ceph: add dir_layout to inode

Add a ceph_dir_layout to the inode, and calculate dentry hash values based
on the parent directory's specified dir_hash function.  This is needed
because the old default Linux dcache hash function is extremely week and
leads to a poor distribution of files among dir fragments.

Signed-off-by: Sage Weil <sage@newdream.net>

4 files changed, 25 insertions, 1 deletions

diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c
index d902948a90d8..562f9884a4d9 100644
--- a/fs/ceph/dir.c
+++ b/fs/ceph/dir.c
@@ -1216,6 +1216,26 @@ void ceph_dentry_lru_del(struct dentry *dn)
         }
 }
 
+/*
+ * Return name hash for a given dentry.  This is dependent on
+ * the parent directory's hash function.
+ */
+unsigned ceph_dentry_hash(struct dentry *dn)
+{
+        struct inode *dir = dn->d_parent->d_inode;
+        struct ceph_inode_info *dci = ceph_inode(dir);
+
+        switch (dci->i_dir_layout.dl_dir_hash) {
+        case 0: /* for backward compat */
+        case CEPH_STR_HASH_LINUX:
+                return dn->d_name.hash;
+
+        default:
+                return ceph_str_hash(dci->i_dir_layout.dl_dir_hash,
+                                     dn->d_name.name, dn->d_name.len);
+        }
+}
+
 const struct file_operations ceph_dir_fops = {
         .read = ceph_read_dir,
         .readdir = ceph_readdir,
diff --git a/fs/ceph/export.c b/fs/ceph/export.c
index 2297d9426992..e41056174bf8 100644
--- a/fs/ceph/export.c
+++ b/fs/ceph/export.c
@@ -59,7 +59,7 @@ static int ceph_encode_fh(struct dentry *dentry, u32 *rawfh, int *max_len,
                 dout("encode_fh %p connectable\n", dentry);
                 cfh->ino = ceph_ino(dentry->d_inode);
                 cfh->parent_ino = ceph_ino(parent->d_inode);
-                cfh->parent_name_hash = parent->d_name.hash;
+                cfh->parent_name_hash = ceph_dentry_hash(parent);
                 *max_len = connected_handle_length;
                 type = 2;
         } else if (*max_len >= handle_length) {
diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
index bf1286588f26..045283ce4413 100644
--- a/fs/ceph/inode.c
+++ b/fs/ceph/inode.c
@@ -297,6 +297,8 @@ struct inode *ceph_alloc_inode(struct super_block *sb)
         ci->i_release_count = 0;
         ci->i_symlink = NULL;
 
+        memset(&ci->i_dir_layout, 0, sizeof(ci->i_dir_layout));
+
         ci->i_fragtree = RB_ROOT;
         mutex_init(&ci->i_fragtree_mutex);
 
diff --git a/fs/ceph/super.h b/fs/ceph/super.h
index 7f01728a4657..6e0826695112 100644
--- a/fs/ceph/super.h
+++ b/fs/ceph/super.h
@@ -239,6 +239,7 @@ struct ceph_inode_info {
         unsigned i_ceph_flags;
         unsigned long i_release_count;
 
+        struct ceph_dir_layout i_dir_layout;
         struct ceph_file_layout i_layout;
         char *i_symlink;
 
@@ -768,6 +769,7 @@ extern void ceph_dentry_lru_add(struct dentry *dn);
 extern void ceph_dentry_lru_touch(struct dentry *dn);
 extern void ceph_dentry_lru_del(struct dentry *dn);
 extern void ceph_invalidate_dentry_lease(struct dentry *dentry);
+extern unsigned ceph_dentry_hash(struct dentry *dn);
 
 /*
  * our d_ops vary depending on whether the inode is live,