NFS: Improve caching while holding a delegation
authorTrond Myklebust <trond.myklebust@hammerspace.com>
Sun, 3 Jun 2018 15:38:38 +0000 (11:38 -0400)
committerTrond Myklebust <trond.myklebust@hammerspace.com>
Mon, 4 Jun 2018 19:03:58 +0000 (15:03 -0400)
Make sure that the client completely ignores change attribute and size
changes on the server when it holds a delegation.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
fs/nfs/inode.c

index f35c5eb09cc9ab2fc2363eb43e30ee80bbda154c..90d9fbfd82db9ca47335234c3789f746d8fdc04b 100644 (file)
@@ -1789,6 +1789,7 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
        bool have_writers = nfs_file_has_buffered_writers(nfsi);
        bool cache_revalidated = true;
        bool attr_changed = false;
+       bool have_delegation;
 
        dfprintk(VFS, "NFS: %s(%s/%lu fh_crc=0x%08x ct=%d info=0x%x)\n",
                        __func__, inode->i_sb->s_id, inode->i_ino,
@@ -1823,6 +1824,9 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
                        !IS_AUTOMOUNT(inode))
                server->fsid = fattr->fsid;
 
+       /* Save the delegation state before clearing cache_validity */
+       have_delegation = nfs_have_delegated_attributes(inode);
+
        /*
         * Update the read time so we don't revalidate too often.
         */
@@ -1845,10 +1849,8 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
        /* More cache consistency checks */
        if (fattr->valid & NFS_ATTR_FATTR_CHANGE) {
                if (!inode_eq_iversion_raw(inode, fattr->change_attr)) {
-                       dprintk("NFS: change_attr change on server for file %s/%ld\n",
-                                       inode->i_sb->s_id, inode->i_ino);
                        /* Could it be a race with writeback? */
-                       if (!have_writers) {
+                       if (!(have_writers || have_delegation)) {
                                invalid |= NFS_INO_INVALID_DATA
                                        | NFS_INO_INVALID_ACCESS
                                        | NFS_INO_INVALID_ACL;
@@ -1859,6 +1861,9 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
                                        | NFS_INO_INVALID_OTHER;
                                if (S_ISDIR(inode->i_mode))
                                        nfs_force_lookup_revalidate(inode);
+                               dprintk("NFS: change_attr change on server for file %s/%ld\n",
+                                               inode->i_sb->s_id,
+                                               inode->i_ino);
                        }
                        inode_set_iversion_raw(inode, fattr->change_attr);
                        attr_changed = true;
@@ -1893,7 +1898,7 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
        if (fattr->valid & NFS_ATTR_FATTR_SIZE) {
                new_isize = nfs_size_to_loff_t(fattr->size);
                cur_isize = i_size_read(inode);
-               if (new_isize != cur_isize) {
+               if (new_isize != cur_isize && !have_delegation) {
                        /* Do we perhaps have any outstanding writes, or has
                         * the file grown beyond our last write? */
                        if (!nfs_have_writebacks(inode) || new_isize > cur_isize) {
@@ -2022,9 +2027,7 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
        if (!(S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode)
                                || S_ISLNK(inode->i_mode)))
                invalid &= ~NFS_INO_INVALID_DATA;
-       if (!NFS_PROTO(inode)->have_delegation(inode, FMODE_READ) ||
-                       (save_cache_validity & NFS_INO_REVAL_FORCED))
-               nfs_set_cache_invalid(inode, invalid);
+       nfs_set_cache_invalid(inode, invalid);
 
        return 0;
  out_err: