Merge tag 'nfs-for-4.8-2' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

Pull NFS client bugfixes from Trond Myklebust:
 "Highlights include:

   - Stable patch from Olga to fix RPCSEC_GSS upcalls when the same user
     needs multiple different security services (e.g.  krb5i and krb5p).

   - Stable patch to fix a regression introduced by the use of
     SO_REUSEPORT, and that prevented the use of multiple different NFS
     versions to the same server.

   - TCP socket reconnection timer fixes.

   - Patch from Neil to disable the use of IPv6 temporary addresses"

* tag 'nfs-for-4.8-2' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
  NFSv4: Cap the transport reconnection timer at 1/2 lease period
  NFSv4: Cleanup the setting of the nfs4 lease period
  SUNRPC: Limit the reconnect backoff timer to the max RPC message timeout
  SUNRPC: Fix reconnection timeouts
  NFSv4.2: LAYOUTSTATS may return NFS4ERR_ADMIN/DELEG_REVOKED
  SUNRPC: disable the use of IPv6 temporary addresses.
  SUNRPC: allow for upcalls for same uid but different gss service
  SUNRPC: Fix up socket autodisconnect
  SUNRPC: Handle EADDRNOTAVAIL on connection failures

diff --git a/fs/nfs/nfs42proc.c b/fs/nfs/nfs42proc.c
index 33da841..6f47527 100644 (file)
--- a/fs/nfs/nfs42proc.c
+++ b/fs/nfs/nfs42proc.c
@@ -338,6 +338,8 @@ nfs42_layoutstat_done(struct rpc_task *task, void *calldata)
        case 0:
                break;
        case -NFS4ERR_EXPIRED:
+       case -NFS4ERR_ADMIN_REVOKED:
+       case -NFS4ERR_DELEG_REVOKED:
        case -NFS4ERR_STALE_STATEID:
        case -NFS4ERR_OLD_STATEID:
        case -NFS4ERR_BAD_STATEID:

diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
index 324bfdc..9bf64ea 100644 (file)
--- a/fs/nfs/nfs4_fs.h
+++ b/fs/nfs/nfs4_fs.h
@@ -396,6 +396,10 @@ extern void nfs4_schedule_state_renewal(struct nfs_client *);
 extern void nfs4_renewd_prepare_shutdown(struct nfs_server *);
 extern void nfs4_kill_renewd(struct nfs_client *);
 extern void nfs4_renew_state(struct work_struct *);
+extern void nfs4_set_lease_period(struct nfs_client *clp,
+               unsigned long lease,
+               unsigned long lastrenewed);
+
 
 /* nfs4state.c */
 struct rpc_cred *nfs4_get_clid_cred(struct nfs_client *clp);

diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index a036e93..1949bbd 100644 (file)
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -4237,12 +4237,9 @@ static int nfs4_do_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle, str
                err = _nfs4_do_fsinfo(server, fhandle, fsinfo);
                trace_nfs4_fsinfo(server, fhandle, fsinfo->fattr, err);
                if (err == 0) {
-                       struct nfs_client *clp = server->nfs_client;
-
-                       spin_lock(&clp->cl_lock);
-                       clp->cl_lease_time = fsinfo->lease_time * HZ;
-                       clp->cl_last_renewal = now;
-                       spin_unlock(&clp->cl_lock);
+                       nfs4_set_lease_period(server->nfs_client,
+                                       fsinfo->lease_time * HZ,
+                                       now);
                        break;
                }
                err = nfs4_handle_exception(server, err, &exception);

diff --git a/fs/nfs/nfs4renewd.c b/fs/nfs/nfs4renewd.c
index e1ba58c..82e7719 100644 (file)
--- a/fs/nfs/nfs4renewd.c
+++ b/fs/nfs/nfs4renewd.c
@@ -136,6 +136,26 @@ nfs4_kill_renewd(struct nfs_client *clp)
        cancel_delayed_work_sync(&clp->cl_renewd);
 }
 
+/**
+ * nfs4_set_lease_period - Sets the lease period on a nfs_client
+ *
+ * @clp: pointer to nfs_client
+ * @lease: new value for lease period
+ * @lastrenewed: time at which lease was last renewed
+ */
+void nfs4_set_lease_period(struct nfs_client *clp,
+               unsigned long lease,
+               unsigned long lastrenewed)
+{
+       spin_lock(&clp->cl_lock);
+       clp->cl_lease_time = lease;
+       clp->cl_last_renewal = lastrenewed;
+       spin_unlock(&clp->cl_lock);
+
+       /* Cap maximum reconnect timeout at 1/2 lease period */
+       rpc_cap_max_reconnect_timeout(clp->cl_rpcclient, lease >> 1);
+}
+
 /*
  * Local variables:
  *   c-basic-offset: 8

diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index 834b875..cada00a 100644 (file)
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -277,20 +277,17 @@ static int nfs41_setup_state_renewal(struct nfs_client *clp)
 {
        int status;
        struct nfs_fsinfo fsinfo;
+       unsigned long now;
 
        if (!test_bit(NFS_CS_CHECK_LEASE_TIME, &clp->cl_res_state)) {
                nfs4_schedule_state_renewal(clp);
                return 0;
        }
 
+       now = jiffies;
        status = nfs4_proc_get_lease_time(clp, &fsinfo);
        if (status == 0) {
-               /* Update lease time and schedule renewal */
-               spin_lock(&clp->cl_lock);
-               clp->cl_lease_time = fsinfo.lease_time * HZ;
-               clp->cl_last_renewal = jiffies;
-               spin_unlock(&clp->cl_lock);
-
+               nfs4_set_lease_period(clp, fsinfo.lease_time * HZ, now);
                nfs4_schedule_state_renewal(clp);
        }
 

diff --git a/include/linux/sunrpc/clnt.h b/include/linux/sunrpc/clnt.h
index b6810c9..5c02b06 100644 (file)
--- a/include/linux/sunrpc/clnt.h
+++ b/include/linux/sunrpc/clnt.h
@@ -195,6 +195,8 @@ int         rpc_clnt_add_xprt(struct rpc_clnt *, struct xprt_create *,
                                struct rpc_xprt *,
                                void *),
                        void *data);
+void           rpc_cap_max_reconnect_timeout(struct rpc_clnt *clnt,
+                       unsigned long timeo);
 
 const char *rpc_proc_name(const struct rpc_task *task);
 #endif /* __KERNEL__ */

diff --git a/include/linux/sunrpc/xprt.h b/include/linux/sunrpc/xprt.h
index 5e3e1b6..a16070d 100644 (file)
--- a/include/linux/sunrpc/xprt.h
+++ b/include/linux/sunrpc/xprt.h
@@ -218,7 +218,8 @@ struct rpc_xprt {
        struct work_struct      task_cleanup;
        struct timer_list       timer;
        unsigned long           last_used,
-                               idle_timeout;
+                               idle_timeout,
+                               max_reconnect_timeout;
 
        /*
         * Send stuff

diff --git a/net/sunrpc/auth_gss/auth_gss.c b/net/sunrpc/auth_gss/auth_gss.c
index 23c8e7c..976c781 100644 (file)
--- a/net/sunrpc/auth_gss/auth_gss.c
+++ b/net/sunrpc/auth_gss/auth_gss.c
@@ -340,12 +340,14 @@ gss_release_msg(struct gss_upcall_msg *gss_msg)
 }
 
 static struct gss_upcall_msg *
-__gss_find_upcall(struct rpc_pipe *pipe, kuid_t uid)
+__gss_find_upcall(struct rpc_pipe *pipe, kuid_t uid, const struct gss_auth *auth)
 {
        struct gss_upcall_msg *pos;
        list_for_each_entry(pos, &pipe->in_downcall, list) {
                if (!uid_eq(pos->uid, uid))
                        continue;
+               if (auth && pos->auth->service != auth->service)
+                       continue;
                atomic_inc(&pos->count);
                dprintk("RPC:       %s found msg %p\n", __func__, pos);
                return pos;
@@ -365,7 +367,7 @@ gss_add_msg(struct gss_upcall_msg *gss_msg)
        struct gss_upcall_msg *old;
 
        spin_lock(&pipe->lock);
-       old = __gss_find_upcall(pipe, gss_msg->uid);
+       old = __gss_find_upcall(pipe, gss_msg->uid, gss_msg->auth);
        if (old == NULL) {
                atomic_inc(&gss_msg->count);
                list_add(&gss_msg->list, &pipe->in_downcall);
@@ -714,7 +716,7 @@ gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
        err = -ENOENT;
        /* Find a matching upcall */
        spin_lock(&pipe->lock);
-       gss_msg = __gss_find_upcall(pipe, uid);
+       gss_msg = __gss_find_upcall(pipe, uid, NULL);
        if (gss_msg == NULL) {
                spin_unlock(&pipe->lock);
                goto err_put_ctx;

diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
index cb49898..7f79fb7 100644 (file)
--- a/net/sunrpc/clnt.c
+++ b/net/sunrpc/clnt.c
@@ -2638,6 +2638,7 @@ int rpc_clnt_add_xprt(struct rpc_clnt *clnt,
 {
        struct rpc_xprt_switch *xps;
        struct rpc_xprt *xprt;
+       unsigned long reconnect_timeout;
        unsigned char resvport;
        int ret = 0;
 
@@ -2649,6 +2650,7 @@ int rpc_clnt_add_xprt(struct rpc_clnt *clnt,
                return -EAGAIN;
        }
        resvport = xprt->resvport;
+       reconnect_timeout = xprt->max_reconnect_timeout;
        rcu_read_unlock();
 
        xprt = xprt_create_transport(xprtargs);
@@ -2657,6 +2659,7 @@ int rpc_clnt_add_xprt(struct rpc_clnt *clnt,
                goto out_put_switch;
        }
        xprt->resvport = resvport;
+       xprt->max_reconnect_timeout = reconnect_timeout;
 
        rpc_xprt_switch_set_roundrobin(xps);
        if (setup) {
@@ -2673,6 +2676,27 @@ out_put_switch:
 }
 EXPORT_SYMBOL_GPL(rpc_clnt_add_xprt);
 
+static int
+rpc_xprt_cap_max_reconnect_timeout(struct rpc_clnt *clnt,
+               struct rpc_xprt *xprt,
+               void *data)
+{
+       unsigned long timeout = *((unsigned long *)data);
+
+       if (timeout < xprt->max_reconnect_timeout)
+               xprt->max_reconnect_timeout = timeout;
+       return 0;
+}
+
+void
+rpc_cap_max_reconnect_timeout(struct rpc_clnt *clnt, unsigned long timeo)
+{
+       rpc_clnt_iterate_for_each_xprt(clnt,
+                       rpc_xprt_cap_max_reconnect_timeout,
+                       &timeo);
+}
+EXPORT_SYMBOL_GPL(rpc_cap_max_reconnect_timeout);
+
 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
 static void rpc_show_header(void)
 {

diff --git a/net/sunrpc/xprt.c b/net/sunrpc/xprt.c
index 8313960..ea244b2 100644 (file)
--- a/net/sunrpc/xprt.c
+++ b/net/sunrpc/xprt.c
@@ -680,6 +680,20 @@ out:
        spin_unlock_bh(&xprt->transport_lock);
 }
 
+static bool
+xprt_has_timer(const struct rpc_xprt *xprt)
+{
+       return xprt->idle_timeout != 0;
+}
+
+static void
+xprt_schedule_autodisconnect(struct rpc_xprt *xprt)
+       __must_hold(&xprt->transport_lock)
+{
+       if (list_empty(&xprt->recv) && xprt_has_timer(xprt))
+               mod_timer(&xprt->timer, xprt->last_used + xprt->idle_timeout);
+}
+
 static void
 xprt_init_autodisconnect(unsigned long data)
 {
@@ -688,6 +702,8 @@ xprt_init_autodisconnect(unsigned long data)
        spin_lock(&xprt->transport_lock);
        if (!list_empty(&xprt->recv))
                goto out_abort;
+       /* Reset xprt->last_used to avoid connect/autodisconnect cycling */
+       xprt->last_used = jiffies;
        if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
                goto out_abort;
        spin_unlock(&xprt->transport_lock);
@@ -725,6 +741,7 @@ void xprt_unlock_connect(struct rpc_xprt *xprt, void *cookie)
                goto out;
        xprt->snd_task =NULL;
        xprt->ops->release_xprt(xprt, NULL);
+       xprt_schedule_autodisconnect(xprt);
 out:
        spin_unlock_bh(&xprt->transport_lock);
        wake_up_bit(&xprt->state, XPRT_LOCKED);
@@ -888,11 +905,6 @@ static void xprt_timer(struct rpc_task *task)
        spin_unlock_bh(&xprt->transport_lock);
 }
 
-static inline int xprt_has_timer(struct rpc_xprt *xprt)
-{
-       return xprt->idle_timeout != 0;
-}
-
 /**
  * xprt_prepare_transmit - reserve the transport before sending a request
  * @task: RPC task about to send a request
@@ -1280,9 +1292,7 @@ void xprt_release(struct rpc_task *task)
        if (!list_empty(&req->rq_list))
                list_del(&req->rq_list);
        xprt->last_used = jiffies;
-       if (list_empty(&xprt->recv) && xprt_has_timer(xprt))
-               mod_timer(&xprt->timer,
-                               xprt->last_used + xprt->idle_timeout);
+       xprt_schedule_autodisconnect(xprt);
        spin_unlock_bh(&xprt->transport_lock);
        if (req->rq_buffer)
                xprt->ops->buf_free(req->rq_buffer);

diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
index 111767a..8ede3bc 100644 (file)
--- a/net/sunrpc/xprtsock.c
+++ b/net/sunrpc/xprtsock.c
@@ -177,7 +177,6 @@ static struct ctl_table sunrpc_table[] = {
  * increase over time if the server is down or not responding.
  */
 #define XS_TCP_INIT_REEST_TO   (3U * HZ)
-#define XS_TCP_MAX_REEST_TO    (5U * 60 * HZ)
 
 /*
  * TCP idle timeout; client drops the transport socket if it is idle
@@ -2173,6 +2172,8 @@ static void xs_udp_finish_connecting(struct rpc_xprt *xprt, struct socket *sock)
                write_unlock_bh(&sk->sk_callback_lock);
        }
        xs_udp_do_set_buffer_size(xprt);
+
+       xprt->stat.connect_start = jiffies;
 }
 
 static void xs_udp_setup_socket(struct work_struct *work)
@@ -2236,6 +2237,7 @@ static int xs_tcp_finish_connecting(struct rpc_xprt *xprt, struct socket *sock)
                unsigned int keepcnt = xprt->timeout->to_retries + 1;
                unsigned int opt_on = 1;
                unsigned int timeo;
+               unsigned int addr_pref = IPV6_PREFER_SRC_PUBLIC;
 
                /* TCP Keepalive options */
                kernel_setsockopt(sock, SOL_SOCKET, SO_KEEPALIVE,
@@ -2247,6 +2249,16 @@ static int xs_tcp_finish_connecting(struct rpc_xprt *xprt, struct socket *sock)
                kernel_setsockopt(sock, SOL_TCP, TCP_KEEPCNT,
                                (char *)&keepcnt, sizeof(keepcnt));
 
+               /* Avoid temporary address, they are bad for long-lived
+                * connections such as NFS mounts.
+                * RFC4941, section 3.6 suggests that:
+                *    Individual applications, which have specific
+                *    knowledge about the normal duration of connections,
+                *    MAY override this as appropriate.
+                */
+               kernel_setsockopt(sock, SOL_IPV6, IPV6_ADDR_PREFERENCES,
+                               (char *)&addr_pref, sizeof(addr_pref));
+
                /* TCP user timeout (see RFC5482) */
                timeo = jiffies_to_msecs(xprt->timeout->to_initval) *
                        (xprt->timeout->to_retries + 1);
@@ -2295,6 +2307,10 @@ static int xs_tcp_finish_connecting(struct rpc_xprt *xprt, struct socket *sock)
                /* SYN_SENT! */
                if (xprt->reestablish_timeout < XS_TCP_INIT_REEST_TO)
                        xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO;
+               break;
+       case -EADDRNOTAVAIL:
+               /* Source port number is unavailable. Try a new one! */
+               transport->srcport = 0;
        }
 out:
        return ret;
@@ -2369,6 +2385,25 @@ out:
        xprt_wake_pending_tasks(xprt, status);
 }
 
+static unsigned long xs_reconnect_delay(const struct rpc_xprt *xprt)
+{
+       unsigned long start, now = jiffies;
+
+       start = xprt->stat.connect_start + xprt->reestablish_timeout;
+       if (time_after(start, now))
+               return start - now;
+       return 0;
+}
+
+static void xs_reconnect_backoff(struct rpc_xprt *xprt)
+{
+       xprt->reestablish_timeout <<= 1;
+       if (xprt->reestablish_timeout > xprt->max_reconnect_timeout)
+               xprt->reestablish_timeout = xprt->max_reconnect_timeout;
+       if (xprt->reestablish_timeout < XS_TCP_INIT_REEST_TO)
+               xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO;
+}
+
 /**
  * xs_connect - connect a socket to a remote endpoint
  * @xprt: pointer to transport structure
@@ -2386,6 +2421,7 @@ out:
 static void xs_connect(struct rpc_xprt *xprt, struct rpc_task *task)
 {
        struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
+       unsigned long delay = 0;
 
        WARN_ON_ONCE(!xprt_lock_connect(xprt, task, transport));
 
@@ -2397,19 +2433,15 @@ static void xs_connect(struct rpc_xprt *xprt, struct rpc_task *task)
                /* Start by resetting any existing state */
                xs_reset_transport(transport);
 
-               queue_delayed_work(xprtiod_workqueue,
-                                  &transport->connect_worker,
-                                  xprt->reestablish_timeout);
-               xprt->reestablish_timeout <<= 1;
-               if (xprt->reestablish_timeout < XS_TCP_INIT_REEST_TO)
-                       xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO;
-               if (xprt->reestablish_timeout > XS_TCP_MAX_REEST_TO)
-                       xprt->reestablish_timeout = XS_TCP_MAX_REEST_TO;
-       } else {
+               delay = xs_reconnect_delay(xprt);
+               xs_reconnect_backoff(xprt);
+
+       } else
                dprintk("RPC:       xs_connect scheduled xprt %p\n", xprt);
-               queue_delayed_work(xprtiod_workqueue,
-                                  &transport->connect_worker, 0);
-       }
+
+       queue_delayed_work(xprtiod_workqueue,
+                       &transport->connect_worker,
+                       delay);
 }
 
 /**
@@ -2961,6 +2993,8 @@ static struct rpc_xprt *xs_setup_tcp(struct xprt_create *args)
        xprt->ops = &xs_tcp_ops;
        xprt->timeout = &xs_tcp_default_timeout;
 
+       xprt->max_reconnect_timeout = xprt->timeout->to_maxval;
+
        INIT_WORK(&transport->recv_worker, xs_tcp_data_receive_workfn);
        INIT_DELAYED_WORK(&transport->connect_worker, xs_tcp_setup_socket);