[PATCH] knfsd: lockd: Make nlm_host_rebooted use the nsm_handle

This patch makes the SM_NOTIFY handling understand and use the nsm_handle.

To make it a bit clear what is happening:

    nlmclent_prepare_reclaim and nlmclnt_finish_reclaim
    get open-coded into 'reclaimer'

The result is tidied up.

Then some of that functionality is moved out into nlm_host_rebooted (which
calls nlmclnt_recovery which starts a thread which runs reclaimer).

Also host_rebooted now finds an nsm_handle rather than a host, then then
iterates over all hosts and deals with each host that shares that nsm_handle.

Signed-off-by: Olaf Kirch <okir@suse.de>
Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

5 files changed, 69 insertions, 57 deletions

diff --git a/fs/lockd/clntlock.c b/fs/lockd/clntlock.c
index 54e63ddef043..a4ab6dd7661f 100644
--- a/fs/lockd/clntlock.c
+++ b/fs/lockd/clntlock.c
@@ -144,43 +144,12 @@ u32 nlmclnt_grant(const struct sockaddr_in *addr, const struct nlm_lock *lock)
  */
 
 /*
- * Someone has sent us an SM_NOTIFY. Ensure we bind to the new port number,
- * that we mark locks for reclaiming, and that we bump the pseudo NSM state.
- */
-static void nlmclnt_prepare_reclaim(struct nlm_host *host)
-{
-        down_write(&host->h_rwsem);
-        if (host->h_nsmhandle)
-                host->h_nsmhandle->sm_monitored = 0;
-        host->h_state++;
-        host->h_nextrebind = 0;
-        nlm_rebind_host(host);
-
-        /*
-         * Mark the locks for reclaiming.
-         */
-        list_splice_init(&host->h_granted, &host->h_reclaim);
-
-        dprintk("NLM: reclaiming locks for host %s\n", host->h_name);
-}
-
-static void nlmclnt_finish_reclaim(struct nlm_host *host)
-{
-        host->h_reclaiming = 0;
-        up_write(&host->h_rwsem);
-        dprintk("NLM: done reclaiming locks for host %s", host->h_name);
-}
-
-/*
  * Reclaim all locks on server host. We do this by spawning a separate
  * reclaimer thread.
  */
 void
-nlmclnt_recovery(struct nlm_host *host, u32 newstate)
+nlmclnt_recovery(struct nlm_host *host)
 {
-        if (host->h_nsmstate == newstate)
-                return;
-        host->h_nsmstate = newstate;
         if (!host->h_reclaiming++) {
                 nlm_get_host(host);
                 __module_get(THIS_MODULE);
@@ -200,18 +169,30 @@ reclaimer(void *ptr)
         daemonize("%s-reclaim", host->h_name);
         allow_signal(SIGKILL);
 
+        down_write(&host->h_rwsem);
+
         /* This one ensures that our parent doesn't terminate while the
          * reclaim is in progress */
         lock_kernel();
         lockd_up(0); /* note: this cannot fail as lockd is already running */
 
-        nlmclnt_prepare_reclaim(host);
-        /* First, reclaim all locks that have been marked. */
+        dprintk("lockd: reclaiming locks for host %s", host->h_name);
+
 restart:
         nsmstate = host->h_nsmstate;
+
+        /* Force a portmap getport - the peer's lockd will
+         * most likely end up on a different port.
+         */
+        host->h_nextrebind = 0;
+        nlm_rebind_host(host);
+
+        /* First, reclaim all locks that have been granted. */
+        list_splice_init(&host->h_granted, &host->h_reclaim);
         list_for_each_entry_safe(fl, next, &host->h_reclaim, fl_u.nfs_fl.list) {
                 list_del_init(&fl->fl_u.nfs_fl.list);
 
+                /* Why are we leaking memory here? --okir */
                 if (signalled())
                         continue;
                 if (nlmclnt_reclaim(host, fl) != 0)
@@ -219,11 +200,13 @@ restart:
                 list_add_tail(&fl->fl_u.nfs_fl.list, &host->h_granted);
                 if (host->h_nsmstate != nsmstate) {
                         /* Argh! The server rebooted again! */
-                        list_splice_init(&host->h_granted, &host->h_reclaim);
                         goto restart;
                 }
         }
-        nlmclnt_finish_reclaim(host);
+
+        host->h_reclaiming = 0;
+        up_write(&host->h_rwsem);
+        dprintk("NLM: done reclaiming locks for host %s", host->h_name);
 
         /* Now, wake up all processes that sleep on a blocked lock */
         list_for_each_entry(block, &nlm_blocked, b_list) {
diff --git a/fs/lockd/host.c b/fs/lockd/host.c
index 4990223a3e18..3cd96e2e1256 100644
--- a/fs/lockd/host.c
+++ b/fs/lockd/host.c
@@ -290,28 +290,57 @@ void nlm_release_host(struct nlm_host *host)
  * has rebooted.
  * Release all resources held by that peer.
  */
-void nlm_host_rebooted(const struct sockaddr_in *sin, const struct nlm_reboot *argp)
+void nlm_host_rebooted(const struct sockaddr_in *sin,
+                                const char *hostname, int hostname_len,
+                                u32 new_state)
 {
-        struct nlm_host *host;
-        int server;
+        struct nsm_handle *nsm;
+        struct nlm_host *host, **hp;
+        int             hash;
 
-        /* Obtain the host pointer for this NFS server and try to
-         * reclaim all locks we hold on this server.
-         */
-        server = (argp->proto & 1)? 1 : 0;
-        host = nlm_lookup_host(server, sin, argp->proto >> 1, argp->vers,
-                        argp->mon, argp->len);
-        if (host == NULL)
+        dprintk("lockd: nlm_host_rebooted(%s, %u.%u.%u.%u)\n",
+                        hostname, NIPQUAD(sin->sin_addr));
+
+        /* Find the NSM handle for this peer */
+        if (!(nsm = __nsm_find(sin, hostname, hostname_len, 0)))
                 return;
 
-        if (server == 0) {
-                /* We are client, he's the server: try to reclaim all locks. */
-                nlmclnt_recovery(host, argp->state);
-        } else {
-                /* He's the client, we're the server: delete all locks held by the client */
-                nlmsvc_free_host_resources(host);
+        /* When reclaiming locks on this peer, make sure that
+         * we set up a new notification */
+        nsm->sm_monitored = 0;
+
+        /* Mark all hosts tied to this NSM state as having rebooted.
+         * We run the loop repeatedly, because we drop the host table
+         * lock for this.
+         * To avoid processing a host several times, we match the nsmstate.
+         */
+again:  mutex_lock(&nlm_host_mutex);
+        for (hash = 0; hash < NLM_HOST_NRHASH; hash++) {
+                for (hp = &nlm_hosts[hash]; (host = *hp); hp = &host->h_next) {
+                        if (host->h_nsmhandle == nsm
+                         && host->h_nsmstate != new_state) {
+                                host->h_nsmstate = new_state;
+                                host->h_state++;
+
+                                nlm_get_host(host);
+                                mutex_unlock(&nlm_host_mutex);
+
+                                if (host->h_server) {
+                                        /* We're server for this guy, just ditch
+                                         * all the locks he held. */
+                                        nlmsvc_free_host_resources(host);
+                                } else {
+                                        /* He's the server, initiate lock recovery. */
+                                        nlmclnt_recovery(host);
+                                }
+
+                                nlm_release_host(host);
+                                goto again;
+                        }
+                }
         }
-        nlm_release_host(host);
+
+        mutex_unlock(&nlm_host_mutex);
 }
 
 /*
diff --git a/fs/lockd/svc4proc.c b/fs/lockd/svc4proc.c
index 7d835ad81074..b8525fb62934 100644
--- a/fs/lockd/svc4proc.c
+++ b/fs/lockd/svc4proc.c
@@ -438,7 +438,7 @@ nlm4svc_proc_sm_notify(struct svc_rqst *rqstp, struct nlm_reboot *argp,
          */