1 files changed, 72 insertions, 9 deletions

diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index 48ac5aad6258..4bdffe0ba025 100644
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -417,6 +417,16 @@ should_free_lseg(struct pnfs_layout_range *lseg_range,
                lo_seg_intersecting(lseg_range, recall_range);
 }
 
+static bool pnfs_lseg_dec_and_remove_zero(struct pnfs_layout_segment *lseg,
+                struct list_head *tmp_list)
+{
+        if (!atomic_dec_and_test(&lseg->pls_refcount))
+                return false;
+        pnfs_layout_remove_lseg(lseg->pls_layout, lseg);
+        list_add(&lseg->pls_list, tmp_list);
+        return true;
+}
+
 /* 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)
@@ -430,11 +440,8 @@ static int mark_lseg_invalid(struct pnfs_layout_segment *lseg,
                  */
                 dprintk("%s: lseg %p ref %d\n", __func__, lseg,
                         atomic_read(&lseg->pls_refcount));
-                if (atomic_dec_and_test(&lseg->pls_refcount)) {
-                        pnfs_layout_remove_lseg(lseg->pls_layout, lseg);
-                        list_add(&lseg->pls_list, tmp_list);
+                if (pnfs_lseg_dec_and_remove_zero(lseg, tmp_list))
                         rv = 1;
-                }
         }
         return rv;
 }
@@ -777,6 +784,21 @@ send_layoutget(struct pnfs_layout_hdr *lo,
         return lseg;
 }
 
+static void pnfs_clear_layoutcommit(struct inode *inode,
+                struct list_head *head)
+{
+        struct nfs_inode *nfsi = NFS_I(inode);
+        struct pnfs_layout_segment *lseg, *tmp;
+
+        if (!test_and_clear_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags))
+                return;
+        list_for_each_entry_safe(lseg, tmp, &nfsi->layout->plh_segs, pls_list) {
+                if (!test_and_clear_bit(NFS_LSEG_LAYOUTCOMMIT, &lseg->pls_flags))
+                        continue;
+                pnfs_lseg_dec_and_remove_zero(lseg, head);
+        }
+}
+
 /*
  * Initiates a LAYOUTRETURN(FILE), and removes the pnfs_layout_hdr
  * when the layout segment list is empty.
@@ -808,6 +830,7 @@ _pnfs_return_layout(struct inode *ino)
         /* Reference matched in nfs4_layoutreturn_release */
         pnfs_get_layout_hdr(lo);
         empty = list_empty(&lo->plh_segs);
+        pnfs_clear_layoutcommit(ino, &tmp_list);
         pnfs_mark_matching_lsegs_invalid(lo, &tmp_list, NULL);
         /* Don't send a LAYOUTRETURN if list was initially empty */
         if (empty) {
@@ -820,8 +843,6 @@ _pnfs_return_layout(struct inode *ino)
         spin_unlock(&ino->i_lock);
         pnfs_free_lseg_list(&tmp_list);
 
-        WARN_ON(test_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags));
-
         lrp = kzalloc(sizeof(*lrp), GFP_KERNEL);
         if (unlikely(lrp == NULL)) {
                 status = -ENOMEM;
@@ -845,6 +866,33 @@ out:
 }
 EXPORT_SYMBOL_GPL(_pnfs_return_layout);
 
+int
+pnfs_commit_and_return_layout(struct inode *inode)
+{
+        struct pnfs_layout_hdr *lo;
+        int ret;
+
+        spin_lock(&inode->i_lock);
+        lo = NFS_I(inode)->layout;
+        if (lo == NULL) {
+                spin_unlock(&inode->i_lock);
+                return 0;
+        }
+        pnfs_get_layout_hdr(lo);
+        /* Block new layoutgets and read/write to ds */
+        lo->plh_block_lgets++;
+        spin_unlock(&inode->i_lock);
+        filemap_fdatawait(inode->i_mapping);
+        ret = pnfs_layoutcommit_inode(inode, true);
+        if (ret == 0)
+                ret = _pnfs_return_layout(inode);
+        spin_lock(&inode->i_lock);
+        lo->plh_block_lgets--;
+        spin_unlock(&inode->i_lock);
+        pnfs_put_layout_hdr(lo);
+        return ret;
+}
+
 bool pnfs_roc(struct inode *ino)
 {
         struct pnfs_layout_hdr *lo;
@@ -1458,7 +1506,6 @@ static void pnfs_ld_handle_write_error(struct nfs_write_data *data)
         dprintk("pnfs write error = %d\n", hdr->pnfs_error);
         if (NFS_SERVER(hdr->inode)->pnfs_curr_ld->flags &
             PNFS_LAYOUTRET_ON_ERROR) {
-                clear_bit(NFS_INO_LAYOUTCOMMIT, &NFS_I(hdr->inode)->flags);
                 pnfs_return_layout(hdr->inode);
         }
         if (!test_and_set_bit(NFS_IOHDR_REDO, &hdr->flags))
@@ -1613,7 +1660,6 @@ static void pnfs_ld_handle_read_error(struct nfs_read_data *data)
         dprintk("pnfs read error = %d\n", hdr->pnfs_error);
         if (NFS_SERVER(hdr->inode)->pnfs_curr_ld->flags &
             PNFS_LAYOUTRET_ON_ERROR) {
-                clear_bit(NFS_INO_LAYOUTCOMMIT, &NFS_I(hdr->inode)->flags);
                 pnfs_return_layout(hdr->inode);
         }
         if (!test_and_set_bit(NFS_IOHDR_REDO, &hdr->flags))
@@ -1746,11 +1792,27 @@ static void pnfs_list_write_lseg(struct inode *inode, struct list_head *listp)
 
         list_for_each_entry(lseg, &NFS_I(inode)->layout->plh_segs, pls_list) {
                 if (lseg->pls_range.iomode == IOMODE_RW &&
-                    test_bit(NFS_LSEG_LAYOUTCOMMIT, &lseg->pls_flags))
+                    test_and_clear_bit(NFS_LSEG_LAYOUTCOMMIT, &lseg->pls_flags))
                         list_add(&lseg->pls_lc_list, listp);
         }
 }
 
+static void pnfs_list_write_lseg_done(struct inode *inode, struct list_head *listp)
+{
+        struct pnfs_layout_segment *lseg, *tmp;
+        unsigned long *bitlock = &NFS_I(inode)->flags;
+
+        /* Matched by references in pnfs_set_layoutcommit */
+        list_for_each_entry_safe(lseg, tmp, listp, pls_lc_list) {
+                list_del_init(&lseg->pls_lc_list);
+                pnfs_put_lseg(lseg);
+        }
+
+        clear_bit_unlock(NFS_INO_LAYOUTCOMMITTING, bitlock);
+        smp_mb__after_clear_bit();
+        wake_up_bit(bitlock, NFS_INO_LAYOUTCOMMITTING);
+}
+
 void pnfs_set_lo_fail(struct pnfs_layout_segment *lseg)
 {
         pnfs_layout_io_set_failed(lseg->pls_layout, lseg->pls_range.iomode);
@@ -1795,6 +1857,7 @@ void pnfs_cleanup_layoutcommit(struct nfs4_layoutcommit_data *data)
 
         if (nfss->pnfs_curr_ld->cleanup_layoutcommit)
                 nfss->pnfs_curr_ld->cleanup_layoutcommit(data);
+        pnfs_list_write_lseg_done(data->args.inode, &data->lseg_list);
 }
 
 /*