From 0cf5537b158caae42bcc03f0f6db10f68585b1ec Mon Sep 17 00:00:00 2001
From: Yehuda Sadeh <yehuda@hq.newdream.net>
Date: Fri, 11 Jun 2010 15:57:06 -0700
Subject: ceph: some endianity fixes

Fix some problems that came up with sparse.

Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/auth_x.c     | 2 +-
 fs/ceph/messenger.c  | 2 +-
 fs/ceph/mon_client.c | 3 ++-
 3 files changed, 4 insertions(+), 3 deletions(-)

(limited to 'fs/ceph')

diff --git a/fs/ceph/auth_x.c b/fs/ceph/auth_x.c
index 83d4d27..3fe4904 100644
--- a/fs/ceph/auth_x.c
+++ b/fs/ceph/auth_x.c
@@ -493,7 +493,7 @@ static int ceph_x_handle_reply(struct ceph_auth_client *ac, int result,
 		return -EAGAIN;
 	}
 
-	op = le32_to_cpu(head->op);
+	op = le16_to_cpu(head->op);
 	result = le32_to_cpu(head->result);
 	dout("handle_reply op %d result %d\n", op, result);
 	switch (op) {
diff --git a/fs/ceph/messenger.c b/fs/ceph/messenger.c
index 64b8b1f..cf1c784 100644
--- a/fs/ceph/messenger.c
+++ b/fs/ceph/messenger.c
@@ -657,7 +657,7 @@ static void prepare_write_connect(struct ceph_messenger *msgr,
 	dout("prepare_write_connect %p cseq=%d gseq=%d proto=%d\n", con,
 	     con->connect_seq, global_seq, proto);
 
-	con->out_connect.features = CEPH_FEATURE_SUPPORTED_CLIENT;
+	con->out_connect.features = cpu_to_le64(CEPH_FEATURE_SUPPORTED_CLIENT);
 	con->out_connect.host_type = cpu_to_le32(CEPH_ENTITY_TYPE_CLIENT);
 	con->out_connect.connect_seq = cpu_to_le32(con->connect_seq);
 	con->out_connect.global_seq = cpu_to_le32(global_seq);
diff --git a/fs/ceph/mon_client.c b/fs/ceph/mon_client.c
index 07a5399..cc115ea 100644
--- a/fs/ceph/mon_client.c
+++ b/fs/ceph/mon_client.c
@@ -725,7 +725,8 @@ static void handle_auth_reply(struct ceph_mon_client *monc,
 		dout("authenticated, starting session\n");
 
 		monc->client->msgr->inst.name.type = CEPH_ENTITY_TYPE_CLIENT;
-		monc->client->msgr->inst.name.num = monc->auth->global_id;
+		monc->client->msgr->inst.name.num =
+					cpu_to_le64(monc->auth->global_id);
 
 		__send_subscribe(monc);
 		__resend_generic_request(monc);
-- 
cgit v1.1


From 4a32f93d29b05cdab63c0e2979bc1524c8ea6bf5 Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Sun, 13 Jun 2010 10:27:53 -0700
Subject: ceph: fix map handler error path

Don't leak message if we receive an unexpected message type.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/osd_client.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

(limited to 'fs/ceph')

diff --git a/fs/ceph/osd_client.c b/fs/ceph/osd_client.c
index d25b4ad..92b7251 100644
--- a/fs/ceph/osd_client.c
+++ b/fs/ceph/osd_client.c
@@ -1344,7 +1344,7 @@ static void dispatch(struct ceph_connection *con, struct ceph_msg *msg)
 	int type = le16_to_cpu(msg->hdr.type);
 
 	if (!osd)
-		return;
+		goto out;
 	osdc = osd->o_osdc;
 
 	switch (type) {
@@ -1359,6 +1359,7 @@ static void dispatch(struct ceph_connection *con, struct ceph_msg *msg)
 		pr_err("received unknown message type %d %s\n", type,
 		       ceph_msg_type_name(type));
 	}
+out:
 	ceph_msg_put(msg);
 }
 
-- 
cgit v1.1


From ae32be31341a5fecfa16c5b3eb78095207182cce Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Sun, 13 Jun 2010 10:30:19 -0700
Subject: ceph: fix message memory leak, uninitialized variable

We need to properly initialize skip, as not all alloc_msg op instances
set it.

Also, BUG if someone says skip but also allocates a message.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/messenger.c | 2 ++
 1 file changed, 2 insertions(+)

(limited to 'fs/ceph')

diff --git a/fs/ceph/messenger.c b/fs/ceph/messenger.c
index cf1c784..9ad43a3 100644
--- a/fs/ceph/messenger.c
+++ b/fs/ceph/messenger.c
@@ -1396,10 +1396,12 @@ static int read_partial_message(struct ceph_connection *con)
 	if (!con->in_msg) {
 		dout("got hdr type %d front %d data %d\n", con->in_hdr.type,
 		     con->in_hdr.front_len, con->in_hdr.data_len);
+		skip = 0;
 		con->in_msg = ceph_alloc_msg(con, &con->in_hdr, &skip);
 		if (skip) {
 			/* skip this message */
 			dout("alloc_msg said skip message\n");
+			BUG_ON(con->in_msg);
 			con->in_base_pos = -front_len - middle_len - data_len -
 				sizeof(m->footer);
 			con->in_tag = CEPH_MSGR_TAG_READY;
-- 
cgit v1.1


From cebc5be6b6c82a99231e9c9af451e9e3d3399ec6 Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Thu, 17 Jun 2010 10:22:48 -0700
Subject: ceph: fix crush map update decoding

If the incremental osdmap has a new crush map, advance the position after
decoding so that we can parse the rest of the osdmap properly.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/osdmap.c | 1 +
 1 file changed, 1 insertion(+)

(limited to 'fs/ceph')

diff --git a/fs/ceph/osdmap.c b/fs/ceph/osdmap.c
index ddc656f..50ce64e 100644
--- a/fs/ceph/osdmap.c
+++ b/fs/ceph/osdmap.c
@@ -707,6 +707,7 @@ struct ceph_osdmap *osdmap_apply_incremental(void **p, void *end,
 		newcrush = crush_decode(*p, min(*p+len, end));
 		if (IS_ERR(newcrush))
 			return ERR_CAST(newcrush);
+		*p += len;
 	}
 
 	/* new flags? */
-- 
cgit v1.1


From d69ed05a80f23b25f06e73af9b7e701ce4900edc Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Mon, 21 Jun 2010 10:38:14 -0700
Subject: ceph: handle splice_dentry/d_materialize_unique error in
 readdir_prepopulate

Handle a splice_dentry failure (due to a d_materialize_unique error)
without crashing.  (Also, report the error code.)

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/inode.c | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)

(limited to 'fs/ceph')

diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
index ab47f46..8f9b9fe 100644
--- a/fs/ceph/inode.c
+++ b/fs/ceph/inode.c
@@ -854,8 +854,8 @@ static struct dentry *splice_dentry(struct dentry *dn, struct inode *in,
 		d_drop(dn);
 	realdn = d_materialise_unique(dn, in);
 	if (IS_ERR(realdn)) {
-		pr_err("splice_dentry error %p inode %p ino %llx.%llx\n",
-		       dn, in, ceph_vinop(in));
+		pr_err("splice_dentry error %ld %p inode %p ino %llx.%llx\n",
+		       PTR_ERR(realdn), dn, in, ceph_vinop(in));
 		if (prehash)
 			*prehash = false; /* don't rehash on error */
 		dn = realdn; /* note realdn contains the error */
@@ -1234,18 +1234,23 @@ retry_lookup:
 				goto out;
 			}
 			dn = splice_dentry(dn, in, NULL);
+			if (IS_ERR(dn))
+				dn = NULL;
 		}
 
 		if (fill_inode(in, &rinfo->dir_in[i], NULL, session,
 			       req->r_request_started, -1,
 			       &req->r_caps_reservation) < 0) {
 			pr_err("fill_inode badness on %p\n", in);
-			dput(dn);
-			continue;
+			goto next_item;
 		}
-		update_dentry_lease(dn, rinfo->dir_dlease[i],
-				    req->r_session, req->r_request_started);
-		dput(dn);
+		if (dn)
+			update_dentry_lease(dn, rinfo->dir_dlease[i],
+					    req->r_session,
+					    req->r_request_started);
+next_item:
+		if (dn)
+			dput(dn);
 	}
 	req->r_did_prepopulate = true;
 
-- 
cgit v1.1


From 17c688c3dfffc274c87be00033da0050bb6eefc0 Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Mon, 21 Jun 2010 16:12:26 -0700
Subject: ceph: delay umount until all mds requests drop inode+dentry refs

This fixes a race between handle_reply finishing an mds request, signalling
completion, and then dropping the request structing and its dentry+inode
refs, and pre_umount function waiting for requests to finish before
letting the vfs tear down the dcache.  If umount was delayed waiting for
mds requests, we could race and BUG in shrink_dcache_for_umount_subtree
because of a slow dput.

This delays umount until the msgr queue flushes, which means handle_reply
will exit and will have dropped the ceph_mds_request struct.  I'm assuming
the VFS has already ensured that its calls have all completed and those
request refs have thus been dropped as well (I haven't seen that race, at
least).

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/mds_client.c | 6 ++++++
 1 file changed, 6 insertions(+)

(limited to 'fs/ceph')

diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
index 1766947..3ab79f6 100644
--- a/fs/ceph/mds_client.c
+++ b/fs/ceph/mds_client.c
@@ -2783,6 +2783,12 @@ void ceph_mdsc_pre_umount(struct ceph_mds_client *mdsc)
 	drop_leases(mdsc);
 	ceph_flush_dirty_caps(mdsc);
 	wait_requests(mdsc);
+
+	/*
+	 * wait for reply handlers to drop their request refs and
+	 * their inode/dcache refs
+	 */
+	ceph_msgr_flush();
 }
 
 /*
-- 
cgit v1.1


From bfaf148eb2e42c00f1c79b2163f0804068ea0c5e Mon Sep 17 00:00:00 2001
From: Yehuda Sadeh <yehuda@hq.newdream.net>
Date: Wed, 23 Jun 2010 15:52:27 -0700
Subject: ceph: fix caps debugfs entry

The ceph client structure was not set correctly.

Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/debugfs.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'fs/ceph')

diff --git a/fs/ceph/debugfs.c b/fs/ceph/debugfs.c
index 3be33fb..f2f5332 100644
--- a/fs/ceph/debugfs.c
+++ b/fs/ceph/debugfs.c
@@ -261,7 +261,7 @@ static int osdc_show(struct seq_file *s, void *pp)
 
 static int caps_show(struct seq_file *s, void *p)
 {
-	struct ceph_client *client = p;
+	struct ceph_client *client = s->private;
 	int total, avail, used, reserved, min;
 
 	ceph_reservation_status(client, &total, &avail, &used, &reserved, &min);
-- 
cgit v1.1


From 55bda7aacd13f5fdfeaafc16934953171405c692 Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Thu, 24 Jun 2010 12:55:48 -0700
Subject: ceph: fix crush recursion

There was a longstanding problem with recursion through intervening
bucket types on complex hierarchies.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/crush/mapper.c | 1 +
 1 file changed, 1 insertion(+)

(limited to 'fs/ceph')

diff --git a/fs/ceph/crush/mapper.c b/fs/ceph/crush/mapper.c
index 9ba54ef..804e6d5 100644
--- a/fs/ceph/crush/mapper.c
+++ b/fs/ceph/crush/mapper.c
@@ -366,6 +366,7 @@ static int crush_choose(struct crush_map *map,
 					BUG_ON(item >= 0 ||
 					       (-1-item) >= map->max_buckets);
 					in = map->buckets[-1-item];
+					retry_bucket = 1;
 					continue;
 				}
 
-- 
cgit v1.1


From a1a31e734241aefcb2b30fb0cc0376977b6d2ba8 Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Thu, 24 Jun 2010 12:58:14 -0700
Subject: ceph: fix crush CHOOSE_LEAF when type is already a leaf

We may not recurse for CHOOSE_LEAF if we start with a leaf node.  When
that happens, the out2 vector needs to be filled in with the result.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/crush/mapper.c | 38 +++++++++++++++++++++++++-------------
 1 file changed, 25 insertions(+), 13 deletions(-)

(limited to 'fs/ceph')

diff --git a/fs/ceph/crush/mapper.c b/fs/ceph/crush/mapper.c
index 804e6d5..374266b 100644
--- a/fs/ceph/crush/mapper.c
+++ b/fs/ceph/crush/mapper.c
@@ -238,7 +238,7 @@ static int bucket_straw_choose(struct crush_bucket_straw *bucket,
 
 static int crush_bucket_choose(struct crush_bucket *in, int x, int r)
 {
-	dprintk("choose %d x=%d r=%d\n", in->id, x, r);
+	dprintk(" crush_bucket_choose %d x=%d r=%d\n", in->id, x, r);
 	switch (in->alg) {
 	case CRUSH_BUCKET_UNIFORM:
 		return bucket_uniform_choose((struct crush_bucket_uniform *)in,
@@ -305,7 +305,9 @@ static int crush_choose(struct crush_map *map,
 	int itemtype;
 	int collide, reject;
 	const int orig_tries = 5; /* attempts before we fall back to search */
-	dprintk("choose bucket %d x %d outpos %d\n", bucket->id, x, outpos);
+
+	dprintk("CHOOSE%s bucket %d x %d outpos %d numrep %d\n", recurse_to_leaf ? "_LEAF" : "",
+		bucket->id, x, outpos, numrep);
 
 	for (rep = outpos; rep < numrep; rep++) {
 		/* keep trying until we get a non-out, non-colliding item */
@@ -378,15 +380,25 @@ static int crush_choose(struct crush_map *map,
 					}
 				}
 
-				if (recurse_to_leaf &&
-				    item < 0 &&
-				    crush_choose(map, map->buckets[-1-item],
-						 weight,
-						 x, outpos+1, 0,
-						 out2, outpos,
-						 firstn, 0, NULL) <= outpos) {
-					reject = 1;
-				} else {
+				reject = 0;
+				if (recurse_to_leaf) {
+					if (item < 0) {
+						if (crush_choose(map,
+							 map->buckets[-1-item],
+							 weight,
+							 x, outpos+1, 0,
+							 out2, outpos,
+							 firstn, 0,
+							 NULL) <= outpos)
+							/* didn't get leaf */
+							reject = 1;
+					} else {
+						/* we already have a leaf! */
+						out2[outpos] = item;
+					}
+				}
+
+				if (!reject) {
 					/* out? */
 					if (itemtype == 0)
 						reject = is_out(map, weight,
@@ -425,12 +437,12 @@ reject:
 			continue;
 		}
 
-		dprintk("choose got %d\n", item);
+		dprintk("CHOOSE got %d\n", item);
 		out[outpos] = item;
 		outpos++;
 	}
 
-	dprintk("choose returns %d\n", outpos);
+	dprintk("CHOOSE returns %d\n", outpos);
 	return outpos;
 }
 
-- 
cgit v1.1


From ec97f88ba6d4256927fde516033ee76d5d85b54a Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Thu, 24 Jun 2010 15:12:37 -0700
Subject: ceph: only release clean, unused caps with mds requests

We can drop caps with an mds request.  Ensure we only drop unused AND
clean caps, since the MDS doesn't support cap writeback in that context,
nor do we track it.  If caps are dirty, and the MDS needs them back, we
it will revoke and we will flush in the normal fashion.

This fixes a possibly loss of metadata.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/caps.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

(limited to 'fs/ceph')

diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
index 619b616..d4fcdda 100644
--- a/fs/ceph/caps.c
+++ b/fs/ceph/caps.c
@@ -2886,18 +2886,19 @@ int ceph_encode_inode_release(void **p, struct inode *inode,
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	struct ceph_cap *cap;
 	struct ceph_mds_request_release *rel = *p;
+	int used, dirty;
 	int ret = 0;
-	int used = 0;
 
 	spin_lock(&inode->i_lock);
 	used = __ceph_caps_used(ci);
+	dirty = __ceph_caps_dirty(ci);
 
-	dout("encode_inode_release %p mds%d used %s drop %s unless %s\n", inode,
-	     mds, ceph_cap_string(used), ceph_cap_string(drop),
+	dout("encode_inode_release %p mds%d used|dirty %s drop %s unless %s\n",
+	     inode, mds, ceph_cap_string(used|dirty), ceph_cap_string(drop),
 	     ceph_cap_string(unless));
 
-	/* only drop unused caps */
-	drop &= ~used;
+	/* only drop unused, clean caps */
+	drop &= ~(used | dirty);
 
 	cap = __get_cap_for_mds(ci, mds);
 	if (cap && __cap_is_valid(cap)) {
-- 
cgit v1.1


From 443b3760a06860187f135c1ecd56c2c7d4ad1022 Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Tue, 29 Jun 2010 09:28:39 -0700
Subject: ceph: fix caps usage accounting for import (non-reserved) case

We need to increase the total and used counters when allocating a new cap
in the non-reserved (cap import) case.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/caps.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

(limited to 'fs/ceph')

diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
index d4fcdda..74144d6 100644
--- a/fs/ceph/caps.c
+++ b/fs/ceph/caps.c
@@ -244,8 +244,14 @@ static struct ceph_cap *get_cap(struct ceph_cap_reservation *ctx)
 	struct ceph_cap *cap = NULL;
 
 	/* temporary, until we do something about cap import/export */
-	if (!ctx)
-		return kmem_cache_alloc(ceph_cap_cachep, GFP_NOFS);
+	if (!ctx) {
+		cap = kmem_cache_alloc(ceph_cap_cachep, GFP_NOFS);
+		if (cap) {
+			caps_use_count++;
+			caps_total_count++;
+		}
+		return cap;
+	}
 
 	spin_lock(&caps_list_lock);
 	dout("get_cap ctx=%p (%d) %d = %d used + %d resv + %d avail\n",
-- 
cgit v1.1


From 153a10939ea6e42e9c0115b0645060d0d7bb4697 Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Mon, 5 Jul 2010 09:44:17 -0700
Subject: ceph: fix crush device 'out' threshold to 1.0, not 0.1

Fix a typo that made any OSD weighted between 0.1 and 1.0 effectively
weighted as 1.0 (fully in).

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/crush/mapper.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'fs/ceph')

diff --git a/fs/ceph/crush/mapper.c b/fs/ceph/crush/mapper.c
index 374266b..a4eec13 100644
--- a/fs/ceph/crush/mapper.c
+++ b/fs/ceph/crush/mapper.c
@@ -264,7 +264,7 @@ static int crush_bucket_choose(struct crush_bucket *in, int x, int r)
  */
 static int is_out(struct crush_map *map, __u32 *weight, int item, int x)
 {
-	if (weight[item] >= 0x1000)
+	if (weight[item] >= 0x10000)
 		return 0;
 	if (weight[item] == 0)
 		return 1;
-- 
cgit v1.1


From ed98adad3d87594c55347824e85137d1829c9e70 Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Mon, 5 Jul 2010 12:15:14 -0700
Subject: ceph: fix message revocation

A message can be on a queue (pending or sent), or out_msg (sending), or
both.  We were assuming that if it's not on a queue it couldn't be out_msg,
but that was false in the case of lossy connections like the OSD.  Fix
ceph_con_revoke() to treat these cases independently.  Also, fix the
out_kvec_is_message check to only trigger if we are currently sending
_this_ message.

This fixes a GPF in tcp_sendpage, triggered by OSD restarts.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/messenger.c | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

(limited to 'fs/ceph')

diff --git a/fs/ceph/messenger.c b/fs/ceph/messenger.c
index 9ad43a3..9692d08 100644
--- a/fs/ceph/messenger.c
+++ b/fs/ceph/messenger.c
@@ -2015,20 +2015,20 @@ void ceph_con_revoke(struct ceph_connection *con, struct ceph_msg *msg)
 {
 	mutex_lock(&con->mutex);
 	if (!list_empty(&msg->list_head)) {
-		dout("con_revoke %p msg %p\n", con, msg);
+		dout("con_revoke %p msg %p - was on queue\n", con, msg);
 		list_del_init(&msg->list_head);
 		ceph_msg_put(msg);
 		msg->hdr.seq = 0;
-		if (con->out_msg == msg) {
-			ceph_msg_put(con->out_msg);
-			con->out_msg = NULL;
-		}
+	}
+	if (con->out_msg == msg) {
+		dout("con_revoke %p msg %p - was sending\n", con, msg);
+		con->out_msg = NULL;
 		if (con->out_kvec_is_msg) {
 			con->out_skip = con->out_kvec_bytes;
 			con->out_kvec_is_msg = false;
 		}
-	} else {
-		dout("con_revoke %p msg %p - not queued (sent?)\n", con, msg);
+		ceph_msg_put(msg);
+		msg->hdr.seq = 0;
 	}
 	mutex_unlock(&con->mutex);
 }
-- 
cgit v1.1


From 22b1de06c9fe128ca3de72560c3e8c2cabf2927a Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Mon, 5 Jul 2010 15:36:49 -0700
Subject: ceph: fix leak of mon authorizer

Fix leak of a struct ceph_buffer on umount.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/auth_x.c | 3 +++
 1 file changed, 3 insertions(+)

(limited to 'fs/ceph')

diff --git a/fs/ceph/auth_x.c b/fs/ceph/auth_x.c
index 3fe4904..6d44053 100644
--- a/fs/ceph/auth_x.c
+++ b/fs/ceph/auth_x.c
@@ -613,6 +613,9 @@ static void ceph_x_destroy(struct ceph_auth_client *ac)
 		remove_ticket_handler(ac, th);
 	}
 
+	if (xi->auth_authorizer.buf)
+		ceph_buffer_put(xi->auth_authorizer.buf);
+
 	kfree(ac->private);
 	ac->private = NULL;
 }
-- 
cgit v1.1


From b0bbb0be8f7fbf6d366b359e034c78a96c4e274d Mon Sep 17 00:00:00 2001
From: Dan Carpenter <error27@gmail.com>
Date: Thu, 8 Jul 2010 14:49:38 +0200
Subject: ceph: add kfree() to error path

We leak a "pi" on this error path.

Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/osdmap.c | 1 +
 1 file changed, 1 insertion(+)

(limited to 'fs/ceph')

diff --git a/fs/ceph/osdmap.c b/fs/ceph/osdmap.c
index 50ce64e..277f8b3 100644
--- a/fs/ceph/osdmap.c
+++ b/fs/ceph/osdmap.c
@@ -568,6 +568,7 @@ struct ceph_osdmap *osdmap_decode(void **p, void *end)
 		if (ev > CEPH_PG_POOL_VERSION) {
 			pr_warning("got unknown v %d > %d of ceph_pg_pool\n",
 				   ev, CEPH_PG_POOL_VERSION);
+			kfree(pi);
 			goto bad;
 		}
 		__decode_pool(p, pi);
-- 
cgit v1.1


From d06dbaf6c2c7187938f3f6745d9e4938a2d0ec47 Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Thu, 8 Jul 2010 10:47:16 -0700
Subject: ceph: fix printing of ipv6 addrs

The buffer was too small.  Make it bigger, use snprintf(), put brackets
around the ipv6 address to avoid mixing it up with the :port, and use the
ever-so-handy %pI[46] formats.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/messenger.c | 24 ++++++------------------
 1 file changed, 6 insertions(+), 18 deletions(-)

(limited to 'fs/ceph')

diff --git a/fs/ceph/messenger.c b/fs/ceph/messenger.c
index 9692d08..e8c5a2d 100644
--- a/fs/ceph/messenger.c
+++ b/fs/ceph/messenger.c
@@ -43,7 +43,8 @@ static void ceph_fault(struct ceph_connection *con);
  * nicely render a sockaddr as a string.
  */
 #define MAX_ADDR_STR 20
-static char addr_str[MAX_ADDR_STR][40];
+#define MAX_ADDR_STR_LEN 60
+static char addr_str[MAX_ADDR_STR][MAX_ADDR_STR_LEN];
 static DEFINE_SPINLOCK(addr_str_lock);
 static int last_addr_str;
 
@@ -52,7 +53,6 @@ const char *pr_addr(const struct sockaddr_storage *ss)
 	int i;
 	char *s;
 	struct sockaddr_in *in4 = (void *)ss;
-	unsigned char *quad = (void *)&in4->sin_addr.s_addr;
 	struct sockaddr_in6 *in6 = (void *)ss;
 
 	spin_lock(&addr_str_lock);
@@ -64,25 +64,13 @@ const char *pr_addr(const struct sockaddr_storage *ss)
 
 	switch (ss->ss_family) {
 	case AF_INET:
-		sprintf(s, "%u.%u.%u.%u:%u",
-			(unsigned int)quad[0],
-			(unsigned int)quad[1],
-			(unsigned int)quad[2],
-			(unsigned int)quad[3],
-			(unsigned int)ntohs(in4->sin_port));
+		snprintf(s, MAX_ADDR_STR_LEN, "%pI4:%u", &in4->sin_addr,
+			 (unsigned int)ntohs(in4->sin_port));
 		break;
 
 	case AF_INET6:
-		sprintf(s, "%04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x:%u",
-			in6->sin6_addr.s6_addr16[0],
-			in6->sin6_addr.s6_addr16[1],
-			in6->sin6_addr.s6_addr16[2],
-			in6->sin6_addr.s6_addr16[3],
-			in6->sin6_addr.s6_addr16[4],
-			in6->sin6_addr.s6_addr16[5],
-			in6->sin6_addr.s6_addr16[6],
-			in6->sin6_addr.s6_addr16[7],
-			(unsigned int)ntohs(in6->sin6_port));
+		snprintf(s, MAX_ADDR_STR_LEN, "[%pI6c]:%u", &in6->sin6_addr,
+			 (unsigned int)ntohs(in6->sin6_port));
 		break;
 
 	default:
-- 
cgit v1.1


From 39139f64e14684cf2370770deb79d929d27cfd9b Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Thu, 8 Jul 2010 09:54:52 -0700
Subject: ceph: fix parsing of ipv6 addresses

Check for brackets around the ipv6 address to avoid ambiguity with the port
number.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/messenger.c | 25 +++++++++++++++++++------
 1 file changed, 19 insertions(+), 6 deletions(-)

(limited to 'fs/ceph')

diff --git a/fs/ceph/messenger.c b/fs/ceph/messenger.c
index e8c5a2d..3ddef15 100644
--- a/fs/ceph/messenger.c
+++ b/fs/ceph/messenger.c
@@ -997,19 +997,32 @@ int ceph_parse_ips(const char *c, const char *end,
 		struct sockaddr_in *in4 = (void *)ss;
 		struct sockaddr_in6 *in6 = (void *)ss;
 		int port;
+		char delim = ',';
+
+		if (*p == '[') {
+			delim = ']';
+			p++;
+		}
 
 		memset(ss, 0, sizeof(*ss));
 		if (in4_pton(p, end - p, (u8 *)&in4->sin_addr.s_addr,
-			     ',', &ipend)) {
+			     delim, &ipend))
 			ss->ss_family = AF_INET;
-		} else if (in6_pton(p, end - p, (u8 *)&in6->sin6_addr.s6_addr,
-				    ',', &ipend)) {
+		else if (in6_pton(p, end - p, (u8 *)&in6->sin6_addr.s6_addr,
+				  delim, &ipend))
 			ss->ss_family = AF_INET6;
-		} else {
+		else
 			goto bad;
-		}
 		p = ipend;
 
+		if (delim == ']') {
+			if (*p != ']') {
+				dout("missing matching ']'\n");
+				goto bad;
+			}
+			p++;
+		}
+
 		/* port? */
 		if (p < end && *p == ':') {
 			port = 0;
@@ -1043,7 +1056,7 @@ int ceph_parse_ips(const char *c, const char *end,
 	return 0;
 
 bad:
-	pr_err("parse_ips bad ip '%s'\n", c);
+	pr_err("parse_ips bad ip '%.*s'\n", (int)(end - c), c);
 	return -EINVAL;
 }
 
-- 
cgit v1.1


From f91d3471ccf1ca9a795f46c94b1ded8dd219940c Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Thu, 1 Jul 2010 15:18:31 -0700
Subject: ceph: fix creation of ipv6 sockets

Use the address family from the peer address instead of assuming IPv4.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/messenger.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

(limited to 'fs/ceph')

diff --git a/fs/ceph/messenger.c b/fs/ceph/messenger.c
index 3ddef15..15167b2 100644
--- a/fs/ceph/messenger.c
+++ b/fs/ceph/messenger.c
@@ -203,12 +203,13 @@ static void set_sock_callbacks(struct socket *sock,
  */
 static struct socket *ceph_tcp_connect(struct ceph_connection *con)
 {
-	struct sockaddr *paddr = (struct sockaddr *)&con->peer_addr.in_addr;
+	struct sockaddr_storage *paddr = &con->peer_addr.in_addr;
 	struct socket *sock;
 	int ret;
 
 	BUG_ON(con->sock);
-	ret = sock_create_kern(AF_INET, SOCK_STREAM, IPPROTO_TCP, &sock);
+	ret = sock_create_kern(con->peer_addr.in_addr.ss_family, SOCK_STREAM,
+			       IPPROTO_TCP, &sock);
 	if (ret)
 		return ERR_PTR(ret);
 	con->sock = sock;
@@ -222,7 +223,8 @@ static struct socket *ceph_tcp_connect(struct ceph_connection *con)
 
 	dout("connect %s\n", pr_addr(&con->peer_addr.in_addr));
 
-	ret = sock->ops->connect(sock, paddr, sizeof(*paddr), O_NONBLOCK);
+	ret = sock->ops->connect(sock, (struct sockaddr *)paddr, sizeof(*paddr),
+				 O_NONBLOCK);
 	if (ret == -EINPROGRESS) {
 		dout("connect %s EINPROGRESS sk_state = %u\n",
 		     pr_addr(&con->peer_addr.in_addr),
-- 
cgit v1.1


From 01a92f174f8a3b99dbb5e02c86e7ee1e576737af Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Thu, 15 Jul 2010 13:24:32 -0700
Subject: ceph: reuse request message when replaying against recovering mds

Replayed rename operations (after an mds failure/recovery) were broken
because the request paths were regenerated from the dentry names, which
get mangled when d_move() is called.

Instead, resend the previous request message when replaying completed
operations.  Just make sure the REPLAY flag is set and the target ino is
filled in.

This fixes problems with workloads doing renames when the MDS restarts,
where the rename operation appears to succeed, but on mds restart then
fails (leading to client confusion, app breakage, etc.).

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/mds_client.c | 27 ++++++++++++++++++++++-----
 1 file changed, 22 insertions(+), 5 deletions(-)

(limited to 'fs/ceph')

diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
index 3ab79f6..23332bc 100644
--- a/fs/ceph/mds_client.c
+++ b/fs/ceph/mds_client.c
@@ -1580,6 +1580,27 @@ static int __prepare_send_request(struct ceph_mds_client *mdsc,
 	dout("prepare_send_request %p tid %lld %s (attempt %d)\n", req,
 	     req->r_tid, ceph_mds_op_name(req->r_op), req->r_attempts);
 
+	if (req->r_got_unsafe) {
+		/*
+		 * Replay.  Do not regenerate message (and rebuild
+		 * paths, etc.); just use the original message.
+		 * Rebuilding paths will break for renames because
+		 * d_move mangles the src name.
+		 */
+		msg = req->r_request;
+		rhead = msg->front.iov_base;
+
+		flags = le32_to_cpu(rhead->flags);
+		flags |= CEPH_MDS_FLAG_REPLAY;
+		rhead->flags = cpu_to_le32(flags);
+
+		if (req->r_target_inode)
+			rhead->ino = cpu_to_le64(ceph_ino(req->r_target_inode));
+
+		rhead->num_retry = req->r_attempts - 1;
+		return 0;
+	}
+
 	if (req->r_request) {
 		ceph_msg_put(req->r_request);
 		req->r_request = NULL;
@@ -1601,13 +1622,9 @@ static int __prepare_send_request(struct ceph_mds_client *mdsc,
 	rhead->flags = cpu_to_le32(flags);
 	rhead->num_fwd = req->r_num_fwd;
 	rhead->num_retry = req->r_attempts - 1;
+	rhead->ino = 0;
 
 	dout(" r_locked_dir = %p\n", req->r_locked_dir);
-
-	if (req->r_target_inode && req->r_got_unsafe)
-		rhead->ino = cpu_to_le64(ceph_ino(req->r_target_inode));
-	else
-		rhead->ino = 0;
 	return 0;
 }
 
-- 
cgit v1.1


From e979cf50395e24c4bdd489f60e2d5dd5ae66d255 Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Thu, 15 Jul 2010 14:58:39 -0700
Subject: ceph: do not include cap/dentry releases in replayed messages

Strip the cap and dentry releases from replayed messages.  They can
cause the shared state to get out of sync because they were generated
(with the request message) earlier, and no longer reflect the current
client state.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/mds_client.c | 8 ++++++++
 fs/ceph/mds_client.h | 1 +
 2 files changed, 9 insertions(+)

(limited to 'fs/ceph')

diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
index 23332bc..416c08d 100644
--- a/fs/ceph/mds_client.c
+++ b/fs/ceph/mds_client.c
@@ -1514,6 +1514,9 @@ static struct ceph_msg *create_request_message(struct ceph_mds_client *mdsc,
 	ceph_encode_filepath(&p, end, ino1, path1);
 	ceph_encode_filepath(&p, end, ino2, path2);
 
+	/* make note of release offset, in case we need to replay */
+	req->r_request_release_offset = p - msg->front.iov_base;
+
 	/* cap releases */
 	releases = 0;
 	if (req->r_inode_drop)
@@ -1598,6 +1601,11 @@ static int __prepare_send_request(struct ceph_mds_client *mdsc,
 			rhead->ino = cpu_to_le64(ceph_ino(req->r_target_inode));
 
 		rhead->num_retry = req->r_attempts - 1;
+
+		/* remove cap/dentry releases from message */
+		rhead->num_releases = 0;
+		msg->hdr.front_len = cpu_to_le32(req->r_request_release_offset);
+		msg->front.iov_len = req->r_request_release_offset;
 		return 0;
 	}
 
diff --git a/fs/ceph/mds_client.h b/fs/ceph/mds_client.h
index b292fa4..952410c 100644
--- a/fs/ceph/mds_client.h
+++ b/fs/ceph/mds_client.h
@@ -188,6 +188,7 @@ struct ceph_mds_request {
 	int r_old_inode_drop, r_old_inode_unless;
 
 	struct ceph_msg  *r_request;  /* original request */
+	int r_request_release_offset;
 	struct ceph_msg  *r_reply;
 	struct ceph_mds_reply_info_parsed r_reply_info;
 	int r_err;
-- 
cgit v1.1


From a0dff78dab0ff8d78bd5c9e33c105cf1292f2282 Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Thu, 22 Jul 2010 13:47:21 -0700
Subject: ceph: avoid dcache readdir for snapdir

We should always go to the MDS for readdir on the hidden snapdir.  The
set of snapshots can change at any time; the client can't trust its cache
for that.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/dir.c | 1 +
 1 file changed, 1 insertion(+)

(limited to 'fs/ceph')

diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c
index f857193..ea36ba9 100644
--- a/fs/ceph/dir.c
+++ b/fs/ceph/dir.c
@@ -266,6 +266,7 @@ static int ceph_readdir(struct file *filp, void *dirent, filldir_t filldir)
 	spin_lock(&inode->i_lock);
 	if ((filp->f_pos == 2 || fi->dentry) &&
 	    !ceph_test_opt(client, NOASYNCREADDIR) &&
+	    ceph_snap(inode) != CEPH_SNAPDIR &&
 	    (ci->i_ceph_flags & CEPH_I_COMPLETE) &&
 	    __ceph_caps_issued_mask(ci, CEPH_CAP_FILE_SHARED, 1)) {
 		err = __dcache_readdir(filp, dirent, filldir);
-- 
cgit v1.1


From 252af5214682191e34e57204e1a31924fb82c207 Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Thu, 22 Jul 2010 13:49:08 -0700
Subject: ceph: fix d_release dop for snapdir, snapped dentries

We need to set the d_release dop for snapdir and snapped dentries so that
the ceph_dentry_info struct gets released.  We also use the dcache to
cache readdir results when possible, which only works if we know when
dentries are dropped from the cache.  Since we don't use the dcache for
readdir in the hidden snapdir, avoid that case in ceph_dentry_release.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/dir.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)

(limited to 'fs/ceph')

diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c
index ea36ba9..f94ed3c 100644
--- a/fs/ceph/dir.c
+++ b/fs/ceph/dir.c
@@ -1014,18 +1014,22 @@ out_touch:
 
 /*
  * When a dentry is released, clear the dir I_COMPLETE if it was part
- * of the current dir gen.
+ * of the current dir gen or if this is in the snapshot namespace.
  */
 static void ceph_dentry_release(struct dentry *dentry)
 {
 	struct ceph_dentry_info *di = ceph_dentry(dentry);
 	struct inode *parent_inode = dentry->d_parent->d_inode;
+	u64 snapid = ceph_snap(parent_inode);
 
-	if (parent_inode) {
+	dout("dentry_release %p parent %p\n", dentry, parent_inode);
+
+	if (parent_inode && snapid != CEPH_SNAPDIR) {
 		struct ceph_inode_info *ci = ceph_inode(parent_inode);
 
 		spin_lock(&parent_inode->i_lock);
-		if (ci->i_shared_gen == di->lease_shared_gen) {
+		if (ci->i_shared_gen == di->lease_shared_gen ||
+		    snapid <= CEPH_MAXSNAP) {
 			dout(" clearing %p complete (d_release)\n",
 			     parent_inode);
 			ci->i_ceph_flags &= ~CEPH_I_COMPLETE;
@@ -1242,7 +1246,9 @@ struct dentry_operations ceph_dentry_ops = {
 
 struct dentry_operations ceph_snapdir_dentry_ops = {
 	.d_revalidate = ceph_snapdir_d_revalidate,
+	.d_release = ceph_dentry_release,
 };
 
 struct dentry_operations ceph_snap_dentry_ops = {
+	.d_release = ceph_dentry_release,
 };
-- 
cgit v1.1


From bc4fdca85734d12cd2c7a25c52323ef6e6e5adef Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Tue, 20 Jul 2010 16:19:56 -0700
Subject: ceph: fix pg_mapping leak on pg_temp updates

Free the ceph_pg_mapping structs when they are removed from the pg_temp
rbtree.  Also fix a leak in the __insert_pg_mapping() error path.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/osdmap.c | 26 +++++++++++++++-----------
 1 file changed, 15 insertions(+), 11 deletions(-)

(limited to 'fs/ceph')

diff --git a/fs/ceph/osdmap.c b/fs/ceph/osdmap.c
index 277f8b3..416d46a 100644
--- a/fs/ceph/osdmap.c
+++ b/fs/ceph/osdmap.c
@@ -831,12 +831,13 @@ struct ceph_osdmap *osdmap_apply_incremental(void **p, void *end,
 		/* remove any? */
 		while (rbp && pgid_cmp(rb_entry(rbp, struct ceph_pg_mapping,
 						node)->pgid, pgid) <= 0) {
-			struct rb_node *cur = rbp;
+			struct ceph_pg_mapping *cur =
+				rb_entry(rbp, struct ceph_pg_mapping, node);
+			
 			rbp = rb_next(rbp);
-			dout(" removed pg_temp %llx\n",
-			     *(u64 *)&rb_entry(cur, struct ceph_pg_mapping,
-					       node)->pgid);
-			rb_erase(cur, &map->pg_temp);
+			dout(" removed pg_temp %llx\n", *(u64 *)&cur->pgid);
+			rb_erase(&cur->node, &map->pg_temp);
+			kfree(cur);
 		}
 
 		if (pglen) {
@@ -852,19 +853,22 @@ struct ceph_osdmap *osdmap_apply_incremental(void **p, void *end,
 			for (j = 0; j < pglen; j++)
 				pg->osds[j] = ceph_decode_32(p);
 			err = __insert_pg_mapping(pg, &map->pg_temp);
-			if (err)
+			if (err) {
+				kfree(pg);
 				goto bad;
+			}
 			dout(" added pg_temp %llx len %d\n", *(u64 *)&pgid,
 			     pglen);
 		}
 	}
 	while (rbp) {
-		struct rb_node *cur = rbp;
+		struct ceph_pg_mapping *cur =
+			rb_entry(rbp, struct ceph_pg_mapping, node);
+
 		rbp = rb_next(rbp);
-		dout(" removed pg_temp %llx\n",
-		     *(u64 *)&rb_entry(cur, struct ceph_pg_mapping,
-				       node)->pgid);
-		rb_erase(cur, &map->pg_temp);
+		dout(" removed pg_temp %llx\n", *(u64 *)&cur->pgid);
+		rb_erase(&cur->node, &map->pg_temp);
+		kfree(cur);
 	}
 
 	/* ignore the rest */
-- 
cgit v1.1


From 8c696737aa61316a252c4514d09dd163f1464d33 Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Thu, 22 Jul 2010 14:11:56 -0700
Subject: ceph: fix leak of dentry in ceph_init_dentry() error path

If we fail to allocate a ceph_dentry_info, don't leak the dn reference.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/inode.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

(limited to 'fs/ceph')

diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
index 8f9b9fe..3582e79 100644
--- a/fs/ceph/inode.c
+++ b/fs/ceph/inode.c
@@ -1199,8 +1199,10 @@ retry_lookup:
 				goto out;
 			}
 			err = ceph_init_dentry(dn);
-			if (err < 0)
+			if (err < 0) {
+				dput(dn);
 				goto out;
+			}
 		} else if (dn->d_inode &&
 			   (ceph_ino(dn->d_inode) != vino.ino ||
 			    ceph_snap(dn->d_inode) != vino.snap)) {
-- 
cgit v1.1


From 1dadcce358a4c4078e1ea0bc4365c3f67b8e373e Mon Sep 17 00:00:00 2001
From: Sage Weil <sage@newdream.net>
Date: Fri, 23 Jul 2010 13:54:21 -0700
Subject: ceph: fix dentry lease release

When we embed a dentry lease release notification in a request, invalidate
our lease so we don't think we still have it.  Otherwise we can get all
sorts of incorrect client behavior when multiple clients are interacting
with the same part of the namespace.

Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/caps.c | 1 +
 1 file changed, 1 insertion(+)

(limited to 'fs/ceph')

diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
index 74144d6..6afc1af 100644
--- a/fs/ceph/caps.c
+++ b/fs/ceph/caps.c
@@ -2984,6 +2984,7 @@ int ceph_encode_dentry_release(void **p, struct dentry *dentry,
 		memcpy(*p, dentry->d_name.name, dentry->d_name.len);
 		*p += dentry->d_name.len;
 		rel->dname_seq = cpu_to_le32(di->lease_seq);
+		__ceph_mdsc_drop_dentry_lease(dentry);
 	}
 	spin_unlock(&dentry->d_lock);
 	return ret;
-- 
cgit v1.1


From 25848b3ec681c7018e3746dd850c1e8ed0a3dd6b Mon Sep 17 00:00:00 2001
From: "Robert P. J. Day" <rpjday@crashcourse.ca>
Date: Sat, 24 Jul 2010 06:41:18 -0400
Subject: ceph: Correct obvious typo of Kconfig variable "CRYPTO_AES"

Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'fs/ceph')

diff --git a/fs/ceph/Kconfig b/fs/ceph/Kconfig
index 04b8280..bc87b9c 100644
--- a/fs/ceph/Kconfig
+++ b/fs/ceph/Kconfig
@@ -2,7 +2,7 @@ config CEPH_FS
         tristate "Ceph distributed file system (EXPERIMENTAL)"
 	depends on INET && EXPERIMENTAL
 	select LIBCRC32C
-	select CONFIG_CRYPTO_AES
+	select CRYPTO_AES
 	help
 	  Choose Y or M here to include support for mounting the
 	  experimental Ceph distributed file system.  Ceph is an extremely
-- 
cgit v1.1


From 03066f23452ff088ad8e2c8acdf4443043f35b51 Mon Sep 17 00:00:00 2001
From: Yehuda Sadeh <yehuda@hq.newdream.net>
Date: Tue, 27 Jul 2010 13:11:08 -0700
Subject: ceph: use complete_all and wake_up_all

This fixes an issue triggered by running concurrent syncs. One of the syncs
would go through while the other would just hang indefinitely. In any case, we
never actually want to wake a single waiter, so the *_all functions should
be used.

Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
---
 fs/ceph/caps.c       | 14 +++++++-------
 fs/ceph/file.c       |  2 +-
 fs/ceph/inode.c      |  2 +-
 fs/ceph/mds_client.c | 10 +++++-----
 fs/ceph/mon_client.c |  6 +++---
 fs/ceph/osd_client.c |  6 +++---
 6 files changed, 20 insertions(+), 20 deletions(-)

(limited to 'fs/ceph')

diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
index 6afc1af..b81be9a 100644
--- a/fs/ceph/caps.c
+++ b/fs/ceph/caps.c
@@ -627,7 +627,7 @@ retry:
 	if (fmode >= 0)
 		__ceph_get_fmode(ci, fmode);
 	spin_unlock(&inode->i_lock);
-	wake_up(&ci->i_cap_wq);
+	wake_up_all(&ci->i_cap_wq);
 	return 0;
 }
 
@@ -1181,7 +1181,7 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap,
 	}
 
 	if (wake)
-		wake_up(&ci->i_cap_wq);
+		wake_up_all(&ci->i_cap_wq);
 
 	return delayed;
 }
@@ -2153,7 +2153,7 @@ void ceph_put_cap_refs(struct ceph_inode_info *ci, int had)
 	else if (flushsnaps)
 		ceph_flush_snaps(ci);
 	if (wake)
-		wake_up(&ci->i_cap_wq);
+		wake_up_all(&ci->i_cap_wq);
 	if (put)
 		iput(inode);
 }
@@ -2229,7 +2229,7 @@ void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
 		iput(inode);
 	} else if (complete_capsnap) {
 		ceph_flush_snaps(ci);
-		wake_up(&ci->i_cap_wq);
+		wake_up_all(&ci->i_cap_wq);
 	}
 	if (drop_capsnap)
 		iput(inode);
@@ -2405,7 +2405,7 @@ static void handle_cap_grant(struct inode *inode, struct ceph_mds_caps *grant,
 	if (queue_invalidate)
 		ceph_queue_invalidate(inode);
 	if (wake)
-		wake_up(&ci->i_cap_wq);
+		wake_up_all(&ci->i_cap_wq);
 
 	if (check_caps == 1)
 		ceph_check_caps(ci, CHECK_CAPS_NODELAY|CHECK_CAPS_AUTHONLY,
@@ -2460,7 +2460,7 @@ static void handle_cap_flush_ack(struct inode *inode, u64 flush_tid,
 					 struct ceph_inode_info,
 					 i_flushing_item)->vfs_inode);
 		mdsc->num_cap_flushing--;
-		wake_up(&mdsc->cap_flushing_wq);
+		wake_up_all(&mdsc->cap_flushing_wq);
 		dout(" inode %p now !flushing\n", inode);
 
 		if (ci->i_dirty_caps == 0) {
@@ -2472,7 +2472,7 @@ static void handle_cap_flush_ack(struct inode *inode, u64 flush_tid,
 		}
 	}
 	spin_unlock(&mdsc->cap_dirty_lock);
-	wake_up(&ci->i_cap_wq);
+	wake_up_all(&ci->i_cap_wq);
 
 out:
 	spin_unlock(&inode->i_lock);
diff --git a/fs/ceph/file.c b/fs/ceph/file.c
index 6251a15..7c08698 100644
--- a/fs/ceph/file.c
+++ b/fs/ceph/file.c
@@ -265,7 +265,7 @@ int ceph_release(struct inode *inode, struct file *file)
 	kmem_cache_free(ceph_file_cachep, cf);
 
 	/* wake up anyone waiting for caps on this inode */
-	wake_up(&ci->i_cap_wq);
+	wake_up_all(&ci->i_cap_wq);
 	return 0;
 }
 
diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
index 3582e79..389f9db 100644
--- a/fs/ceph/inode.c
+++ b/fs/ceph/inode.c
@@ -1501,7 +1501,7 @@ retry:
 	if (wrbuffer_refs == 0)
 		ceph_check_caps(ci, CHECK_CAPS_AUTHONLY, NULL);
 	if (wake)
-		wake_up(&ci->i_cap_wq);
+		wake_up_all(&ci->i_cap_wq);
 }
 
 
diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
index 416c08d..dd440bd 100644
--- a/fs/ceph/mds_client.c
+++ b/fs/ceph/mds_client.c
@@ -868,7 +868,7 @@ static int wake_up_session_cb(struct inode *inode, struct ceph_cap *cap,
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
 
-	wake_up(&ci->i_cap_wq);
+	wake_up_all(&ci->i_cap_wq);
 	if (arg) {
 		spin_lock(&inode->i_lock);
 		ci->i_wanted_max_size = 0;
@@ -1564,7 +1564,7 @@ static void complete_request(struct ceph_mds_client *mdsc,
 	if (req->r_callback)
 		req->r_callback(mdsc, req);
 	else
-		complete(&req->r_completion);
+		complete_all(&req->r_completion);
 }
 
 /*
@@ -1932,7 +1932,7 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg)
 	if (head->safe) {
 		req->r_got_safe = true;
 		__unregister_request(mdsc, req);
-		complete(&req->r_safe_completion);
+		complete_all(&req->r_safe_completion);
 
 		if (req->r_got_unsafe) {
 			/*
@@ -1947,7 +1947,7 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg)
 
 			/* last unsafe request during umount? */
 			if (mdsc->stopping && !__get_oldest_req(mdsc))
-				complete(&mdsc->safe_umount_waiters);
+				complete_all(&mdsc->safe_umount_waiters);
 			mutex_unlock(&mdsc->mutex);
 			goto out;
 		}
@@ -2126,7 +2126,7 @@ static void handle_session(struct ceph_mds_session *session,
 			pr_info("mds%d reconnect denied\n", session->s_mds);
 		remove_session_caps(session);
 		wake = 1; /* for good measure */
-		complete(&mdsc->session_close_waiters);
+		complete_all(&mdsc->session_close_waiters);
 		kick_requests(mdsc, mds);
 		break;
 
diff --git a/fs/ceph/mon_client.c b/fs/ceph/mon_client.c
index cc115ea..54fe01c 100644
--- a/fs/ceph/mon_client.c
+++ b/fs/ceph/mon_client.c
@@ -345,7 +345,7 @@ static void ceph_monc_handle_map(struct ceph_mon_client *monc,
 
 out:
 	mutex_unlock(&monc->mutex);
-	wake_up(&client->auth_wq);
+	wake_up_all(&client->auth_wq);
 }
 
 /*
@@ -462,7 +462,7 @@ static void handle_statfs_reply(struct ceph_mon_client *monc,
 	}
 	mutex_unlock(&monc->mutex);
 	if (req) {
-		complete(&req->completion);
+		complete_all(&req->completion);
 		put_generic_request(req);
 	}
 	return;
@@ -718,7 +718,7 @@ static void handle_auth_reply(struct ceph_mon_client *monc,
 				     monc->m_auth->front_max);
 	if (ret < 0) {
 		monc->client->auth_err = ret;
-		wake_up(&monc->client->auth_wq);
+		wake_up_all(&monc->client->auth_wq);
 	} else if (ret > 0) {
 		__send_prepared_auth_request(monc, ret);
 	} else if (!was_auth && monc->auth->ops->is_authenticated(monc->auth)) {
diff --git a/fs/ceph/osd_client.c b/fs/ceph/osd_client.c
index 92b7251..e385223 100644
--- a/fs/ceph/osd_client.c
+++ b/fs/ceph/osd_client.c
@@ -862,12 +862,12 @@ static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
 	if (req->r_callback)
 		req->r_callback(req, msg);
 	else
-		complete(&req->r_completion);
+		complete_all(&req->r_completion);
 
 	if (flags & CEPH_OSD_FLAG_ONDISK) {
 		if (req->r_safe_callback)
 			req->r_safe_callback(req, msg);
-		complete(&req->r_safe_completion);  /* fsync waiter */
+		complete_all(&req->r_safe_completion);  /* fsync waiter */
 	}
 
 done:
@@ -1083,7 +1083,7 @@ done:
 	if (newmap)
 		kick_requests(osdc, NULL);
 	up_read(&osdc->map_sem);
-	wake_up(&osdc->client->auth_wq);
+	wake_up_all(&osdc->client->auth_wq);
 	return;
 
 bad:
-- 
cgit v1.1