pvc_p = &hdlc->first_pvc;
+	pvc_device *pvc;
+	int result, create = 1;	/* Create or delete PVC */
+
+	if(!capable(CAP_NET_ADMIN))
+		return -EPERM;
+
+	if(dlci<0) {
+		dlci = -dlci;
+		create = 0;
+	}
+
+	if(dlci <= 0 || dlci >= 1024)
+		return -EINVAL;	/* Only 10 bits for DLCI, DLCI=0 is reserved */
+
+	if(!mode_is(hdlc, MODE_FR))
+		return -EINVAL;	/* Only meaningfull on FR */
+
+	while(*pvc_p) {
+		if (netdev_dlci(&(*pvc_p)->netdev) == dlci)
+			break;
+		pvc_p = &(*pvc_p)->next;
+	}
+
+	if (create) {		/* Create PVC */
+		if (*pvc_p != NULL)
+			return -EEXIST;
+
+		pvc = *pvc_p = kmalloc(sizeof(pvc_device), GFP_KERNEL);
+		if (!pvc) {
+			printk(KERN_WARNING "%s: Memory squeeze on "
+			       "hdlc_fr_pvc()\n", hdlc_to_name(hdlc));
+			return -ENOBUFS;
+		}
+		memset(pvc, 0, sizeof(pvc_device));
+
+		pvc->netdev.hard_start_xmit = pvc_xmit;
+		pvc->netdev.get_stats = pvc_get_stats;
+		pvc->netdev.open = pvc_open;
+		pvc->netdev.stop = pvc_close;
+		pvc->netdev.change_mtu = pvc_change_mtu;
+		pvc->netdev.mtu = HDLC_MAX_MTU;
+
+		pvc->netdev.type = ARPHRD_DLCI;
+		pvc->netdev.hard_header_len = 16;
+		pvc->netdev.hard_header = fr_hard_header;
+		pvc->netdev.tx_queue_len = 0;
+		pvc->netdev.flags = IFF_POINTOPOINT;
+
+		dev_init_buffers(&pvc->netdev);
+
+		pvc->master = hdlc;
+		*(u16*)pvc->netdev.dev_addr = htons(dlci);
+		dlci_to_q922(pvc->netdev.broadcast, dlci);
+		pvc->netdev.addr_len = 2;
+		pvc->netdev.irq = hdlc_to_dev(hdlc)->irq;
+
+		result = dev_alloc_name(&pvc->netdev, "pvc%d");
+		if (result < 0) {
+			kfree(pvc);
+			*pvc_p = NULL;
+			return result;
+		}
+
+		if (register_netdevice(&pvc->netdev) != 0) {
+			kfree(pvc);
+			*pvc_p = NULL;
+			return -EIO;
+		}
+
+		if (!mode_is(hdlc, MODE_SOFT) && hdlc->create_pvc) {
+			result = hdlc->create_pvc(pvc);
+			if (result) {
+				unregister_netdevice(&pvc->netdev);
+				kfree(pvc);
+				*pvc_p = NULL;
+				return result;
+			}
+		}
+
+		hdlc->lmi.state |= LINK_STATE_CHANGED;
+		hdlc->pvc_count++;
+		return 0;
+	}
+
+	if (*pvc_p == NULL)		/* Delete PVC */
+		return -ENOENT;
+
+	pvc = *pvc_p;
+
+	if (pvc->netdev.flags & IFF_UP)
+		return -EBUSY;		/* PVC in use */
+
+	if (!mode_is(hdlc, MODE_SOFT) && hdlc->destroy_pvc)
+		hdlc->destroy_pvc(pvc);
+
+	hdlc->lmi.state |= LINK_STATE_CHANGED;
+	hdlc->pvc_count--;
+	*pvc_p = pvc->next;
+	unregister_netdevice(&pvc->netdev);
+	kfree(pvc);
+	return 0;
+}
+
+
+
+static int hdlc_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
+{
+	hdlc_device *hdlc = dev_to_hdlc(dev);
+
+	switch(cmd) {
+	case HDLCGMODE:
+		ifr->ifr_ifru.ifru_ivalue = hdlc->mode;
+		return 0;
+
+	case HDLCSMODE:
+		return hdlc_set_mode(hdlc, ifr->ifr_ifru.ifru_ivalue);
+
+	case HDLCPVC:
+		return hdlc_fr_pvc(hdlc, ifr->ifr_ifru.ifru_ivalue);
+
+	default:
+		if (hdlc->ioctl != NULL)
+			return hdlc->ioctl(hdlc, ifr, cmd);
+	}
+
+	return -EINVAL;
+}
+
+
+
+static int hdlc_init(struct net_device *dev)
+{
+	hdlc_device *hdlc = dev_to_hdlc(dev);
+
+	memset(&(hdlc->stats), 0, sizeof(struct net_device_stats));
+
+	dev->get_stats = hdlc_get_stats;
+	dev->open = hdlc_open;
+	dev->stop = hdlc_close;
+	dev->hard_start_xmit = hdlc_xmit;
+	dev->do_ioctl = hdlc_ioctl;
+	dev->change_mtu = hdlc_change_mtu;
+	dev->mtu = HDLC_MAX_MTU;
+
+	dev->type = ARPHRD_RAWHDLC;
+	dev->hard_header_len = 16;
+
+	dev->flags = IFF_POINTOPOINT | IFF_NOARP;
+
+	dev_init_buffers(dev);
+	return 0;
+}
+
+
+
+int register_hdlc_device(hdlc_device *hdlc)
+{
+	int result;
+	struct net_device *dev = hdlc_to_dev(hdlc);
+
+	dev->init = hdlc_init;
+	dev->priv = &hdlc->syncppp_ptr;
+	hdlc->syncppp_ptr = &hdlc->pppdev;
+	hdlc->pppdev.dev = dev;
+	hdlc->mode = MODE_NONE;
+	hdlc->lmi.T391 = 10;	/* polling verification timer */
+	hdlc->lmi.T392 = 15;	/* link integrity verification polling timer */
+	hdlc->lmi.N391 = 6;	/* full status polling counter */
+	hdlc->lmi.N392 = 3;	/* error threshold */
+	hdlc->lmi.N393 = 4;	/* monitored events count */
+
+	result = dev_alloc_name(dev, "hdlc%d");
+	if (result<0)
+		return result;
+
+	result = register_netdev(dev);
+	if (result != 0)
+		return -EIO;
+
+	dev_init_buffers(dev);
+	MOD_INC_USE_COUNT;
+	return 0;
+}
+
+
+
+void unregister_hdlc_device(hdlc_device *hdlc)
+{
+	destroy_pvc_list(hdlc);
+	unregister_netdev(hdlc_to_dev(hdlc));
+	MOD_DEC_USE_COUNT;
+}
+
+MODULE_AUTHOR("Krzysztof Halasa <khc@pm.waw.pl>");
+MODULE_DESCRIPTION("HDLC support module");
+
+EXPORT_SYMBOL(hdlc_netif_rx);
+EXPORT_SYMBOL(register_hdlc_device);
+EXPORT_SYMBOL(unregister_hdlc_device);
+
+static int __init hdlc_module_init(void)
+{
+	printk(KERN_INFO "%s\n", version);
+	return 0;
+}
+
+
+module_init(hdlc_module_init);

FUNET's LINUX-ADM group, linux-adm@nic.funet.fi
TCL-scripts by Sam Shen (who was at: slshen@lbl.gov)