ceph: negotiate authentication protocol; implement AUTH_NONE protocol

When we open a monitor session, we send an initial AUTH message listing
the auth protocols we support, our entity name, and (possibly) a previously
assigned global_id.  The monitor chooses a protocol and responds with an
initial message.

Initially implement AUTH_NONE, a dummy protocol that provides no security,
but works within the new framework.  It generates 'authorizers' that are
used when connecting to (mds, osd) services that simply state our entity
name and global_id.

This is a wire protocol change.
Signed-off-by: Sage Weil <sage@newdream.net>

fs/ceph/Makefile

@@ -13,6 +13,7 @@ ceph-objs := super.o inode.o dir.o file.o addr.o ioctl.o \
 	mon_client.o \
 	osd_client.o osdmap.o crush/crush.o crush/mapper.o crush/hash.o \
 	debugfs.o \
+	auth.o auth_none.o \
 	ceph_fs.o ceph_strings.o ceph_hash.o ceph_frag.o
 
 else

fs/ceph/auth.c

+#include "ceph_debug.h"
+
+#include <linux/module.h>
+#include <linux/err.h>
+
+#include "types.h"
+#include "auth_none.h"
+#include "decode.h"
+#include "super.h"
+
+#include "messenger.h"
+
+/*
+ * get protocol handler
+ */
+static u32 supported_protocols[] = {
+	CEPH_AUTH_NONE
+};
+
+int ceph_auth_init_protocol(struct ceph_auth_client *ac, int protocol)
+{
+	switch (protocol) {
+	case CEPH_AUTH_NONE:
+		return ceph_auth_none_init(ac);
+	default:
+		return -ENOENT;
+	}
+}
+
+/*
+ * setup, teardown.
+ */
+struct ceph_auth_client *ceph_auth_init(const char *name, const char *secret)
+{
+	struct ceph_auth_client *ac;
+	int ret;
+
+	dout("auth_init name '%s' secret '%s'\n", name, secret);
+
+	ret = -ENOMEM;
+	ac = kzalloc(sizeof(*ac), GFP_NOFS);
+	if (!ac)
+		goto out;
+
+	ac->negotiating = true;
+	if (name)
+		ac->name = name;
+	else
+		ac->name = CEPH_AUTH_NAME_DEFAULT;
+	dout("auth_init name %s secret %s\n", ac->name, secret);
+	ac->secret = secret;
+	return ac;
+
+out:
+	return ERR_PTR(ret);
+}
+
+void ceph_auth_destroy(struct ceph_auth_client *ac)
+{
+	dout("auth_destroy %p\n", ac);
+	if (ac->ops)
+		ac->ops->destroy(ac);
+	kfree(ac);
+}
+
+/*
+ * Reset occurs when reconnecting to the monitor.
+ */
+void ceph_auth_reset(struct ceph_auth_client *ac)
+{
+	dout("auth_reset %p\n", ac);
+	if (ac->ops && !ac->negotiating)
+		ac->ops->reset(ac);
+	ac->negotiating = true;
+}
+
+int ceph_entity_name_encode(const char *name, void **p, void *end)
+{
+	int len = strlen(name);
+
+	if (*p + 2*sizeof(u32) + len > end)
+		return -ERANGE;
+	ceph_encode_32(p, CEPH_ENTITY_TYPE_CLIENT);
+	ceph_encode_32(p, len);
+	ceph_encode_copy(p, name, len);
+	return 0;
+}
+
+/*
+ * Initiate protocol negotiation with monitor.  Include entity name
+ * and list supported protocols.
+ */
+int ceph_auth_build_hello(struct ceph_auth_client *ac, void *buf, size_t len)
+{
+	struct ceph_mon_request_header *monhdr = buf;
+	void *p = monhdr + 1, *end = buf + len, *lenp;
+	int i, num;
+	int ret;
+
+	dout("auth_build_hello\n");
+	monhdr->have_version = 0;
+	monhdr->session_mon = cpu_to_le16(-1);
+	monhdr->session_mon_tid = 0;
+
+	ceph_encode_32(&p, 0);  /* no protocol, yet */
+
+	lenp = p;
+	p += sizeof(u32);
+
+	num = ARRAY_SIZE(supported_protocols);
+	ceph_encode_32(&p, num);
+	for (i = 0; i < num; i++)
+		ceph_encode_32(&p, supported_protocols[i]);
+
+	ret = ceph_entity_name_encode(ac->name, &p, end);
+	if (ret < 0)
+		return ret;
+	ceph_decode_need(&p, end, sizeof(u64), bad);
+	ceph_encode_64(&p, ac->global_id);
+
+	ceph_encode_32(&lenp, p - lenp - sizeof(u32));
+	return p - buf;
+
+bad:
+	return -ERANGE;
+}
+
+/*
+ * Handle auth message from monitor.
+ */
+int ceph_handle_auth_reply(struct ceph_auth_client *ac,
+			   void *buf, size_t len,
+			   void *reply_buf, size_t reply_len)
+{
+	void *p = buf;
+	void *end = buf + len;
+	int protocol;
+	s32 result;
+	u64 global_id;
+	void *payload, *payload_end;
+	int payload_len;
+	char *result_msg;
+	int result_msg_len;
+	int ret = -EINVAL;
+
+	dout("handle_auth_reply %p %p\n", p, end);
+	ceph_decode_need(&p, end, sizeof(u32) * 3 + sizeof(u64), bad);
+	protocol = ceph_decode_32(&p);
+	result = ceph_decode_32(&p);
+	global_id = ceph_decode_64(&p);
+	payload_len = ceph_decode_32(&p);
+	payload = p;
+	p += payload_len;
+	ceph_decode_need(&p, end, sizeof(u32), bad);
+	result_msg_len = ceph_decode_32(&p);
+	result_msg = p;
+	p += result_msg_len;
+	if (p != end)
+		goto bad;
+
+	dout(" result %d '%.*s' gid %llu len %d\n", result, result_msg_len,
+	     result_msg, global_id, payload_len);
+
+	payload_end = payload + payload_len;
+
+	if (global_id && ac->global_id != global_id) {
+		dout(" set global_id %lld -> %lld\n", ac->global_id, global_id);
+		ac->global_id = global_id;
+	}
+
+	if (ac->negotiating) {
+		/* set up (new) protocol handler? */
+		if (ac->protocol && ac->protocol != protocol) {
+			ac->ops->destroy(ac);
+			ac->protocol = 0;
+			ac->ops = NULL;
+		}
+		if (ac->protocol != protocol) {
+			ret = ceph_auth_init_protocol(ac, protocol);
+			if (ret) {
+				pr_err("error %d on auth protocol %d init\n",
+				       ret, protocol);
+				goto out;
+			}
+		}
+	}
+
+	ret = ac->ops->handle_reply(ac, result, payload, payload_end);
+	if (ret == -EAGAIN) {
+		struct ceph_mon_request_header *monhdr = reply_buf;
+		void *p = reply_buf + 1;
+		void *end = reply_buf + reply_len;
+
+		monhdr->have_version = 0;
+		monhdr->session_mon = cpu_to_le16(-1);
+		monhdr->session_mon_tid = 0;
+
+		ceph_encode_32(&p, ac->protocol);
+
+		ret = ac->ops->build_request(ac, p + sizeof(u32), end);
+		if (ret < 0) {
+			pr_err("error %d building request\n", ret);
+			goto out;
+		}
+		dout(" built request %d bytes\n", ret);
+		ceph_encode_32(&p, ret);
+		return p + ret - reply_buf;
+	} else if (ret) {
+		pr_err("authentication error %d\n", ret);
+		return ret;
+	}
+	return 0;
+
+bad:
+	pr_err("failed to decode auth msg\n");
+out:
+	return ret;
+}
+
+

fs/ceph/auth.h

+#ifndef _FS_CEPH_AUTH_H
+#define _FS_CEPH_AUTH_H
+
+#include "types.h"
+#include "buffer.h"
+
+/*
+ * Abstract interface for communicating with the authenticate module.
+ * There is some handshake that takes place between us and the monitor
+ * to acquire the necessary keys.  These are used to generate an
+ * 'authorizer' that we use when connecting to a service (mds, osd).
+ */
+
+struct ceph_auth_client;
+struct ceph_authorizer;
+
+struct ceph_auth_client_ops {
+	/*
+	 * true if we are authenticated and can connect to
+	 * services.
+	 */
+	int (*is_authenticated)(struct ceph_auth_client *ac);
+
+	/*
+	 * build requests and process replies during monitor
+	 * handshake.  if handle_reply returns -EAGAIN, we build
+	 * another request.
+	 */
+	int (*build_request)(struct ceph_auth_client *ac, void *buf, void *end);
+	int (*handle_reply)(struct ceph_auth_client *ac, int result,
+			    void *buf, void *end);
+
+	/*
+	 * Create authorizer for connecting to a service, and verify
+	 * the response to authenticate the service.
+	 */
+	int (*create_authorizer)(struct ceph_auth_client *ac, int peer_type,
+				 struct ceph_authorizer **a,
+				 void **buf, size_t *len,
+				 void **reply_buf, size_t *reply_len);
+	int (*verify_authorizer_reply)(struct ceph_auth_client *ac,
+				       struct ceph_authorizer *a, size_t len);
+	void (*destroy_authorizer)(struct ceph_auth_client *ac,
+				   struct ceph_authorizer *a);
+
+	/* reset when we (re)connect to a monitor */
+	void (*reset)(struct ceph_auth_client *ac);
+
+	void (*destroy)(struct ceph_auth_client *ac);
+};
+
+struct ceph_auth_client {
+	u32 protocol;           /* CEPH_AUTH_* */
+	void *private;          /* for use by protocol implementation */
+	const struct ceph_auth_client_ops *ops;  /* null iff protocol==0 */
+
+	bool negotiating;       /* true if negotiating protocol */
+	const char *name;       /* entity name */
+	u64 global_id;          /* our unique id in system */
+	const char *secret;     /* our secret key */
+	unsigned want_keys;     /* which services we want */
+};
+
+extern struct ceph_auth_client *ceph_auth_init(const char *name,
+					       const char *secret);
+extern void ceph_auth_destroy(struct ceph_auth_client *ac);
+
+extern void ceph_auth_reset(struct ceph_auth_client *ac);
+
+extern int ceph_auth_build_hello(struct ceph_auth_client *ac,
+				 void *buf, size_t len);
+extern int ceph_handle_auth_reply(struct ceph_auth_client *ac,
+				  void *buf, size_t len,
+				  void *reply_buf, size_t reply_len);
+extern int ceph_entity_name_encode(const char *name, void **p, void *end);
+
+#endif

fs/ceph/auth_none.c

+
+#include "ceph_debug.h"
+
+#include <linux/err.h>
+#include <linux/module.h>
+#include <linux/random.h>
+
+#include "auth_none.h"
+#include "auth.h"
+#include "decode.h"
+
+static void reset(struct ceph_auth_client *ac)
+{
+	struct ceph_auth_none_info *xi = ac->private;
+
+	xi->starting = true;
+	xi->built_authorizer = false;
+}
+
+static void destroy(struct ceph_auth_client *ac)
+{
+	kfree(ac->private);
+	ac->private = NULL;
+}
+
+static int is_authenticated(struct ceph_auth_client *ac)
+{
+	struct ceph_auth_none_info *xi = ac->private;
+
+	return !xi->starting;
+}
+
+/*
+ * the generic auth code decode the global_id, and we carry no actual
+ * authenticate state, so nothing happens here.
+ */
+static int handle_reply(struct ceph_auth_client *ac, int result,
+			void *buf, void *end)
+{
+	struct ceph_auth_none_info *xi = ac->private;
+
+	xi->starting = false;
+	return result;
+}
+
+/*
+ * build an 'authorizer' with our entity_name and global_id.  we can
+ * reuse a single static copy since it is identical for all services
+ * we connect to.
+ */
+static int ceph_auth_none_create_authorizer(
+	struct ceph_auth_client *ac, int peer_type,
+	struct ceph_authorizer **a,
+	void **buf, size_t *len,
+	void **reply_buf, size_t *reply_len)
+{
+	struct ceph_auth_none_info *ai = ac->private;
+	struct ceph_none_authorizer *au = &ai->au;
+	void *p, *end;
+	int ret;
+
+	if (!ai->built_authorizer) {
+		p = au->buf;
+		end = p + sizeof(au->buf);
+		ret = ceph_entity_name_encode(ac->name, &p, end - 8);
+		if (ret < 0)
+			goto bad;
+		ceph_decode_need(&p, end, sizeof(u64), bad2);
+		ceph_encode_64(&p, ac->global_id);
+		au->buf_len = p - (void *)au->buf;
+		ai->built_authorizer = true;
+		dout("built authorizer len %d\n", au->buf_len);
+	}
+
+	*a = (struct ceph_authorizer *)au;
+	*buf = au->buf;
+	*len = au->buf_len;
+	*reply_buf = au->reply_buf;
+	*reply_len = sizeof(au->reply_buf);
+	return 0;
+
+bad2:
+	ret = -ERANGE;
+bad:
+	return ret;
+}
+
+static void ceph_auth_none_destroy_authorizer(struct ceph_auth_client *ac,
+				      struct ceph_authorizer *a)
+{
+	/* nothing to do */
+}
+
+static const struct ceph_auth_client_ops ceph_auth_none_ops = {
+	.reset = reset,
+	.destroy = destroy,
+	.is_authenticated = is_authenticated,
+	.handle_reply = handle_reply,
+	.create_authorizer = ceph_auth_none_create_authorizer,
+	.destroy_authorizer = ceph_auth_none_destroy_authorizer,
+};
+
+int ceph_auth_none_init(struct ceph_auth_client *ac)
+{
+	struct ceph_auth_none_info *xi;
+
+	dout("ceph_auth_none_init %p\n", ac);
+	xi = kzalloc(sizeof(*xi), GFP_NOFS);
+	if (!xi)
+		return -ENOMEM;
+
+	xi->starting = true;
+	xi->built_authorizer = false;
+
+	ac->protocol = CEPH_AUTH_NONE;
+	ac->private = xi;
+	ac->ops = &ceph_auth_none_ops;
+	return 0;
+}
+

fs/ceph/auth_none.h

+#ifndef _FS_CEPH_AUTH_NONE_H
+#define _FS_CEPH_AUTH_NONE_H
+
+#include "auth.h"
+
+/*
+ * null security mode.
+ *
+ * we use a single static authorizer that simply encodes our entity name
+ * and global id.
+ */
+
+struct ceph_none_authorizer {
+	char buf[128];
+	int buf_len;
+	char reply_buf[0];
+};
+
+struct ceph_auth_none_info {
+	bool starting;
+	bool built_authorizer;
+	struct ceph_none_authorizer au;   /* we only need one; it's static */
+};
+
+extern int ceph_auth_none_init(struct ceph_auth_client *ac);
+
+#endif
+

fs/ceph/ceph_fs.h

@@ -75,6 +75,16 @@ struct ceph_file_layout {
 int ceph_file_layout_is_valid(const struct ceph_file_layout *layout);
 
 
+/* crypto algorithms */
+#define CEPH_CRYPTO_NONE 0x0
+#define CEPH_CRYPTO_AES  0x1
+
+/* security/authentication protocols */
+#define CEPH_AUTH_UNKNOWN	0x0
+#define CEPH_AUTH_NONE	 	0x1
+#define CEPH_AUTH_CEPHX	 	0x2
+
+
 /*********************************************
  * message layer
  */
@@ -90,12 +100,12 @@ int ceph_file_layout_is_valid(const struct ceph_file_layout *layout);
 /* client <-> monitor */
 #define CEPH_MSG_MON_MAP                4
 #define CEPH_MSG_MON_GET_MAP            5
-#define CEPH_MSG_CLIENT_MOUNT           10
-#define CEPH_MSG_CLIENT_MOUNT_ACK       11
 #define CEPH_MSG_STATFS                 13
 #define CEPH_MSG_STATFS_REPLY           14
 #define CEPH_MSG_MON_SUBSCRIBE          15
 #define CEPH_MSG_MON_SUBSCRIBE_ACK      16
+#define CEPH_MSG_AUTH			17
+#define CEPH_MSG_AUTH_REPLY		18
 
 /* client <-> mds */
 #define CEPH_MSG_MDS_MAP                21

fs/ceph/ceph_strings.c

@@ -3,6 +3,19 @@
  */
 #include "types.h"
 
+const char *ceph_entity_type_name(int type)
+{
+	switch (type) {
+	case CEPH_ENTITY_TYPE_MDS: return "mds";
+	case CEPH_ENTITY_TYPE_OSD: return "osd";
+	case CEPH_ENTITY_TYPE_MON: return "mon";
+	case CEPH_ENTITY_TYPE_CLIENT: return "client";
+	case CEPH_ENTITY_TYPE_ADMIN: return "admin";
+	case CEPH_ENTITY_TYPE_AUTH: return "auth";
+	default: return "unknown";
+	}
+}
+
 const char *ceph_osd_op_name(int op)
 {
 	switch (op) {

fs/ceph/decode.h

@@ -98,6 +98,7 @@ static inline void ceph_encode_addr(struct ceph_entity_addr *a)
 static inline void ceph_decode_addr(struct ceph_entity_addr *a)
 {
 	a->in_addr.ss_family = ntohs(a->in_addr.ss_family);
+	WARN_ON(a->in_addr.ss_family == 512);
 }
 
 /*
@@ -123,6 +124,11 @@ static inline void ceph_encode_8(void **p, u8 v)
 	*(u8 *)*p = v;
 	(*p)++;
 }
+static inline void ceph_encode_copy(void **p, const void *s, int len)
+{
+	memcpy(*p, s, len);
+	*p += len;
+}
 
 /*
  * filepath, string encoders

fs/ceph/mds_client.c

@@ -8,6 +8,7 @@
 #include "super.h"
 #include "messenger.h"
 #include "decode.h"
+#include "auth.h"
 
 /*
  * A cluster of MDS (metadata server) daemons is responsible for
@@ -274,8 +275,12 @@ void ceph_put_mds_session(struct ceph_mds_session *s)
 {
 	dout("mdsc put_session %p %d -> %d\n", s,
 	     atomic_read(&s->s_ref), atomic_read(&s->s_ref)-1);
-	if (atomic_dec_and_test(&s->s_ref))
+	if (atomic_dec_and_test(&s->s_ref)) {
+		if (s->s_authorizer)
+			s->s_mdsc->client->monc.auth->ops->destroy_authorizer(
+				s->s_mdsc->client->monc.auth, s->s_authorizer);
 		kfree(s);
+	}
 }
 
 /*
@@ -2777,10 +2782,16 @@ void ceph_mdsc_handle_map(struct ceph_mds_client *mdsc, struct ceph_msg *msg)
 
 	ceph_decode_need(&p, end, sizeof(fsid)+2*sizeof(u32), bad);
 	ceph_decode_copy(&p, &fsid, sizeof(fsid));
-	if (ceph_fsid_compare(&fsid, &mdsc->client->monc.monmap->fsid)) {
+        if (mdsc->client->monc.have_fsid) {
+		if (ceph_fsid_compare(&fsid,
+				      &mdsc->client->monc.monmap->fsid)) {
 			pr_err("got mdsmap with wrong fsid\n");
 			return;
 		}
+	} else {
+		ceph_fsid_set(&mdsc->client->monc.monmap->fsid, &fsid);
+		mdsc->client->monc.have_fsid = true;
+	}
 	epoch = ceph_decode_32(&p);
 	maplen = ceph_decode_32(&p);
 	dout("handle_map epoch %u len %d\n", epoch, (int)maplen);
@@ -2895,10 +2906,60 @@ static void dispatch(struct ceph_connection *con, struct ceph_msg *msg)
 	ceph_msg_put(msg);
 }
 
+/*
+ * authentication
+ */
+static int get_authorizer(struct ceph_connection *con,
+			  void **buf, int *len, int *proto,
+			  void **reply_buf, int *reply_len, int force_new)
+{
+	struct ceph_mds_session *s = con->private;
+	struct ceph_mds_client *mdsc = s->s_mdsc;
+	struct ceph_auth_client *ac = mdsc->client->monc.auth;
+	int ret = 0;
+
+	if (force_new && s->s_authorizer) {
+		ac->ops->destroy_authorizer(ac, s->s_authorizer);
+		s->s_authorizer = NULL;
+	}
+	if (s->s_authorizer == NULL) {
+		if (ac->ops->create_authorizer) {
+			ret = ac->ops->create_authorizer(
+				ac, CEPH_ENTITY_TYPE_MDS,
+				&s->s_authorizer,
+				&s->s_authorizer_buf,
+				&s->s_authorizer_buf_len,
+				&s->s_authorizer_reply_buf,
+				&s->s_authorizer_reply_buf_len);
+			if (ret)
+				return ret;
+		}
+	}
+
+	*proto = ac->protocol;
+	*buf = s->s_authorizer_buf;
+	*len = s->s_authorizer_buf_len;
+	*reply_buf = s->s_authorizer_reply_buf;
+	*reply_len = s->s_authorizer_reply_buf_len;
+	return 0;
+}
+
+
+static int verify_authorizer_reply(struct ceph_connection *con, int len)
+{
+	struct ceph_mds_session *s = con->private;
+	struct ceph_mds_client *mdsc = s->s_mdsc;
+	struct ceph_auth_client *ac = mdsc->client->monc.auth;
+
+	return ac->ops->verify_authorizer_reply(ac, s->s_authorizer, len);
+}
+
 const static struct ceph_connection_operations mds_con_ops = {
 	.get = con_get,
 	.put = con_put,
 	.dispatch = dispatch,
+	.get_authorizer = get_authorizer,
+	.verify_authorizer_reply = verify_authorizer_reply,
 	.peer_reset = peer_reset,
 	.alloc_msg = ceph_alloc_msg,
 	.alloc_middle = ceph_alloc_middle,

fs/ceph/mds_client.h

@@ -100,6 +100,10 @@ struct ceph_mds_session {
 
 	struct ceph_connection s_con;
 
+	struct ceph_authorizer *s_authorizer;
+	void             *s_authorizer_buf, *s_authorizer_reply_buf;
+	size_t            s_authorizer_buf_len, s_authorizer_reply_buf_len;
+
 	/* protected by s_cap_lock */
 	spinlock_t        s_cap_lock;
 	u32               s_cap_gen;  /* inc each time we get mds stale msg */

fs/ceph/messenger.c

@@ -550,6 +550,27 @@ static void prepare_write_keepalive(struct ceph_connection *con)
  * Connection negotiation.
  */
 
+static void prepare_connect_authorizer(struct ceph_connection *con)
+{
+	void *auth_buf;
+	int auth_len = 0;
+	int auth_protocol = 0;
+
+	if (con->ops->get_authorizer)
+		con->ops->get_authorizer(con, &auth_buf, &auth_len,
+					 &auth_protocol, &con->auth_reply_buf,
+					 &con->auth_reply_buf_len,
+					 con->auth_retry);
+
+	con->out_connect.authorizer_protocol = cpu_to_le32(auth_protocol);
+	con->out_connect.authorizer_len = cpu_to_le32(auth_len);
+
+	con->out_kvec[con->out_kvec_left].iov_base = auth_buf;
+	con->out_kvec[con->out_kvec_left].iov_len = auth_len;
+	con->out_kvec_left++;
+	con->out_kvec_bytes += auth_len;
+}
+
 /*
  * We connected to a peer and are saying hello.
  */
@@ -592,6 +613,7 @@ 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.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);
@@ -611,6 +633,8 @@ static void prepare_write_connect(struct ceph_messenger *msgr,
 	con->out_kvec_cur = con->out_kvec;
 	con->out_more = 0;
 	set_bit(WRITE_PENDING, &con->state);
+
+	prepare_connect_authorizer(con);
 }
 
 
@@ -777,6 +801,13 @@ static void prepare_read_connect(struct ceph_connection *con)
 	con->in_base_pos = 0;
 }
 
+static void prepare_read_connect_retry(struct ceph_connection *con)
+{
+	dout("prepare_read_connect_retry %p\n", con);
+	con->in_base_pos = strlen(CEPH_BANNER) + sizeof(con->actual_peer_addr)
+		+ sizeof(con->peer_addr_for_me);
+}
+
 static void prepare_read_ack(struct ceph_connection *con)
 {
 	dout("prepare_read_ack %p\n", con);
@@ -853,9 +884,14 @@ static int read_partial_connect(struct ceph_connection *con)
 	ret = read_partial(con, &to, sizeof(con->in_reply), &con->in_reply);
 	if (ret <= 0)
 		goto out;
+	ret = read_partial(con, &to, le32_to_cpu(con->in_reply.authorizer_len),
+			   con->auth_reply_buf);
+	if (ret <= 0)
+		goto out;
 
-	dout("read_partial_connect %p connect_seq = %u, global_seq = %u\n",
-	     con, le32_to_cpu(con->in_reply.connect_seq),
+	dout("read_partial_connect %p tag %d, con_seq = %u, g_seq = %u\n",
+	     con, (int)con->in_reply.tag,
+	     le32_to_cpu(con->in_reply.connect_seq),
 	     le32_to_cpu(con->in_reply.global_seq));
 out:
 	return ret;
@@ -1051,6 +1087,20 @@ static int process_connect(struct ceph_connection *con)
 		set_bit(CLOSED, &con->state);  /* in case there's queued work */
 		return -1;
 
+	case CEPH_MSGR_TAG_BADAUTHORIZER:
+		con->auth_retry++;
+		dout("process_connect %p got BADAUTHORIZER attempt %d\n", con,
+		     con->auth_retry);
+		if (con->auth_retry == 2) {
+			con->error_msg = "connect authorization failure";
+			reset_connection(con);
+			set_bit(CLOSED, &con->state);
+			return -1;
+		}
+		con->auth_retry = 1;
+		prepare_write_connect(con->msgr, con, 0);
+		prepare_read_connect_retry(con);
+		break;
 
 	case CEPH_MSGR_TAG_RESETSESSION:
 		/*

fs/ceph/messenger.h

@@ -26,6 +26,12 @@ struct ceph_connection_operations {
 	/* handle an incoming message. */
 	void (*dispatch) (struct ceph_connection *con, struct ceph_msg *m);
 
+	/* authorize an outgoing connection */
+	int (*get_authorizer) (struct ceph_connection *con,
+			       void **buf, int *len, int *proto,
+			       void **reply_buf, int *reply_len, int force_new);
+	int (*verify_authorizer_reply) (struct ceph_connection *con, int len);
+
 	/* protocol version mismatch */
 	void (*bad_proto) (struct ceph_connection *con);
 
@@ -144,6 +150,10 @@ struct ceph_connection {
 				 attempt for this connection, client */
 	u32 peer_global_seq;  /* peer's global seq for this connection */
 
+	int auth_retry;       /* true if we need a newer authorizer */
+	void *auth_reply_buf;   /* where to put the authorizer reply */
+	int auth_reply_buf_len;
+
 	/* out queue */
 	struct mutex out_mutex;
 	struct list_head out_queue;

fs/ceph/mon_client.h

@@ -9,6 +9,7 @@
 
 struct ceph_client;
 struct ceph_mount_args;
+struct ceph_auth_client;
 
 /*
  * The monitor map enumerates the set of all monitors.
@@ -58,23 +59,26 @@ struct ceph_mon_client {
 	struct mutex mutex;
 	struct delayed_work delayed_work;
 
+	struct ceph_auth_client *auth;
+	struct ceph_msg *m_auth;
+
 	bool hunting;
 	int cur_mon;                       /* last monitor i contacted */
 	unsigned long sub_sent, sub_renew_after;
 	struct ceph_connection *con;
+	bool have_fsid;
 
 	/* msg pools */
-	struct ceph_msgpool msgpool_mount_ack;
 	struct ceph_msgpool msgpool_subscribe_ack;
 	struct ceph_msgpool msgpool_statfs_reply;
+	struct ceph_msgpool msgpool_auth_reply;
 
 	/* pending statfs requests */
 	struct radix_tree_root statfs_request_tree;
 	int num_statfs_requests;
 	u64 last_tid;
 
-	/* mds/osd map or mount requests */
-	bool want_mount;
+	/* mds/osd map */
 	int want_next_osdmap; /* 1 = want, 2 = want+asked */
 	u32 have_osdmap, have_mdsmap;
 
@@ -101,11 +105,11 @@ extern int ceph_monc_got_osdmap(struct ceph_mon_client *monc, u32 have);
 
 extern void ceph_monc_request_next_osdmap(struct ceph_mon_client *monc);
 
-extern int ceph_monc_request_mount(struct ceph_mon_client *monc);
-
 extern int ceph_monc_do_statfs(struct ceph_mon_client *monc,
 			       struct ceph_statfs *buf);
 
+extern int ceph_monc_open_session(struct ceph_mon_client *monc);
+
 
 
 #endif

fs/ceph/msgr.h

@@ -21,7 +21,7 @@
  * whenever the wire protocol changes.  try to keep this string length
  * constant.
  */
-#define CEPH_BANNER "ceph v023"
+#define CEPH_BANNER "ceph v024"
 #define CEPH_BANNER_MAX_LEN 30
 
 
@@ -46,11 +46,16 @@ struct ceph_entity_name {
 	__le64 num;
 } __attribute__ ((packed));
 
-#define CEPH_ENTITY_TYPE_MON    1
-#define CEPH_ENTITY_TYPE_MDS    2
-#define CEPH_ENTITY_TYPE_OSD    3
-#define CEPH_ENTITY_TYPE_CLIENT 4
-#define CEPH_ENTITY_TYPE_ADMIN  5
+#define CEPH_ENTITY_TYPE_MON    0x01
+#define CEPH_ENTITY_TYPE_MDS    0x02
+#define CEPH_ENTITY_TYPE_OSD    0x04
+#define CEPH_ENTITY_TYPE_CLIENT 0x08
+#define CEPH_ENTITY_TYPE_ADMIN  0x10
+#define CEPH_ENTITY_TYPE_AUTH   0x20
+
+#define CEPH_ENTITY_TYPE_ANY    0xFF
+
+extern const char *ceph_entity_type_name(int type);
 
 /*
  * entity_addr -- network address
@@ -94,6 +99,7 @@ struct ceph_entity_inst {
 #define CEPH_MSGR_TAG_ACK           8  /* message ack */
 #define CEPH_MSGR_TAG_KEEPALIVE     9  /* just a keepalive byte! */
 #define CEPH_MSGR_TAG_BADPROTOVER  10  /* bad protocol version */
+#define CEPH_MSGR_TAG_BADAUTHORIZER 11 /* bad authorizer */
 
 
 /*
@@ -104,6 +110,8 @@ struct ceph_msg_connect {
 	__le32 global_seq;   /* count connections initiated by this host */
 	__le32 connect_seq;  /* count connections initiated in this session */
 	__le32 protocol_version;
+	__le32 authorizer_protocol;
+	__le32 authorizer_len;
 	__u8  flags;         /* CEPH_MSG_CONNECT_* */
 } __attribute__ ((packed));
 
@@ -112,6 +120,7 @@ struct ceph_msg_connect_reply {
 	__le32 global_seq;
 	__le32 connect_seq;
 	__le32 protocol_version;
+	__le32 authorizer_len;
 	__u8 flags;
 } __attribute__ ((packed));
 

fs/ceph/osd_client.c

@@ -11,6 +11,7 @@
 #include "osd_client.h"
 #include "messenger.h"
 #include "decode.h"
+#include "auth.h"
 
 const static struct ceph_connection_operations osd_con_ops;
 
@@ -331,6 +332,7 @@ static struct ceph_osd *create_osd(struct ceph_osd_client *osdc)
 	osd->o_con.private = osd;
 	osd->o_con.ops = &osd_con_ops;
 	osd->o_con.peer_name.type = CEPH_ENTITY_TYPE_OSD;
+
 	return osd;
 }
 
@@ -880,10 +882,16 @@ void ceph_osdc_handle_map(struct ceph_osd_client *osdc, struct ceph_msg *msg)
 	/* verify fsid */
 	ceph_decode_need(&p, end, sizeof(fsid), bad);
 	ceph_decode_copy(&p, &fsid, sizeof(fsid));
-	if (ceph_fsid_compare(&fsid, &osdc->client->monc.monmap->fsid)) {
+        if (osdc->client->monc.have_fsid) {
+		if (ceph_fsid_compare(&fsid,
+			              &osdc->client->monc.monmap->fsid)) {
 			pr_err("got osdmap with wrong fsid, ignoring\n");
 			return;
 		}
+	} else {
+		ceph_fsid_set(&osdc->client->monc.monmap->fsid, &fsid);
+		osdc->client->monc.have_fsid = true;
+	}
 
 	down_write(&osdc->map_sem);
 
@@ -1302,10 +1310,59 @@ static void put_osd_con(struct ceph_connection *con)
 	put_osd(osd);
 }
 
+/*
+ * authentication
+ */
+static int get_authorizer(struct ceph_connection *con,
+                          void **buf, int *len, int *proto,
+                          void **reply_buf, int *reply_len, int force_new)
+{
+	struct ceph_osd *o = con->private;
+	struct ceph_osd_client *osdc = o->o_osdc;
+	struct ceph_auth_client *ac = osdc->client->monc.auth;
+	int ret = 0;
+
+	if (force_new && o->o_authorizer) {
+		ac->ops->destroy_authorizer(ac, o->o_authorizer);
+		o->o_authorizer = NULL;
+	}
+	if (o->o_authorizer == NULL) {
+		ret = ac->ops->create_authorizer(
+			ac, CEPH_ENTITY_TYPE_OSD,
+			&o->o_authorizer,
+			&o->o_authorizer_buf,
+			&o->o_authorizer_buf_len,
+			&o->o_authorizer_reply_buf,
+			&o->o_authorizer_reply_buf_len);
+		if (ret)
+		return ret;
+	}
+
+	*proto = ac->protocol;
+	*buf = o->o_authorizer_buf;
+	*len = o->o_authorizer_buf_len;
+	*reply_buf = o->o_authorizer_reply_buf;
+	*reply_len = o->o_authorizer_reply_buf_len;
+	return 0;
+}
+
+
+static int verify_authorizer_reply(struct ceph_connection *con, int len)
+{
+	struct ceph_osd *o = con->private;
+	struct ceph_osd_client *osdc = o->o_osdc;
+	struct ceph_auth_client *ac = osdc->client->monc.auth;
+
+	return ac->ops->verify_authorizer_reply(ac, o->o_authorizer, len);
+}
+
+
 const static struct ceph_connection_operations osd_con_ops = {
 	.get = get_osd_con,
 	.put = put_osd_con,
 	.dispatch = dispatch,
+	.get_authorizer = get_authorizer,
+	.verify_authorizer_reply = verify_authorizer_reply,
 	.alloc_msg = alloc_msg,
 	.fault = osd_reset,
 	.alloc_middle = ceph_alloc_middle,

fs/ceph/osd_client.h

@@ -13,6 +13,7 @@ struct ceph_msg;
 struct ceph_snap_context;
 struct ceph_osd_request;
 struct ceph_osd_client;
+struct ceph_authorizer;
 
 /*
  * completion callback for async writepages
@@ -29,6 +30,9 @@ struct ceph_osd {
 	struct rb_node o_node;
 	struct ceph_connection o_con;
 	struct list_head o_requests;
+	struct ceph_authorizer *o_authorizer;
+	void *o_authorizer_buf, *o_authorizer_reply_buf;
+	size_t o_authorizer_buf_len, o_authorizer_reply_buf_len;
 };
 
 /* an in-flight request */

fs/ceph/rados.h

@@ -157,7 +157,6 @@ struct ceph_eversion {
 #define CEPH_OSD_OP_MODE_WR    0x2000
 #define CEPH_OSD_OP_MODE_RMW   0x3000
 #define CEPH_OSD_OP_MODE_SUB   0x4000
-#define CEPH_OSD_OP_MODE_EXEC  0x8000
 
 #define CEPH_OSD_OP_TYPE       0x0f00
 #define CEPH_OSD_OP_TYPE_LOCK  0x0100
@@ -285,6 +284,7 @@ enum {
 	CEPH_OSD_FLAG_BALANCE_READS = 256,
 	CEPH_OSD_FLAG_PARALLELEXEC = 512, /* execute op in parallel */
 	CEPH_OSD_FLAG_PGOP = 1024,      /* pg op, no object */
+	CEPH_OSD_FLAG_EXEC = 2048,      /* op may exec */
 };
 
 enum {

fs/ceph/super.c

@@ -128,6 +128,8 @@ static int ceph_show_options(struct seq_file *m, struct vfsmount *mnt)
 		seq_puts(m, ",noasyncreaddir");
 	if (strcmp(args->snapdir_name, CEPH_SNAPDIRNAME_DEFAULT))
 		seq_printf(m, ",snapdirname=%s", args->snapdir_name);
+	if (args->name)
+		seq_printf(m, ",name=%s", args->name);
 	if (args->secret)
 		seq_puts(m, ",secret=<hidden>");
 	return 0;
@@ -224,12 +226,12 @@ const char *ceph_msg_type_name(int type)
 	switch (type) {
 	case CEPH_MSG_SHUTDOWN: return "shutdown";
 	case CEPH_MSG_PING: return "ping";
+	case CEPH_MSG_AUTH: return "auth";
+	case CEPH_MSG_AUTH_REPLY: return "auth_reply";
 	case CEPH_MSG_MON_MAP: return "mon_map";
 	case CEPH_MSG_MON_GET_MAP: return "mon_get_map";
 	case CEPH_MSG_MON_SUBSCRIBE: return "mon_subscribe";
 	case CEPH_MSG_MON_SUBSCRIBE_ACK: return "mon_subscribe_ack";
-	case CEPH_MSG_CLIENT_MOUNT: return "client_mount";
-	case CEPH_MSG_CLIENT_MOUNT_ACK: return "client_mount_ack";
 	case CEPH_MSG_STATFS: return "statfs";
 	case CEPH_MSG_STATFS_REPLY: return "statfs_reply";
 	case CEPH_MSG_MDS_MAP: return "mds_map";
@@ -267,6 +269,7 @@ enum {
 	Opt_last_int,
 	/* int args above */
 	Opt_snapdirname,
+	Opt_name,
 	Opt_secret,
 	Opt_last_string,
 	/* string args above */
@@ -293,6 +296,7 @@ static match_table_t arg_tokens = {
 	{Opt_readdir_max_entries, "readdir_max_entries=%d"},
 	/* int args above */
 	{Opt_snapdirname, "snapdirname=%s"},
+	{Opt_name, "name=%s"},
 	{Opt_secret, "secret=%s"},
 	/* string args above */
 	{Opt_ip, "ip=%s"},
@@ -407,6 +411,11 @@ static struct ceph_mount_args *parse_mount_args(int flags, char *options,
 					      argstr[0].to-argstr[0].from,
 					      GFP_KERNEL);
 			break;
+		case Opt_name:
+			args->name = kstrndup(argstr[0].from,
+					      argstr[0].to-argstr[0].from,
+					      GFP_KERNEL);
+			break;
 		case Opt_secret:
 			args->secret = kstrndup(argstr[0].from,
 						argstr[0].to-argstr[0].from,
@@ -476,6 +485,8 @@ static void destroy_mount_args(struct ceph_mount_args *args)
 	dout("destroy_mount_args %p\n", args);
 	kfree(args->snapdir_name);
 	args->snapdir_name = NULL;
+	kfree(args->name);
+	args->name = NULL;
 	kfree(args->secret);
 	args->secret = NULL;
 	kfree(args);
@@ -657,27 +668,23 @@ static int ceph_mount(struct ceph_client *client, struct vfsmount *mnt,
 		client->msgr->nocrc = ceph_test_opt(client, NOCRC);
 	}
 
-	/* send mount request, and wait for mon, mds, and osd maps */
-	err = ceph_monc_request_mount(&client->monc);
+	/* open session, and wait for mon, mds, and osd maps */
+	err = ceph_monc_open_session(&client->monc);
 	if (err < 0)
 		goto out;
 
-	while (!have_mon_map(client) && !client->mount_err) {
+	while (!have_mon_map(client)) {
 		err = -EIO;
 		if (timeout && time_after_eq(jiffies, started + timeout))
 			goto out;
 
 		/* wait */
-		dout("mount waiting for mount\n");
-		err = wait_event_interruptible_timeout(client->mount_wq,
-			       client->mount_err || have_mon_map(client),
+		dout("mount waiting for mon_map\n");
+		err = wait_event_interruptible_timeout(client->mount_wq, /* FIXME */
+			       have_mon_map(client),
 			       timeout);
 		if (err == -EINTR || err == -ERESTARTSYS)
 			goto out;
-		if (client->mount_err) {
-			err = client->mount_err;
-			goto out;
-		}
 	}
 
 	dout("mount opening root\n");
@@ -795,7 +802,6 @@ static int ceph_register_bdi(struct super_block *sb, struct ceph_client *client)
 		client->backing_dev_info.ra_pages =
 			(client->mount_args->rsize + PAGE_CACHE_SIZE - 1)
 			>> PAGE_SHIFT;
-
 	err = bdi_register_dev(&client->backing_dev_info, sb->s_dev);
 	return err;
 }

fs/ceph/super.h

@@ -61,6 +61,7 @@ struct ceph_mount_args {
 	int max_readdir;      /* max readdir size */
 	int osd_timeout;
 	char *snapdir_name;   /* default ".snap" */
+	char *name;
 	char *secret;
 	int cap_release_safety;
 };
@@ -75,6 +76,7 @@ struct ceph_mount_args {
 #define CEPH_MSG_MAX_DATA_LEN	(16*1024*1024)
 
 #define CEPH_SNAPDIRNAME_DEFAULT ".snap"
+#define CEPH_AUTH_NAME_DEFAULT   "guest"
 
 /*
  * Delay telling the MDS we no longer want caps, in case we reopen