/*
 * Implement receive operations on a socket.
 * We depend on the way that records are added to the sockbuf
 * by sbappend*.  In particular, each record (mbufs linked through m_next)
 * must begin with an address if the protocol so specifies,
 * followed by an optional mbuf or mbufs containing ancillary data,
 * and then zero or more mbufs of data.
 * In order to avoid blocking network interrupts for the entire time here,
 * we splx() while doing the actual copy to user space.
 * Although the sockbuf is locked, new data may still be appended,
 * and thus we must maintain consistency of the sockbuf during that time.
 *
 * The caller may receive the data as a single mbuf chain by supplying
 * an mbuf **mp0 for use in returning the chain.  The uio is then used
 * only for the count in uio_resid.
 */
int
soreceive(so, psa, uio, mp0, controlp, flagsp)
	struct socket *so;
	struct sockaddr **psa;
	struct uio *uio;
	struct mbuf **mp0;
	struct mbuf **controlp;
	int *flagsp;
{
	struct mbuf *m, **mp;
	int flags, len, error, offset;
	struct protosw *pr = so->so_proto;
	struct mbuf *nextrecord;
	int moff, type = 0;
	int orig_resid = uio->uio_resid;

	mp = mp0;
	if (psa != NULL)
		*psa = 0;
	if (controlp != NULL)
		*controlp = 0;
	if (flagsp != NULL)
		flags = *flagsp &~ MSG_EOR;
	else
		flags = 0;
	if (flags & MSG_OOB) {
		m = m_get(M_TRYWAIT, MT_DATA);
		if (m == NULL)
			return (ENOBUFS);
		error = (*pr->pr_usrreqs->pru_rcvoob)(so, m, flags & MSG_PEEK);
		if (error)
			goto bad;
		do {
#ifdef ZERO_COPY_SOCKETS
			if (so_zero_copy_receive) {
				vm_page_t pg;
				int disposable;

				if ((m->m_flags & M_EXT)
				 && (m->m_ext.ext_type == EXT_DISPOSABLE))
					disposable = 1;
				else
					disposable = 0;

				pg = PHYS_TO_VM_PAGE(vtophys(mtod(m, caddr_t)));
				if (uio->uio_offset == -1)
					uio->uio_offset =IDX_TO_OFF(pg->pindex);

				error = uiomoveco(mtod(m, void *),
						  min(uio->uio_resid, m->m_len),
						  uio, pg->object,
						  disposable);
			} else
#endif /* ZERO_COPY_SOCKETS */
			error = uiomove(mtod(m, void *),
			    (int) min(uio->uio_resid, m->m_len), uio);
			m = m_free(m);
		} while (uio->uio_resid && error == 0 && m);
bad:
		if (m != NULL)
			m_freem(m);
		return (error);
	}
	if (mp != NULL)
		*mp = NULL;
	if (so->so_state & SS_ISCONFIRMING && uio->uio_resid)
		(*pr->pr_usrreqs->pru_rcvd)(so, 0);

	SOCKBUF_LOCK(&so->so_rcv);
restart:
	SOCKBUF_LOCK_ASSERT(&so->so_rcv);
	error = sblock(&so->so_rcv, SBLOCKWAIT(flags));
	if (error)
		goto out;

	m = so->so_rcv.sb_mb;
	/*
	 * If we have less data than requested, block awaiting more
	 * (subject to any timeout) if:
	 *   1. the current count is less than the low water mark, or
	 *   2. MSG_WAITALL is set, and it is possible to do the entire
	 *	receive operation at once if we block (resid <= hiwat).
	 *   3. MSG_DONTWAIT is not set
	 * If MSG_WAITALL is set but resid is larger than the receive buffer,
	 * we have to do the receive in sections, and thus risk returning
	 * a short count if a timeout or signal occurs after we start.
	 */
	if (m == NULL || (((flags & MSG_DONTWAIT) == 0 &&
	    so->so_rcv.sb_cc < uio->uio_resid) &&
	    (so->so_rcv.sb_cc < so->so_rcv.sb_lowat ||
	    ((flags & MSG_WAITALL) && uio->uio_resid <= so->so_rcv.sb_hiwat)) &&
	    m->m_nextpkt == NULL && (pr->pr_flags & PR_ATOMIC) == 0)) {
		KASSERT(m != NULL || !so->so_rcv.sb_cc,
		    ("receive: m == %p so->so_rcv.sb_cc == %u",
		    m, so->so_rcv.sb_cc));
		if (so->so_error) {
			if (m != NULL)
				goto dontblock;
			error = so->so_error;
			if ((flags & MSG_PEEK) == 0)
				so->so_error = 0;
			goto release;
		}
		if (so->so_rcv.sb_state & SBS_CANTRCVMORE) {
			if (m)
				goto dontblock;
			else
				goto release;
		}
		for (; m != NULL; m = m->m_next)
			if (m->m_type == MT_OOBDATA  || (m->m_flags & M_EOR)) {
				m = so->so_rcv.sb_mb;
				goto dontblock;
			}
		if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 &&
		    (so->so_proto->pr_flags & PR_CONNREQUIRED)) {
			error = ENOTCONN;
			goto release;
		}
		if (uio->uio_resid == 0)
			goto release;
		if ((so->so_state & SS_NBIO) ||
		    (flags & (MSG_DONTWAIT|MSG_NBIO))) {
			error = EWOULDBLOCK;
			goto release;
		}
		SBLASTRECORDCHK(&so->so_rcv);
		SBLASTMBUFCHK(&so->so_rcv);
		sbunlock(&so->so_rcv);
		error = sbwait(&so->so_rcv);
		if (error)
			goto out;
		goto restart;
	}
dontblock:
	SOCKBUF_LOCK_ASSERT(&so->so_rcv);
	if (uio->uio_td)
		uio->uio_td->td_proc->p_stats->p_ru.ru_msgrcv++;
	SBLASTRECORDCHK(&so->so_rcv);
	SBLASTMBUFCHK(&so->so_rcv);
	nextrecord = m->m_nextpkt;
	if (pr->pr_flags & PR_ADDR) {
		KASSERT(m->m_type == MT_SONAME,
		    ("m->m_type == %d", m->m_type));
		orig_resid = 0;
		if (psa != NULL)
			*psa = sodupsockaddr(mtod(m, struct sockaddr *),
			    M_NOWAIT);
		if (flags & MSG_PEEK) {
			m = m->m_next;
		} else {
			sbfree(&so->so_rcv, m);
			so->so_rcv.sb_mb = m_free(m);
			m = so->so_rcv.sb_mb;
		}
	}
	while (m != NULL && m->m_type == MT_CONTROL && error == 0) {
		if (flags & MSG_PEEK) {
			if (controlp != NULL)
				*controlp = m_copy(m, 0, m->m_len);
			m = m->m_next;
		} else {
			sbfree(&so->so_rcv, m);
			so->so_rcv.sb_mb = m->m_next;
			m->m_next = NULL;
			if (pr->pr_domain->dom_externalize) {
				SOCKBUF_UNLOCK(&so->so_rcv);
				error = (*pr->pr_domain->dom_externalize)
				    (m, controlp);
				SOCKBUF_LOCK(&so->so_rcv);
			} else if (controlp != NULL)
				*controlp = m;
			else
				m_freem(m);
			m = so->so_rcv.sb_mb;
		}
		if (controlp != NULL) {
			orig_resid = 0;
			while (*controlp != NULL)
				controlp = &(*controlp)->m_next;
		}
	}
	if (m != NULL) {
		if ((flags & MSG_PEEK) == 0) {
			m->m_nextpkt = nextrecord;
			/*
			 * If nextrecord == NULL (this is a single chain),
			 * then sb_lastrecord may not be valid here if m
			 * was changed earlier.
			 */
			if (nextrecord == NULL) {
				KASSERT(so->so_rcv.sb_mb == m,
					("receive tailq 1"));
				so->so_rcv.sb_lastrecord = m;
			}
		}
		type = m->m_type;
		if (type == MT_OOBDATA)
			flags |= MSG_OOB;
	} else {
		if ((flags & MSG_PEEK) == 0) {
			KASSERT(so->so_rcv.sb_mb == m,("receive tailq 2"));
			so->so_rcv.sb_mb = nextrecord;
			SB_EMPTY_FIXUP(&so->so_rcv);
		}
	}
	SOCKBUF_LOCK_ASSERT(&so->so_rcv);
	SBLASTRECORDCHK(&so->so_rcv);
	SBLASTMBUFCHK(&so->so_rcv);

	moff = 0;
	offset = 0;
	while (m != NULL && uio->uio_resid > 0 && error == 0) {
		SOCKBUF_LOCK_ASSERT(&so->so_rcv);
		if (m->m_type == MT_OOBDATA) {
			if (type != MT_OOBDATA)
				break;
		} else if (type == MT_OOBDATA)
			break;
		else
		    KASSERT(m->m_type == MT_DATA || m->m_type == MT_HEADER,
			("m->m_type == %d", m->m_type));
		so->so_rcv.sb_state &= ~SBS_RCVATMARK;
		len = uio->uio_resid;
		if (so->so_oobmark && len > so->so_oobmark - offset)
			len = so->so_oobmark - offset;
		if (len > m->m_len - moff)
			len = m->m_len - moff;
		/*
		 * If mp is set, just pass back the mbufs.
		 * Otherwise copy them out via the uio, then free.
		 * Sockbuf must be consistent here (points to current mbuf,
		 * it points to next record) when we drop priority;
		 * we must note any additions to the sockbuf when we
		 * block interrupts again.
		 */
		if (mp == NULL) {
			SOCKBUF_LOCK_ASSERT(&so->so_rcv);
			SBLASTRECORDCHK(&so->so_rcv);
			SBLASTMBUFCHK(&so->so_rcv);
			SOCKBUF_UNLOCK(&so->so_rcv);
#ifdef ZERO_COPY_SOCKETS
			if (so_zero_copy_receive) {
				vm_page_t pg;
				int disposable;

				if ((m->m_flags & M_EXT)
				 && (m->m_ext.ext_type == EXT_DISPOSABLE))
					disposable = 1;
				else
					disposable = 0;

				pg = PHYS_TO_VM_PAGE(vtophys(mtod(m, caddr_t) +
					moff));

				if (uio->uio_offset == -1)
					uio->uio_offset =IDX_TO_OFF(pg->pindex);

				error = uiomoveco(mtod(m, char *) + moff,
						  (int)len, uio,pg->object,
						  disposable);
			} else
#endif /* ZERO_COPY_SOCKETS */
			error = uiomove(mtod(m, char *) + moff, (int)len, uio);
			SOCKBUF_LOCK(&so->so_rcv);
			if (error)
				goto release;
		} else
			uio->uio_resid -= len;
		if (len == m->m_len - moff) {
			if (m->m_flags & M_EOR)
				flags |= MSG_EOR;
			if (flags & MSG_PEEK) {
				m = m->m_next;
				moff = 0;
			} else {
				nextrecord = m->m_nextpkt;
				sbfree(&so->so_rcv, m);
				if (mp != NULL) {
					*mp = m;
					mp = &m->m_next;
					so->so_rcv.sb_mb = m = m->m_next;
					*mp = NULL;
				} else {
					so->so_rcv.sb_mb = m_free(m);
					m = so->so_rcv.sb_mb;
				}
				if (m != NULL) {
					m->m_nextpkt = nextrecord;
					if (nextrecord == NULL)
						so->so_rcv.sb_lastrecord = m;
				} else {
					so->so_rcv.sb_mb = nextrecord;
					SB_EMPTY_FIXUP(&so->so_rcv);
				}
				SBLASTRECORDCHK(&so->so_rcv);
				SBLASTMBUFCHK(&so->so_rcv);
			}
		} else {
			if (flags & MSG_PEEK)
				moff += len;
			else {
				if (mp != NULL)
					*mp = m_copym(m, 0, len, M_TRYWAIT);
				m->m_data += len;
				m->m_len -= len;
				so->so_rcv.sb_cc -= len;
			}
		}
		if (so->so_oobmark) {
			if ((flags & MSG_PEEK) == 0) {
				so->so_oobmark -= len;
				if (so->so_oobmark == 0) {
					SOCKBUF_LOCK_ASSERT(&so->so_rcv);
					so->so_rcv.sb_state |= SBS_RCVATMARK;
					break;
				}
			} else {
				offset += len;
				if (offset == so->so_oobmark)
					break;
			}
		}
		if (flags & MSG_EOR)
			break;
		/*
		 * If the MSG_WAITALL flag is set (for non-atomic socket),
		 * we must not quit until "uio->uio_resid == 0" or an error
		 * termination.  If a signal/timeout occurs, return
		 * with a short count but without error.
		 * Keep sockbuf locked against other readers.
		 */
		while (flags & MSG_WAITALL && m == NULL && uio->uio_resid > 0 &&
		    !sosendallatonce(so) && nextrecord == NULL) {
			SOCKBUF_LOCK_ASSERT(&so->so_rcv);
			if (so->so_error || so->so_rcv.sb_state & SBS_CANTRCVMORE)
				break;
			/*
			 * Notify the protocol that some data has been
			 * drained before blocking.
			 */
			if (pr->pr_flags & PR_WANTRCVD && so->so_pcb != NULL) {
				SOCKBUF_UNLOCK(&so->so_rcv);
				(*pr->pr_usrreqs->pru_rcvd)(so, flags);
				SOCKBUF_LOCK(&so->so_rcv);
			}
			SBLASTRECORDCHK(&so->so_rcv);
			SBLASTMBUFCHK(&so->so_rcv);
			error = sbwait(&so->so_rcv);
			if (error)
				goto release;
			m = so->so_rcv.sb_mb;
			if (m != NULL)
				nextrecord = m->m_nextpkt;
		}
	}

	if (m != NULL && pr->pr_flags & PR_ATOMIC) {
		flags |= MSG_TRUNC;
		if ((flags & MSG_PEEK) == 0) {
			SOCKBUF_LOCK_ASSERT(&so->so_rcv);
			(void) sbdroprecord_locked(&so->so_rcv);
		}
	}
	if ((flags & MSG_PEEK) == 0) {
		if (m == NULL) {
			/*
			 * First part is an inline SB_EMPTY_FIXUP().  Second
			 * part makes sure sb_lastrecord is up-to-date if
			 * there is still data in the socket buffer.
			 */
			so->so_rcv.sb_mb = nextrecord;
			if (so->so_rcv.sb_mb == NULL) {
				so->so_rcv.sb_mbtail = NULL;
				so->so_rcv.sb_lastrecord = NULL;
			} else if (nextrecord->m_nextpkt == NULL)
				so->so_rcv.sb_lastrecord = nextrecord;
		}
		SBLASTRECORDCHK(&so->so_rcv);
		SBLASTMBUFCHK(&so->so_rcv);
		if (pr->pr_flags & PR_WANTRCVD && so->so_pcb) {
			SOCKBUF_UNLOCK(&so->so_rcv);
			(*pr->pr_usrreqs->pru_rcvd)(so, flags);
			SOCKBUF_LOCK(&so->so_rcv);
		}
	}
	SOCKBUF_LOCK_ASSERT(&so->so_rcv);
	if (orig_resid == uio->uio_resid && orig_resid &&
	    (flags & MSG_EOR) == 0 && (so->so_rcv.sb_state & SBS_CANTRCVMORE) == 0) {
		sbunlock(&so->so_rcv);
		goto restart;
	}

	if (flagsp != NULL)
		*flagsp |= flags;
release:
	SOCKBUF_LOCK_ASSERT(&so->so_rcv);
	sbunlock(&so->so_rcv);
out:
	SOCKBUF_UNLOCK(&so->so_rcv);
	return (error);
}