From 9d8e4e8a384d2f443694a816d41f414e3e9715f6 Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Wed, 20 Aug 2008 09:36:24 -0700
Subject: [PATCH] kclient: introduce ceph_cap_snap

---
 src/kernel/caps.c       | 58 ++++++++++++++++++++---------------------
 src/kernel/mds_client.c | 12 ++++-----
 src/kernel/mds_client.h |  7 +----
 src/kernel/super.c      |  1 +
 src/kernel/super.h      | 37 +++++++++++++++++++-------
 5 files changed, 64 insertions(+), 51 deletions(-)

diff --git a/src/kernel/caps.c b/src/kernel/caps.c
index 695c04a9700b3..c977b443f9653 100644
--- a/src/kernel/caps.c
+++ b/src/kernel/caps.c
@@ -12,14 +12,14 @@ int ceph_debug_caps = -1;
 #include "messenger.h"
 
 
-static struct ceph_inode_cap *__get_cap_for_mds(struct inode *inode, int mds)
+static struct ceph_cap *__get_cap_for_mds(struct inode *inode, int mds)
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
-	struct ceph_inode_cap *cap;
+	struct ceph_cap *cap;
 	struct rb_node *n = ci->i_caps.rb_node;
 
 	while (n) {
-		cap = rb_entry(n, struct ceph_inode_cap, ci_node);
+		cap = rb_entry(n, struct ceph_cap, ci_node);
 		if (mds < cap->mds)
 			n = n->rb_left;
 		else if (mds > cap->mds)
@@ -31,15 +31,15 @@ static struct ceph_inode_cap *__get_cap_for_mds(struct inode *inode, int mds)
 }
 
 static void __insert_cap_node(struct ceph_inode_info *ci,
-			      struct ceph_inode_cap *new)
+			      struct ceph_cap *new)
 {
 	struct rb_node **p = &ci->i_caps.rb_node;
 	struct rb_node *parent = NULL;
-	struct ceph_inode_cap *cap = 0;
+	struct ceph_cap *cap = 0;
 
 	while (*p) {
 		parent = *p;
-		cap = rb_entry(parent, struct ceph_inode_cap, ci_node);
+		cap = rb_entry(parent, struct ceph_cap, ci_node);
 		if (new->mds < cap->mds)
 			p = &(*p)->rb_left;
 		else if (new->mds > cap->mds)
@@ -55,7 +55,7 @@ static void __insert_cap_node(struct ceph_inode_info *ci,
 int ceph_get_cap_mds(struct inode *inode)
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
-	struct ceph_inode_cap *cap;
+	struct ceph_cap *cap;
 	int mds = -1;
 
 	/*
@@ -63,7 +63,7 @@ int ceph_get_cap_mds(struct inode *inode)
 	 */
 	spin_lock(&inode->i_lock);
 	if (!RB_EMPTY_ROOT(&ci->i_caps)) {
-		cap = rb_entry(ci->i_caps.rb_node, struct ceph_inode_cap,
+		cap = rb_entry(ci->i_caps.rb_node, struct ceph_cap,
 			       ci_node);
 		mds = cap->mds;
 	}
@@ -84,7 +84,7 @@ int ceph_add_cap(struct inode *inode,
 {
 	int mds = session->s_mds;
 	struct ceph_inode_info *ci = ceph_inode(inode);
-	struct ceph_inode_cap *cap, *new_cap = 0;
+	struct ceph_cap *cap, *new_cap = 0;
 	int i;
 	int is_first = 0;
 	struct ceph_snaprealm *realm = 0;
@@ -166,13 +166,13 @@ retry:
 int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented)
 {
 	int have = ci->i_snap_caps;
-	struct ceph_inode_cap *cap;
+	struct ceph_cap *cap;
 	u32 gen;
 	unsigned long ttl;
 	struct rb_node *p;
 
 	for (p = rb_first(&ci->i_caps); p; p = rb_next(p)) {
-		cap = rb_entry(p, struct ceph_inode_cap, ci_node);
+		cap = rb_entry(p, struct ceph_cap, ci_node);
 
 		spin_lock(&cap->session->s_cap_lock);
 		gen = cap->session->s_cap_gen;
@@ -198,7 +198,7 @@ int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented)
  * caller should hold i_lock, snap_rwsem, and session s_mutex.
  * returns true if this is the last cap.  if so, caller should iput.
  */
-int __ceph_remove_cap(struct ceph_inode_cap *cap)
+int __ceph_remove_cap(struct ceph_cap *cap)
 {
 	struct ceph_mds_session *session = cap->session;
 	struct ceph_inode_info *ci = cap->ci;
@@ -228,7 +228,7 @@ int __ceph_remove_cap(struct ceph_inode_cap *cap)
 /*
  * caller should hold snap_rwsem and session s_mutex.
  */
-void ceph_remove_cap(struct ceph_inode_cap *cap)
+void ceph_remove_cap(struct ceph_cap *cap)
 {
 	struct inode *inode = &cap->ci->vfs_inode;
 	int was_last;
@@ -272,7 +272,7 @@ void ceph_check_caps(struct ceph_inode_info *ci, int is_delayed, int flush_snap)
 	struct ceph_client *client = ceph_inode_to_client(&ci->vfs_inode);
 	struct ceph_mds_client *mdsc = &client->mdsc;
 	struct inode *inode = &ci->vfs_inode;
-	struct ceph_inode_cap *cap;
+	struct ceph_cap *cap;
 	int wanted, used;
 	struct ceph_mds_session *session = 0;  /* if non-NULL, i hold s_mutex */
 	int took_snap_rwsem = 0;             /* true if mdsc->snap_rwsem held */
@@ -291,7 +291,7 @@ retry:
 		__ceph_cap_delay_requeue(mdsc, ci);
 	
 	for (p = rb_first(&ci->i_caps); p; p = rb_next(p)) {
-		cap = rb_entry(p, struct ceph_inode_cap, ci_node);
+		cap = rb_entry(p, struct ceph_cap, ci_node);
 		if (mds >= cap->mds)
 			continue;
 
@@ -374,8 +374,8 @@ ack:
 		mds = cap->mds;  /* remember mds, so we don't repeat */
 
 		/* send_cap drops i_lock */
-		__ceph_mdsc_send_cap(mdsc, session, cap,
-				     used, wanted, flush_snap);
+		__ceph_send_cap(mdsc, session, cap,
+				used, wanted, flush_snap);
 
 		goto retry; /* retake i_lock and restart our cap scan. */
 	}
@@ -561,7 +561,7 @@ static void send_cap(struct ceph_mds_client *mdsc, __u64 ino, int op,
 static int handle_cap_grant(struct inode *inode, struct ceph_mds_caps *grant,
 			    struct ceph_mds_session *session)
 {
-	struct ceph_inode_cap *cap;
+	struct ceph_cap *cap;
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	int mds = session->s_mds;
 	int seq = le32_to_cpu(grant->seq);
@@ -707,7 +707,7 @@ static void handle_cap_released(struct inode *inode,
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	int seq = le32_to_cpu(m->seq);
 	int removed_last;
-	struct ceph_inode_cap *cap;
+	struct ceph_cap *cap;
 
 	dout(10, "handle_cap_released inode %p ci %p mds%d seq %d\n", inode, ci,
 	     session->s_mds, seq);
@@ -797,7 +797,7 @@ static void handle_cap_export(struct inode *inode, struct ceph_mds_caps *ex,
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	int mds = session->s_mds;
 	unsigned mseq = le32_to_cpu(ex->migrate_seq);
-	struct ceph_inode_cap *cap = 0, *t;
+	struct ceph_cap *cap = 0, *t;
 	struct rb_node *p;
 	int was_last = 0;
 
@@ -808,7 +808,7 @@ static void handle_cap_export(struct inode *inode, struct ceph_mds_caps *ex,
 
 	/* make sure we haven't seen a higher mseq */
 	for (p = rb_first(&ci->i_caps); p; p = rb_next(p)) {
-		t = rb_entry(p, struct ceph_inode_cap, ci_node);
+		t = rb_entry(p, struct ceph_cap, ci_node);
 		if (t->mseq > mseq) {
 			dout(10, " higher mseq on cap from mds%d\n",
 			     t->session->s_mds);
@@ -982,11 +982,11 @@ bad:
  *
  * returns true if we removed the last cap on this inode.
  */
-int __ceph_mdsc_send_cap(struct ceph_mds_client *mdsc,
-			 struct ceph_mds_session *session,
-			 struct ceph_inode_cap *cap,
-			 int used, int wanted,
-			 int flush_snap)
+int __ceph_send_cap(struct ceph_mds_client *mdsc,
+		    struct ceph_mds_session *session,
+		    struct ceph_cap *cap,
+		    int used, int wanted,
+		    int flush_snap)
 {
 	struct ceph_inode_info *ci = cap->ci;
 	struct inode *inode = &ci->vfs_inode;
@@ -1078,8 +1078,8 @@ void ceph_flush_write_caps(struct ceph_mds_client *mdsc,
 	struct list_head *p, *n;
 
 	list_for_each_safe (p, n, &session->s_caps) {
-		struct ceph_inode_cap *cap =
-			list_entry(p, struct ceph_inode_cap, session_caps);
+		struct ceph_cap *cap =
+			list_entry(p, struct ceph_cap, session_caps);
 		struct inode *inode = &cap->ci->vfs_inode;
 		int used, wanted;
 
@@ -1099,7 +1099,7 @@ void ceph_flush_write_caps(struct ceph_mds_client *mdsc,
 			used = wanted = 0;
 		}
 
-		__ceph_mdsc_send_cap(mdsc, session, cap, used, wanted, 0);
+		__ceph_send_cap(mdsc, session, cap, used, wanted, 0);
 	}
 }
 
diff --git a/src/kernel/mds_client.c b/src/kernel/mds_client.c
index 54b2aebed8e1d..c4ba343b8fd6a 100644
--- a/src/kernel/mds_client.c
+++ b/src/kernel/mds_client.c
@@ -606,7 +606,7 @@ static int open_session(struct ceph_mds_client *mdsc,
  */
 static void remove_session_caps(struct ceph_mds_session *session)
 {
-	struct ceph_inode_cap *cap;
+	struct ceph_cap *cap;
 	struct ceph_inode_info *ci;
 
 	/*
@@ -615,7 +615,7 @@ static void remove_session_caps(struct ceph_mds_session *session)
 	 */
 	dout(10, "remove_session_caps on %p\n", session);
 	while (session->s_nr_caps > 0) {
-		cap = list_entry(session->s_caps.next, struct ceph_inode_cap,
+		cap = list_entry(session->s_caps.next, struct ceph_cap,
 				 session_caps);
 		ci = cap->ci;
 		dout(10, "removing cap %p, ci is %p, inode is %p\n",
@@ -752,11 +752,11 @@ static void remove_session_leases(struct ceph_mds_session *session)
 static void wake_up_session_caps(struct ceph_mds_session *session)
 {
 	struct list_head *p;
-	struct ceph_inode_cap *cap;
+	struct ceph_cap *cap;
 
 	dout(10, "wake_up_session_caps %p mds%d\n", session, session->s_mds);
 	list_for_each(p, &session->s_caps) {
-		cap = list_entry(p, struct ceph_inode_cap, session_caps);
+		cap = list_entry(p, struct ceph_cap, session_caps);
 		dout(20, "waking up waiters on %p\n", &cap->ci->vfs_inode);
 		wake_up(&cap->ci->i_cap_wq);
 	}
@@ -1285,7 +1285,7 @@ static void send_mds_reconnect(struct ceph_mds_client *mdsc, int mds)
 	int newlen, len = 4 + 1;
 	void *p, *end;
 	struct list_head *cp;
-	struct ceph_inode_cap *cap;
+	struct ceph_cap *cap;
 	char *path;
 	int pathlen, err;
 	u64 pathbase;
@@ -1340,7 +1340,7 @@ retry:
 	ceph_encode_32(&p, session->s_nr_caps);
 	count = 0;
 	list_for_each(cp, &session->s_caps) {
-		cap = list_entry(cp, struct ceph_inode_cap, session_caps);
+		cap = list_entry(cp, struct ceph_cap, session_caps);
 		ci = cap->ci;
 		ceph_decode_need(&p, end, sizeof(u64) +
 				 sizeof(struct ceph_mds_cap_reconnect),
diff --git a/src/kernel/mds_client.h b/src/kernel/mds_client.h
index 768ad0aa7ebba..9610f88a4ac9a 100644
--- a/src/kernel/mds_client.h
+++ b/src/kernel/mds_client.h
@@ -12,7 +12,7 @@
 #include "mdsmap.h"
 
 struct ceph_client;
-struct ceph_inode_cap;
+struct ceph_cap;
 
 /*
  * for mds reply parsing
@@ -186,11 +186,6 @@ extern int ceph_mdsc_do_request(struct ceph_mds_client *mdsc,
 				struct ceph_mds_request *req);
 extern void ceph_mdsc_put_request(struct ceph_mds_request *req);
 
-extern int __ceph_mdsc_send_cap(struct ceph_mds_client *mdsc,
-				struct ceph_mds_session *session,
-				struct ceph_inode_cap *cap,
-				int used, int wanted,
-				int flush_snap);
 extern void ceph_mdsc_pre_umount(struct ceph_mds_client *mdsc);
 
 #endif
diff --git a/src/kernel/super.c b/src/kernel/super.c
index 3115c9fdf84ba..5318ff43f15ce 100644
--- a/src/kernel/super.c
+++ b/src/kernel/super.c
@@ -173,6 +173,7 @@ static struct inode *ceph_alloc_inode(struct super_block *sb)
 	for (i = 0; i < CEPH_FILE_MODE_NUM; i++)
 		ci->i_nr_by_mode[i] = 0;
 	init_waitqueue_head(&ci->i_cap_wq);
+	INIT_LIST_HEAD(&ci->i_cap_snaps);
 
 	ci->i_wanted_max_size = 0;
 	ci->i_requested_max_size = 0;
diff --git a/src/kernel/super.h b/src/kernel/super.h
index c8e6c2ba94491..78d83bccf9007 100644
--- a/src/kernel/super.h
+++ b/src/kernel/super.h
@@ -166,10 +166,11 @@ static inline struct ceph_client *ceph_client(struct super_block *sb)
 }
 
 /*
- * CEPH file system in-core inode info
+ * file i/o capability
  */
+#define STATIC_CAPS 1
 
-struct ceph_inode_cap {
+struct ceph_cap {
 	struct ceph_inode_info *ci;
 	struct rb_node ci_node;         /* per-ci cap tree */
 	struct ceph_mds_session *session;
@@ -182,12 +183,23 @@ struct ceph_inode_cap {
 	u64 flushed_snap;
 };
 
-#define MAX_DIRFRAG_REP 4
+/*
+ * snapped cap state, pending flush to mds
+ */
+struct ceph_cap_snap {
+	struct list_head ci_item;
+	int mds;
+	u64 follows;
+	u64 size;
+	struct timespec mtime, atime;
+};
 
 /*
  * a _leaf_ frag will be present in the i_fragtree IFF there is
  * delegation info.  that is, if mds >= 0 || ndist > 0.
  */
+#define MAX_DIRFRAG_REP 4
+
 struct ceph_inode_frag {
 	struct rb_node node;
 
@@ -201,7 +213,6 @@ struct ceph_inode_frag {
 	int dist[MAX_DIRFRAG_REP];
 };
 
-#define STATIC_CAPS 1
 
 struct ceph_dir_info {
 	u64 nfiles, nsubdirs;
@@ -234,19 +245,20 @@ struct ceph_inode_info {
 	char *i_xattr_data;
 
 	struct rb_root i_caps;
-	struct ceph_inode_cap i_static_caps[STATIC_CAPS];
+	struct ceph_cap i_static_caps[STATIC_CAPS];
 	wait_queue_head_t i_cap_wq;
 	unsigned long i_hold_caps_until; /* jiffies */
 	struct list_head i_cap_delay_list;
 	int i_cap_exporting_mds;
 	unsigned i_cap_exporting_mseq;
 	unsigned i_cap_exporting_issued;
-	unsigned i_snap_caps;
+	struct list_head i_cap_snaps;
+	unsigned i_snap_caps;         /* cap bits for snap i/o */
 
 	int i_nr_by_mode[CEPH_FILE_MODE_NUM];
-	loff_t i_max_size;      /* size authorized by mds */
+	loff_t i_max_size;            /* size authorized by mds */
 	loff_t i_reported_size; /* (max_)size reported to or requested of mds */
-	loff_t i_wanted_max_size;  /* offset we'd like to write too */
+	loff_t i_wanted_max_size;     /* offset we'd like to write too */
 	loff_t i_requested_max_size;  /* max_size we've requested */
 	struct timespec i_old_atime;
 
@@ -564,8 +576,8 @@ extern int ceph_add_cap(struct inode *inode,
 			int fmode, unsigned issued,
 			unsigned cap, unsigned seq,
 			void *snapblob, int snapblob_len);
-extern int __ceph_remove_cap(struct ceph_inode_cap *cap);
-extern void ceph_remove_cap(struct ceph_inode_cap *cap);
+extern int __ceph_remove_cap(struct ceph_cap *cap);
+extern void ceph_remove_cap(struct ceph_cap *cap);
 extern void ceph_remove_all_caps(struct ceph_inode_info *ci);
 extern int ceph_get_cap_mds(struct inode *inode);
 extern int ceph_get_cap_refs(struct ceph_inode_info *ci, int need, int want, int *got, loff_t offset);
@@ -577,6 +589,11 @@ extern void ceph_check_delayed_caps(struct ceph_mds_client *mdsc);
 extern void ceph_flush_write_caps(struct ceph_mds_client *mdsc,
 				  struct ceph_mds_session *session,
 				  int purge);
+extern int __ceph_send_cap(struct ceph_mds_client *mdsc,
+			   struct ceph_mds_session *session,
+			   struct ceph_cap *cap,
+			   int used, int wanted,
+			   int flush_snap);
 
 /* addr.c */
 extern const struct address_space_operations ceph_aops;
-- 
2.39.5