uipc_usrreq.c

早期freebsd实现

	vattr.va_type = VSOCK;
	vattr.va_mode = ACCESSPERMS;
	LEASE_CHECK(nd.ni_dvp, p, p->p_ucred, LEASE_WRITE);
	if (error = VOP_CREATE(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr))
		return (error);
	vp = nd.ni_vp;
	vp->v_socket = unp->unp_socket;
	unp->unp_vnode = vp;
	unp->unp_addr = m_copy(nam, 0, (int)M_COPYALL);
	VOP_UNLOCK(vp);
	return (0);
}

unp_connect(so, nam, p)
	struct socket *so;
	struct mbuf *nam;
	struct proc *p;
{
	register struct sockaddr_un *soun = mtod(nam, struct sockaddr_un *);
	register struct vnode *vp;
	register struct socket *so2, *so3;
	struct unpcb *unp2, *unp3;
	int error;
	struct nameidata nd;

	NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE, soun->sun_path, p);
	if (nam->m_data + nam->m_len == &nam->m_dat[MLEN]) {	/* XXX */
		if (*(mtod(nam, caddr_t) + nam->m_len - 1) != 0)
			return (EMSGSIZE);
	} else
		*(mtod(nam, caddr_t) + nam->m_len) = 0;
	if (error = namei(&nd))
		return (error);
	vp = nd.ni_vp;
	if (vp->v_type != VSOCK) {
		error = ENOTSOCK;
		goto bad;
	}
	if (error = VOP_ACCESS(vp, VWRITE, p->p_ucred, p))
		goto bad;
	so2 = vp->v_socket;
	if (so2 == 0) {
		error = ECONNREFUSED;
		goto bad;
	}
	if (so->so_type != so2->so_type) {
		error = EPROTOTYPE;
		goto bad;
	}
	if (so->so_proto->pr_flags & PR_CONNREQUIRED) {
		if ((so2->so_options & SO_ACCEPTCONN) == 0 ||
		    (so3 = sonewconn(so2, 0)) == 0) {
			error = ECONNREFUSED;
			goto bad;
		}
		unp2 = sotounpcb(so2);
		unp3 = sotounpcb(so3);
		if (unp2->unp_addr)
			unp3->unp_addr =
				  m_copy(unp2->unp_addr, 0, (int)M_COPYALL);
		so2 = so3;
	}
	error = unp_connect2(so, so2);
bad:
	vput(vp);
	return (error);
}

unp_connect2(so, so2)
	register struct socket *so;
	register struct socket *so2;
{
	register struct unpcb *unp = sotounpcb(so);
	register struct unpcb *unp2;

	if (so2->so_type != so->so_type)
		return (EPROTOTYPE);
	unp2 = sotounpcb(so2);
	unp->unp_conn = unp2;
	switch (so->so_type) {

	case SOCK_DGRAM:
		unp->unp_nextref = unp2->unp_refs;
		unp2->unp_refs = unp;
		soisconnected(so);
		break;

	case SOCK_STREAM:
		unp2->unp_conn = unp;
		soisconnected(so);
		soisconnected(so2);
		break;

	default:
		panic("unp_connect2");
	}
	return (0);
}

unp_disconnect(unp)
	struct unpcb *unp;
{
	register struct unpcb *unp2 = unp->unp_conn;

	if (unp2 == 0)
		return;
	unp->unp_conn = 0;
	switch (unp->unp_socket->so_type) {

	case SOCK_DGRAM:
		if (unp2->unp_refs == unp)
			unp2->unp_refs = unp->unp_nextref;
		else {
			unp2 = unp2->unp_refs;
			for (;;) {
				if (unp2 == 0)
					panic("unp_disconnect");
				if (unp2->unp_nextref == unp)
					break;
				unp2 = unp2->unp_nextref;
			}
			unp2->unp_nextref = unp->unp_nextref;
		}
		unp->unp_nextref = 0;
		unp->unp_socket->so_state &= ~SS_ISCONNECTED;
		break;

	case SOCK_STREAM:
		soisdisconnected(unp->unp_socket);
		unp2->unp_conn = 0;
		soisdisconnected(unp2->unp_socket);
		break;
	}
}

#ifdef notdef
unp_abort(unp)
	struct unpcb *unp;
{

	unp_detach(unp);
}
#endif

unp_shutdown(unp)
	struct unpcb *unp;
{
	struct socket *so;

	if (unp->unp_socket->so_type == SOCK_STREAM && unp->unp_conn &&
	    (so = unp->unp_conn->unp_socket))
		socantrcvmore(so);
}

unp_drop(unp, errno)
	struct unpcb *unp;
	int errno;
{
	struct socket *so = unp->unp_socket;

	so->so_error = errno;
	unp_disconnect(unp);
	if (so->so_head) {
		so->so_pcb = (caddr_t) 0;
		m_freem(unp->unp_addr);
		(void) m_free(dtom(unp));
		sofree(so);
	}
}

#ifdef notdef
unp_drain()
{

}
#endif

unp_externalize(rights)
	struct mbuf *rights;
{
	struct proc *p = curproc;		/* XXX */
	register int i;
	register struct cmsghdr *cm = mtod(rights, struct cmsghdr *);
	register struct file **rp = (struct file **)(cm + 1);
	register struct file *fp;
	int newfds = (cm->cmsg_len - sizeof(*cm)) / sizeof (int);
	int f;

	if (!fdavail(p, newfds)) {
		for (i = 0; i < newfds; i++) {
			fp = *rp;
			unp_discard(fp);
			*rp++ = 0;
		}
		return (EMSGSIZE);
	}
	for (i = 0; i < newfds; i++) {
		if (fdalloc(p, 0, &f))
			panic("unp_externalize");
		fp = *rp;
		p->p_fd->fd_ofiles[f] = fp;
		fp->f_msgcount--;
		unp_rights--;
		*(int *)rp++ = f;
	}
	return (0);
}

unp_internalize(control, p)
	struct mbuf *control;
	struct proc *p;
{
	struct filedesc *fdp = p->p_fd;
	register struct cmsghdr *cm = mtod(control, struct cmsghdr *);
	register struct file **rp;
	register struct file *fp;
	register int i, fd;
	int oldfds;

	if (cm->cmsg_type != SCM_RIGHTS || cm->cmsg_level != SOL_SOCKET ||
	    cm->cmsg_len != control->m_len)
		return (EINVAL);
	oldfds = (cm->cmsg_len - sizeof (*cm)) / sizeof (int);
	rp = (struct file **)(cm + 1);
	for (i = 0; i < oldfds; i++) {
		fd = *(int *)rp++;
		if ((unsigned)fd >= fdp->fd_nfiles ||
		    fdp->fd_ofiles[fd] == NULL)
			return (EBADF);
	}
	rp = (struct file **)(cm + 1);
	for (i = 0; i < oldfds; i++) {
		fp = fdp->fd_ofiles[*(int *)rp];
		*rp++ = fp;
		fp->f_count++;
		fp->f_msgcount++;
		unp_rights++;
	}
	return (0);
}

int	unp_defer, unp_gcing;
int	unp_mark();
extern	struct domain unixdomain;

unp_gc()
{
	register struct file *fp, *nextfp;
	register struct socket *so;
	struct file **extra_ref, **fpp;
	int nunref, i;

	if (unp_gcing)
		return;
	unp_gcing = 1;
	unp_defer = 0;
	for (fp = filehead; fp; fp = fp->f_filef)
		fp->f_flag &= ~(FMARK|FDEFER);
	do {
		for (fp = filehead; fp; fp = fp->f_filef) {
			if (fp->f_count == 0)
				continue;
			if (fp->f_flag & FDEFER) {
				fp->f_flag &= ~FDEFER;
				unp_defer--;
			} else {
				if (fp->f_flag & FMARK)
					continue;
				if (fp->f_count == fp->f_msgcount)
					continue;
				fp->f_flag |= FMARK;
			}
			if (fp->f_type != DTYPE_SOCKET ||
			    (so = (struct socket *)fp->f_data) == 0)
				continue;
			if (so->so_proto->pr_domain != &unixdomain ||
			    (so->so_proto->pr_flags&PR_RIGHTS) == 0)
				continue;
#ifdef notdef
			if (so->so_rcv.sb_flags & SB_LOCK) {
				/*
				 * This is problematical; it's not clear
				 * we need to wait for the sockbuf to be
				 * unlocked (on a uniprocessor, at least),
				 * and it's also not clear what to do
				 * if sbwait returns an error due to receipt
				 * of a signal.  If sbwait does return
				 * an error, we'll go into an infinite
				 * loop.  Delete all of this for now.
				 */
				(void) sbwait(&so->so_rcv);
				goto restart;
			}
#endif
			unp_scan(so->so_rcv.sb_mb, unp_mark);
		}
	} while (unp_defer);
	/*
	 * We grab an extra reference to each of the file table entries
	 * that are not otherwise accessible and then free the rights
	 * that are stored in messages on them.
	 *
	 * The bug in the orginal code is a little tricky, so I'll describe
	 * what's wrong with it here.
	 *
	 * It is incorrect to simply unp_discard each entry for f_msgcount
	 * times -- consider the case of sockets A and B that contain
	 * references to each other.  On a last close of some other socket,
	 * we trigger a gc since the number of outstanding rights (unp_rights)
	 * is non-zero.  If during the sweep phase the gc code un_discards,
	 * we end up doing a (full) closef on the descriptor.  A closef on A
	 * results in the following chain.  Closef calls soo_close, which
	 * calls soclose.   Soclose calls first (through the switch
	 * uipc_usrreq) unp_detach, which re-invokes unp_gc.  Unp_gc simply
	 * returns because the previous instance had set unp_gcing, and
	 * we return all the way back to soclose, which marks the socket
	 * with SS_NOFDREF, and then calls sofree.  Sofree calls sorflush
	 * to free up the rights that are queued in messages on the socket A,
	 * i.e., the reference on B.  The sorflush calls via the dom_dispose
	 * switch unp_dispose, which unp_scans with unp_discard.  This second
	 * instance of unp_discard just calls closef on B.
	 *
	 * Well, a similar chain occurs on B, resulting in a sorflush on B,
	 * which results in another closef on A.  Unfortunately, A is already
	 * being closed, and the descriptor has already been marked with
	 * SS_NOFDREF, and soclose panics at this point.
	 *
	 * Here, we first take an extra reference to each inaccessible
	 * descriptor.  Then, we call sorflush ourself, since we know
	 * it is a Unix domain socket anyhow.  After we destroy all the
	 * rights carried in messages, we do a last closef to get rid
	 * of our extra reference.  This is the last close, and the
	 * unp_detach etc will shut down the socket.
	 *
	 * 91/09/19, [e-mail]
	 */
	extra_ref = malloc(nfiles * sizeof(struct file *), M_FILE, M_WAITOK);
	for (nunref = 0, fp = filehead, fpp = extra_ref; fp; fp = nextfp) {
		nextfp = fp->f_filef;
		if (fp->f_count == 0)
			continue;
		if (fp->f_count == fp->f_msgcount && !(fp->f_flag & FMARK)) {
			*fpp++ = fp;
			nunref++;
			fp->f_count++;
		}
	}
	for (i = nunref, fpp = extra_ref; --i >= 0; ++fpp)
		sorflush((struct socket *)(*fpp)->f_data);
	for (i = nunref, fpp = extra_ref; --i >= 0; ++fpp)
		closef(*fpp);
	free((caddr_t)extra_ref, M_FILE);
	unp_gcing = 0;
}

unp_dispose(m)
	struct mbuf *m;
{
	int unp_discard();

	if (m)
		unp_scan(m, unp_discard);
}

unp_scan(m0, op)
	register struct mbuf *m0;
	int (*op)();
{
	register struct mbuf *m;
	register struct file **rp;
	register struct cmsghdr *cm;
	register int i;
	int qfds;

	while (m0) {
		for (m = m0; m; m = m->m_next)
			if (m->m_type == MT_CONTROL &&
			    m->m_len >= sizeof(*cm)) {
				cm = mtod(m, struct cmsghdr *);
				if (cm->cmsg_level != SOL_SOCKET ||
				    cm->cmsg_type != SCM_RIGHTS)
					continue;
				qfds = (cm->cmsg_len - sizeof *cm)
						/ sizeof (struct file *);
				rp = (struct file **)(cm + 1);
				for (i = 0; i < qfds; i++)
					(*op)(*rp++);
				break;		/* XXX, but saves time */
			}
		m0 = m0->m_act;
	}
}

unp_mark(fp)
	struct file *fp;
{

	if (fp->f_flag & FMARK)
		return;
	unp_defer++;
	fp->f_flag |= (FMARK|FDEFER);
}

unp_discard(fp)
	struct file *fp;
{

	fp->f_msgcount--;
	unp_rights--;
	(void) closef(fp, (struct proc *)NULL);
}