uipc_socket2.c

RTEMS (Real-Time Executive for Multiprocessor Systems) is a free open source real-time operating sys

/*
 * This file has undergone several changes to reflect the
 * differences between the RTEMS and FreeBSD kernels.
 */

/*
 * Copyright (c) 1982, 1986, 1988, 1990, 1993
 *	The Regents of the University of California.  All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. All advertising materials mentioning features or use of this software
 *    must display the following acknowledgement:
 *	This product includes software developed by the University of
 *	California, Berkeley and its contributors.
 * 4. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 *
 *	@(#)uipc_socket2.c	8.1 (Berkeley) 6/10/93
 * $Id: uipc_socket2.c,v 1.2 2003/01/03 18:09:25 joel Exp $
 */

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/proc.h>
#include <sys/file.h>
#include <sys/buf.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/protosw.h>
#include <sys/stat.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/signalvar.h>
#include <sys/sysctl.h>

/*
 * Primitive routines for operating on sockets and socket buffers
 */

u_long	sb_max = SB_MAX;		/* XXX should be static */
SYSCTL_INT(_kern, KIPC_MAXSOCKBUF, maxsockbuf, CTLFLAG_RW, &sb_max, 0, "");

static	u_long sb_efficiency = 8;	/* parameter for sbreserve() */
SYSCTL_INT(_kern, OID_AUTO, sockbuf_waste_factor, CTLFLAG_RW, &sb_efficiency,
	   0, "");

/*
 * Procedures to manipulate state flags of socket
 * and do appropriate wakeups.  Normal sequence from the
 * active (originating) side is that soisconnecting() is
 * called during processing of connect() call,
 * resulting in an eventual call to soisconnected() if/when the
 * connection is established.  When the connection is torn down
 * soisdisconnecting() is called during processing of disconnect() call,
 * and soisdisconnected() is called when the connection to the peer
 * is totally severed.  The semantics of these routines are such that
 * connectionless protocols can call soisconnected() and soisdisconnected()
 * only, bypassing the in-progress calls when setting up a ``connection''
 * takes no time.
 *
 * From the passive side, a socket is created with
 * two queues of sockets: so_q0 for connections in progress
 * and so_q for connections already made and awaiting user acceptance.
 * As a protocol is preparing incoming connections, it creates a socket
 * structure queued on so_q0 by calling sonewconn().  When the connection
 * is established, soisconnected() is called, and transfers the
 * socket structure to so_q, making it available to accept().
 *
 * If a socket is closed with sockets on either
 * so_q0 or so_q, these sockets are dropped.
 *
 * If higher level protocols are implemented in
 * the kernel, the wakeups done here will sometimes
 * cause software-interrupt process scheduling.
 */

void
soisconnecting(so)
	register struct socket *so;
{

	so->so_state &= ~(SS_ISCONNECTED|SS_ISDISCONNECTING);
	so->so_state |= SS_ISCONNECTING;
}

void
soisconnected(so)
	register struct socket *so;
{
	register struct socket *head = so->so_head;

	so->so_state &= ~(SS_ISCONNECTING|SS_ISDISCONNECTING|SS_ISCONFIRMING);
	so->so_state |= SS_ISCONNECTED;
	if (head && (so->so_state & SS_INCOMP)) {
		TAILQ_REMOVE(&head->so_incomp, so, so_list);
		head->so_incqlen--;
		so->so_state &= ~SS_INCOMP;
		TAILQ_INSERT_TAIL(&head->so_comp, so, so_list);
		so->so_state |= SS_COMP;
		sorwakeup(head);
		soconnwakeup(head);
	} else {
		soconnwakeup(so);
		sorwakeup(so);
		sowwakeup(so);
	}
}

void
soisdisconnecting(so)
	register struct socket *so;
{

	so->so_state &= ~SS_ISCONNECTING;
	so->so_state |= (SS_ISDISCONNECTING|SS_CANTRCVMORE|SS_CANTSENDMORE);
	soconnwakeup(so);
	sowwakeup(so);
	sorwakeup(so);
}

void
soisdisconnected(so)
	register struct socket *so;
{

	so->so_state &= ~(SS_ISCONNECTING|SS_ISCONNECTED|SS_ISDISCONNECTING);
	so->so_state |= (SS_CANTRCVMORE|SS_CANTSENDMORE);
	soconnwakeup(so);
	sowwakeup(so);
	sorwakeup(so);
}

/*
 * Return a random connection that hasn't been serviced yet and
 * is eligible for discard.  There is a one in qlen chance that
 * we will return a null, saying that there are no dropable
 * requests.  In this case, the protocol specific code should drop
 * the new request.  This insures fairness.
 *
 * This may be used in conjunction with protocol specific queue
 * congestion routines.
 */
struct socket *
sodropablereq(head)
	register struct socket *head;
{
	register struct socket *so;
	unsigned int i, j, qlen, m;

	static int rnd;
	static long old_mono_secs;
	static unsigned int cur_cnt, old_cnt;

	if ((i = (m = rtems_bsdnet_seconds_since_boot()) - old_mono_secs) != 0) {
		old_mono_secs = m;
		old_cnt = cur_cnt / i;
		cur_cnt = 0;
	}

	so = TAILQ_FIRST(&head->so_incomp);
	if (!so)
		return (so);

	qlen = head->so_incqlen;
	if (++cur_cnt > qlen || old_cnt > qlen) {
		rnd = (314159 * rnd + 66329) & 0xffff;
		j = ((qlen + 1) * rnd) >> 16;

		while (j-- && so)
		    so = TAILQ_NEXT(so, so_list);
	}

	return (so);
}

/*
 * When an attempt at a new connection is noted on a socket
 * which accepts connections, sonewconn is called.  If the
 * connection is possible (subject to space constraints, etc.)
 * then we allocate a new structure, propoerly linked into the
 * data structure of the original socket, and return this.
 * Connstatus may be 0, or SO_ISCONFIRMING, or SO_ISCONNECTED.
 *
 * Currently, sonewconn() is defined as sonewconn1() in socketvar.h
 * to catch calls that are missing the (new) second parameter.
 */
struct socket *
sonewconn1(head, connstatus)
	register struct socket *head;
	int connstatus;
{
	register struct socket *so;

	if (head->so_qlen > 3 * head->so_qlimit / 2)
		return ((struct socket *)0);
	MALLOC(so, struct socket *, sizeof(*so), M_SOCKET, M_DONTWAIT);
	if (so == NULL)
		return ((struct socket *)0);
	bzero((caddr_t)so, sizeof(*so));
	so->so_head = head;
	so->so_type = head->so_type;
	so->so_options = head->so_options &~ SO_ACCEPTCONN;
	so->so_linger = head->so_linger;
	so->so_state = head->so_state | SS_NOFDREF;
	so->so_proto = head->so_proto;
	so->so_timeo = head->so_timeo;
	so->so_pgid = head->so_pgid;
	so->so_uid = head->so_uid;
	(void) soreserve(so, head->so_snd.sb_hiwat, head->so_rcv.sb_hiwat);
	if (connstatus) {
		TAILQ_INSERT_TAIL(&head->so_comp, so, so_list);
		so->so_state |= SS_COMP;
	} else {
		TAILQ_INSERT_TAIL(&head->so_incomp, so, so_list);
		so->so_state |= SS_INCOMP;
		head->so_incqlen++;
	}
	head->so_qlen++;
	if ((*so->so_proto->pr_usrreqs->pru_attach)(so, 0)) {
		if (so->so_state & SS_COMP) {
			TAILQ_REMOVE(&head->so_comp, so, so_list);
		} else {
			TAILQ_REMOVE(&head->so_incomp, so, so_list);
			head->so_incqlen--;
		}
		head->so_qlen--;
		(void) free((caddr_t)so, M_SOCKET);
		return ((struct socket *)0);
	}
	if (connstatus) {
		sorwakeup(head);
		soconnwakeup(head);
		so->so_state |= connstatus;
	}
	return (so);
}

/*
 * Socantsendmore indicates that no more data will be sent on the
 * socket; it would normally be applied to a socket when the user
 * informs the system that no more data is to be sent, by the protocol
 * code (in case PRU_SHUTDOWN).  Socantrcvmore indicates that no more data
 * will be received, and will normally be applied to the socket by a
 * protocol when it detects that the peer will send no more data.
 * Data queued for reading in the socket may yet be read.
 */

void
socantsendmore(so)
	struct socket *so;
{

	so->so_state |= SS_CANTSENDMORE;
	sowwakeup(so);
}

void
socantrcvmore(so)
	struct socket *so;
{

	so->so_state |= SS_CANTRCVMORE;
	sorwakeup(so);
}

/*
 * Socket buffer (struct sockbuf) utility routines.
 *
 * Each socket contains two socket buffers: one for sending data and
 * one for receiving data.  Each buffer contains a queue of mbufs,
 * information about the number of mbufs and amount of data in the
 * queue, and other fields allowing select() statements and notification
 * on data availability to be implemented.
 *
 * Data stored in a socket buffer is maintained as a list of records.
 * Each record is a list of mbufs chained together with the m_next
 * field.  Records are chained together with the m_nextpkt field. The upper
 * level routine soreceive() expects the following conventions to be
 * observed when placing information in the receive buffer:
 *
 * 1. If the protocol requires each message be preceded by the sender's
 *    name, then a record containing that name must be present before
 *    any associated data (mbuf's must be of type MT_SONAME).
 * 2. If the protocol supports the exchange of ``access rights'' (really
 *    just additional data associated with the message), and there are
 *    ``rights'' to be received, then a record containing this data
 *    should be present (mbuf's must be of type MT_RIGHTS).
 * 3. If a name or rights record exists, then it must be followed by
 *    a data record, perhaps of zero length.
 *
 * Before using a new socket structure it is first necessary to reserve
 * buffer space to the socket, by calling sbreserve().  This should commit
 * some of the available buffer space in the system buffer pool for the
 * socket (currently, it does nothing but enforce limits).  The space
 * should be released by calling sbrelease() when the socket is destroyed.
 */

int
soreserve(so, sndcc, rcvcc)
	register struct socket *so;
	u_long sndcc, rcvcc;
{

	if (sbreserve(&so->so_snd, sndcc) == 0)
		goto bad;
	if (sbreserve(&so->so_rcv, rcvcc) == 0)
		goto bad2;
	if (so->so_rcv.sb_lowat == 0)
		so->so_rcv.sb_lowat = 1;
	if (so->so_snd.sb_lowat == 0)
		so->so_snd.sb_lowat = MCLBYTES;
	if (so->so_snd.sb_lowat > so->so_snd.sb_hiwat)
		so->so_snd.sb_lowat = so->so_snd.sb_hiwat;
	return (0);
bad2:
	sbrelease(&so->so_snd);
bad:
	return (ENOBUFS);
}

/*
 * Allot mbufs to a sockbuf.
 * Attempt to scale mbmax so that mbcnt doesn't become limiting
 * if buffering efficiency is near the normal case.
 */
int
sbreserve(sb, cc)
	struct sockbuf *sb;
	u_long cc;
{

	if (cc > sb_max * MCLBYTES / (MSIZE + MCLBYTES))
		return (0);
	sb->sb_hiwat = cc;
	sb->sb_mbmax = min(cc * sb_efficiency, sb_max);
	if (sb->sb_lowat > sb->sb_hiwat)
		sb->sb_lowat = sb->sb_hiwat;
	return (1);
}

/*
 * Free mbufs held by a socket, and reserved mbuf space.
 */
void
sbrelease(sb)
	struct sockbuf *sb;
{

	sbflush(sb);
	sb->sb_hiwat = sb->sb_mbmax = 0;
}

/*
 * Routines to add and remove
 * data from an mbuf queue.
 *
 * The routines sbappend() or sbappendrecord() are normally called to
 * append new mbufs to a socket buffer, after checking that adequate
 * space is available, comparing the function sbspace() with the amount
 * of data to be added.  sbappendrecord() differs from sbappend() in
 * that data supplied is treated as the beginning of a new record.
 * To place a sender's address, optional access rights, and data in a
 * socket receive buffer, sbappendaddr() should be used.  To place
 * access rights and data in a socket receive buffer, sbappendrights()
 * should be used.  In either case, the new data begins a new record.
 * Note that unlike sbappend() and sbappendrecord(), these routines check
 * for the caller that there will be enough space to store the data.
 * Each fails if there is not enough space, or if it cannot find mbufs
 * to store additional information in.
 *
 * Reliable protocols may use the socket send buffer to hold data
 * awaiting acknowledgement.  Data is normally copied from a socket
 * send buffer in a protocol with m_copy for output to a peer,
 * and then removing the data from the socket buffer with sbdrop()
 * or sbdroprecord() when the data is acknowledged by the peer.
 */

/*
 * Append mbuf chain m to the last record in the
 * socket buffer sb.  The additional space associated
 * the mbuf chain is recorded in sb.  Empty mbufs are
 * discarded and mbufs are compacted where possible.
 */
void
sbappend(sb, m)
	struct sockbuf *sb;
	struct mbuf *m;
{
	register struct mbuf *n;

	if (m == 0)
		return;
	n = sb->sb_mb;
	if (n) {
		while (n->m_nextpkt)
			n = n->m_nextpkt;
		do {
			if (n->m_flags & M_EOR) {
				sbappendrecord(sb, m); /* XXXXXX!!!! */
				return;
			}
		} while (n->m_next && (n = n->m_next));
	}
	sbcompress(sb, m, n);
}

#ifdef SOCKBUF_DEBUG
void
sbcheck(sb)
	register struct sockbuf *sb;
{
	register struct mbuf *m;
	register int len = 0, mbcnt = 0;

	for (m = sb->sb_mb; m; m = m->m_next) {
		len += m->m_len;
		mbcnt += MSIZE;
		if (m->m_flags & M_EXT) /*XXX*/ /* pretty sure this is bogus */
			mbcnt += m->m_ext.ext_size;
		if (m->m_nextpkt)
			panic("sbcheck nextpkt");
	}
	if (len != sb->sb_cc || mbcnt != sb->sb_mbcnt) {
		printf("cc %d != %d || mbcnt %d != %d\n", len, sb->sb_cc,
		    mbcnt, sb->sb_mbcnt);
		panic("sbcheck");
	}
}
#endif

/*
 * As above, except the mbuf chain
 * begins a new record.
 */
void
sbappendrecord(sb, m0)
	register struct sockbuf *sb;
	register struct mbuf *m0;
{
	register struct mbuf *m;

	if (m0 == 0)
		return;
	m = sb->sb_mb;
	if (m)
		while (m->m_nextpkt)
			m = m->m_nextpkt;
	/*
	 * Put the first mbuf on the queue.
	 * Note this permits zero length records.
	 */
	sballoc(sb, m0);
	if (m)
		m->m_nextpkt = m0;
	else
		sb->sb_mb = m0;
	m = m0->m_next;
	m0->m_next = 0;
	if (m && (m0->m_flags & M_EOR)) {
		m0->m_flags &= ~M_EOR;
		m->m_flags |= M_EOR;
	}