ceph: make seeky readdir more efficient

Current cephfs client uses string to indicate start position of
readdir. The string is last entry of previous readdir reply.
This approach does not work for seeky readdir because we can
not easily convert the new postion to a string. For seeky readdir,
mds needs to return dentries from the beginning. Client keeps
retrying if the reply does not contain the dentry it wants.

In current version of ceph, mds sorts CDentry in its cache in
hash order. Client also uses dentry hash to compose dir postion.
For seeky readdir, if client passes the hash part of dir postion
to mds. mds can avoid replying useless dentries.

Signed-off-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

5 files changed, 21 insertions, 6 deletions

diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c
index 3e9ad501addf..ae61cdf7d489 100644
--- a/fs/ceph/dir.c
+++ b/fs/ceph/dir.c
@@ -378,7 +378,11 @@ more:
                                 ceph_mdsc_put_request(req);
                                 return -ENOMEM;
                         }
+                } else if (is_hash_order(ctx->pos)) {
+                        req->r_args.readdir.offset_hash =
+                                cpu_to_le32(fpos_hash(ctx->pos));
                 }
+
                 req->r_dir_release_cnt = fi->dir_release_count;
                 req->r_dir_ordered_cnt = fi->dir_ordered_count;
                 req->r_readdir_cache_idx = fi->readdir_cache_idx;
diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
index d3119fe3ab45..dcce79b84406 100644
--- a/fs/ceph/inode.c
+++ b/fs/ceph/inode.c
@@ -1482,10 +1482,17 @@ int ceph_readdir_prepopulate(struct ceph_mds_request *req,
         if (test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags))
                 return readdir_prepopulate_inodes_only(req, session);
 
-        if (rinfo->hash_order && req->r_path2) {
-                last_hash = ceph_str_hash(ci->i_dir_layout.dl_dir_hash,
-                                          req->r_path2, strlen(req->r_path2));
-                last_hash = ceph_frag_value(last_hash);
+        if (rinfo->hash_order) {
+                if (req->r_path2) {
+                        last_hash = ceph_str_hash(ci->i_dir_layout.dl_dir_hash,
+                                                  req->r_path2,
+                                                  strlen(req->r_path2));
+                        last_hash = ceph_frag_value(last_hash);
+                } else if (rinfo->offset_hash) {
+                        /* mds understands offset_hash */
+                        WARN_ON_ONCE(req->r_readdir_offset != 2);
+                        last_hash = le32_to_cpu(rhead->args.readdir.offset_hash);
+                }
         }
 
         if (rinfo->dir_dir &&
@@ -1510,7 +1517,7 @@ int ceph_readdir_prepopulate(struct ceph_mds_request *req,
         }
 
         if (ceph_frag_is_leftmost(frag) && req->r_readdir_offset == 2 &&
-            !(rinfo->hash_order && req->r_path2)) {
+            !(rinfo->hash_order && last_hash)) {
                 /* note dir version at start of readdir so we can tell
                  * if any dentries get dropped */
                 req->r_dir_release_cnt = atomic64_read(&ci->i_release_count);
diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
index a22688873ec3..8cc4d4e8b077 100644
--- a/fs/ceph/mds_client.c
+++ b/fs/ceph/mds_client.c
@@ -189,6 +189,7 @@ static int parse_reply_info_dir(void **p, void *end,
                 info->dir_end = !!(flags & CEPH_READDIR_FRAG_END);
                 info->dir_complete = !!(flags & CEPH_READDIR_FRAG_COMPLETE);
                 info->hash_order = !!(flags & CEPH_READDIR_HASH_ORDER);
+                info->offset_hash = !!(flags & CEPH_READDIR_OFFSET_HASH);
         }
         if (num == 0)
                 goto done;
diff --git a/fs/ceph/mds_client.h b/fs/ceph/mds_client.h
index bbebcd55d79e..3e67dd2169fa 100644
--- a/fs/ceph/mds_client.h
+++ b/fs/ceph/mds_client.h
@@ -83,9 +83,10 @@ struct ceph_mds_reply_info_parsed {
                         struct ceph_mds_reply_dirfrag *dir_dir;
                         size_t                        dir_buf_size;
                         int                           dir_nr;
-                        bool                          dir_complete;
                         bool                          dir_end;
+                        bool                          dir_complete;
                         bool                          hash_order;
+                        bool                          offset_hash;
                         struct ceph_mds_reply_dir_entry  *dir_entries;
                 };
 
diff --git a/include/linux/ceph/ceph_fs.h b/include/linux/ceph/ceph_fs.h
index f4b2ee18f38c..1787e4a8e251 100644
--- a/include/linux/ceph/ceph_fs.h
+++ b/include/linux/ceph/ceph_fs.h
@@ -365,6 +365,7 @@ extern const char *ceph_mds_op_name(int op);
 #define CEPH_READDIR_FRAG_END           (1<<0)
 #define CEPH_READDIR_FRAG_COMPLETE      (1<<8)
 #define CEPH_READDIR_HASH_ORDER         (1<<9)
+#define CEPH_READDIR_OFFSET_HASH        (1<<10)
 
 union ceph_mds_request_args {
         struct {
@@ -384,6 +385,7 @@ union ceph_mds_request_args {
                 __le32 max_entries;          /* how many dentries to grab */
                 __le32 max_bytes;
                 __le16 flags;
+                __le32 offset_hash;
         } __attribute__ ((packed)) readdir;
         struct {
                 __le32 mode;