Add the rt linux 4.1.3-rt3 as base
[kvmfornfv.git] / kernel / fs / nfsd / nfs4layouts.c
diff --git a/kernel/fs/nfsd/nfs4layouts.c b/kernel/fs/nfsd/nfs4layouts.c
new file mode 100644 (file)
index 0000000..6904213
--- /dev/null
@@ -0,0 +1,723 @@
+/*
+ * Copyright (c) 2014 Christoph Hellwig.
+ */
+#include <linux/kmod.h>
+#include <linux/file.h>
+#include <linux/jhash.h>
+#include <linux/sched.h>
+#include <linux/sunrpc/addr.h>
+
+#include "pnfs.h"
+#include "netns.h"
+#include "trace.h"
+
+#define NFSDDBG_FACILITY                NFSDDBG_PNFS
+
+struct nfs4_layout {
+       struct list_head                lo_perstate;
+       struct nfs4_layout_stateid      *lo_state;
+       struct nfsd4_layout_seg         lo_seg;
+};
+
+static struct kmem_cache *nfs4_layout_cache;
+static struct kmem_cache *nfs4_layout_stateid_cache;
+
+static struct nfsd4_callback_ops nfsd4_cb_layout_ops;
+static const struct lock_manager_operations nfsd4_layouts_lm_ops;
+
+const struct nfsd4_layout_ops *nfsd4_layout_ops[LAYOUT_TYPE_MAX] =  {
+       [LAYOUT_BLOCK_VOLUME]   = &bl_layout_ops,
+};
+
+/* pNFS device ID to export fsid mapping */
+#define DEVID_HASH_BITS        8
+#define DEVID_HASH_SIZE        (1 << DEVID_HASH_BITS)
+#define DEVID_HASH_MASK        (DEVID_HASH_SIZE - 1)
+static u64 nfsd_devid_seq = 1;
+static struct list_head nfsd_devid_hash[DEVID_HASH_SIZE];
+static DEFINE_SPINLOCK(nfsd_devid_lock);
+
+static inline u32 devid_hashfn(u64 idx)
+{
+       return jhash_2words(idx, idx >> 32, 0) & DEVID_HASH_MASK;
+}
+
+static void
+nfsd4_alloc_devid_map(const struct svc_fh *fhp)
+{
+       const struct knfsd_fh *fh = &fhp->fh_handle;
+       size_t fsid_len = key_len(fh->fh_fsid_type);
+       struct nfsd4_deviceid_map *map, *old;
+       int i;
+
+       map = kzalloc(sizeof(*map) + fsid_len, GFP_KERNEL);
+       if (!map)
+               return;
+
+       map->fsid_type = fh->fh_fsid_type;
+       memcpy(&map->fsid, fh->fh_fsid, fsid_len);
+
+       spin_lock(&nfsd_devid_lock);
+       if (fhp->fh_export->ex_devid_map)
+               goto out_unlock;
+
+       for (i = 0; i < DEVID_HASH_SIZE; i++) {
+               list_for_each_entry(old, &nfsd_devid_hash[i], hash) {
+                       if (old->fsid_type != fh->fh_fsid_type)
+                               continue;
+                       if (memcmp(old->fsid, fh->fh_fsid,
+                                       key_len(old->fsid_type)))
+                               continue;
+
+                       fhp->fh_export->ex_devid_map = old;
+                       goto out_unlock;
+               }
+       }
+
+       map->idx = nfsd_devid_seq++;
+       list_add_tail_rcu(&map->hash, &nfsd_devid_hash[devid_hashfn(map->idx)]);
+       fhp->fh_export->ex_devid_map = map;
+       map = NULL;
+
+out_unlock:
+       spin_unlock(&nfsd_devid_lock);
+       kfree(map);
+}
+
+struct nfsd4_deviceid_map *
+nfsd4_find_devid_map(int idx)
+{
+       struct nfsd4_deviceid_map *map, *ret = NULL;
+
+       rcu_read_lock();
+       list_for_each_entry_rcu(map, &nfsd_devid_hash[devid_hashfn(idx)], hash)
+               if (map->idx == idx)
+                       ret = map;
+       rcu_read_unlock();
+
+       return ret;
+}
+
+int
+nfsd4_set_deviceid(struct nfsd4_deviceid *id, const struct svc_fh *fhp,
+               u32 device_generation)
+{
+       if (!fhp->fh_export->ex_devid_map) {
+               nfsd4_alloc_devid_map(fhp);
+               if (!fhp->fh_export->ex_devid_map)
+                       return -ENOMEM;
+       }
+
+       id->fsid_idx = fhp->fh_export->ex_devid_map->idx;
+       id->generation = device_generation;
+       id->pad = 0;
+       return 0;
+}
+
+void nfsd4_setup_layout_type(struct svc_export *exp)
+{
+       struct super_block *sb = exp->ex_path.mnt->mnt_sb;
+
+       if (!(exp->ex_flags & NFSEXP_PNFS))
+               return;
+
+       if (sb->s_export_op->get_uuid &&
+           sb->s_export_op->map_blocks &&
+           sb->s_export_op->commit_blocks)
+               exp->ex_layout_type = LAYOUT_BLOCK_VOLUME;
+}
+
+static void
+nfsd4_free_layout_stateid(struct nfs4_stid *stid)
+{
+       struct nfs4_layout_stateid *ls = layoutstateid(stid);
+       struct nfs4_client *clp = ls->ls_stid.sc_client;
+       struct nfs4_file *fp = ls->ls_stid.sc_file;
+
+       trace_layoutstate_free(&ls->ls_stid.sc_stateid);
+
+       spin_lock(&clp->cl_lock);
+       list_del_init(&ls->ls_perclnt);
+       spin_unlock(&clp->cl_lock);
+
+       spin_lock(&fp->fi_lock);
+       list_del_init(&ls->ls_perfile);
+       spin_unlock(&fp->fi_lock);
+
+       vfs_setlease(ls->ls_file, F_UNLCK, NULL, (void **)&ls);
+       fput(ls->ls_file);
+
+       if (ls->ls_recalled)
+               atomic_dec(&ls->ls_stid.sc_file->fi_lo_recalls);
+
+       kmem_cache_free(nfs4_layout_stateid_cache, ls);
+}
+
+static int
+nfsd4_layout_setlease(struct nfs4_layout_stateid *ls)
+{
+       struct file_lock *fl;
+       int status;
+
+       fl = locks_alloc_lock();
+       if (!fl)
+               return -ENOMEM;
+       locks_init_lock(fl);
+       fl->fl_lmops = &nfsd4_layouts_lm_ops;
+       fl->fl_flags = FL_LAYOUT;
+       fl->fl_type = F_RDLCK;
+       fl->fl_end = OFFSET_MAX;
+       fl->fl_owner = ls;
+       fl->fl_pid = current->tgid;
+       fl->fl_file = ls->ls_file;
+
+       status = vfs_setlease(fl->fl_file, fl->fl_type, &fl, NULL);
+       if (status) {
+               locks_free_lock(fl);
+               return status;
+       }
+       BUG_ON(fl != NULL);
+       return 0;
+}
+
+static struct nfs4_layout_stateid *
+nfsd4_alloc_layout_stateid(struct nfsd4_compound_state *cstate,
+               struct nfs4_stid *parent, u32 layout_type)
+{
+       struct nfs4_client *clp = cstate->clp;
+       struct nfs4_file *fp = parent->sc_file;
+       struct nfs4_layout_stateid *ls;
+       struct nfs4_stid *stp;
+
+       stp = nfs4_alloc_stid(cstate->clp, nfs4_layout_stateid_cache);
+       if (!stp)
+               return NULL;
+       stp->sc_free = nfsd4_free_layout_stateid;
+       get_nfs4_file(fp);
+       stp->sc_file = fp;
+
+       ls = layoutstateid(stp);
+       INIT_LIST_HEAD(&ls->ls_perclnt);
+       INIT_LIST_HEAD(&ls->ls_perfile);
+       spin_lock_init(&ls->ls_lock);
+       INIT_LIST_HEAD(&ls->ls_layouts);
+       ls->ls_layout_type = layout_type;
+       nfsd4_init_cb(&ls->ls_recall, clp, &nfsd4_cb_layout_ops,
+                       NFSPROC4_CLNT_CB_LAYOUT);
+
+       if (parent->sc_type == NFS4_DELEG_STID)
+               ls->ls_file = get_file(fp->fi_deleg_file);
+       else
+               ls->ls_file = find_any_file(fp);
+       BUG_ON(!ls->ls_file);
+
+       if (nfsd4_layout_setlease(ls)) {
+               put_nfs4_file(fp);
+               kmem_cache_free(nfs4_layout_stateid_cache, ls);
+               return NULL;
+       }
+
+       spin_lock(&clp->cl_lock);
+       stp->sc_type = NFS4_LAYOUT_STID;
+       list_add(&ls->ls_perclnt, &clp->cl_lo_states);
+       spin_unlock(&clp->cl_lock);
+
+       spin_lock(&fp->fi_lock);
+       list_add(&ls->ls_perfile, &fp->fi_lo_states);
+       spin_unlock(&fp->fi_lock);
+
+       trace_layoutstate_alloc(&ls->ls_stid.sc_stateid);
+       return ls;
+}
+
+__be32
+nfsd4_preprocess_layout_stateid(struct svc_rqst *rqstp,
+               struct nfsd4_compound_state *cstate, stateid_t *stateid,
+               bool create, u32 layout_type, struct nfs4_layout_stateid **lsp)
+{
+       struct nfs4_layout_stateid *ls;
+       struct nfs4_stid *stid;
+       unsigned char typemask = NFS4_LAYOUT_STID;
+       __be32 status;
+
+       if (create)
+               typemask |= (NFS4_OPEN_STID | NFS4_LOCK_STID | NFS4_DELEG_STID);
+
+       status = nfsd4_lookup_stateid(cstate, stateid, typemask, &stid,
+                       net_generic(SVC_NET(rqstp), nfsd_net_id));
+       if (status)
+               goto out;
+
+       if (!fh_match(&cstate->current_fh.fh_handle,
+                     &stid->sc_file->fi_fhandle)) {
+               status = nfserr_bad_stateid;
+               goto out_put_stid;
+       }
+
+       if (stid->sc_type != NFS4_LAYOUT_STID) {
+               ls = nfsd4_alloc_layout_stateid(cstate, stid, layout_type);
+               nfs4_put_stid(stid);
+
+               status = nfserr_jukebox;
+               if (!ls)
+                       goto out;
+       } else {
+               ls = container_of(stid, struct nfs4_layout_stateid, ls_stid);
+
+               status = nfserr_bad_stateid;
+               if (stateid->si_generation > stid->sc_stateid.si_generation)
+                       goto out_put_stid;
+               if (layout_type != ls->ls_layout_type)
+                       goto out_put_stid;
+       }
+
+       *lsp = ls;
+       return 0;
+
+out_put_stid:
+       nfs4_put_stid(stid);
+out:
+       return status;
+}
+
+static void
+nfsd4_recall_file_layout(struct nfs4_layout_stateid *ls)
+{
+       spin_lock(&ls->ls_lock);
+       if (ls->ls_recalled)
+               goto out_unlock;
+
+       ls->ls_recalled = true;
+       atomic_inc(&ls->ls_stid.sc_file->fi_lo_recalls);
+       if (list_empty(&ls->ls_layouts))
+               goto out_unlock;
+
+       trace_layout_recall(&ls->ls_stid.sc_stateid);
+
+       atomic_inc(&ls->ls_stid.sc_count);
+       update_stateid(&ls->ls_stid.sc_stateid);
+       memcpy(&ls->ls_recall_sid, &ls->ls_stid.sc_stateid, sizeof(stateid_t));
+       nfsd4_run_cb(&ls->ls_recall);
+
+out_unlock:
+       spin_unlock(&ls->ls_lock);
+}
+
+static inline u64
+layout_end(struct nfsd4_layout_seg *seg)
+{
+       u64 end = seg->offset + seg->length;
+       return end >= seg->offset ? end : NFS4_MAX_UINT64;
+}
+
+static void
+layout_update_len(struct nfsd4_layout_seg *lo, u64 end)
+{
+       if (end == NFS4_MAX_UINT64)
+               lo->length = NFS4_MAX_UINT64;
+       else
+               lo->length = end - lo->offset;
+}
+
+static bool
+layouts_overlapping(struct nfs4_layout *lo, struct nfsd4_layout_seg *s)
+{
+       if (s->iomode != IOMODE_ANY && s->iomode != lo->lo_seg.iomode)
+               return false;
+       if (layout_end(&lo->lo_seg) <= s->offset)
+               return false;
+       if (layout_end(s) <= lo->lo_seg.offset)
+               return false;
+       return true;
+}
+
+static bool
+layouts_try_merge(struct nfsd4_layout_seg *lo, struct nfsd4_layout_seg *new)
+{
+       if (lo->iomode != new->iomode)
+               return false;
+       if (layout_end(new) < lo->offset)
+               return false;
+       if (layout_end(lo) < new->offset)
+               return false;
+
+       lo->offset = min(lo->offset, new->offset);
+       layout_update_len(lo, max(layout_end(lo), layout_end(new)));
+       return true;
+}
+
+static __be32
+nfsd4_recall_conflict(struct nfs4_layout_stateid *ls)
+{
+       struct nfs4_file *fp = ls->ls_stid.sc_file;
+       struct nfs4_layout_stateid *l, *n;
+       __be32 nfserr = nfs_ok;
+
+       assert_spin_locked(&fp->fi_lock);
+
+       list_for_each_entry_safe(l, n, &fp->fi_lo_states, ls_perfile) {
+               if (l != ls) {
+                       nfsd4_recall_file_layout(l);
+                       nfserr = nfserr_recallconflict;
+               }
+       }
+
+       return nfserr;
+}
+
+__be32
+nfsd4_insert_layout(struct nfsd4_layoutget *lgp, struct nfs4_layout_stateid *ls)
+{
+       struct nfsd4_layout_seg *seg = &lgp->lg_seg;
+       struct nfs4_file *fp = ls->ls_stid.sc_file;
+       struct nfs4_layout *lp, *new = NULL;
+       __be32 nfserr;
+
+       spin_lock(&fp->fi_lock);
+       nfserr = nfsd4_recall_conflict(ls);
+       if (nfserr)
+               goto out;
+       spin_lock(&ls->ls_lock);
+       list_for_each_entry(lp, &ls->ls_layouts, lo_perstate) {
+               if (layouts_try_merge(&lp->lo_seg, seg))
+                       goto done;
+       }
+       spin_unlock(&ls->ls_lock);
+       spin_unlock(&fp->fi_lock);
+
+       new = kmem_cache_alloc(nfs4_layout_cache, GFP_KERNEL);
+       if (!new)
+               return nfserr_jukebox;
+       memcpy(&new->lo_seg, seg, sizeof(lp->lo_seg));
+       new->lo_state = ls;
+
+       spin_lock(&fp->fi_lock);
+       nfserr = nfsd4_recall_conflict(ls);
+       if (nfserr)
+               goto out;
+       spin_lock(&ls->ls_lock);
+       list_for_each_entry(lp, &ls->ls_layouts, lo_perstate) {
+               if (layouts_try_merge(&lp->lo_seg, seg))
+                       goto done;
+       }
+
+       atomic_inc(&ls->ls_stid.sc_count);
+       list_add_tail(&new->lo_perstate, &ls->ls_layouts);
+       new = NULL;
+done:
+       update_stateid(&ls->ls_stid.sc_stateid);
+       memcpy(&lgp->lg_sid, &ls->ls_stid.sc_stateid, sizeof(stateid_t));
+       spin_unlock(&ls->ls_lock);
+out:
+       spin_unlock(&fp->fi_lock);
+       if (new)
+               kmem_cache_free(nfs4_layout_cache, new);
+       return nfserr;
+}
+
+static void
+nfsd4_free_layouts(struct list_head *reaplist)
+{
+       while (!list_empty(reaplist)) {
+               struct nfs4_layout *lp = list_first_entry(reaplist,
+                               struct nfs4_layout, lo_perstate);
+
+               list_del(&lp->lo_perstate);
+               nfs4_put_stid(&lp->lo_state->ls_stid);
+               kmem_cache_free(nfs4_layout_cache, lp);
+       }
+}
+
+static void
+nfsd4_return_file_layout(struct nfs4_layout *lp, struct nfsd4_layout_seg *seg,
+               struct list_head *reaplist)
+{
+       struct nfsd4_layout_seg *lo = &lp->lo_seg;
+       u64 end = layout_end(lo);
+
+       if (seg->offset <= lo->offset) {
+               if (layout_end(seg) >= end) {
+                       list_move_tail(&lp->lo_perstate, reaplist);
+                       return;
+               }
+               lo->offset = layout_end(seg);
+       } else {
+               /* retain the whole layout segment on a split. */
+               if (layout_end(seg) < end) {
+                       dprintk("%s: split not supported\n", __func__);
+                       return;
+               }
+               end = seg->offset;
+       }
+
+       layout_update_len(lo, end);
+}
+
+__be32
+nfsd4_return_file_layouts(struct svc_rqst *rqstp,
+               struct nfsd4_compound_state *cstate,
+               struct nfsd4_layoutreturn *lrp)
+{
+       struct nfs4_layout_stateid *ls;
+       struct nfs4_layout *lp, *n;
+       LIST_HEAD(reaplist);
+       __be32 nfserr;
+       int found = 0;
+
+       nfserr = nfsd4_preprocess_layout_stateid(rqstp, cstate, &lrp->lr_sid,
+                                               false, lrp->lr_layout_type,
+                                               &ls);
+       if (nfserr) {
+               trace_layout_return_lookup_fail(&lrp->lr_sid);
+               return nfserr;
+       }
+
+       spin_lock(&ls->ls_lock);
+       list_for_each_entry_safe(lp, n, &ls->ls_layouts, lo_perstate) {
+               if (layouts_overlapping(lp, &lrp->lr_seg)) {
+                       nfsd4_return_file_layout(lp, &lrp->lr_seg, &reaplist);
+                       found++;
+               }
+       }
+       if (!list_empty(&ls->ls_layouts)) {
+               if (found) {
+                       update_stateid(&ls->ls_stid.sc_stateid);
+                       memcpy(&lrp->lr_sid, &ls->ls_stid.sc_stateid,
+                               sizeof(stateid_t));
+               }
+               lrp->lrs_present = 1;
+       } else {
+               trace_layoutstate_unhash(&ls->ls_stid.sc_stateid);
+               nfs4_unhash_stid(&ls->ls_stid);
+               lrp->lrs_present = 0;
+       }
+       spin_unlock(&ls->ls_lock);
+
+       nfs4_put_stid(&ls->ls_stid);
+       nfsd4_free_layouts(&reaplist);
+       return nfs_ok;
+}
+
+__be32
+nfsd4_return_client_layouts(struct svc_rqst *rqstp,
+               struct nfsd4_compound_state *cstate,
+               struct nfsd4_layoutreturn *lrp)
+{
+       struct nfs4_layout_stateid *ls, *n;
+       struct nfs4_client *clp = cstate->clp;
+       struct nfs4_layout *lp, *t;
+       LIST_HEAD(reaplist);
+
+       lrp->lrs_present = 0;
+
+       spin_lock(&clp->cl_lock);
+       list_for_each_entry_safe(ls, n, &clp->cl_lo_states, ls_perclnt) {
+               if (ls->ls_layout_type != lrp->lr_layout_type)
+                       continue;
+
+               if (lrp->lr_return_type == RETURN_FSID &&
+                   !fh_fsid_match(&ls->ls_stid.sc_file->fi_fhandle,
+                                  &cstate->current_fh.fh_handle))
+                       continue;
+
+               spin_lock(&ls->ls_lock);
+               list_for_each_entry_safe(lp, t, &ls->ls_layouts, lo_perstate) {
+                       if (lrp->lr_seg.iomode == IOMODE_ANY ||
+                           lrp->lr_seg.iomode == lp->lo_seg.iomode)
+                               list_move_tail(&lp->lo_perstate, &reaplist);
+               }
+               spin_unlock(&ls->ls_lock);
+       }
+       spin_unlock(&clp->cl_lock);
+
+       nfsd4_free_layouts(&reaplist);
+       return 0;
+}
+
+static void
+nfsd4_return_all_layouts(struct nfs4_layout_stateid *ls,
+               struct list_head *reaplist)
+{
+       spin_lock(&ls->ls_lock);
+       list_splice_init(&ls->ls_layouts, reaplist);
+       spin_unlock(&ls->ls_lock);
+}
+
+void
+nfsd4_return_all_client_layouts(struct nfs4_client *clp)
+{
+       struct nfs4_layout_stateid *ls, *n;
+       LIST_HEAD(reaplist);
+
+       spin_lock(&clp->cl_lock);
+       list_for_each_entry_safe(ls, n, &clp->cl_lo_states, ls_perclnt)
+               nfsd4_return_all_layouts(ls, &reaplist);
+       spin_unlock(&clp->cl_lock);
+
+       nfsd4_free_layouts(&reaplist);
+}
+
+void
+nfsd4_return_all_file_layouts(struct nfs4_client *clp, struct nfs4_file *fp)
+{
+       struct nfs4_layout_stateid *ls, *n;
+       LIST_HEAD(reaplist);
+
+       spin_lock(&fp->fi_lock);
+       list_for_each_entry_safe(ls, n, &fp->fi_lo_states, ls_perfile) {
+               if (ls->ls_stid.sc_client == clp)
+                       nfsd4_return_all_layouts(ls, &reaplist);
+       }
+       spin_unlock(&fp->fi_lock);
+
+       nfsd4_free_layouts(&reaplist);
+}
+
+static void
+nfsd4_cb_layout_fail(struct nfs4_layout_stateid *ls)
+{
+       struct nfs4_client *clp = ls->ls_stid.sc_client;
+       char addr_str[INET6_ADDRSTRLEN];
+       static char *envp[] = {
+               "HOME=/",
+               "TERM=linux",
+               "PATH=/sbin:/usr/sbin:/bin:/usr/bin",
+               NULL
+       };
+       char *argv[8];
+       int error;
+
+       rpc_ntop((struct sockaddr *)&clp->cl_addr, addr_str, sizeof(addr_str));
+
+       trace_layout_recall_fail(&ls->ls_stid.sc_stateid);
+
+       printk(KERN_WARNING
+               "nfsd: client %s failed to respond to layout recall. "
+               "  Fencing..\n", addr_str);
+
+       argv[0] = "/sbin/nfsd-recall-failed";
+       argv[1] = addr_str;
+       argv[2] = ls->ls_file->f_path.mnt->mnt_sb->s_id;
+       argv[3] = NULL;
+
+       error = call_usermodehelper(argv[0], argv, envp, UMH_WAIT_PROC);
+       if (error) {
+               printk(KERN_ERR "nfsd: fence failed for client %s: %d!\n",
+                       addr_str, error);
+       }
+}
+
+static int
+nfsd4_cb_layout_done(struct nfsd4_callback *cb, struct rpc_task *task)
+{
+       struct nfs4_layout_stateid *ls =
+               container_of(cb, struct nfs4_layout_stateid, ls_recall);
+       LIST_HEAD(reaplist);
+
+       switch (task->tk_status) {
+       case 0:
+               return 1;
+       case -NFS4ERR_NOMATCHING_LAYOUT:
+               trace_layout_recall_done(&ls->ls_stid.sc_stateid);
+               task->tk_status = 0;
+               return 1;
+       case -NFS4ERR_DELAY:
+               /* Poll the client until it's done with the layout */
+               /* FIXME: cap number of retries.
+                * The pnfs standard states that we need to only expire
+                * the client after at-least "lease time" .eg lease-time * 2
+                * when failing to communicate a recall
+                */
+               rpc_delay(task, HZ/100); /* 10 mili-seconds */
+               return 0;
+       default:
+               /*
+                * Unknown error or non-responding client, we'll need to fence.
+                */
+               nfsd4_cb_layout_fail(ls);
+               return -1;
+       }
+}
+
+static void
+nfsd4_cb_layout_release(struct nfsd4_callback *cb)
+{
+       struct nfs4_layout_stateid *ls =
+               container_of(cb, struct nfs4_layout_stateid, ls_recall);
+       LIST_HEAD(reaplist);
+
+       trace_layout_recall_release(&ls->ls_stid.sc_stateid);
+
+       nfsd4_return_all_layouts(ls, &reaplist);
+       nfsd4_free_layouts(&reaplist);
+       nfs4_put_stid(&ls->ls_stid);
+}
+
+static struct nfsd4_callback_ops nfsd4_cb_layout_ops = {
+       .done           = nfsd4_cb_layout_done,
+       .release        = nfsd4_cb_layout_release,
+};
+
+static bool
+nfsd4_layout_lm_break(struct file_lock *fl)
+{
+       /*
+        * We don't want the locks code to timeout the lease for us;
+        * we'll remove it ourself if a layout isn't returned
+        * in time:
+        */
+       fl->fl_break_time = 0;
+       nfsd4_recall_file_layout(fl->fl_owner);
+       return false;
+}
+
+static int
+nfsd4_layout_lm_change(struct file_lock *onlist, int arg,
+               struct list_head *dispose)
+{
+       BUG_ON(!(arg & F_UNLCK));
+       return lease_modify(onlist, arg, dispose);
+}
+
+static const struct lock_manager_operations nfsd4_layouts_lm_ops = {
+       .lm_break       = nfsd4_layout_lm_break,
+       .lm_change      = nfsd4_layout_lm_change,
+};
+
+int
+nfsd4_init_pnfs(void)
+{
+       int i;
+
+       for (i = 0; i < DEVID_HASH_SIZE; i++)
+               INIT_LIST_HEAD(&nfsd_devid_hash[i]);
+
+       nfs4_layout_cache = kmem_cache_create("nfs4_layout",
+                       sizeof(struct nfs4_layout), 0, 0, NULL);
+       if (!nfs4_layout_cache)
+               return -ENOMEM;
+
+       nfs4_layout_stateid_cache = kmem_cache_create("nfs4_layout_stateid",
+                       sizeof(struct nfs4_layout_stateid), 0, 0, NULL);
+       if (!nfs4_layout_stateid_cache) {
+               kmem_cache_destroy(nfs4_layout_cache);
+               return -ENOMEM;
+       }
+       return 0;
+}
+
+void
+nfsd4_exit_pnfs(void)
+{
+       int i;
+
+       kmem_cache_destroy(nfs4_layout_cache);
+       kmem_cache_destroy(nfs4_layout_stateid_cache);
+
+       for (i = 0; i < DEVID_HASH_SIZE; i++) {
+               struct nfsd4_deviceid_map *map, *n;
+
+               list_for_each_entry_safe(map, n, &nfsd_devid_hash[i], hash)
+                       kfree(map);
+       }
+}