tcp.c

GNU Hurd 源代码

				}
				if (signal_pending(current)) {
					err = -ERESTARTSYS;
					goto do_interrupted;
				}
				tcp_push_pending_frames(sk, tp);
				wait_for_tcp_memory(sk);

				/* If SACK's were formed or PMTU events happened,
				 * we must find out about it.
				 */
				mss_now = tcp_current_mss(sk);
				continue;
			}

			seglen -= copy;

			/* Prepare control bits for TCP header creation engine. */
			TCP_SKB_CB(skb)->flags = (TCPCB_FLAG_ACK |
						  ((PSH_NEEDED || psh) ?
						   TCPCB_FLAG_PSH : 0));
			TCP_SKB_CB(skb)->sacked = 0;
			if (flags & MSG_OOB) {
				TCP_SKB_CB(skb)->flags |= TCPCB_FLAG_URG;
				TCP_SKB_CB(skb)->urg_ptr = copy;
			} else
				TCP_SKB_CB(skb)->urg_ptr = 0;

			/* TCP data bytes are SKB_PUT() on top, later
			 * TCP+IP+DEV headers are SKB_PUSH()'d beneath.
			 * Reserve header space and checksum the data.
			 */
			skb_reserve(skb, MAX_HEADER + sk->prot->max_header);
			skb->csum = csum_and_copy_from_user(from,
					skb_put(skb, copy), copy, 0, &err);

			if (err)
				goto do_fault;

			from += copy;
			copied += copy;

			TCP_SKB_CB(skb)->seq = tp->write_seq;
			TCP_SKB_CB(skb)->end_seq = TCP_SKB_CB(skb)->seq + copy;

			/* This advances tp->write_seq for us. */
			tcp_send_skb(sk, skb, queue_it);
		}
	}
	sk->err = 0;
	err = copied;
	goto out;

do_sock_err:
	if(copied)
		err = copied;
	else
		err = sock_error(sk);
	goto out;
do_shutdown:
	if(copied)
		err = copied;
	else {
		if (!(flags&MSG_NOSIGNAL))
			send_sig(SIGPIPE, current, 0);
		err = -EPIPE;
	}
	goto out;
do_interrupted:
	if(copied)
		err = copied;
	goto out;
do_fault:
	kfree_skb(skb);
do_fault2:
	err = -EFAULT;
out:
	tcp_push_pending_frames(sk, tp);
	release_sock(sk);
	return err;
}

#undef PSH_NEEDED

/*
 *	Send an ack if one is backlogged at this point. Ought to merge
 *	this with tcp_send_ack().
 *      This is called for delayed acks also.
 */

void tcp_read_wakeup(struct sock *sk)
{
	/* If we're closed, don't send an ack, or we'll get a RST
	 * from the closed destination.
	 */
	if (sk->state != TCP_CLOSE)
		tcp_send_ack(sk);
}

/*
 *	Handle reading urgent data. BSD has very simple semantics for
 *	this, no blocking and very strange errors 8)
 */

static int tcp_recv_urg(struct sock * sk, int nonblock,
			struct msghdr *msg, int len, int flags,
			int *addr_len)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);

	/* No URG data to read. */
	if (sk->urginline || !tp->urg_data || tp->urg_data == URG_READ)
		return -EINVAL;	/* Yes this is right ! */

	if (sk->err)
		return sock_error(sk);

	if (sk->done)
		return -ENOTCONN;

	if (sk->state == TCP_CLOSE || (sk->shutdown & RCV_SHUTDOWN)) {
		sk->done = 1;
		return 0;
	}

	lock_sock(sk);
	if (tp->urg_data & URG_VALID) {
		int err = 0;
		char c = tp->urg_data;

		if (!(flags & MSG_PEEK))
			tp->urg_data = URG_READ;

		if(msg->msg_name)
			tp->af_specific->addr2sockaddr(sk, (struct sockaddr *)
						       msg->msg_name);

		if(addr_len)
			*addr_len = tp->af_specific->sockaddr_len;

		/* Read urgent data. */
		msg->msg_flags|=MSG_OOB;
		release_sock(sk);

		if(len>0)
		{
			err = memcpy_toiovec(msg->msg_iov, &c, 1);
			/* N.B. already set above ... */
			msg->msg_flags|=MSG_OOB;
		}
		else
			msg->msg_flags|=MSG_TRUNC;

		/* N.B. Is this right?? If len == 0 we didn't read any data */
		return err ? -EFAULT : 1;
	}
	release_sock(sk);

	/* Fixed the recv(..., MSG_OOB) behaviour.  BSD docs and
	 * the available implementations agree in this case:
	 * this call should never block, independent of the
	 * blocking state of the socket.
	 * Mike <[e-mail]>
	 */
	return -EAGAIN;
}

/*
 *	Release a skb if it is no longer needed. This routine
 *	must be called with interrupts disabled or with the
 *	socket locked so that the sk_buff queue operation is ok.
 */

static inline void tcp_eat_skb(struct sock *sk, struct sk_buff * skb)
{
	__skb_unlink(skb, &sk->receive_queue);
	kfree_skb(skb);
}

/* Clean up the receive buffer for full frames taken by the user,
 * then send an ACK if necessary.  COPIED is the number of bytes
 * tcp_recvmsg has given to the user so far, it speeds up the
 * calculation of whether or not we must ACK for the sake of
 * a window update.
 */
static void cleanup_rbuf(struct sock *sk, int copied)
{
	struct sk_buff *skb;

	/* NOTE! The socket must be locked, so that we don't get
	 * a messed-up receive queue.
	 */
	while ((skb=skb_peek(&sk->receive_queue)) != NULL) {
		if (!skb->used || atomic_read(&skb->users) > 1)
			break;
		tcp_eat_skb(sk, skb);
	}

  	/* We send an ACK if we can now advertise a non-zero window
	 * which has been raised "significantly".
  	 */
	if(copied > 0) {
		struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
		__u32 rcv_window_now = tcp_receive_window(tp);
		__u32 new_window = __tcp_select_window(sk);

		/* We won't be raising the window any further than
		 * the window-clamp allows.  Our window selection
		 * also keeps things a nice multiple of MSS.  These
		 * checks are necessary to prevent spurious ACKs
		 * which don't advertize a larger window.
		 */
		if((new_window && (new_window >= rcv_window_now * 2)) &&
		   ((rcv_window_now + tp->mss_cache) <= tp->window_clamp))
			tcp_read_wakeup(sk);
	}
}


/*
 *	This routine copies from a sock struct into the user buffer.
 */

int tcp_recvmsg(struct sock *sk, struct msghdr *msg,
		int len, int nonblock, int flags, int *addr_len)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
	struct wait_queue wait = { current, NULL };
	int copied = 0;
	u32 peek_seq;
	volatile u32 *seq;	/* So gcc doesn't overoptimise */
	unsigned long used;
	int err = 0;
	int target = 1;		/* Read at least this many bytes */

	if (sk->err)
		return sock_error(sk);

	if (sk->state == TCP_LISTEN)
		return -ENOTCONN;

	/* Urgent data needs to be handled specially. */
	if (flags & MSG_OOB)
		return tcp_recv_urg(sk, nonblock, msg, len, flags, addr_len);

	/*	Copying sequence to update. This is volatile to handle
	 *	the multi-reader case neatly (memcpy_to/fromfs might be
	 *	inline and thus not flush cached variables otherwise).
	 */
	peek_seq = tp->copied_seq;
	seq = &tp->copied_seq;
	if (flags & MSG_PEEK)
		seq = &peek_seq;

	/* Handle the POSIX bogosity MSG_WAITALL. */
	if (flags & MSG_WAITALL)
		target=len;

	add_wait_queue(sk->sleep, &wait);
	lock_sock(sk);

	/*
	 *	BUG BUG BUG
	 *	This violates 1003.1g compliance. We must wait for
	 *	data to exist even if we read none!
	 */

	while (len > 0) {
		struct sk_buff * skb;
		u32 offset;

		/* Are we at urgent data? Stop if we have read anything. */
		if (copied && tp->urg_data && tp->urg_seq == *seq)
			break;

		/* We need to check signals first, to get correct SIGURG
		 * handling. FIXME: Need to check this doesnt impact 1003.1g
		 * and move it down to the bottom of the loop
		 */
		if (signal_pending(current)) {
			if (copied)
				break;
			copied = -ERESTARTSYS;
			if (nonblock)
				copied = -EAGAIN;
			break;
		}

		/* Next get a buffer. */
		current->state = TASK_INTERRUPTIBLE;

		skb = skb_peek(&sk->receive_queue);
		do {
			if (!skb)
				break;

			/* Now that we have two receive queues this
			 * shouldn't happen.
			 */
			if (before(*seq, TCP_SKB_CB(skb)->seq)) {
				printk(KERN_INFO "recvmsg bug: copied %X seq %X\n",
				       *seq, TCP_SKB_CB(skb)->seq);
				break;
			}
			offset = *seq - TCP_SKB_CB(skb)->seq;
			if (skb->h.th->syn)
				offset--;
			if (offset < skb->len)
				goto found_ok_skb;
			if (skb->h.th->fin)
				goto found_fin_ok;
			if (!(flags & MSG_PEEK))
				skb->used = 1;
			skb = skb->next;
		} while (skb != (struct sk_buff *)&sk->receive_queue);

		if (copied >= target)
			break;

		/*
		   These three lines and clause if (sk->state == TCP_CLOSE)
		   are unlikely to be correct, if target > 1.
		   I DO NOT FIX IT, because I have no idea, what
		   POSIX prescribes to make here. Probably, it really
		   wants to lose data 8), if not all target is received.
		                                                 --ANK
		 */
		if (sk->err && !(flags&MSG_PEEK)) {
			copied = sock_error(sk);
			break;
		}

		if (sk->shutdown & RCV_SHUTDOWN) {
			sk->done = 1;
			break;
		}

		if (sk->state == TCP_CLOSE) {
			if (!sk->done) {
				sk->done = 1;
				break;
			}
			copied = -ENOTCONN;
			break;
		}

		if (nonblock) {
			copied = -EAGAIN;
			break;
		}

		cleanup_rbuf(sk, copied);
		release_sock(sk);
		sk->socket->flags |= SO_WAITDATA;
		schedule();
		sk->socket->flags &= ~SO_WAITDATA;
		lock_sock(sk);
		continue;

	found_ok_skb:
		/*	Lock the buffer. We can be fairly relaxed as
		 *	an interrupt will never steal a buffer we are
		 *	using unless I've missed something serious in
		 *	tcp_data.
		 */
		atomic_inc(&skb->users);

		/* Ok so how much can we use? */
		used = skb->len - offset;
		if (len < used)
			used = len;

		/* Do we have urgent data here? */
		if (tp->urg_data) {
			u32 urg_offset = tp->urg_seq - *seq;
			if (urg_offset < used) {
				if (!urg_offset) {
					if (!sk->urginline) {
						++*seq;
						offset++;
						used--;
					}
				} else
					used = urg_offset;
			}
		}

		/*	Copy it - We _MUST_ update *seq first so that we
		 *	don't ever double read when we have dual readers
		 */
		*seq += used;

		/*	This memcpy_toiovec can sleep. If it sleeps and we
		 *	do a second read it relies on the skb->users to avoid
		 *	a crash when cleanup_rbuf() gets called.
		 */
		err = memcpy_toiovec(msg->msg_iov, ((unsigned char *)skb->h.th) + skb->h.th->doff*4 + offset, used);
		if (err) {
			/* Exception. Bailout! */
			atomic_dec(&skb->users);
			copied = -EFAULT;
			break;
		}

		copied += used;
		len -= used;

		/*	We now will not sleep again until we are finished
		 *	with skb. Sorry if you are doing the SMP port
		 *	but you'll just have to fix it neatly ;)
		 */
		atomic_dec(&skb->users);

		if (after(tp->copied_seq,tp->urg_seq))
			tp->urg_data = 0;
		if (used + offset < skb->len)
			continue;

		/*	Process the FIN. We may also need to handle PSH
		 *	here and make it break out of MSG_WAITALL.
		 */
		if (skb->h.th->fin)
			goto found_fin_ok;
		if (flags & MSG_PEEK)
			continue;
		skb->used = 1;
		if (atomic_read(&skb->users) == 1)
			tcp_eat_skb(sk, skb);
		continue;

	found_fin_ok:
		++*seq;
		if (flags & MSG_PEEK)
			break;

		/* All is done. */
		skb->used = 1;
		sk->shutdown |= RCV_SHUTDOWN;
		break;
	}

	if(copied >= 0 && msg->msg_name) {
		tp->af_specific->addr2sockaddr(sk, (struct sockaddr *)
					       msg->msg_name);
		if(addr_len)
			*addr_len = tp->af_specific->sockaddr_len;
	}

	remove_wait_queue(sk->sleep, &wait);
	current->state = TASK_RUNNING;

	/* Clean up data we have read: This will do ACK frames. */
	cleanup_rbuf(sk, copied);
	release_sock(sk);
	return copied;
}

/*
 * Check whether to renew the timer.
 */
static inline void tcp_check_fin_timer(struct sock *sk)
{
	if (sk->state == TCP_FIN_WAIT2 && !sk->timer.prev)
		tcp_reset_msl_timer(sk, TIME_CLOSE, sysctl_tcp_fin_timeout);
}

/*
 *	State processing on a close. This implements the state shift for
 *	sending our FIN frame. Note that we only send a FIN for some
 *	states. A shutdown() may have already sent the FIN, or we may be
 *	closed.
 */