Listening sockets improvements.

o Separate fields of struct socket that belong to listening from
  fields that belong to normal dataflow, and unionize them.  This
  shrinks the structure a bit.
  - Take out selinfo's from the socket buffers into the socket. The
    first reason is to support braindamaged scenario when a socket is
    added to kevent(2) and then listen(2) is cast on it. The second
    reason is that there is future plan to make socket buffers pluggable,
    so that for a dataflow socket a socket buffer can be changed, and
    in this case we also want to keep same selinfos through the lifetime
    of a socket.
  - Remove struct struct so_accf. Since now listening stuff no longer
    affects struct socket size, just move its fields into listening part
    of the union.
  - Provide sol_upcall field and enforce that so_upcall_set() may be called
    only on a dataflow socket, which has buffers, and for listening sockets
    provide solisten_upcall_set().

o Remove ACCEPT_LOCK() global.
  - Add a mutex to socket, to be used instead of socket buffer lock to lock
    fields of struct socket that don't belong to a socket buffer.
  - Allow to acquire two socket locks, but the first one must belong to a
    listening socket.
  - Make soref()/sorele() to use atomic(9).  This allows in some situations
    to do soref() without owning socket lock.  There is place for improvement
    here, it is possible to make sorele() also to lock optionally.
  - Most protocols aren't touched by this change, except UNIX local sockets.
    See below for more information.

o Reduce copy-and-paste in kernel modules that accept connections from
  listening sockets: provide function solisten_dequeue(), and use it in
  the following modules: ctl(4), iscsi(4), ng_btsocket(4), ng_ksocket(4),
  infiniband, rpc.

o UNIX local sockets.
  - Removal of ACCEPT_LOCK() global uncovered several races in the UNIX
    local sockets.  Most races exist around spawning a new socket, when we
    are connecting to a local listening socket.  To cover them, we need to
    hold locks on both PCBs when spawning a third one.  This means holding
    them across sonewconn().  This creates a LOR between pcb locks and
    unp_list_lock.
  - To fix the new LOR, abandon the global unp_list_lock in favor of global
    unp_link_lock.  Indeed, separating these two locks didn't provide us any
    extra parralelism in the UNIX sockets.
  - Now call into uipc_attach() may happen with unp_link_lock hold if, we
    are accepting, or without unp_link_lock in case if we are just creating
    a socket.
  - Another problem in UNIX sockets is that uipc_close() basicly did nothing
    for a listening socket.  The vnode remained opened for connections.  This
    is fixed by removing vnode in uipc_close().  Maybe the right way would be
    to do it for all sockets (not only listening), simply move the vnode
    teardown from uipc_detach() to uipc_close()?

Sponsored by:		Netflix
Differential Revision:	https://reviews.freebsd.org/D9770

2 files changed, 23 insertions, 55 deletions

diff --git a/sys/ofed/drivers/infiniband/core/iwcm.c b/sys/ofed/drivers/infiniband/core/iwcm.c
index fa6b674bb0db..a40d9dc8bbe9 100644
--- a/sys/ofed/drivers/infiniband/core/iwcm.c
+++ b/sys/ofed/drivers/infiniband/core/iwcm.c
@@ -416,34 +416,19 @@ dequeue_socket(struct socket *head)
 {
 	struct socket *so;
 	struct sockaddr_in *remote;
+	int error;
 
-	ACCEPT_LOCK();
-	so = TAILQ_FIRST(&head->so_comp);
-	if (!so) {
-		ACCEPT_UNLOCK();
-		return NULL;
-	}
-
-	SOCK_LOCK(so);
-	/*
-	 * Before changing the flags on the socket, we have to bump the
-	 * reference count.  Otherwise, if the protocol calls sofree(),
-	 * the socket will be released due to a zero refcount.
-	 */
-	soref(so);
-	TAILQ_REMOVE(&head->so_comp, so, so_list);
-	head->so_qlen--;
-	so->so_qstate &= ~SQ_COMP;
-	so->so_head = NULL;
-	so->so_state |= SS_NBIO;
-	SOCK_UNLOCK(so);
-	ACCEPT_UNLOCK();
+	SOLISTEN_LOCK(head);
+	error = solisten_dequeue(head, &so, SOCK_NONBLOCK);
+	if (error == EWOULDBLOCK)
+		return (NULL);
 	remote = NULL;
 	soaccept(so, (struct sockaddr **)&remote);
 
 	free(remote, M_SONAME);
 	return so;
 }
+
 static void
 iw_so_event_handler(struct work_struct *_work)
 {
@@ -485,18 +470,17 @@ err:
 #endif
 	return;
 }
+
 static int
 iw_so_upcall(struct socket *parent_so, void *arg, int waitflag)
 {
 	struct iwcm_listen_work *work;
-	struct socket *so;
 	struct iw_cm_id *cm_id = arg;
 
 	/* check whether iw_so_event_handler() already dequeued this 'so' */
-	so = TAILQ_FIRST(&parent_so->so_comp);
-	if (!so)
+	if (TAILQ_EMPTY(&parent_so->sol_comp))
 		return SU_OK;
-	work = kzalloc(sizeof(*work), M_NOWAIT);
+	work = kzalloc(sizeof(*work), waitflag);
 	if (!work)
 		return -ENOMEM;
 	work->cm_id = cm_id;
@@ -507,17 +491,21 @@ iw_so_upcall(struct socket *parent_so, void *arg, int waitflag)
 	return SU_OK;
 }
 
-static void
-iw_init_sock(struct iw_cm_id *cm_id)
+static int
+iw_create_listen(struct iw_cm_id *cm_id, int backlog)
 {
 	struct sockopt sopt;
 	struct socket *so = cm_id->so;
 	int on = 1;
+	int rc;
 
-	SOCK_LOCK(so);
-	soupcall_set(so, SO_RCV, iw_so_upcall, cm_id);
+	rc = -solisten(cm_id->so, backlog, curthread);
+	if (rc != 0)
+		return (rc);
+	SOLISTEN_LOCK(so);
+	solisten_upcall_set(so, iw_so_upcall, cm_id);
 	so->so_state |= SS_NBIO;
-	SOCK_UNLOCK(so);
+	SOLISTEN_UNLOCK(so);
 	sopt.sopt_dir = SOPT_SET;
 	sopt.sopt_level = IPPROTO_TCP;
 	sopt.sopt_name = TCP_NODELAY;
@@ -525,37 +513,18 @@ iw_init_sock(struct iw_cm_id *cm_id)
 	sopt.sopt_valsize = sizeof(on);
 	sopt.sopt_td = NULL;
 	sosetopt(so, &sopt);
-}
-
-static int
-iw_uninit_socket(struct iw_cm_id *cm_id)
-{
-	struct socket *so = cm_id->so;
-
-	SOCK_LOCK(so);
-	soupcall_clear(so, SO_RCV);
-	SOCK_UNLOCK(so);
-
 	return (0);
 }
 
 static int
-iw_create_listen(struct iw_cm_id *cm_id, int backlog)
-{
-	int rc;
-
-	iw_init_sock(cm_id);
-	rc = -solisten(cm_id->so, backlog, curthread);
-	if (rc != 0)
-		iw_uninit_socket(cm_id);
-	return (rc);
-}
-
-static int
 iw_destroy_listen(struct iw_cm_id *cm_id)
 {
+	struct socket *so = cm_id->so;
 
-	return (iw_uninit_socket(cm_id));
+	SOLISTEN_LOCK(so);
+	solisten_upcall_set(so, NULL, NULL);
+	SOLISTEN_UNLOCK(so);
+	return (0);
 }
 
 
diff --git a/sys/ofed/drivers/infiniband/ulp/sdp/sdp_main.c b/sys/ofed/drivers/infiniband/ulp/sdp/sdp_main.c
index f822466e4abc..a1ae56e52832 100644
--- a/sys/ofed/drivers/infiniband/ulp/sdp/sdp_main.c
+++ b/sys/ofed/drivers/infiniband/ulp/sdp/sdp_main.c
@@ -310,7 +310,6 @@ sdp_closed(struct sdp_sock *ssk)
 		    ("sdp_closed: !SS_PROTOREF"));
 		ssk->flags &= ~SDP_SOCKREF;
 		SDP_WUNLOCK(ssk);
-		ACCEPT_LOCK();
 		SOCK_LOCK(so);
 		so->so_state &= ~SS_PROTOREF;
 		sofree(so);