[deliverable/linux.git] / net / core / netpoll.c

/*
 * Common framework for low-level network console, dump, and debugger code
 *
 * Sep 8 2003  Matt Mackall <mpm@selenic.com>
 *
 * based on the netconsole code from:
 *
 * Copyright (C) 2001  Ingo Molnar <mingo@redhat.com>
 * Copyright (C) 2002  Red Hat, Inc.
 */

#include <linux/smp_lock.h>
#include <linux/netdevice.h>
#include <linux/etherdevice.h>
#include <linux/string.h>
#include <linux/inetdevice.h>
#include <linux/inet.h>
#include <linux/interrupt.h>
#include <linux/netpoll.h>
#include <linux/sched.h>
#include <linux/delay.h>
#include <linux/rcupdate.h>
#include <linux/workqueue.h>
#include <net/tcp.h>
#include <net/udp.h>
#include <asm/unaligned.h>

/*
 * We maintain a small pool of fully-sized skbs, to make sure the
 * message gets out even in extreme OOM situations.
 */

#define MAX_UDP_CHUNK 1460
#define MAX_SKBS 32
#define MAX_QUEUE_DEPTH (MAX_SKBS / 2)

static DEFINE_SPINLOCK(skb_list_lock);
static int nr_skbs;
static struct sk_buff *skbs;

static DEFINE_SPINLOCK(queue_lock);
static int queue_depth;
static struct sk_buff *queue_head, *queue_tail;

static atomic_t trapped;

#define NETPOLL_RX_ENABLED  1
#define NETPOLL_RX_DROP     2

#define MAX_SKB_SIZE \
		(MAX_UDP_CHUNK + sizeof(struct udphdr) + \
				sizeof(struct iphdr) + sizeof(struct ethhdr))

static void zap_completion_queue(void);

static void queue_process(void *p)
{
	unsigned long flags;
	struct sk_buff *skb;

	while (queue_head) {
		spin_lock_irqsave(&queue_lock, flags);

		skb = queue_head;
		queue_head = skb->next;
		if (skb == queue_tail)
			queue_head = NULL;

		queue_depth--;

		spin_unlock_irqrestore(&queue_lock, flags);

		dev_queue_xmit(skb);
	}
}

static DECLARE_WORK(send_queue, queue_process, NULL);

void netpoll_queue(struct sk_buff *skb)
{
	unsigned long flags;

	if (queue_depth == MAX_QUEUE_DEPTH) {
		__kfree_skb(skb);
		return;
	}

	spin_lock_irqsave(&queue_lock, flags);
	if (!queue_head)
		queue_head = skb;
	else
		queue_tail->next = skb;
	queue_tail = skb;
	queue_depth++;
	spin_unlock_irqrestore(&queue_lock, flags);

	schedule_work(&send_queue);
}

static int checksum_udp(struct sk_buff *skb, struct udphdr *uh,
			     unsigned short ulen, u32 saddr, u32 daddr)
{
	if (uh->check == 0)
		return 0;

	if (skb->ip_summed == CHECKSUM_HW)
		return csum_tcpudp_magic(
			saddr, daddr, ulen, IPPROTO_UDP, skb->csum);

	skb->csum = csum_tcpudp_nofold(saddr, daddr, ulen, IPPROTO_UDP, 0);

	return csum_fold(skb_checksum(skb, 0, skb->len, skb->csum));
}

/*
 * Check whether delayed processing was scheduled for our NIC. If so,
 * we attempt to grab the poll lock and use ->poll() to pump the card.
 * If this fails, either we've recursed in ->poll() or it's already
 * running on another CPU.
 *
 * Note: we don't mask interrupts with this lock because we're using
 * trylock here and interrupts are already disabled in the softirq
 * case. Further, we test the poll_owner to avoid recursion on UP
 * systems where the lock doesn't exist.
 *
 * In cases where there is bi-directional communications, reading only
 * one message at a time can lead to packets being dropped by the
 * network adapter, forcing superfluous retries and possibly timeouts.
 * Thus, we set our budget to greater than 1.
 */
static void poll_napi(struct netpoll *np)
{
	struct netpoll_info *npinfo = np->dev->npinfo;
	int budget = 16;

	if (test_bit(__LINK_STATE_RX_SCHED, &np->dev->state) &&
	    npinfo->poll_owner != smp_processor_id() &&
	    spin_trylock(&npinfo->poll_lock)) {
		npinfo->rx_flags |= NETPOLL_RX_DROP;
		atomic_inc(&trapped);

		np->dev->poll(np->dev, &budget);

		atomic_dec(&trapped);
		npinfo->rx_flags &= ~NETPOLL_RX_DROP;
		spin_unlock(&npinfo->poll_lock);
	}
}

void netpoll_poll(struct netpoll *np)
{
	if(!np->dev || !netif_running(np->dev) || !np->dev->poll_controller)
		return;

	/* Process pending work on NIC */
	np->dev->poll_controller(np->dev);
	if (np->dev->poll)
		poll_napi(np);

	zap_completion_queue();
}

static void refill_skbs(void)
{
	struct sk_buff *skb;
	unsigned long flags;

	spin_lock_irqsave(&skb_list_lock, flags);
	while (nr_skbs < MAX_SKBS) {
		skb = alloc_skb(MAX_SKB_SIZE, GFP_ATOMIC);
		if (!skb)
			break;

		skb->next = skbs;
		skbs = skb;
		nr_skbs++;
	}
	spin_unlock_irqrestore(&skb_list_lock, flags);
}

static void zap_completion_queue(void)
{
	unsigned long flags;
	struct softnet_data *sd = &get_cpu_var(softnet_data);

	if (sd->completion_queue) {
		struct sk_buff *clist;

		local_irq_save(flags);
		clist = sd->completion_queue;
		sd->completion_queue = NULL;
		local_irq_restore(flags);

		while (clist != NULL) {
			struct sk_buff *skb = clist;
			clist = clist->next;
			if(skb->destructor)
				dev_kfree_skb_any(skb); /* put this one back */
			else
				__kfree_skb(skb);
		}
	}

	put_cpu_var(softnet_data);
}

static struct sk_buff * find_skb(struct netpoll *np, int len, int reserve)
{
	int once = 1, count = 0;
	unsigned long flags;
	struct sk_buff *skb = NULL;

	zap_completion_queue();
repeat:
	if (nr_skbs < MAX_SKBS)
		refill_skbs();

	skb = alloc_skb(len, GFP_ATOMIC);

	if (!skb) {
		spin_lock_irqsave(&skb_list_lock, flags);
		skb = skbs;
		if (skb) {
			skbs = skb->next;
			skb->next = NULL;
			nr_skbs--;
		}
		spin_unlock_irqrestore(&skb_list_lock, flags);
	}

	if(!skb) {
		count++;
		if (once && (count == 1000000)) {
			printk("out of netpoll skbs!\n");
			once = 0;
		}
		netpoll_poll(np);
		goto repeat;
	}

	atomic_set(&skb->users, 1);
	skb_reserve(skb, reserve);
	return skb;
}

static void netpoll_send_skb(struct netpoll *np, struct sk_buff *skb)
{
	int status;
	struct netpoll_info *npinfo;

repeat:
	if(!np || !np->dev || !netif_running(np->dev)) {
		__kfree_skb(skb);
		return;
	}

	/* avoid recursion */
	npinfo = np->dev->npinfo;
	if (npinfo->poll_owner == smp_processor_id() ||
	    np->dev->xmit_lock_owner == smp_processor_id()) {
		if (np->drop)
			np->drop(skb);
		else
			__kfree_skb(skb);
		return;
	}

	spin_lock(&np->dev->xmit_lock);
	np->dev->xmit_lock_owner = smp_processor_id();

	/*
	 * network drivers do not expect to be called if the queue is
	 * stopped.
	 */
	if (netif_queue_stopped(np->dev)) {
		np->dev->xmit_lock_owner = -1;
		spin_unlock(&np->dev->xmit_lock);

		netpoll_poll(np);
		goto repeat;
	}

	status = np->dev->hard_start_xmit(skb, np->dev);
	np->dev->xmit_lock_owner = -1;
	spin_unlock(&np->dev->xmit_lock);

	/* transmit busy */
	if(status) {
		netpoll_poll(np);
		goto repeat;
	}
}

void netpoll_send_udp(struct netpoll *np, const char *msg, int len)
{
	int total_len, eth_len, ip_len, udp_len;
	struct sk_buff *skb;
	struct udphdr *udph;
	struct iphdr *iph;
	struct ethhdr *eth;

	udp_len = len + sizeof(*udph);
	ip_len = eth_len = udp_len + sizeof(*iph);
	total_len = eth_len + ETH_HLEN + NET_IP_ALIGN;

	skb = find_skb(np, total_len, total_len - len);
	if (!skb)
		return;

	memcpy(skb->data, msg, len);
	skb->len += len;

	udph = (struct udphdr *) skb_push(skb, sizeof(*udph));
	udph->source = htons(np->local_port);
	udph->dest = htons(np->remote_port);
	udph->len = htons(udp_len);
	udph->check = 0;

	iph = (struct iphdr *)skb_push(skb, sizeof(*iph));

	/* iph->version = 4; iph->ihl = 5; */
	put_unaligned(0x45, (unsigned char *)iph);
	iph->tos      = 0;
	put_unaligned(htons(ip_len), &(iph->tot_len));
	iph->id       = 0;
	iph->frag_off = 0;
	iph->ttl      = 64;
	iph->protocol = IPPROTO_UDP;
	iph->check    = 0;
	put_unaligned(htonl(np->local_ip), &(iph->saddr));
	put_unaligned(htonl(np->remote_ip), &(iph->daddr));
	iph->check    = ip_fast_csum((unsigned char *)iph, iph->ihl);

	eth = (struct ethhdr *) skb_push(skb, ETH_HLEN);

	eth->h_proto = htons(ETH_P_IP);
	memcpy(eth->h_source, np->local_mac, 6);
	memcpy(eth->h_dest, np->remote_mac, 6);

	skb->dev = np->dev;

	netpoll_send_skb(np, skb);
}

static void arp_reply(struct sk_buff *skb)
{
	struct netpoll_info *npinfo = skb->dev->npinfo;
	struct arphdr *arp;
	unsigned char *arp_ptr;
	int size, type = ARPOP_REPLY, ptype = ETH_P_ARP;
	u32 sip, tip;
	struct sk_buff *send_skb;
	struct netpoll *np = NULL;

	if (npinfo)
		np = npinfo->np;
	if (!np)
		return;

	/* No arp on this interface */
	if (skb->dev->flags & IFF_NOARP)
		return;

	if (!pskb_may_pull(skb, (sizeof(struct arphdr) +
				 (2 * skb->dev->addr_len) +
				 (2 * sizeof(u32)))))
		return;

	skb->h.raw = skb->nh.raw = skb->data;
	arp = skb->nh.arph;

	if ((arp->ar_hrd != htons(ARPHRD_ETHER) &&
	     arp->ar_hrd != htons(ARPHRD_IEEE802)) ||
	    arp->ar_pro != htons(ETH_P_IP) ||
	    arp->ar_op != htons(ARPOP_REQUEST))
		return;

	arp_ptr = (unsigned char *)(arp+1) + skb->dev->addr_len;
	memcpy(&sip, arp_ptr, 4);
	arp_ptr += 4 + skb->dev->addr_len;
	memcpy(&tip, arp_ptr, 4);

	/* Should we ignore arp? */
	if (tip != htonl(np->local_ip) || LOOPBACK(tip) || MULTICAST(tip))
		return;

	size = sizeof(struct arphdr) + 2 * (skb->dev->addr_len + 4);
	send_skb = find_skb(np, size + LL_RESERVED_SPACE(np->dev),
			    LL_RESERVED_SPACE(np->dev));

	if (!send_skb)
		return;

	send_skb->nh.raw = send_skb->data;
	arp = (struct arphdr *) skb_put(send_skb, size);
	send_skb->dev = skb->dev;
	send_skb->protocol = htons(ETH_P_ARP);

	/* Fill the device header for the ARP frame */

	if (np->dev->hard_header &&
	    np->dev->hard_header(send_skb, skb->dev, ptype,
				       np->remote_mac, np->local_mac,
				       send_skb->len) < 0) {
		kfree_skb(send_skb);
		return;
	}

	/*
	 * Fill out the arp protocol part.
	 *
	 * we only support ethernet device type,
	 * which (according to RFC 1390) should always equal 1 (Ethernet).
	 */

	arp->ar_hrd = htons(np->dev->type);
	arp->ar_pro = htons(ETH_P_IP);
	arp->ar_hln = np->dev->addr_len;
	arp->ar_pln = 4;
	arp->ar_op = htons(type);

	arp_ptr=(unsigned char *)(arp + 1);
	memcpy(arp_ptr, np->dev->dev_addr, np->dev->addr_len);
	arp_ptr += np->dev->addr_len;
	memcpy(arp_ptr, &tip, 4);
	arp_ptr += 4;
	memcpy(arp_ptr, np->remote_mac, np->dev->addr_len);
	arp_ptr += np->dev->addr_len;
	memcpy(arp_ptr, &sip, 4);

	netpoll_send_skb(np, send_skb);
}

int __netpoll_rx(struct sk_buff *skb)
{
	int proto, len, ulen;
	struct iphdr *iph;
	struct udphdr *uh;
	struct netpoll *np = skb->dev->npinfo->np;

	if (!np->rx_hook)
		goto out;
	if (skb->dev->type != ARPHRD_ETHER)
		goto out;

	/* check if netpoll clients need ARP */
	if (skb->protocol == __constant_htons(ETH_P_ARP) &&
	    atomic_read(&trapped)) {
		arp_reply(skb);
		return 1;
	}

	proto = ntohs(eth_hdr(skb)->h_proto);
	if (proto != ETH_P_IP)
		goto out;
	if (skb->pkt_type == PACKET_OTHERHOST)
		goto out;
	if (skb_shared(skb))
		goto out;

	iph = (struct iphdr *)skb->data;
	if (!pskb_may_pull(skb, sizeof(struct iphdr)))
		goto out;
	if (iph->ihl < 5 || iph->version != 4)
		goto out;
	if (!pskb_may_pull(skb, iph->ihl*4))
		goto out;
	if (ip_fast_csum((u8 *)iph, iph->ihl) != 0)
		goto out;

	len = ntohs(iph->tot_len);
	if (skb->len < len || len < iph->ihl*4)
		goto out;

	if (iph->protocol != IPPROTO_UDP)
		goto out;

	len -= iph->ihl*4;
	uh = (struct udphdr *)(((char *)iph) + iph->ihl*4);
	ulen = ntohs(uh->len);

	if (ulen != len)
		goto out;
	if (checksum_udp(skb, uh, ulen, iph->saddr, iph->daddr) < 0)
		goto out;
	if (np->local_ip && np->local_ip != ntohl(iph->daddr))
		goto out;
	if (np->remote_ip && np->remote_ip != ntohl(iph->saddr))
		goto out;
	if (np->local_port && np->local_port != ntohs(uh->dest))
		goto out;

	np->rx_hook(np, ntohs(uh->source),
		    (char *)(uh+1),
		    ulen - sizeof(struct udphdr));

	kfree_skb(skb);
	return 1;

out:
	if (atomic_read(&trapped)) {
		kfree_skb(skb);
		return 1;
	}

	return 0;
}

int netpoll_parse_options(struct netpoll *np, char *opt)
{
	char *cur=opt, *delim;

	if(*cur != '@') {
		if ((delim = strchr(cur, '@')) == NULL)
			goto parse_failed;
		*delim=0;
		np->local_port=simple_strtol(cur, NULL, 10);
		cur=delim;
	}
	cur++;
	printk(KERN_INFO "%s: local port %d\n", np->name, np->local_port);

	if(*cur != '/') {
		if ((delim = strchr(cur, '/')) == NULL)
			goto parse_failed;
		*delim=0;
		np->local_ip=ntohl(in_aton(cur));
		cur=delim;

		printk(KERN_INFO "%s: local IP %d.%d.%d.%d\n",
		       np->name, HIPQUAD(np->local_ip));
	}
	cur++;

	if ( *cur != ',') {
		/* parse out dev name */
		if ((delim = strchr(cur, ',')) == NULL)
			goto parse_failed;
		*delim=0;
		strlcpy(np->dev_name, cur, sizeof(np->dev_name));
		cur=delim;
	}
	cur++;

	printk(KERN_INFO "%s: interface %s\n", np->name, np->dev_name);

	if ( *cur != '@' ) {
		/* dst port */
		if ((delim = strchr(cur, '@')) == NULL)
			goto parse_failed;
		*delim=0;
		np->remote_port=simple_strtol(cur, NULL, 10);
		cur=delim;
	}
	cur++;
	printk(KERN_INFO "%s: remote port %d\n", np->name, np->remote_port);

	/* dst ip */
	if ((delim = strchr(cur, '/')) == NULL)
		goto parse_failed;
	*delim=0;
	np->remote_ip=ntohl(in_aton(cur));
	cur=delim+1;

	printk(KERN_INFO "%s: remote IP %d.%d.%d.%d\n",
		       np->name, HIPQUAD(np->remote_ip));

	if( *cur != 0 )
	{
		/* MAC address */
		if ((delim = strchr(cur, ':')) == NULL)
			goto parse_failed;
		*delim=0;
		np->remote_mac[0]=simple_strtol(cur, NULL, 16);
		cur=delim+1;
		if ((delim = strchr(cur, ':')) == NULL)
			goto parse_failed;
		*delim=0;
		np->remote_mac[1]=simple_strtol(cur, NULL, 16);
		cur=delim+1;
		if ((delim = strchr(cur, ':')) == NULL)
			goto parse_failed;
		*delim=0;
		np->remote_mac[2]=simple_strtol(cur, NULL, 16);
		cur=delim+1;
		if ((delim = strchr(cur, ':')) == NULL)
			goto parse_failed;
		*delim=0;
		np->remote_mac[3]=simple_strtol(cur, NULL, 16);
		cur=delim+1;
		if ((delim = strchr(cur, ':')) == NULL)
			goto parse_failed;
		*delim=0;
		np->remote_mac[4]=simple_strtol(cur, NULL, 16);
		cur=delim+1;
		np->remote_mac[5]=simple_strtol(cur, NULL, 16);
	}

	printk(KERN_INFO "%s: remote ethernet address "
	       "%02x:%02x:%02x:%02x:%02x:%02x\n",
	       np->name,
	       np->remote_mac[0],
	       np->remote_mac[1],
	       np->remote_mac[2],
	       np->remote_mac[3],
	       np->remote_mac[4],
	       np->remote_mac[5]);

	return 0;

 parse_failed:
	printk(KERN_INFO "%s: couldn't parse config at %s!\n",
	       np->name, cur);
	return -1;
}

int netpoll_setup(struct netpoll *np)
{
	struct net_device *ndev = NULL;
	struct in_device *in_dev;
	struct netpoll_info *npinfo;

	if (np->dev_name)
		ndev = dev_get_by_name(np->dev_name);
	if (!ndev) {
		printk(KERN_ERR "%s: %s doesn't exist, aborting.\n",
		       np->name, np->dev_name);
		return -1;
	}

	np->dev = ndev;
	if (!ndev->npinfo) {
		npinfo = kmalloc(sizeof(*npinfo), GFP_KERNEL);
		if (!npinfo)
			goto release;

		npinfo->np = NULL;
		npinfo->poll_lock = SPIN_LOCK_UNLOCKED;
		npinfo->poll_owner = -1;
	} else
		npinfo = ndev->npinfo;

	if (!ndev->poll_controller) {
		printk(KERN_ERR "%s: %s doesn't support polling, aborting.\n",
		       np->name, np->dev_name);
		goto release;
	}

	if (!netif_running(ndev)) {
		unsigned long atmost, atleast;

		printk(KERN_INFO "%s: device %s not up yet, forcing it\n",
		       np->name, np->dev_name);

		rtnl_shlock();
		if (dev_change_flags(ndev, ndev->flags | IFF_UP) < 0) {
			printk(KERN_ERR "%s: failed to open %s\n",
			       np->name, np->dev_name);
			rtnl_shunlock();
			goto release;
		}
		rtnl_shunlock();

		atleast = jiffies + HZ/10;
 		atmost = jiffies + 4*HZ;
		while (!netif_carrier_ok(ndev)) {
			if (time_after(jiffies, atmost)) {
				printk(KERN_NOTICE
				       "%s: timeout waiting for carrier\n",
				       np->name);
				break;
			}
			cond_resched();
		}

		/* If carrier appears to come up instantly, we don't
		 * trust it and pause so that we don't pump all our
		 * queued console messages into the bitbucket.
		 */

		if (time_before(jiffies, atleast)) {
			printk(KERN_NOTICE "%s: carrier detect appears"
			       " untrustworthy, waiting 4 seconds\n",
			       np->name);
			msleep(4000);
		}
	}

	if (!memcmp(np->local_mac, "\0\0\0\0\0\0", 6) && ndev->dev_addr)
		memcpy(np->local_mac, ndev->dev_addr, 6);

	if (!np->local_ip) {
		rcu_read_lock();
		in_dev = __in_dev_get(ndev);

		if (!in_dev || !in_dev->ifa_list) {
			rcu_read_unlock();
			printk(KERN_ERR "%s: no IP address for %s, aborting\n",
			       np->name, np->dev_name);
			goto release;
		}

		np->local_ip = ntohl(in_dev->ifa_list->ifa_local);
		rcu_read_unlock();
		printk(KERN_INFO "%s: local IP %d.%d.%d.%d\n",
		       np->name, HIPQUAD(np->local_ip));
	}

	if(np->rx_hook)
		npinfo->rx_flags = NETPOLL_RX_ENABLED;
	npinfo->np = np;
	ndev->npinfo = npinfo;

	return 0;

 release:
	if (!ndev->npinfo)
		kfree(npinfo);
	np->dev = NULL;
	dev_put(ndev);
	return -1;
}

void netpoll_cleanup(struct netpoll *np)
{
	if (np->dev) {
		if (np->dev->npinfo)
			np->dev->npinfo->np = NULL;
		dev_put(np->dev);
	}
	np->dev = NULL;
}

int netpoll_trap(void)
{
	return atomic_read(&trapped);
}

void netpoll_set_trap(int trap)
{
	if (trap)
		atomic_inc(&trapped);
	else
		atomic_dec(&trapped);
}

EXPORT_SYMBOL(netpoll_set_trap);
EXPORT_SYMBOL(netpoll_trap);
EXPORT_SYMBOL(netpoll_parse_options);
EXPORT_SYMBOL(netpoll_setup);
EXPORT_SYMBOL(netpoll_cleanup);
EXPORT_SYMBOL(netpoll_send_udp);
EXPORT_SYMBOL(netpoll_poll);
EXPORT_SYMBOL(netpoll_queue);
Commit	Line	Data
1da177e4 LT	1	/*
	2	* Common framework for low-level network console, dump, and debugger code
	3	*
	4	* Sep 8 2003 Matt Mackall <mpm@selenic.com>
	5	*
	6	* based on the netconsole code from:
	7	*
	8	* Copyright (C) 2001 Ingo Molnar <mingo@redhat.com>
	9	* Copyright (C) 2002 Red Hat, Inc.
	10	*/
	11
	12	#include <linux/smp_lock.h>
	13	#include <linux/netdevice.h>
	14	#include <linux/etherdevice.h>
	15	#include <linux/string.h>
	16	#include <linux/inetdevice.h>
	17	#include <linux/inet.h>
	18	#include <linux/interrupt.h>
	19	#include <linux/netpoll.h>
	20	#include <linux/sched.h>
	21	#include <linux/delay.h>
	22	#include <linux/rcupdate.h>
	23	#include <linux/workqueue.h>
	24	#include <net/tcp.h>
	25	#include <net/udp.h>
	26	#include <asm/unaligned.h>
	27
	28	/*
	29	* We maintain a small pool of fully-sized skbs, to make sure the
	30	* message gets out even in extreme OOM situations.
	31	*/
	32
	33	#define MAX_UDP_CHUNK 1460
	34	#define MAX_SKBS 32
	35	#define MAX_QUEUE_DEPTH (MAX_SKBS / 2)
	36
	37	static DEFINE_SPINLOCK(skb_list_lock);
	38	static int nr_skbs;
	39	static struct sk_buff *skbs;
	40
	41	static DEFINE_SPINLOCK(queue_lock);
	42	static int queue_depth;
	43	static struct sk_buff queue_head, queue_tail;
	44
	45	static atomic_t trapped;
	46
	47	#define NETPOLL_RX_ENABLED 1
	48	#define NETPOLL_RX_DROP 2
	49
	50	#define MAX_SKB_SIZE \
	51	(MAX_UDP_CHUNK + sizeof(struct udphdr) + \
	52	sizeof(struct iphdr) + sizeof(struct ethhdr))
	53
	54	static void zap_completion_queue(void);
	55
	56	static void queue_process(void *p)
	57	{
	58	unsigned long flags;
	59	struct sk_buff *skb;
	60
	61	while (queue_head) {
	62	spin_lock_irqsave(&queue_lock, flags);
	63
	64	skb = queue_head;
65	queue_head = skb->next;
66	if (skb == queue_tail)
67	queue_head = NULL;
68
69	queue_depth--;
70
71	spin_unlock_irqrestore(&queue_lock, flags);
72
73	dev_queue_xmit(skb);
74	}
75	}
76
77	static DECLARE_WORK(send_queue, queue_process, NULL);
78
79	void netpoll_queue(struct sk_buff *skb)
80	{
81	unsigned long flags;
82
83	if (queue_depth == MAX_QUEUE_DEPTH) {
84	__kfree_skb(skb);
85	return;
86	}
87
88	spin_lock_irqsave(&queue_lock, flags);
89	if (!queue_head)
90	queue_head = skb;
91	else
92	queue_tail->next = skb;
93	queue_tail = skb;
94	queue_depth++;
95	spin_unlock_irqrestore(&queue_lock, flags);
96
97	schedule_work(&send_queue);
98	}
99
100	static int checksum_udp(struct sk_buff skb, struct udphdr uh,
101	unsigned short ulen, u32 saddr, u32 daddr)
102	{
103	if (uh->check == 0)
104	return 0;
105
106	if (skb->ip_summed == CHECKSUM_HW)
107	return csum_tcpudp_magic(
108	saddr, daddr, ulen, IPPROTO_UDP, skb->csum);
109
110	skb->csum = csum_tcpudp_nofold(saddr, daddr, ulen, IPPROTO_UDP, 0);
111
112	return csum_fold(skb_checksum(skb, 0, skb->len, skb->csum));
113	}
114
115	/*
116	* Check whether delayed processing was scheduled for our NIC. If so,
117	* we attempt to grab the poll lock and use ->poll() to pump the card.
118	* If this fails, either we've recursed in ->poll() or it's already
119	* running on another CPU.
120	*
121	* Note: we don't mask interrupts with this lock because we're using
122	* trylock here and interrupts are already disabled in the softirq
123	* case. Further, we test the poll_owner to avoid recursion on UP
124	* systems where the lock doesn't exist.
125	*
126	* In cases where there is bi-directional communications, reading only
127	* one message at a time can lead to packets being dropped by the
128	* network adapter, forcing superfluous retries and possibly timeouts.
129	* Thus, we set our budget to greater than 1.
130	*/
131	static void poll_napi(struct netpoll *np)
132	{
115c1d6e	133	struct netpoll_info *npinfo = np->dev->npinfo;
1da177e4 LT	134	int budget = 16;
	135
	136	if (test_bit(__LINK_STATE_RX_SCHED, &np->dev->state) &&
115c1d6e JM	137	npinfo->poll_owner != smp_processor_id() &&
	138	spin_trylock(&npinfo->poll_lock)) {
	139	npinfo->rx_flags \|= NETPOLL_RX_DROP;
1da177e4 LT	140	atomic_inc(&trapped);
	141
	142	np->dev->poll(np->dev, &budget);
	143
	144	atomic_dec(&trapped);
115c1d6e JM	145	npinfo->rx_flags &= ~NETPOLL_RX_DROP;
115c1d6e JM	146	spin_unlock(&npinfo->poll_lock);
1da177e4 LT	147	}
	148	}
	149
	150	void netpoll_poll(struct netpoll *np)
	151	{
	152	if(!np->dev \|\| !netif_running(np->dev) \|\| !np->dev->poll_controller)
	153	return;
	154
	155	/* Process pending work on NIC */
	156	np->dev->poll_controller(np->dev);
	157	if (np->dev->poll)
	158	poll_napi(np);
	159
	160	zap_completion_queue();
	161	}
	162
	163	static void refill_skbs(void)
	164	{
	165	struct sk_buff *skb;
	166	unsigned long flags;
	167
	168	spin_lock_irqsave(&skb_list_lock, flags);
	169	while (nr_skbs < MAX_SKBS) {
	170	skb = alloc_skb(MAX_SKB_SIZE, GFP_ATOMIC);
	171	if (!skb)
	172	break;
	173
	174	skb->next = skbs;
	175	skbs = skb;
	176	nr_skbs++;
	177	}
	178	spin_unlock_irqrestore(&skb_list_lock, flags);
	179	}
	180
	181	static void zap_completion_queue(void)
	182	{
	183	unsigned long flags;
	184	struct softnet_data *sd = &get_cpu_var(softnet_data);
	185
	186	if (sd->completion_queue) {
	187	struct sk_buff *clist;
	188
	189	local_irq_save(flags);
	190	clist = sd->completion_queue;
	191	sd->completion_queue = NULL;
	192	local_irq_restore(flags);
	193
	194	while (clist != NULL) {
	195	struct sk_buff *skb = clist;
	196	clist = clist->next;
	197	if(skb->destructor)
	198	dev_kfree_skb_any(skb); /* put this one back */
	199	else
	200	__kfree_skb(skb);
	201	}
	202	}
	203
	204	put_cpu_var(softnet_data);
	205	}
	206
	207	static struct sk_buff * find_skb(struct netpoll *np, int len, int reserve)
	208	{
	209	int once = 1, count = 0;
	210	unsigned long flags;
211	struct sk_buff *skb = NULL;
212
213	zap_completion_queue();
214	repeat:
215	if (nr_skbs < MAX_SKBS)
216	refill_skbs();
217
218	skb = alloc_skb(len, GFP_ATOMIC);
219
220	if (!skb) {
221	spin_lock_irqsave(&skb_list_lock, flags);
222	skb = skbs;
223	if (skb) {
224	skbs = skb->next;
225	skb->next = NULL;
226	nr_skbs--;
227	}
228	spin_unlock_irqrestore(&skb_list_lock, flags);
229	}
230
231	if(!skb) {
232	count++;
233	if (once && (count == 1000000)) {
234	printk("out of netpoll skbs!\n");
235	once = 0;
236	}
237	netpoll_poll(np);
238	goto repeat;
239	}
240
241	atomic_set(&skb->users, 1);
242	skb_reserve(skb, reserve);
243	return skb;
244	}
245
246	static void netpoll_send_skb(struct netpoll np, struct sk_buff skb)
247	{
248	int status;
115c1d6e	249	struct netpoll_info *npinfo;
1da177e4 LT	250
	251	repeat:
	252	if(!np \|\| !np->dev \|\| !netif_running(np->dev)) {
	253	__kfree_skb(skb);
	254	return;
	255	}
	256
	257	/* avoid recursion */
115c1d6e JM	258	npinfo = np->dev->npinfo;
	259	if (npinfo->poll_owner == smp_processor_id() \|\|
	260	np->dev->xmit_lock_owner == smp_processor_id()) {
1da177e4 LT	261	if (np->drop)
	262	np->drop(skb);
	263	else
	264	__kfree_skb(skb);
	265	return;
	266	}
	267
	268	spin_lock(&np->dev->xmit_lock);
	269	np->dev->xmit_lock_owner = smp_processor_id();
	270
	271	/*
	272	* network drivers do not expect to be called if the queue is
	273	* stopped.
	274	*/
	275	if (netif_queue_stopped(np->dev)) {
	276	np->dev->xmit_lock_owner = -1;
	277	spin_unlock(&np->dev->xmit_lock);
	278
	279	netpoll_poll(np);
	280	goto repeat;
	281	}
	282
	283	status = np->dev->hard_start_xmit(skb, np->dev);
	284	np->dev->xmit_lock_owner = -1;
	285	spin_unlock(&np->dev->xmit_lock);
	286
	287	/* transmit busy */
	288	if(status) {
	289	netpoll_poll(np);
	290	goto repeat;
	291	}
	292	}
	293
	294	void netpoll_send_udp(struct netpoll np, const char msg, int len)
	295	{
	296	int total_len, eth_len, ip_len, udp_len;
	297	struct sk_buff *skb;
	298	struct udphdr *udph;
	299	struct iphdr *iph;
	300	struct ethhdr *eth;
	301
	302	udp_len = len + sizeof(*udph);
	303	ip_len = eth_len = udp_len + sizeof(*iph);
	304	total_len = eth_len + ETH_HLEN + NET_IP_ALIGN;
	305
	306	skb = find_skb(np, total_len, total_len - len);
	307	if (!skb)
	308	return;
	309
	310	memcpy(skb->data, msg, len);
	311	skb->len += len;
	312
	313	udph = (struct udphdr ) skb_push(skb, sizeof(udph));
	314	udph->source = htons(np->local_port);
	315	udph->dest = htons(np->remote_port);
	316	udph->len = htons(udp_len);
	317	udph->check = 0;
	318
	319	iph = (struct iphdr )skb_push(skb, sizeof(iph));
	320
	321	/* iph->version = 4; iph->ihl = 5; */
	322	put_unaligned(0x45, (unsigned char *)iph);
	323	iph->tos = 0;
	324	put_unaligned(htons(ip_len), &(iph->tot_len));
325	iph->id = 0;
326	iph->frag_off = 0;
327	iph->ttl = 64;
328	iph->protocol = IPPROTO_UDP;
329	iph->check = 0;
330	put_unaligned(htonl(np->local_ip), &(iph->saddr));
331	put_unaligned(htonl(np->remote_ip), &(iph->daddr));
332	iph->check = ip_fast_csum((unsigned char *)iph, iph->ihl);
333
334	eth = (struct ethhdr *) skb_push(skb, ETH_HLEN);
335
336	eth->h_proto = htons(ETH_P_IP);
337	memcpy(eth->h_source, np->local_mac, 6);
338	memcpy(eth->h_dest, np->remote_mac, 6);
339
340	skb->dev = np->dev;
341
342	netpoll_send_skb(np, skb);
343	}
344
345	static void arp_reply(struct sk_buff *skb)
346	{
115c1d6e	347	struct netpoll_info *npinfo = skb->dev->npinfo;
1da177e4 LT	348	struct arphdr *arp;
	349	unsigned char *arp_ptr;
	350	int size, type = ARPOP_REPLY, ptype = ETH_P_ARP;
	351	u32 sip, tip;
	352	struct sk_buff *send_skb;
115c1d6e	353	struct netpoll *np = NULL;
1da177e4	354
115c1d6e JM	355	if (npinfo)
	356	np = npinfo->np;
	357	if (!np)
	358	return;
1da177e4 LT	359
	360	/* No arp on this interface */
	361	if (skb->dev->flags & IFF_NOARP)
	362	return;
	363
	364	if (!pskb_may_pull(skb, (sizeof(struct arphdr) +
	365	(2 * skb->dev->addr_len) +
	366	(2 * sizeof(u32)))))
	367	return;
	368
	369	skb->h.raw = skb->nh.raw = skb->data;
	370	arp = skb->nh.arph;
	371
	372	if ((arp->ar_hrd != htons(ARPHRD_ETHER) &&
	373	arp->ar_hrd != htons(ARPHRD_IEEE802)) \|\|
	374	arp->ar_pro != htons(ETH_P_IP) \|\|
	375	arp->ar_op != htons(ARPOP_REQUEST))
	376	return;
	377
	378	arp_ptr = (unsigned char *)(arp+1) + skb->dev->addr_len;
	379	memcpy(&sip, arp_ptr, 4);
	380	arp_ptr += 4 + skb->dev->addr_len;
	381	memcpy(&tip, arp_ptr, 4);
	382
	383	/* Should we ignore arp? */
	384	if (tip != htonl(np->local_ip) \|\| LOOPBACK(tip) \|\| MULTICAST(tip))
	385	return;
	386
	387	size = sizeof(struct arphdr) + 2 * (skb->dev->addr_len + 4);
	388	send_skb = find_skb(np, size + LL_RESERVED_SPACE(np->dev),
	389	LL_RESERVED_SPACE(np->dev));
	390
	391	if (!send_skb)
	392	return;
	393
	394	send_skb->nh.raw = send_skb->data;
	395	arp = (struct arphdr *) skb_put(send_skb, size);
	396	send_skb->dev = skb->dev;
	397	send_skb->protocol = htons(ETH_P_ARP);
	398
	399	/* Fill the device header for the ARP frame */
	400
	401	if (np->dev->hard_header &&
	402	np->dev->hard_header(send_skb, skb->dev, ptype,
	403	np->remote_mac, np->local_mac,
	404	send_skb->len) < 0) {
	405	kfree_skb(send_skb);
	406	return;
	407	}
	408
	409	/*
	410	* Fill out the arp protocol part.
	411	*
	412	* we only support ethernet device type,
	413	* which (according to RFC 1390) should always equal 1 (Ethernet).
	414	*/
	415
	416	arp->ar_hrd = htons(np->dev->type);
	417	arp->ar_pro = htons(ETH_P_IP);
	418	arp->ar_hln = np->dev->addr_len;
	419	arp->ar_pln = 4;
	420	arp->ar_op = htons(type);
	421
	422	arp_ptr=(unsigned char *)(arp + 1);
423	memcpy(arp_ptr, np->dev->dev_addr, np->dev->addr_len);
424	arp_ptr += np->dev->addr_len;
425	memcpy(arp_ptr, &tip, 4);
426	arp_ptr += 4;
427	memcpy(arp_ptr, np->remote_mac, np->dev->addr_len);
428	arp_ptr += np->dev->addr_len;
429	memcpy(arp_ptr, &sip, 4);
430
431	netpoll_send_skb(np, send_skb);
432	}
433
434	int __netpoll_rx(struct sk_buff *skb)
435	{
436	int proto, len, ulen;
437	struct iphdr *iph;
438	struct udphdr *uh;
115c1d6e	439	struct netpoll *np = skb->dev->npinfo->np;
1da177e4 LT	440
	441	if (!np->rx_hook)
	442	goto out;
	443	if (skb->dev->type != ARPHRD_ETHER)
	444	goto out;
	445
	446	/* check if netpoll clients need ARP */
	447	if (skb->protocol == __constant_htons(ETH_P_ARP) &&
	448	atomic_read(&trapped)) {
	449	arp_reply(skb);
	450	return 1;
	451	}
	452
	453	proto = ntohs(eth_hdr(skb)->h_proto);
	454	if (proto != ETH_P_IP)
	455	goto out;
	456	if (skb->pkt_type == PACKET_OTHERHOST)
	457	goto out;
	458	if (skb_shared(skb))
	459	goto out;
	460
	461	iph = (struct iphdr *)skb->data;
	462	if (!pskb_may_pull(skb, sizeof(struct iphdr)))
	463	goto out;
	464	if (iph->ihl < 5 \|\| iph->version != 4)
	465	goto out;
	466	if (!pskb_may_pull(skb, iph->ihl*4))
	467	goto out;
	468	if (ip_fast_csum((u8 *)iph, iph->ihl) != 0)
	469	goto out;
	470
	471	len = ntohs(iph->tot_len);
	472	if (skb->len < len \|\| len < iph->ihl*4)
	473	goto out;
	474
	475	if (iph->protocol != IPPROTO_UDP)
	476	goto out;
	477
	478	len -= iph->ihl*4;
	479	uh = (struct udphdr )(((char )iph) + iph->ihl*4);
	480	ulen = ntohs(uh->len);
	481
	482	if (ulen != len)
	483	goto out;
	484	if (checksum_udp(skb, uh, ulen, iph->saddr, iph->daddr) < 0)
	485	goto out;
	486	if (np->local_ip && np->local_ip != ntohl(iph->daddr))
	487	goto out;
	488	if (np->remote_ip && np->remote_ip != ntohl(iph->saddr))
	489	goto out;
	490	if (np->local_port && np->local_port != ntohs(uh->dest))
	491	goto out;
	492
	493	np->rx_hook(np, ntohs(uh->source),
	494	(char *)(uh+1),
	495	ulen - sizeof(struct udphdr));
	496
	497	kfree_skb(skb);
	498	return 1;
	499
	500	out:
	501	if (atomic_read(&trapped)) {
	502	kfree_skb(skb);
	503	return 1;
504	}
505
506	return 0;
507	}
508
509	int netpoll_parse_options(struct netpoll np, char opt)
510	{
511	char cur=opt, delim;
512
513	if(*cur != '@') {
514	if ((delim = strchr(cur, '@')) == NULL)
515	goto parse_failed;
516	*delim=0;
517	np->local_port=simple_strtol(cur, NULL, 10);
518	cur=delim;
519	}
520	cur++;
521	printk(KERN_INFO "%s: local port %d\n", np->name, np->local_port);
522
523	if(*cur != '/') {
524	if ((delim = strchr(cur, '/')) == NULL)
525	goto parse_failed;
526	*delim=0;
527	np->local_ip=ntohl(in_aton(cur));
528	cur=delim;
529
530	printk(KERN_INFO "%s: local IP %d.%d.%d.%d\n",
531	np->name, HIPQUAD(np->local_ip));
532	}
533	cur++;
534
535	if ( *cur != ',') {
536	/* parse out dev name */
537	if ((delim = strchr(cur, ',')) == NULL)
538	goto parse_failed;
539	*delim=0;
540	strlcpy(np->dev_name, cur, sizeof(np->dev_name));
541	cur=delim;
542	}
543	cur++;
544
545	printk(KERN_INFO "%s: interface %s\n", np->name, np->dev_name);
546
547	if ( *cur != '@' ) {
548	/* dst port */
549	if ((delim = strchr(cur, '@')) == NULL)
550	goto parse_failed;
551	*delim=0;
552	np->remote_port=simple_strtol(cur, NULL, 10);
553	cur=delim;
554	}
555	cur++;
556	printk(KERN_INFO "%s: remote port %d\n", np->name, np->remote_port);
557
558	/* dst ip */
559	if ((delim = strchr(cur, '/')) == NULL)
560	goto parse_failed;
561	*delim=0;
562	np->remote_ip=ntohl(in_aton(cur));
563	cur=delim+1;
564
565	printk(KERN_INFO "%s: remote IP %d.%d.%d.%d\n",
566	np->name, HIPQUAD(np->remote_ip));
567
568	if( *cur != 0 )
569	{
570	/* MAC address */
571	if ((delim = strchr(cur, ':')) == NULL)
572	goto parse_failed;
573	*delim=0;
574	np->remote_mac[0]=simple_strtol(cur, NULL, 16);
575	cur=delim+1;
576	if ((delim = strchr(cur, ':')) == NULL)
577	goto parse_failed;
578	*delim=0;
579	np->remote_mac[1]=simple_strtol(cur, NULL, 16);
580	cur=delim+1;
581	if ((delim = strchr(cur, ':')) == NULL)
582	goto parse_failed;
583	*delim=0;
584	np->remote_mac[2]=simple_strtol(cur, NULL, 16);
585	cur=delim+1;
586	if ((delim = strchr(cur, ':')) == NULL)
587	goto parse_failed;
588	*delim=0;
589	np->remote_mac[3]=simple_strtol(cur, NULL, 16);
590	cur=delim+1;
591	if ((delim = strchr(cur, ':')) == NULL)
592	goto parse_failed;
593	*delim=0;
594	np->remote_mac[4]=simple_strtol(cur, NULL, 16);
595	cur=delim+1;
596	np->remote_mac[5]=simple_strtol(cur, NULL, 16);
597	}
598
599	printk(KERN_INFO "%s: remote ethernet address "
600	"%02x:%02x:%02x:%02x:%02x:%02x\n",
601	np->name,
602	np->remote_mac[0],
603	np->remote_mac[1],
604	np->remote_mac[2],
605	np->remote_mac[3],
606	np->remote_mac[4],
607	np->remote_mac[5]);
608
609	return 0;
610
611	parse_failed:
612	printk(KERN_INFO "%s: couldn't parse config at %s!\n",
613	np->name, cur);
614	return -1;
615	}
616
617	int netpoll_setup(struct netpoll *np)
618	{
619	struct net_device *ndev = NULL;
620	struct in_device *in_dev;
115c1d6e	621	struct netpoll_info *npinfo;
1da177e4 LT	622
	623	if (np->dev_name)
	624	ndev = dev_get_by_name(np->dev_name);
	625	if (!ndev) {
	626	printk(KERN_ERR "%s: %s doesn't exist, aborting.\n",
	627	np->name, np->dev_name);
	628	return -1;
	629	}
	630
	631	np->dev = ndev;
115c1d6e JM	632	if (!ndev->npinfo) {
	633	npinfo = kmalloc(sizeof(*npinfo), GFP_KERNEL);
	634	if (!npinfo)
	635	goto release;
	636
	637	npinfo->np = NULL;
	638	npinfo->poll_lock = SPIN_LOCK_UNLOCKED;
	639	npinfo->poll_owner = -1;
	640	} else
	641	npinfo = ndev->npinfo;
1da177e4 LT	642
	643	if (!ndev->poll_controller) {
	644	printk(KERN_ERR "%s: %s doesn't support polling, aborting.\n",
	645	np->name, np->dev_name);
	646	goto release;
	647	}
	648
	649	if (!netif_running(ndev)) {
	650	unsigned long atmost, atleast;
	651
	652	printk(KERN_INFO "%s: device %s not up yet, forcing it\n",
	653	np->name, np->dev_name);
	654
	655	rtnl_shlock();
	656	if (dev_change_flags(ndev, ndev->flags \| IFF_UP) < 0) {
	657	printk(KERN_ERR "%s: failed to open %s\n",
	658	np->name, np->dev_name);
	659	rtnl_shunlock();
	660	goto release;
	661	}
	662	rtnl_shunlock();
	663
	664	atleast = jiffies + HZ/10;
	665	atmost = jiffies + 4*HZ;
	666	while (!netif_carrier_ok(ndev)) {
	667	if (time_after(jiffies, atmost)) {
	668	printk(KERN_NOTICE
	669	"%s: timeout waiting for carrier\n",
	670	np->name);
	671	break;
	672	}
	673	cond_resched();
	674	}
	675
	676	/* If carrier appears to come up instantly, we don't
	677	* trust it and pause so that we don't pump all our
	678	* queued console messages into the bitbucket.
	679	*/
	680
	681	if (time_before(jiffies, atleast)) {
	682	printk(KERN_NOTICE "%s: carrier detect appears"
	683	" untrustworthy, waiting 4 seconds\n",
	684	np->name);
	685	msleep(4000);
	686	}
	687	}
	688
	689	if (!memcmp(np->local_mac, "\0\0\0\0\0\0", 6) && ndev->dev_addr)
	690	memcpy(np->local_mac, ndev->dev_addr, 6);
	691
	692	if (!np->local_ip) {
	693	rcu_read_lock();
	694	in_dev = __in_dev_get(ndev);
	695
	696	if (!in_dev \|\| !in_dev->ifa_list) {
	697	rcu_read_unlock();
	698	printk(KERN_ERR "%s: no IP address for %s, aborting\n",
	699	np->name, np->dev_name);
	700	goto release;
	701	}
	702
	703	np->local_ip = ntohl(in_dev->ifa_list->ifa_local);
	704	rcu_read_unlock();
	705	printk(KERN_INFO "%s: local IP %d.%d.%d.%d\n",
706	np->name, HIPQUAD(np->local_ip));
707	}
708
709	if(np->rx_hook)
115c1d6e JM	710	npinfo->rx_flags = NETPOLL_RX_ENABLED;
	711	npinfo->np = np;
	712	ndev->npinfo = npinfo;
1da177e4 LT	713
	714	return 0;
	715
	716	release:
115c1d6e JM	717	if (!ndev->npinfo)
115c1d6e JM	718	kfree(npinfo);
1da177e4 LT	719	np->dev = NULL;
	720	dev_put(ndev);
	721	return -1;
	722	}
	723
	724	void netpoll_cleanup(struct netpoll *np)
	725	{
115c1d6e JM	726	if (np->dev) {
	727	if (np->dev->npinfo)
	728	np->dev->npinfo->np = NULL;
	729	dev_put(np->dev);
	730	}
1da177e4 LT	731	np->dev = NULL;
	732	}
	733
	734	int netpoll_trap(void)
	735	{
	736	return atomic_read(&trapped);
	737	}
	738
	739	void netpoll_set_trap(int trap)
	740	{
	741	if (trap)
	742	atomic_inc(&trapped);
	743	else
	744	atomic_dec(&trapped);
	745	}
	746
	747	EXPORT_SYMBOL(netpoll_set_trap);
	748	EXPORT_SYMBOL(netpoll_trap);
	749	EXPORT_SYMBOL(netpoll_parse_options);
	750	EXPORT_SYMBOL(netpoll_setup);
	751	EXPORT_SYMBOL(netpoll_cleanup);
	752	EXPORT_SYMBOL(netpoll_send_udp);
	753	EXPORT_SYMBOL(netpoll_poll);
	754	EXPORT_SYMBOL(netpoll_queue);