pnfs: change how lsegs are removed from layout list

This is to prepare the way for sensible io draining.  Instead of just
removing the lseg from the list, we instead clear the VALID flag
(preventing new io from grabbing references to the lseg) and remove
the reference holding it in the list.  Thus the lseg will be removed
once any io in progress completes and any references still held are
dropped.

Signed-off-by: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

3 files changed, 96 insertions, 44 deletions

diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index c7782b278e8b..790b786e1ae1 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -1410,9 +1410,9 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
  */
 void nfs4_evict_inode(struct inode *inode)
 {
+        pnfs_destroy_layout(NFS_I(inode));
         truncate_inode_pages(&inode->i_data, 0);
         end_writeback(inode);
-        pnfs_destroy_layout(NFS_I(inode));
         /* If we are holding a delegation, return it! */
         nfs_inode_return_delegation_noreclaim(inode);
         /* First call standard NFS clear_inode() code */
diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index 08313f536b45..212cbc22c59d 100644
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -211,68 +211,109 @@ static void
 init_lseg(struct pnfs_layout_hdr *lo, struct pnfs_layout_segment *lseg)
 {
         INIT_LIST_HEAD(&lseg->pls_list);
-        kref_init(&lseg->pls_refcount);
+        atomic_set(&lseg->pls_refcount, 1);
+        smp_mb();
+        set_bit(NFS_LSEG_VALID, &lseg->pls_flags);
         lseg->pls_layout = lo;
 }
 
-/* Called without i_lock held, as the free_lseg call may sleep */
-static void
-destroy_lseg(struct kref *kref)
+static void free_lseg(struct pnfs_layout_segment *lseg)
 {
-        struct pnfs_layout_segment *lseg =
-                container_of(kref, struct pnfs_layout_segment, pls_refcount);
         struct inode *ino = lseg->pls_layout->plh_inode;
 
-        dprintk("--> %s\n", __func__);
         NFS_SERVER(ino)->pnfs_curr_ld->free_lseg(lseg);
         /* Matched by get_layout_hdr in pnfs_insert_layout */
         put_layout_hdr(ino);
 }
 
-static void
-put_lseg(struct pnfs_layout_segment *lseg)
+/* The use of tmp_list is necessary because pnfs_curr_ld->free_lseg
+ * could sleep, so must be called outside of the lock.
+ * Returns 1 if object was removed, otherwise return 0.
+ */
+static int
+put_lseg_locked(struct pnfs_layout_segment *lseg,
+                struct list_head *tmp_list)
 {
-        if (!lseg)
-                return;
+        dprintk("%s: lseg %p ref %d valid %d\n", __func__, lseg,
+                atomic_read(&lseg->pls_refcount),
+                test_bit(NFS_LSEG_VALID, &lseg->pls_flags));
+        if (atomic_dec_and_test(&lseg->pls_refcount)) {
+                struct inode *ino = lseg->pls_layout->plh_inode;
 
-        dprintk("%s: lseg %p ref %d\n", __func__, lseg,
-                atomic_read(&lseg->pls_refcount.refcount));
-        kref_put(&lseg->pls_refcount, destroy_lseg);
+                BUG_ON(test_bit(NFS_LSEG_VALID, &lseg->pls_flags));
+                list_del(&lseg->pls_list);
+                if (list_empty(&lseg->pls_layout->plh_segs)) {
+                        struct nfs_client *clp;
+
+                        clp = NFS_SERVER(ino)->nfs_client;
+                        spin_lock(&clp->cl_lock);
+                        /* List does not take a reference, so no need for put here */
+                        list_del_init(&lseg->pls_layout->plh_layouts);
+                        spin_unlock(&clp->cl_lock);
+                }
+                list_add(&lseg->pls_list, tmp_list);
+                return 1;
+        }
+        return 0;
 }
 
-static void
-pnfs_clear_lseg_list(struct pnfs_layout_hdr *lo, struct list_head *tmp_list)
+static bool
+should_free_lseg(u32 lseg_iomode, u32 recall_iomode)
 {
-        struct pnfs_layout_segment *lseg, *next;
-        struct nfs_client *clp;
+        return (recall_iomode == IOMODE_ANY ||
+                lseg_iomode == recall_iomode);
+}
 
-        dprintk("%s:Begin lo %p\n", __func__, lo);
+/* Returns 1 if lseg is removed from list, 0 otherwise */
+static int mark_lseg_invalid(struct pnfs_layout_segment *lseg,
+                             struct list_head *tmp_list)
+{
+        int rv = 0;
 
-        assert_spin_locked(&lo->plh_inode->i_lock);
-        list_for_each_entry_safe(lseg, next, &lo->plh_segs, pls_list) {
-                dprintk("%s: freeing lseg %p\n", __func__, lseg);
-                list_move(&lseg->pls_list, tmp_list);
+        if (test_and_clear_bit(NFS_LSEG_VALID, &lseg->pls_flags)) {
+                /* Remove the reference keeping the lseg in the
+                 * list.  It will now be removed when all
+                 * outstanding io is finished.
+                 */
+                rv = put_lseg_locked(lseg, tmp_list);
         }
-        clp = NFS_SERVER(lo->plh_inode)->nfs_client;
-        spin_lock(&clp->cl_lock);
-        /* List does not take a reference, so no need for put here */
-        list_del_init(&lo->plh_layouts);
-        spin_unlock(&clp->cl_lock);
+        return rv;
+}
 
-        dprintk("%s:Return\n", __func__);
+/* Returns count of number of matching invalid lsegs remaining in list
+ * after call.
+ */
+static int
+mark_matching_lsegs_invalid(struct pnfs_layout_hdr *lo,
+                            struct list_head *tmp_list,
+                            u32 iomode)
+{
+        struct pnfs_layout_segment *lseg, *next;
+        int invalid = 0, removed = 0;
+
+        dprintk("%s:Begin lo %p\n", __func__, lo);
+
+        list_for_each_entry_safe(lseg, next, &lo->plh_segs, pls_list)
+                if (should_free_lseg(lseg->pls_range.iomode, iomode)) {
+                        dprintk("%s: freeing lseg %p iomode %d "
+                                "offset %llu length %llu\n", __func__,
+                                lseg, lseg->pls_range.iomode, lseg->pls_range.offset,
+                                lseg->pls_range.length);
+                        invalid++;
+                        removed += mark_lseg_invalid(lseg, tmp_list);
+                }
+        dprintk("%s:Return %i\n", __func__, invalid - removed);
+        return invalid - removed;
 }
 
 static void
-pnfs_free_lseg_list(struct list_head *tmp_list)
+pnfs_free_lseg_list(struct list_head *free_me)
 {
-        struct pnfs_layout_segment *lseg;
+        struct pnfs_layout_segment *lseg, *tmp;
 
-        while (!list_empty(tmp_list)) {
-                lseg = list_entry(tmp_list->next, struct pnfs_layout_segment,
-                                pls_list);
-                dprintk("%s calling put_lseg on %p\n", __func__, lseg);
+        list_for_each_entry_safe(lseg, tmp, free_me, pls_list) {
                 list_del(&lseg->pls_list);
-                put_lseg(lseg);
+                free_lseg(lseg);
         }
 }
 
@@ -285,7 +326,8 @@ pnfs_destroy_layout(struct nfs_inode *nfsi)
         spin_lock(&nfsi->vfs_inode.i_lock);
         lo = nfsi->layout;
         if (lo) {
-                pnfs_clear_lseg_list(lo, &tmp_list);
+                set_bit(NFS_LAYOUT_DESTROYED, &nfsi->layout->plh_flags);
+                mark_matching_lsegs_invalid(lo, &tmp_list, IOMODE_ANY);
                 /* Matched by refcount set to 1 in alloc_init_layout_hdr */
                 put_layout_hdr_locked(lo);
         }
@@ -477,9 +519,12 @@ pnfs_find_alloc_layout(struct inode *ino)
         dprintk("%s Begin ino=%p layout=%p\n", __func__, ino, nfsi->layout);
 
         assert_spin_locked(&ino->i_lock);
-        if (nfsi->layout)
-                return nfsi->layout;
-
+        if (nfsi->layout) {
+                if (test_bit(NFS_LAYOUT_DESTROYED, &nfsi->layout->plh_flags))
+                        return NULL;
+                else
+                        return nfsi->layout;
+        }
         spin_unlock(&ino->i_lock);
         new = alloc_init_layout_hdr(ino);
         spin_lock(&ino->i_lock);
@@ -520,7 +565,8 @@ pnfs_has_layout(struct pnfs_layout_hdr *lo, u32 iomode)
 
         assert_spin_locked(&lo->plh_inode->i_lock);
         list_for_each_entry(lseg, &lo->plh_segs, pls_list) {
-                if (is_matching_lseg(lseg, iomode)) {
+                if (test_bit(NFS_LSEG_VALID, &lseg->pls_flags) &&
+                    is_matching_lseg(lseg, iomode)) {
                         ret = lseg;
                         break;
                 }
@@ -529,7 +575,7 @@ pnfs_has_layout(struct pnfs_layout_hdr *lo, u32 iomode)
         }
 
         dprintk("%s:Return lseg %p ref %d\n",
-                __func__, ret, ret ? atomic_read(&ret->pls_refcount.refcount) : 0);
+                __func__, ret, ret ? atomic_read(&ret->pls_refcount) : 0);
         return ret;
 }
 
diff --git a/fs/nfs/pnfs.h b/fs/nfs/pnfs.h
index 10937203d236..787253e6fca3 100644
--- a/fs/nfs/pnfs.h
+++ b/fs/nfs/pnfs.h
@@ -30,10 +30,15 @@
 #ifndef FS_NFS_PNFS_H
 #define FS_NFS_PNFS_H
 
+enum {
+        NFS_LSEG_VALID = 0,     /* cleared when lseg is recalled/returned */
+};
+
 struct pnfs_layout_segment {
         struct list_head pls_list;
         struct pnfs_layout_range pls_range;
-        struct kref pls_refcount;
+        atomic_t pls_refcount;
+        unsigned long pls_flags;
         struct pnfs_layout_hdr *pls_layout;
 };
 
@@ -44,6 +49,7 @@ struct pnfs_layout_segment {
 enum {
         NFS_LAYOUT_RO_FAILED = 0,       /* get ro layout failed stop trying */
         NFS_LAYOUT_RW_FAILED,           /* get rw layout failed stop trying */
+        NFS_LAYOUT_DESTROYED,           /* no new use of layout allowed */
 };
 
 /* Per-layout driver specific registration structure */