NFS: Adapt readdirplus to application usage patterns

While the use of READDIRPLUS is significantly more efficient than
READDIR followed by many LOOKUP calls, it is still less efficient
than just READDIR if the attributes are not required.

This patch tracks when lookups are attempted on the directory,
and uses that information to selectively disable READDIRPLUS
on that directory.
The first 'readdir' call is always served using READDIRPLUS.
Subsequent calls only use READDIRPLUS if there was a successful
lookup or revalidation on a child in the mean time.

Credit for the original idea should go to Neil Brown. See:
      http://www.spinics.net/lists/linux-nfs/msg19996.html
However, the implementation in this patch differs from Neil's
in that it focuses on tracking lookups rather than calls to
stat().

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: Neil Brown <neilb@suse.de>

diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index 82b42e2ea65c92ee326d128ff3f7817202e609fd..d0884c0d946452ef04c2f9cc8d2d54319b34842b 100644 (file)
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -474,6 +474,29 @@ different:
        return 0;
 }
 
+static
+bool nfs_use_readdirplus(struct inode *dir, struct file *filp)
+{
+       if (!nfs_server_capable(dir, NFS_CAP_READDIRPLUS))
+               return false;
+       if (test_and_clear_bit(NFS_INO_ADVISE_RDPLUS, &NFS_I(dir)->flags))
+               return true;
+       if (filp->f_pos == 0)
+               return true;
+       return false;
+}
+
+/*
+ * This function is called by the lookup code to request the use of
+ * readdirplus to accelerate any future lookups in the same
+ * directory.
+ */
+static
+void nfs_advise_use_readdirplus(struct inode *dir)
+{
+       set_bit(NFS_INO_ADVISE_RDPLUS, &NFS_I(dir)->flags);
+}
+
 static
 void nfs_prime_dcache(struct dentry *parent, struct nfs_entry *entry)
 {
@@ -874,7 +897,7 @@ static int nfs_readdir(struct file *filp, void *dirent, filldir_t filldir)
        desc->file = filp;
        desc->dir_cookie = &dir_ctx->dir_cookie;
        desc->decode = NFS_PROTO(inode)->decode_dirent;
-       desc->plus = NFS_USE_READDIRPLUS(inode);
+       desc->plus = nfs_use_readdirplus(inode, filp) ? 1 : 0;
 
        nfs_block_sillyrename(dentry);
        res = nfs_revalidate_mapping(inode, filp->f_mapping);
@@ -1114,7 +1137,7 @@ static int nfs_lookup_revalidate(struct dentry *dentry, struct nameidata *nd)
        if (!inode) {
                if (nfs_neg_need_reval(dir, dentry, nd))
                        goto out_bad;
-               goto out_valid;
+               goto out_valid_noent;
        }
 
        if (is_bad_inode(inode)) {
@@ -1156,6 +1179,9 @@ static int nfs_lookup_revalidate(struct dentry *dentry, struct nameidata *nd)
 out_set_verifier:
        nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
  out_valid:
+       /* Success: notify readdir to use READDIRPLUS */
+       nfs_advise_use_readdirplus(dir);
+ out_valid_noent:
        dput(parent);
        dfprintk(LOOKUPCACHE, "NFS: %s(%s/%s) is valid\n",
                        __func__, dentry->d_parent->d_name.name,
@@ -1311,6 +1337,9 @@ static struct dentry *nfs_lookup(struct inode *dir, struct dentry * dentry, stru
        if (IS_ERR(res))
                goto out_unblock_sillyrename;
 
+       /* Success: notify readdir to use READDIRPLUS */
+       nfs_advise_use_readdirplus(dir);
+
 no_entry:
        res = d_materialise_unique(dentry, inode);
        if (res != NULL) {

diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 0d53113207e53de0ad232a92d1c9e3e247d9d812..9f17cd19e7105a3c9c9e153e15de330b515d6771 100644 (file)
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -298,8 +298,6 @@ nfs_fhget(struct super_block *sb, struct nfs_fh *fh, struct nfs_fattr *fattr)
                        inode->i_op = NFS_SB(sb)->nfs_client->rpc_ops->dir_inode_ops;
                        inode->i_fop = &nfs_dir_operations;
                        inode->i_data.a_ops = &nfs_dir_aops;
-                       if (nfs_server_capable(inode, NFS_CAP_READDIRPLUS))
-                               set_bit(NFS_INO_ADVISE_RDPLUS, &NFS_I(inode)->flags);
                        /* Deal with crossing mountpoints */
                        if (fattr->valid & NFS_ATTR_FATTR_MOUNTPOINT ||
                                        fattr->valid & NFS_ATTR_FATTR_V4_REFERRAL) {

diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h
index 8a88c16662c5379eac1a8afbbfca7fc0a9e61dab..6cc7dbaf0695727214a542d6f21f3859ea811dc1 100644 (file)
--- a/include/linux/nfs_fs.h
+++ b/include/linux/nfs_fs.h
@@ -314,11 +314,6 @@ static inline int nfs_server_capable(struct inode *inode, int cap)
        return NFS_SERVER(inode)->caps & cap;
 }
 
-static inline int NFS_USE_READDIRPLUS(struct inode *inode)
-{
-       return test_bit(NFS_INO_ADVISE_RDPLUS, &NFS_I(inode)->flags);
-}
-
 static inline void nfs_set_verifier(struct dentry * dentry, unsigned long verf)
 {
        dentry->d_time = verf;