svcsock.c

linux 内核源代码

/*
 * linux/net/sunrpc/svcsock.c
 *
 * These are the RPC server socket internals.
 *
 * The server scheduling algorithm does not always distribute the load
 * evenly when servicing a single client. May need to modify the
 * svc_sock_enqueue procedure...
 *
 * TCP support is largely untested and may be a little slow. The problem
 * is that we currently do two separate recvfrom's, one for the 4-byte
 * record length, and the second for the actual record. This could possibly
 * be improved by always reading a minimum size of around 100 bytes and
 * tucking any superfluous bytes away in a temporary store. Still, that
 * leaves write requests out in the rain. An alternative may be to peek at
 * the first skb in the queue, and if it matches the next TCP sequence
 * number, to extract the record marker. Yuck.
 *
 * Copyright (C) 1995, 1996 Olaf Kirch <[e-mail]>
 */

#include <linux/kernel.h>
#include <linux/sched.h>
#include <linux/errno.h>
#include <linux/fcntl.h>
#include <linux/net.h>
#include <linux/in.h>
#include <linux/inet.h>
#include <linux/udp.h>
#include <linux/tcp.h>
#include <linux/unistd.h>
#include <linux/slab.h>
#include <linux/netdevice.h>
#include <linux/skbuff.h>
#include <linux/file.h>
#include <linux/freezer.h>
#include <net/sock.h>
#include <net/checksum.h>
#include <net/ip.h>
#include <net/ipv6.h>
#include <net/tcp_states.h>
#include <asm/uaccess.h>
#include <asm/ioctls.h>

#include <linux/sunrpc/types.h>
#include <linux/sunrpc/clnt.h>
#include <linux/sunrpc/xdr.h>
#include <linux/sunrpc/svcsock.h>
#include <linux/sunrpc/stats.h>

/* SMP locking strategy:
 *
 *	svc_pool->sp_lock protects most of the fields of that pool.
 * 	svc_serv->sv_lock protects sv_tempsocks, sv_permsocks, sv_tmpcnt.
 *	when both need to be taken (rare), svc_serv->sv_lock is first.
 *	BKL protects svc_serv->sv_nrthread.
 *	svc_sock->sk_lock protects the svc_sock->sk_deferred list
 *             and the ->sk_info_authunix cache.
 *	svc_sock->sk_flags.SK_BUSY prevents a svc_sock being enqueued multiply.
 *
 *	Some flags can be set to certain values at any time
 *	providing that certain rules are followed:
 *
 *	SK_CONN, SK_DATA, can be set or cleared at any time.
 *		after a set, svc_sock_enqueue must be called.
 *		after a clear, the socket must be read/accepted
 *		 if this succeeds, it must be set again.
 *	SK_CLOSE can set at any time. It is never cleared.
 *      sk_inuse contains a bias of '1' until SK_DEAD is set.
 *             so when sk_inuse hits zero, we know the socket is dead
 *             and no-one is using it.
 *      SK_DEAD can only be set while SK_BUSY is held which ensures
 *             no other thread will be using the socket or will try to
 *	       set SK_DEAD.
 *
 */

#define RPCDBG_FACILITY	RPCDBG_SVCSOCK


static struct svc_sock *svc_setup_socket(struct svc_serv *, struct socket *,
					 int *errp, int flags);
static void		svc_delete_socket(struct svc_sock *svsk);
static void		svc_udp_data_ready(struct sock *, int);
static int		svc_udp_recvfrom(struct svc_rqst *);
static int		svc_udp_sendto(struct svc_rqst *);
static void		svc_close_socket(struct svc_sock *svsk);

static struct svc_deferred_req *svc_deferred_dequeue(struct svc_sock *svsk);
static int svc_deferred_recv(struct svc_rqst *rqstp);
static struct cache_deferred_req *svc_defer(struct cache_req *req);

/* apparently the "standard" is that clients close
 * idle connections after 5 minutes, servers after
 * 6 minutes
 *   http://www.connectathon.org/talks96/nfstcp.pdf
 */
static int svc_conn_age_period = 6*60;

#ifdef CONFIG_DEBUG_LOCK_ALLOC
static struct lock_class_key svc_key[2];
static struct lock_class_key svc_slock_key[2];

static inline void svc_reclassify_socket(struct socket *sock)
{
	struct sock *sk = sock->sk;
	BUG_ON(sock_owned_by_user(sk));
	switch (sk->sk_family) {
	case AF_INET:
		sock_lock_init_class_and_name(sk, "slock-AF_INET-NFSD",
		    &svc_slock_key[0], "sk_lock-AF_INET-NFSD", &svc_key[0]);
		break;

	case AF_INET6:
		sock_lock_init_class_and_name(sk, "slock-AF_INET6-NFSD",
		    &svc_slock_key[1], "sk_lock-AF_INET6-NFSD", &svc_key[1]);
		break;

	default:
		BUG();
	}
}
#else
static inline void svc_reclassify_socket(struct socket *sock)
{
}
#endif

static char *__svc_print_addr(struct sockaddr *addr, char *buf, size_t len)
{
	switch (addr->sa_family) {
	case AF_INET:
		snprintf(buf, len, "%u.%u.%u.%u, port=%u",
			NIPQUAD(((struct sockaddr_in *) addr)->sin_addr),
			ntohs(((struct sockaddr_in *) addr)->sin_port));
		break;

	case AF_INET6:
		snprintf(buf, len, "%x:%x:%x:%x:%x:%x:%x:%x, port=%u",
			NIP6(((struct sockaddr_in6 *) addr)->sin6_addr),
			ntohs(((struct sockaddr_in6 *) addr)->sin6_port));
		break;

	default:
		snprintf(buf, len, "unknown address type: %d", addr->sa_family);
		break;
	}
	return buf;
}

/**
 * svc_print_addr - Format rq_addr field for printing
 * @rqstp: svc_rqst struct containing address to print
 * @buf: target buffer for formatted address
 * @len: length of target buffer
 *
 */
char *svc_print_addr(struct svc_rqst *rqstp, char *buf, size_t len)
{
	return __svc_print_addr(svc_addr(rqstp), buf, len);
}
EXPORT_SYMBOL_GPL(svc_print_addr);

/*
 * Queue up an idle server thread.  Must have pool->sp_lock held.
 * Note: this is really a stack rather than a queue, so that we only
 * use as many different threads as we need, and the rest don't pollute
 * the cache.
 */
static inline void
svc_thread_enqueue(struct svc_pool *pool, struct svc_rqst *rqstp)
{
	list_add(&rqstp->rq_list, &pool->sp_threads);
}

/*
 * Dequeue an nfsd thread.  Must have pool->sp_lock held.
 */
static inline void
svc_thread_dequeue(struct svc_pool *pool, struct svc_rqst *rqstp)
{
	list_del(&rqstp->rq_list);
}

/*
 * Release an skbuff after use
 */
static inline void
svc_release_skb(struct svc_rqst *rqstp)
{
	struct sk_buff *skb = rqstp->rq_skbuff;
	struct svc_deferred_req *dr = rqstp->rq_deferred;

	if (skb) {
		rqstp->rq_skbuff = NULL;

		dprintk("svc: service %p, releasing skb %p\n", rqstp, skb);
		skb_free_datagram(rqstp->rq_sock->sk_sk, skb);
	}
	if (dr) {
		rqstp->rq_deferred = NULL;
		kfree(dr);
	}
}

/*
 * Any space to write?
 */
static inline unsigned long
svc_sock_wspace(struct svc_sock *svsk)
{
	int wspace;

	if (svsk->sk_sock->type == SOCK_STREAM)
		wspace = sk_stream_wspace(svsk->sk_sk);
	else
		wspace = sock_wspace(svsk->sk_sk);

	return wspace;
}

/*
 * Queue up a socket with data pending. If there are idle nfsd
 * processes, wake 'em up.
 *
 */
static void
svc_sock_enqueue(struct svc_sock *svsk)
{
	struct svc_serv	*serv = svsk->sk_server;
	struct svc_pool *pool;
	struct svc_rqst	*rqstp;
	int cpu;

	if (!(svsk->sk_flags &
	      ( (1<<SK_CONN)|(1<<SK_DATA)|(1<<SK_CLOSE)|(1<<SK_DEFERRED)) ))
		return;
	if (test_bit(SK_DEAD, &svsk->sk_flags))
		return;

	cpu = get_cpu();
	pool = svc_pool_for_cpu(svsk->sk_server, cpu);
	put_cpu();

	spin_lock_bh(&pool->sp_lock);

	if (!list_empty(&pool->sp_threads) &&
	    !list_empty(&pool->sp_sockets))
		printk(KERN_ERR
			"svc_sock_enqueue: threads and sockets both waiting??\n");

	if (test_bit(SK_DEAD, &svsk->sk_flags)) {
		/* Don't enqueue dead sockets */
		dprintk("svc: socket %p is dead, not enqueued\n", svsk->sk_sk);
		goto out_unlock;
	}

	/* Mark socket as busy. It will remain in this state until the
	 * server has processed all pending data and put the socket back
	 * on the idle list.  We update SK_BUSY atomically because
	 * it also guards against trying to enqueue the svc_sock twice.
	 */
	if (test_and_set_bit(SK_BUSY, &svsk->sk_flags)) {
		/* Don't enqueue socket while already enqueued */
		dprintk("svc: socket %p busy, not enqueued\n", svsk->sk_sk);
		goto out_unlock;
	}
	BUG_ON(svsk->sk_pool != NULL);
	svsk->sk_pool = pool;

	set_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);
	if (((atomic_read(&svsk->sk_reserved) + serv->sv_max_mesg)*2
	     > svc_sock_wspace(svsk))
	    && !test_bit(SK_CLOSE, &svsk->sk_flags)
	    && !test_bit(SK_CONN, &svsk->sk_flags)) {
		/* Don't enqueue while not enough space for reply */
		dprintk("svc: socket %p  no space, %d*2 > %ld, not enqueued\n",
			svsk->sk_sk, atomic_read(&svsk->sk_reserved)+serv->sv_max_mesg,
			svc_sock_wspace(svsk));
		svsk->sk_pool = NULL;
		clear_bit(SK_BUSY, &svsk->sk_flags);
		goto out_unlock;
	}
	clear_bit(SOCK_NOSPACE, &svsk->sk_sock->flags);


	if (!list_empty(&pool->sp_threads)) {
		rqstp = list_entry(pool->sp_threads.next,
				   struct svc_rqst,
				   rq_list);
		dprintk("svc: socket %p served by daemon %p\n",
			svsk->sk_sk, rqstp);
		svc_thread_dequeue(pool, rqstp);
		if (rqstp->rq_sock)
			printk(KERN_ERR
				"svc_sock_enqueue: server %p, rq_sock=%p!\n",
				rqstp, rqstp->rq_sock);
		rqstp->rq_sock = svsk;
		atomic_inc(&svsk->sk_inuse);
		rqstp->rq_reserved = serv->sv_max_mesg;
		atomic_add(rqstp->rq_reserved, &svsk->sk_reserved);
		BUG_ON(svsk->sk_pool != pool);
		wake_up(&rqstp->rq_wait);
	} else {
		dprintk("svc: socket %p put into queue\n", svsk->sk_sk);
		list_add_tail(&svsk->sk_ready, &pool->sp_sockets);
		BUG_ON(svsk->sk_pool != pool);
	}

out_unlock:
	spin_unlock_bh(&pool->sp_lock);
}

/*
 * Dequeue the first socket.  Must be called with the pool->sp_lock held.
 */
static inline struct svc_sock *
svc_sock_dequeue(struct svc_pool *pool)
{
	struct svc_sock	*svsk;

	if (list_empty(&pool->sp_sockets))
		return NULL;

	svsk = list_entry(pool->sp_sockets.next,
			  struct svc_sock, sk_ready);
	list_del_init(&svsk->sk_ready);

	dprintk("svc: socket %p dequeued, inuse=%d\n",
		svsk->sk_sk, atomic_read(&svsk->sk_inuse));

	return svsk;
}

/*
 * Having read something from a socket, check whether it
 * needs to be re-enqueued.
 * Note: SK_DATA only gets cleared when a read-attempt finds
 * no (or insufficient) data.
 */
static inline void
svc_sock_received(struct svc_sock *svsk)
{
	svsk->sk_pool = NULL;
	clear_bit(SK_BUSY, &svsk->sk_flags);
	svc_sock_enqueue(svsk);
}


/**
 * svc_reserve - change the space reserved for the reply to a request.
 * @rqstp:  The request in question
 * @space: new max space to reserve
 *
 * Each request reserves some space on the output queue of the socket
 * to make sure the reply fits.  This function reduces that reserved
 * space to be the amount of space used already, plus @space.
 *
 */
void svc_reserve(struct svc_rqst *rqstp, int space)
{
	space += rqstp->rq_res.head[0].iov_len;

	if (space < rqstp->rq_reserved) {
		struct svc_sock *svsk = rqstp->rq_sock;
		atomic_sub((rqstp->rq_reserved - space), &svsk->sk_reserved);
		rqstp->rq_reserved = space;

		svc_sock_enqueue(svsk);
	}
}

/*
 * Release a socket after use.
 */
static inline void
svc_sock_put(struct svc_sock *svsk)
{
	if (atomic_dec_and_test(&svsk->sk_inuse)) {
		BUG_ON(! test_bit(SK_DEAD, &svsk->sk_flags));

		dprintk("svc: releasing dead socket\n");
		if (svsk->sk_sock->file)
			sockfd_put(svsk->sk_sock);
		else
			sock_release(svsk->sk_sock);
		if (svsk->sk_info_authunix != NULL)
			svcauth_unix_info_release(svsk->sk_info_authunix);
		kfree(svsk);
	}
}

static void
svc_sock_release(struct svc_rqst *rqstp)
{
	struct svc_sock	*svsk = rqstp->rq_sock;

	svc_release_skb(rqstp);

	svc_free_res_pages(rqstp);
	rqstp->rq_res.page_len = 0;
	rqstp->rq_res.page_base = 0;


	/* Reset response buffer and release
	 * the reservation.
	 * But first, check that enough space was reserved
	 * for the reply, otherwise we have a bug!
	 */
	if ((rqstp->rq_res.len) >  rqstp->rq_reserved)
		printk(KERN_ERR "RPC request reserved %d but used %d\n",
		       rqstp->rq_reserved,
		       rqstp->rq_res.len);

	rqstp->rq_res.head[0].iov_len = 0;
	svc_reserve(rqstp, 0);
	rqstp->rq_sock = NULL;

	svc_sock_put(svsk);
}

/*
 * External function to wake up a server waiting for data
 * This really only makes sense for services like lockd
 * which have exactly one thread anyway.
 */
void
svc_wake_up(struct svc_serv *serv)
{
	struct svc_rqst	*rqstp;
	unsigned int i;
	struct svc_pool *pool;

	for (i = 0; i < serv->sv_nrpools; i++) {
		pool = &serv->sv_pools[i];

		spin_lock_bh(&pool->sp_lock);
		if (!list_empty(&pool->sp_threads)) {
			rqstp = list_entry(pool->sp_threads.next,
					   struct svc_rqst,
					   rq_list);
			dprintk("svc: daemon %p woken up.\n", rqstp);
			/*
			svc_thread_dequeue(pool, rqstp);
			rqstp->rq_sock = NULL;
			 */
			wake_up(&rqstp->rq_wait);
		}
		spin_unlock_bh(&pool->sp_lock);
	}
}

union svc_pktinfo_u {
	struct in_pktinfo pkti;
	struct in6_pktinfo pkti6;
};
#define SVC_PKTINFO_SPACE \
	CMSG_SPACE(sizeof(union svc_pktinfo_u))

static void svc_set_cmsg_data(struct svc_rqst *rqstp, struct cmsghdr *cmh)
{
	switch (rqstp->rq_sock->sk_sk->sk_family) {
	case AF_INET: {
			struct in_pktinfo *pki = CMSG_DATA(cmh);

			cmh->cmsg_level = SOL_IP;
			cmh->cmsg_type = IP_PKTINFO;
			pki->ipi_ifindex = 0;
			pki->ipi_spec_dst.s_addr = rqstp->rq_daddr.addr.s_addr;
			cmh->cmsg_len = CMSG_LEN(sizeof(*pki));
		}
		break;

	case AF_INET6: {
			struct in6_pktinfo *pki = CMSG_DATA(cmh);

			cmh->cmsg_level = SOL_IPV6;
			cmh->cmsg_type = IPV6_PKTINFO;
			pki->ipi6_ifindex = 0;
			ipv6_addr_copy(&pki->ipi6_addr,
					&rqstp->rq_daddr.addr6);
			cmh->cmsg_len = CMSG_LEN(sizeof(*pki));
		}
		break;
	}
	return;
}

/*
 * Generic sendto routine
 */
static int