[deliverable/linux.git] / net / ceph / ceph_common.c


#include <linux/ceph/ceph_debug.h>
#include <linux/backing-dev.h>
#include <linux/ctype.h>
#include <linux/fs.h>
#include <linux/inet.h>
#include <linux/in6.h>
#include <linux/key.h>
#include <keys/ceph-type.h>
#include <linux/module.h>
#include <linux/mount.h>
#include <linux/parser.h>
#include <linux/sched.h>
#include <linux/seq_file.h>
#include <linux/slab.h>
#include <linux/statfs.h>
#include <linux/string.h>
#include <linux/vmalloc.h>
#include <linux/nsproxy.h>
#include <net/net_namespace.h>


#include <linux/ceph/ceph_features.h>
#include <linux/ceph/libceph.h>
#include <linux/ceph/debugfs.h>
#include <linux/ceph/decode.h>
#include <linux/ceph/mon_client.h>
#include <linux/ceph/auth.h>
#include "crypto.h"


/*
 * Module compatibility interface.  For now it doesn't do anything,
 * but its existence signals a certain level of functionality.
 *
 * The data buffer is used to pass information both to and from
 * libceph.  The return value indicates whether libceph determines
 * it is compatible with the caller (from another kernel module),
 * given the provided data.
 *
 * The data pointer can be null.
 */
bool libceph_compatible(void *data)
{
	return true;
}
EXPORT_SYMBOL(libceph_compatible);

/*
 * find filename portion of a path (/foo/bar/baz -> baz)
 */
const char *ceph_file_part(const char *s, int len)
{
	const char *e = s + len;

	while (e != s && *(e-1) != '/')
		e--;
	return e;
}
EXPORT_SYMBOL(ceph_file_part);

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_STATFS: return "statfs";
	case CEPH_MSG_STATFS_REPLY: return "statfs_reply";
	case CEPH_MSG_MON_GET_VERSION: return "mon_get_version";
	case CEPH_MSG_MON_GET_VERSION_REPLY: return "mon_get_version_reply";
	case CEPH_MSG_MDS_MAP: return "mds_map";
	case CEPH_MSG_CLIENT_SESSION: return "client_session";
	case CEPH_MSG_CLIENT_RECONNECT: return "client_reconnect";
	case CEPH_MSG_CLIENT_REQUEST: return "client_request";
	case CEPH_MSG_CLIENT_REQUEST_FORWARD: return "client_request_forward";
	case CEPH_MSG_CLIENT_REPLY: return "client_reply";
	case CEPH_MSG_CLIENT_CAPS: return "client_caps";
	case CEPH_MSG_CLIENT_CAPRELEASE: return "client_cap_release";
	case CEPH_MSG_CLIENT_SNAP: return "client_snap";
	case CEPH_MSG_CLIENT_LEASE: return "client_lease";
	case CEPH_MSG_OSD_MAP: return "osd_map";
	case CEPH_MSG_OSD_OP: return "osd_op";
	case CEPH_MSG_OSD_OPREPLY: return "osd_opreply";
	case CEPH_MSG_WATCH_NOTIFY: return "watch_notify";
	default: return "unknown";
	}
}
EXPORT_SYMBOL(ceph_msg_type_name);

/*
 * Initially learn our fsid, or verify an fsid matches.
 */
int ceph_check_fsid(struct ceph_client *client, struct ceph_fsid *fsid)
{
	if (client->have_fsid) {
		if (ceph_fsid_compare(&client->fsid, fsid)) {
			pr_err("bad fsid, had %pU got %pU",
			       &client->fsid, fsid);
			return -1;
		}
	} else {
		memcpy(&client->fsid, fsid, sizeof(*fsid));
	}
	return 0;
}
EXPORT_SYMBOL(ceph_check_fsid);

static int strcmp_null(const char *s1, const char *s2)
{
	if (!s1 && !s2)
		return 0;
	if (s1 && !s2)
		return -1;
	if (!s1 && s2)
		return 1;
	return strcmp(s1, s2);
}

int ceph_compare_options(struct ceph_options *new_opt,
			 struct ceph_client *client)
{
	struct ceph_options *opt1 = new_opt;
	struct ceph_options *opt2 = client->options;
	int ofs = offsetof(struct ceph_options, mon_addr);
	int i;
	int ret;

	ret = memcmp(opt1, opt2, ofs);
	if (ret)
		return ret;

	ret = strcmp_null(opt1->name, opt2->name);
	if (ret)
		return ret;

	if (opt1->key && !opt2->key)
		return -1;
	if (!opt1->key && opt2->key)
		return 1;
	if (opt1->key && opt2->key) {
		if (opt1->key->type != opt2->key->type)
			return -1;
		if (opt1->key->created.tv_sec != opt2->key->created.tv_sec)
			return -1;
		if (opt1->key->created.tv_nsec != opt2->key->created.tv_nsec)
			return -1;
		if (opt1->key->len != opt2->key->len)
			return -1;
		if (opt1->key->key && !opt2->key->key)
			return -1;
		if (!opt1->key->key && opt2->key->key)
			return 1;
		if (opt1->key->key && opt2->key->key) {
			ret = memcmp(opt1->key->key, opt2->key->key, opt1->key->len);
			if (ret)
				return ret;
		}
	}

	/* any matching mon ip implies a match */
	for (i = 0; i < opt1->num_mon; i++) {
		if (ceph_monmap_contains(client->monc.monmap,
				 &opt1->mon_addr[i]))
			return 0;
	}
	return -1;
}
EXPORT_SYMBOL(ceph_compare_options);

void *ceph_kvmalloc(size_t size, gfp_t flags)
{
	if (size <= (PAGE_SIZE << PAGE_ALLOC_COSTLY_ORDER)) {
		void *ptr = kmalloc(size, flags | __GFP_NOWARN);
		if (ptr)
			return ptr;
	}

	return __vmalloc(size, flags | __GFP_HIGHMEM, PAGE_KERNEL);
}

void ceph_kvfree(const void *ptr)
{
	if (is_vmalloc_addr(ptr))
		vfree(ptr);
	else
		kfree(ptr);
}


static int parse_fsid(const char *str, struct ceph_fsid *fsid)
{
	int i = 0;
	char tmp[3];
	int err = -EINVAL;
	int d;

	dout("parse_fsid '%s'\n", str);
	tmp[2] = 0;
	while (*str && i < 16) {
		if (ispunct(*str)) {
			str++;
			continue;
		}
		if (!isxdigit(str[0]) || !isxdigit(str[1]))
			break;
		tmp[0] = str[0];
		tmp[1] = str[1];
		if (sscanf(tmp, "%x", &d) < 1)
			break;
		fsid->fsid[i] = d & 0xff;
		i++;
		str += 2;
	}

	if (i == 16)
		err = 0;
	dout("parse_fsid ret %d got fsid %pU", err, fsid);
	return err;
}

/*
 * ceph options
 */
enum {
	Opt_osdtimeout,
	Opt_osdkeepalivetimeout,
	Opt_mount_timeout,
	Opt_osd_idle_ttl,
	Opt_last_int,
	/* int args above */
	Opt_fsid,
	Opt_name,
	Opt_secret,
	Opt_key,
	Opt_ip,
	Opt_last_string,
	/* string args above */
	Opt_share,
	Opt_noshare,
	Opt_crc,
	Opt_nocrc,
};

static match_table_t opt_tokens = {
	{Opt_osdtimeout, "osdtimeout=%d"},
	{Opt_osdkeepalivetimeout, "osdkeepalive=%d"},
	{Opt_mount_timeout, "mount_timeout=%d"},
	{Opt_osd_idle_ttl, "osd_idle_ttl=%d"},
	/* int args above */
	{Opt_fsid, "fsid=%s"},
	{Opt_name, "name=%s"},
	{Opt_secret, "secret=%s"},
	{Opt_key, "key=%s"},
	{Opt_ip, "ip=%s"},
	/* string args above */
	{Opt_share, "share"},
	{Opt_noshare, "noshare"},
	{Opt_crc, "crc"},
	{Opt_nocrc, "nocrc"},
	{-1, NULL}
};

void ceph_destroy_options(struct ceph_options *opt)
{
	dout("destroy_options %p\n", opt);
	kfree(opt->name);
	if (opt->key) {
		ceph_crypto_key_destroy(opt->key);
		kfree(opt->key);
	}
	kfree(opt->mon_addr);
	kfree(opt);
}
EXPORT_SYMBOL(ceph_destroy_options);

/* get secret from key store */
static int get_secret(struct ceph_crypto_key *dst, const char *name) {
	struct key *ukey;
	int key_err;
	int err = 0;
	struct ceph_crypto_key *ckey;

	ukey = request_key(&key_type_ceph, name, NULL);
	if (!ukey || IS_ERR(ukey)) {
		/* request_key errors don't map nicely to mount(2)
		   errors; don't even try, but still printk */
		key_err = PTR_ERR(ukey);
		switch (key_err) {
		case -ENOKEY:
			pr_warn("ceph: Mount failed due to key not found: %s\n",
				name);
			break;
		case -EKEYEXPIRED:
			pr_warn("ceph: Mount failed due to expired key: %s\n",
				name);
			break;
		case -EKEYREVOKED:
			pr_warn("ceph: Mount failed due to revoked key: %s\n",
				name);
			break;
		default:
			pr_warn("ceph: Mount failed due to unknown key error %d: %s\n",
				key_err, name);
		}
		err = -EPERM;
		goto out;
	}

	ckey = ukey->payload.data;
	err = ceph_crypto_key_clone(dst, ckey);
	if (err)
		goto out_key;
	/* pass through, err is 0 */

out_key:
	key_put(ukey);
out:
	return err;
}

struct ceph_options *
ceph_parse_options(char *options, const char *dev_name,
			const char *dev_name_end,
			int (*parse_extra_token)(char *c, void *private),
			void *private)
{
	struct ceph_options *opt;
	const char *c;
	int err = -ENOMEM;
	substring_t argstr[MAX_OPT_ARGS];

	if (current->nsproxy->net_ns != &init_net)
		return ERR_PTR(-EINVAL);

	opt = kzalloc(sizeof(*opt), GFP_KERNEL);
	if (!opt)
		return ERR_PTR(-ENOMEM);
	opt->mon_addr = kcalloc(CEPH_MAX_MON, sizeof(*opt->mon_addr),
				GFP_KERNEL);
	if (!opt->mon_addr)
		goto out;

	dout("parse_options %p options '%s' dev_name '%s'\n", opt, options,
	     dev_name);

	/* start with defaults */
	opt->flags = CEPH_OPT_DEFAULT;
	opt->osd_keepalive_timeout = CEPH_OSD_KEEPALIVE_DEFAULT;
	opt->mount_timeout = CEPH_MOUNT_TIMEOUT_DEFAULT; /* seconds */
	opt->osd_idle_ttl = CEPH_OSD_IDLE_TTL_DEFAULT;   /* seconds */

	/* get mon ip(s) */
	/* ip1[:port1][,ip2[:port2]...] */
	err = ceph_parse_ips(dev_name, dev_name_end, opt->mon_addr,
			     CEPH_MAX_MON, &opt->num_mon);
	if (err < 0)
		goto out;

	/* parse mount options */
	while ((c = strsep(&options, ",")) != NULL) {
		int token, intval, ret;
		if (!*c)
			continue;
		err = -EINVAL;
		token = match_token((char *)c, opt_tokens, argstr);
		if (token < 0 && parse_extra_token) {
			/* extra? */
			err = parse_extra_token((char *)c, private);
			if (err < 0) {
				pr_err("bad option at '%s'\n", c);
				goto out;
			}
			continue;
		}
		if (token < Opt_last_int) {
			ret = match_int(&argstr[0], &intval);
			if (ret < 0) {
				pr_err("bad mount option arg (not int) "
				       "at '%s'\n", c);
				continue;
			}
			dout("got int token %d val %d\n", token, intval);
		} else if (token > Opt_last_int && token < Opt_last_string) {
			dout("got string token %d val %s\n", token,
			     argstr[0].from);
		} else {
			dout("got token %d\n", token);
		}
		switch (token) {
		case Opt_ip:
			err = ceph_parse_ips(argstr[0].from,
					     argstr[0].to,
					     &opt->my_addr,
					     1, NULL);
			if (err < 0)
				goto out;
			opt->flags |= CEPH_OPT_MYIP;
			break;

		case Opt_fsid:
			err = parse_fsid(argstr[0].from, &opt->fsid);
			if (err == 0)
				opt->flags |= CEPH_OPT_FSID;
			break;
		case Opt_name:
			opt->name = kstrndup(argstr[0].from,
					      argstr[0].to-argstr[0].from,
					      GFP_KERNEL);
			break;
		case Opt_secret:
		        opt->key = kzalloc(sizeof(*opt->key), GFP_KERNEL);
			if (!opt->key) {
				err = -ENOMEM;
				goto out;
			}
			err = ceph_crypto_key_unarmor(opt->key, argstr[0].from);
			if (err < 0)
				goto out;
			break;
		case Opt_key:
		        opt->key = kzalloc(sizeof(*opt->key), GFP_KERNEL);
			if (!opt->key) {
				err = -ENOMEM;
				goto out;
			}
			err = get_secret(opt->key, argstr[0].from);
			if (err < 0)
				goto out;
			break;

			/* misc */
		case Opt_osdtimeout:
			pr_warn("ignoring deprecated osdtimeout option\n");
			break;
		case Opt_osdkeepalivetimeout:
			opt->osd_keepalive_timeout = intval;
			break;
		case Opt_osd_idle_ttl:
			opt->osd_idle_ttl = intval;
			break;
		case Opt_mount_timeout:
			opt->mount_timeout = intval;
			break;

		case Opt_share:
			opt->flags &= ~CEPH_OPT_NOSHARE;
			break;
		case Opt_noshare:
			opt->flags |= CEPH_OPT_NOSHARE;
			break;

		case Opt_crc:
			opt->flags &= ~CEPH_OPT_NOCRC;
			break;
		case Opt_nocrc:
			opt->flags |= CEPH_OPT_NOCRC;
			break;

		default:
			BUG_ON(token);
		}
	}

	/* success */
	return opt;

out:
	ceph_destroy_options(opt);
	return ERR_PTR(err);
}
EXPORT_SYMBOL(ceph_parse_options);

u64 ceph_client_id(struct ceph_client *client)
{
	return client->monc.auth->global_id;
}
EXPORT_SYMBOL(ceph_client_id);

/*
 * create a fresh client instance
 */
struct ceph_client *ceph_create_client(struct ceph_options *opt, void *private,
				       u64 supported_features,
				       u64 required_features)
{
	struct ceph_client *client;
	struct ceph_entity_addr *myaddr = NULL;
	int err = -ENOMEM;

	client = kzalloc(sizeof(*client), GFP_KERNEL);
	if (client == NULL)
		return ERR_PTR(-ENOMEM);

	client->private = private;
	client->options = opt;

	mutex_init(&client->mount_mutex);
	init_waitqueue_head(&client->auth_wq);
	client->auth_err = 0;

	client->extra_mon_dispatch = NULL;
	client->supported_features = CEPH_FEATURES_SUPPORTED_DEFAULT |
		supported_features;
	client->required_features = CEPH_FEATURES_REQUIRED_DEFAULT |
		required_features;

	/* msgr */
	if (ceph_test_opt(client, MYIP))
		myaddr = &client->options->my_addr;
	ceph_messenger_init(&client->msgr, myaddr,
		client->supported_features,
		client->required_features,
		ceph_test_opt(client, NOCRC));

	/* subsystems */
	err = ceph_monc_init(&client->monc, client);
	if (err < 0)
		goto fail;
	err = ceph_osdc_init(&client->osdc, client);
	if (err < 0)
		goto fail_monc;

	return client;

fail_monc:
	ceph_monc_stop(&client->monc);
fail:
	kfree(client);
	return ERR_PTR(err);
}
EXPORT_SYMBOL(ceph_create_client);

void ceph_destroy_client(struct ceph_client *client)
{
	dout("destroy_client %p\n", client);

	atomic_set(&client->msgr.stopping, 1);

	/* unmount */
	ceph_osdc_stop(&client->osdc);

	ceph_monc_stop(&client->monc);

	ceph_debugfs_client_cleanup(client);

	ceph_destroy_options(client->options);

	kfree(client);
	dout("destroy_client %p done\n", client);
}
EXPORT_SYMBOL(ceph_destroy_client);

/*
 * true if we have the mon map (and have thus joined the cluster)
 */
static int have_mon_and_osd_map(struct ceph_client *client)
{
	return client->monc.monmap && client->monc.monmap->epoch &&
	       client->osdc.osdmap && client->osdc.osdmap->epoch;
}

/*
 * mount: join the ceph cluster, and open root directory.
 */
int __ceph_open_session(struct ceph_client *client, unsigned long started)
{
	int err;
	unsigned long timeout = client->options->mount_timeout * HZ;

	/* open session, and wait for mon and osd maps */
	err = ceph_monc_open_session(&client->monc);
	if (err < 0)
		return err;

	while (!have_mon_and_osd_map(client)) {
		err = -EIO;
		if (timeout && time_after_eq(jiffies, started + timeout))
			return err;

		/* wait */
		dout("mount waiting for mon_map\n");
		err = wait_event_interruptible_timeout(client->auth_wq,
			have_mon_and_osd_map(client) || (client->auth_err < 0),
			timeout);
		if (err == -EINTR || err == -ERESTARTSYS)
			return err;
		if (client->auth_err < 0)
			return client->auth_err;
	}

	return 0;
}
EXPORT_SYMBOL(__ceph_open_session);


int ceph_open_session(struct ceph_client *client)
{
	int ret;
	unsigned long started = jiffies;  /* note the start time */

	dout("open_session start\n");
	mutex_lock(&client->mount_mutex);

	ret = __ceph_open_session(client, started);

	mutex_unlock(&client->mount_mutex);
	return ret;
}
EXPORT_SYMBOL(ceph_open_session);


static int __init init_ceph_lib(void)
{
	int ret = 0;

	ret = ceph_debugfs_init();
	if (ret < 0)
		goto out;

	ret = ceph_crypto_init();
	if (ret < 0)
		goto out_debugfs;

	ret = ceph_msgr_init();
	if (ret < 0)
		goto out_crypto;

	ret = ceph_osdc_setup();
	if (ret < 0)
		goto out_msgr;

	pr_info("loaded (mon/osd proto %d/%d)\n",
		CEPH_MONC_PROTOCOL, CEPH_OSDC_PROTOCOL);

	return 0;

out_msgr:
	ceph_msgr_exit();
out_crypto:
	ceph_crypto_shutdown();
out_debugfs:
	ceph_debugfs_cleanup();
out:
	return ret;
}

static void __exit exit_ceph_lib(void)
{
	dout("exit_ceph_lib\n");
	ceph_osdc_cleanup();
	ceph_msgr_exit();
	ceph_crypto_shutdown();
	ceph_debugfs_cleanup();
}

module_init(init_ceph_lib);
module_exit(exit_ceph_lib);

MODULE_AUTHOR("Sage Weil <sage@newdream.net>");
MODULE_AUTHOR("Yehuda Sadeh <yehuda@hq.newdream.net>");
MODULE_AUTHOR("Patience Warnick <patience@newdream.net>");
MODULE_DESCRIPTION("Ceph filesystem for Linux");
MODULE_LICENSE("GPL");
Commit	Line	Data
3d14c5d2 YS	1
	2	#include <linux/ceph/ceph_debug.h>
	3	#include <linux/backing-dev.h>
	4	#include <linux/ctype.h>
	5	#include <linux/fs.h>
	6	#include <linux/inet.h>
	7	#include <linux/in6.h>
e2c3d29b	8	#include <linux/key.h>
4b2a58ab	9	#include <keys/ceph-type.h>
3d14c5d2 YS	10	#include <linux/module.h>
	11	#include <linux/mount.h>
	12	#include <linux/parser.h>
	13	#include <linux/sched.h>
	14	#include <linux/seq_file.h>
	15	#include <linux/slab.h>
	16	#include <linux/statfs.h>
	17	#include <linux/string.h>
eeb0bed5	18	#include <linux/vmalloc.h>
eea553c2 EB	19	#include <linux/nsproxy.h>
eea553c2 EB	20	#include <net/net_namespace.h>
3d14c5d2 YS	21
3d14c5d2 YS	22
1fe60e51	23	#include <linux/ceph/ceph_features.h>
3d14c5d2 YS	24	#include <linux/ceph/libceph.h>
	25	#include <linux/ceph/debugfs.h>
	26	#include <linux/ceph/decode.h>
	27	#include <linux/ceph/mon_client.h>
	28	#include <linux/ceph/auth.h>
8323c3aa	29	#include "crypto.h"
3d14c5d2 YS	30
3d14c5d2 YS	31
72fe25e3 AE	32	/*
	33	* Module compatibility interface. For now it doesn't do anything,
	34	* but its existence signals a certain level of functionality.
	35	*
	36	* The data buffer is used to pass information both to and from
	37	* libceph. The return value indicates whether libceph determines
	38	* it is compatible with the caller (from another kernel module),
	39	* given the provided data.
	40	*
	41	* The data pointer can be null.
	42	*/
	43	bool libceph_compatible(void *data)
	44	{
1e32d34c	45	return true;
72fe25e3 AE	46	}
72fe25e3 AE	47	EXPORT_SYMBOL(libceph_compatible);
3d14c5d2 YS	48
	49	/*
	50	* find filename portion of a path (/foo/bar/baz -> baz)
	51	*/
	52	const char ceph_file_part(const char s, int len)
	53	{
	54	const char *e = s + len;
	55
	56	while (e != s && *(e-1) != '/')
	57	e--;
	58	return e;
	59	}
	60	EXPORT_SYMBOL(ceph_file_part);
	61
	62	const char *ceph_msg_type_name(int type)
	63	{
	64	switch (type) {
	65	case CEPH_MSG_SHUTDOWN: return "shutdown";
	66	case CEPH_MSG_PING: return "ping";
	67	case CEPH_MSG_AUTH: return "auth";
	68	case CEPH_MSG_AUTH_REPLY: return "auth_reply";
	69	case CEPH_MSG_MON_MAP: return "mon_map";
	70	case CEPH_MSG_MON_GET_MAP: return "mon_get_map";
	71	case CEPH_MSG_MON_SUBSCRIBE: return "mon_subscribe";
	72	case CEPH_MSG_MON_SUBSCRIBE_ACK: return "mon_subscribe_ack";
	73	case CEPH_MSG_STATFS: return "statfs";
	74	case CEPH_MSG_STATFS_REPLY: return "statfs_reply";
513a8243 ID	75	case CEPH_MSG_MON_GET_VERSION: return "mon_get_version";
513a8243 ID	76	case CEPH_MSG_MON_GET_VERSION_REPLY: return "mon_get_version_reply";
3d14c5d2 YS	77	case CEPH_MSG_MDS_MAP: return "mds_map";
	78	case CEPH_MSG_CLIENT_SESSION: return "client_session";
	79	case CEPH_MSG_CLIENT_RECONNECT: return "client_reconnect";
	80	case CEPH_MSG_CLIENT_REQUEST: return "client_request";
	81	case CEPH_MSG_CLIENT_REQUEST_FORWARD: return "client_request_forward";
	82	case CEPH_MSG_CLIENT_REPLY: return "client_reply";
	83	case CEPH_MSG_CLIENT_CAPS: return "client_caps";
	84	case CEPH_MSG_CLIENT_CAPRELEASE: return "client_cap_release";
	85	case CEPH_MSG_CLIENT_SNAP: return "client_snap";
	86	case CEPH_MSG_CLIENT_LEASE: return "client_lease";
	87	case CEPH_MSG_OSD_MAP: return "osd_map";
	88	case CEPH_MSG_OSD_OP: return "osd_op";
	89	case CEPH_MSG_OSD_OPREPLY: return "osd_opreply";
a40c4f10	90	case CEPH_MSG_WATCH_NOTIFY: return "watch_notify";
3d14c5d2 YS	91	default: return "unknown";
	92	}
	93	}
	94	EXPORT_SYMBOL(ceph_msg_type_name);
	95
	96	/*
	97	* Initially learn our fsid, or verify an fsid matches.
	98	*/
	99	int ceph_check_fsid(struct ceph_client client, struct ceph_fsid fsid)
	100	{
	101	if (client->have_fsid) {
	102	if (ceph_fsid_compare(&client->fsid, fsid)) {
	103	pr_err("bad fsid, had %pU got %pU",
	104	&client->fsid, fsid);
	105	return -1;
	106	}
	107	} else {
3d14c5d2	108	memcpy(&client->fsid, fsid, sizeof(*fsid));
3d14c5d2 YS	109	}
	110	return 0;
	111	}
	112	EXPORT_SYMBOL(ceph_check_fsid);
	113
	114	static int strcmp_null(const char s1, const char s2)
	115	{
	116	if (!s1 && !s2)
	117	return 0;
	118	if (s1 && !s2)
	119	return -1;
	120	if (!s1 && s2)
	121	return 1;
	122	return strcmp(s1, s2);
	123	}
	124
	125	int ceph_compare_options(struct ceph_options *new_opt,
	126	struct ceph_client *client)
	127	{
	128	struct ceph_options *opt1 = new_opt;
	129	struct ceph_options *opt2 = client->options;
	130	int ofs = offsetof(struct ceph_options, mon_addr);
	131	int i;
	132	int ret;
	133
	134	ret = memcmp(opt1, opt2, ofs);
	135	if (ret)
	136	return ret;
	137
	138	ret = strcmp_null(opt1->name, opt2->name);
	139	if (ret)
	140	return ret;
	141
8323c3aa TV	142	if (opt1->key && !opt2->key)
	143	return -1;
	144	if (!opt1->key && opt2->key)
	145	return 1;
	146	if (opt1->key && opt2->key) {
	147	if (opt1->key->type != opt2->key->type)
	148	return -1;
	149	if (opt1->key->created.tv_sec != opt2->key->created.tv_sec)
	150	return -1;
	151	if (opt1->key->created.tv_nsec != opt2->key->created.tv_nsec)
	152	return -1;
	153	if (opt1->key->len != opt2->key->len)
	154	return -1;
	155	if (opt1->key->key && !opt2->key->key)
	156	return -1;
	157	if (!opt1->key->key && opt2->key->key)
	158	return 1;
	159	if (opt1->key->key && opt2->key->key) {
	160	ret = memcmp(opt1->key->key, opt2->key->key, opt1->key->len);
	161	if (ret)
	162	return ret;
	163	}
	164	}
3d14c5d2 YS	165
	166	/* any matching mon ip implies a match */
	167	for (i = 0; i < opt1->num_mon; i++) {
	168	if (ceph_monmap_contains(client->monc.monmap,
	169	&opt1->mon_addr[i]))
	170	return 0;
	171	}
	172	return -1;
	173	}
	174	EXPORT_SYMBOL(ceph_compare_options);
	175
eeb0bed5 ID	176	void *ceph_kvmalloc(size_t size, gfp_t flags)
	177	{
	178	if (size <= (PAGE_SIZE << PAGE_ALLOC_COSTLY_ORDER)) {
	179	void *ptr = kmalloc(size, flags \| __GFP_NOWARN);
	180	if (ptr)
	181	return ptr;
	182	}
	183
	184	return __vmalloc(size, flags \| __GFP_HIGHMEM, PAGE_KERNEL);
	185	}
	186
	187	void ceph_kvfree(const void *ptr)
	188	{
	189	if (is_vmalloc_addr(ptr))
	190	vfree(ptr);
	191	else
	192	kfree(ptr);
	193	}
	194
3d14c5d2 YS	195
	196	static int parse_fsid(const char str, struct ceph_fsid fsid)
	197	{
	198	int i = 0;
	199	char tmp[3];
	200	int err = -EINVAL;
	201	int d;
	202
	203	dout("parse_fsid '%s'\n", str);
	204	tmp[2] = 0;
	205	while (*str && i < 16) {
	206	if (ispunct(*str)) {
	207	str++;
	208	continue;
	209	}
	210	if (!isxdigit(str[0]) \|\| !isxdigit(str[1]))
	211	break;
	212	tmp[0] = str[0];
	213	tmp[1] = str[1];
	214	if (sscanf(tmp, "%x", &d) < 1)
	215	break;
	216	fsid->fsid[i] = d & 0xff;
	217	i++;
	218	str += 2;
	219	}
	220
	221	if (i == 16)
	222	err = 0;
	223	dout("parse_fsid ret %d got fsid %pU", err, fsid);
	224	return err;
	225	}
	226
	227	/*
	228	* ceph options
	229	*/
	230	enum {
	231	Opt_osdtimeout,
	232	Opt_osdkeepalivetimeout,
	233	Opt_mount_timeout,
	234	Opt_osd_idle_ttl,
	235	Opt_last_int,
	236	/* int args above */
	237	Opt_fsid,
	238	Opt_name,
	239	Opt_secret,
e2c3d29b	240	Opt_key,
3d14c5d2 YS	241	Opt_ip,
	242	Opt_last_string,
	243	/* string args above */
cffaba15	244	Opt_share,
3d14c5d2	245	Opt_noshare,
cffaba15	246	Opt_crc,
3d14c5d2 YS	247	Opt_nocrc,
	248	};
	249
	250	static match_table_t opt_tokens = {
	251	{Opt_osdtimeout, "osdtimeout=%d"},
	252	{Opt_osdkeepalivetimeout, "osdkeepalive=%d"},
	253	{Opt_mount_timeout, "mount_timeout=%d"},
	254	{Opt_osd_idle_ttl, "osd_idle_ttl=%d"},
	255	/* int args above */
	256	{Opt_fsid, "fsid=%s"},
	257	{Opt_name, "name=%s"},
	258	{Opt_secret, "secret=%s"},
e2c3d29b	259	{Opt_key, "key=%s"},
3d14c5d2 YS	260	{Opt_ip, "ip=%s"},
3d14c5d2 YS	261	/* string args above */
cffaba15	262	{Opt_share, "share"},
3d14c5d2	263	{Opt_noshare, "noshare"},
cffaba15	264	{Opt_crc, "crc"},
3d14c5d2 YS	265	{Opt_nocrc, "nocrc"},
	266	{-1, NULL}
	267	};
	268
	269	void ceph_destroy_options(struct ceph_options *opt)
	270	{
	271	dout("destroy_options %p\n", opt);
	272	kfree(opt->name);
8323c3aa TV	273	if (opt->key) {
	274	ceph_crypto_key_destroy(opt->key);
	275	kfree(opt->key);
	276	}
1cad7893	277	kfree(opt->mon_addr);
3d14c5d2 YS	278	kfree(opt);
	279	}
	280	EXPORT_SYMBOL(ceph_destroy_options);
	281
e2c3d29b TV	282	/* get secret from key store */
	283	static int get_secret(struct ceph_crypto_key dst, const char name) {
	284	struct key *ukey;
	285	int key_err;
	286	int err = 0;
4b2a58ab	287	struct ceph_crypto_key *ckey;
e2c3d29b	288
4b2a58ab	289	ukey = request_key(&key_type_ceph, name, NULL);
e2c3d29b TV	290	if (!ukey \|\| IS_ERR(ukey)) {
	291	/* request_key errors don't map nicely to mount(2)
	292	errors; don't even try, but still printk */
	293	key_err = PTR_ERR(ukey);
	294	switch (key_err) {
	295	case -ENOKEY:
b9a67899 JP	296	pr_warn("ceph: Mount failed due to key not found: %s\n",
b9a67899 JP	297	name);
e2c3d29b TV	298	break;
e2c3d29b TV	299	case -EKEYEXPIRED:
b9a67899 JP	300	pr_warn("ceph: Mount failed due to expired key: %s\n",
b9a67899 JP	301	name);
e2c3d29b TV	302	break;
e2c3d29b TV	303	case -EKEYREVOKED:
b9a67899 JP	304	pr_warn("ceph: Mount failed due to revoked key: %s\n",
b9a67899 JP	305	name);
e2c3d29b TV	306	break;
e2c3d29b TV	307	default:
b9a67899 JP	308	pr_warn("ceph: Mount failed due to unknown key error %d: %s\n",
b9a67899 JP	309	key_err, name);
e2c3d29b TV	310	}
	311	err = -EPERM;
	312	goto out;
	313	}
	314
4b2a58ab TV	315	ckey = ukey->payload.data;
4b2a58ab TV	316	err = ceph_crypto_key_clone(dst, ckey);
e2c3d29b TV	317	if (err)
	318	goto out_key;
	319	/* pass through, err is 0 */
	320
	321	out_key:
	322	key_put(ukey);
	323	out:
	324	return err;
	325	}
	326
ee57741c AE	327	struct ceph_options *
	328	ceph_parse_options(char options, const char dev_name,
	329	const char *dev_name_end,
	330	int (parse_extra_token)(char c, void *private),
	331	void *private)
3d14c5d2 YS	332	{
	333	struct ceph_options *opt;
	334	const char *c;
	335	int err = -ENOMEM;
	336	substring_t argstr[MAX_OPT_ARGS];
	337
eea553c2 EB	338	if (current->nsproxy->net_ns != &init_net)
	339	return ERR_PTR(-EINVAL);
	340
3d14c5d2 YS	341	opt = kzalloc(sizeof(*opt), GFP_KERNEL);
3d14c5d2 YS	342	if (!opt)
ee57741c	343	return ERR_PTR(-ENOMEM);
3d14c5d2 YS	344	opt->mon_addr = kcalloc(CEPH_MAX_MON, sizeof(*opt->mon_addr),
	345	GFP_KERNEL);
	346	if (!opt->mon_addr)
	347	goto out;
	348
	349	dout("parse_options %p options '%s' dev_name '%s'\n", opt, options,
	350	dev_name);
	351
	352	/* start with defaults */
	353	opt->flags = CEPH_OPT_DEFAULT;
3d14c5d2 YS	354	opt->osd_keepalive_timeout = CEPH_OSD_KEEPALIVE_DEFAULT;
	355	opt->mount_timeout = CEPH_MOUNT_TIMEOUT_DEFAULT; /* seconds */
	356	opt->osd_idle_ttl = CEPH_OSD_IDLE_TTL_DEFAULT; /* seconds */
	357
	358	/* get mon ip(s) */
	359	/* ip1[:port1][,ip2[:port2]...] */
	360	err = ceph_parse_ips(dev_name, dev_name_end, opt->mon_addr,
	361	CEPH_MAX_MON, &opt->num_mon);
	362	if (err < 0)
	363	goto out;
	364
	365	/* parse mount options */
	366	while ((c = strsep(&options, ",")) != NULL) {
	367	int token, intval, ret;
	368	if (!*c)
	369	continue;
	370	err = -EINVAL;
	371	token = match_token((char *)c, opt_tokens, argstr);
010e3b48	372	if (token < 0 && parse_extra_token) {
3d14c5d2 YS	373	/* extra? */
	374	err = parse_extra_token((char *)c, private);
	375	if (err < 0) {
	376	pr_err("bad option at '%s'\n", c);
	377	goto out;
	378	}
	379	continue;
	380	}
	381	if (token < Opt_last_int) {
	382	ret = match_int(&argstr[0], &intval);
	383	if (ret < 0) {
	384	pr_err("bad mount option arg (not int) "
	385	"at '%s'\n", c);
	386	continue;
	387	}
	388	dout("got int token %d val %d\n", token, intval);
	389	} else if (token > Opt_last_int && token < Opt_last_string) {
	390	dout("got string token %d val %s\n", token,
	391	argstr[0].from);
	392	} else {
	393	dout("got token %d\n", token);
	394	}
	395	switch (token) {
	396	case Opt_ip:
	397	err = ceph_parse_ips(argstr[0].from,
	398	argstr[0].to,
	399	&opt->my_addr,
	400	1, NULL);
	401	if (err < 0)
	402	goto out;
	403	opt->flags \|= CEPH_OPT_MYIP;
	404	break;
	405
	406	case Opt_fsid:
	407	err = parse_fsid(argstr[0].from, &opt->fsid);
	408	if (err == 0)
	409	opt->flags \|= CEPH_OPT_FSID;
	410	break;
	411	case Opt_name:
	412	opt->name = kstrndup(argstr[0].from,
	413	argstr[0].to-argstr[0].from,
	414	GFP_KERNEL);
	415	break;
	416	case Opt_secret:
8323c3aa TV	417	opt->key = kzalloc(sizeof(*opt->key), GFP_KERNEL);
	418	if (!opt->key) {
	419	err = -ENOMEM;
	420	goto out;
	421	}
	422	err = ceph_crypto_key_unarmor(opt->key, argstr[0].from);
	423	if (err < 0)
	424	goto out;
3d14c5d2	425	break;
e2c3d29b TV	426	case Opt_key:
	427	opt->key = kzalloc(sizeof(*opt->key), GFP_KERNEL);
	428	if (!opt->key) {
	429	err = -ENOMEM;
	430	goto out;
	431	}
	432	err = get_secret(opt->key, argstr[0].from);
	433	if (err < 0)
	434	goto out;
	435	break;
3d14c5d2 YS	436
	437	/* misc */
	438	case Opt_osdtimeout:
b9a67899	439	pr_warn("ignoring deprecated osdtimeout option\n");
3d14c5d2 YS	440	break;
	441	case Opt_osdkeepalivetimeout:
	442	opt->osd_keepalive_timeout = intval;
	443	break;
	444	case Opt_osd_idle_ttl:
	445	opt->osd_idle_ttl = intval;
	446	break;
	447	case Opt_mount_timeout:
	448	opt->mount_timeout = intval;
	449	break;
	450
cffaba15 AE	451	case Opt_share:
	452	opt->flags &= ~CEPH_OPT_NOSHARE;
	453	break;
3d14c5d2 YS	454	case Opt_noshare:
	455	opt->flags \|= CEPH_OPT_NOSHARE;
	456	break;
	457
cffaba15 AE	458	case Opt_crc:
	459	opt->flags &= ~CEPH_OPT_NOCRC;
	460	break;
3d14c5d2 YS	461	case Opt_nocrc:
	462	opt->flags \|= CEPH_OPT_NOCRC;
	463	break;
	464
	465	default:
	466	BUG_ON(token);
	467	}
	468	}
	469
	470	/* success */
ee57741c	471	return opt;
3d14c5d2 YS	472
	473	out:
	474	ceph_destroy_options(opt);
ee57741c	475	return ERR_PTR(err);
3d14c5d2 YS	476	}
	477	EXPORT_SYMBOL(ceph_parse_options);
	478
	479	u64 ceph_client_id(struct ceph_client *client)
	480	{
	481	return client->monc.auth->global_id;
	482	}
	483	EXPORT_SYMBOL(ceph_client_id);
	484
	485	/*
	486	* create a fresh client instance
	487	*/
6ab00d46	488	struct ceph_client ceph_create_client(struct ceph_options opt, void *private,
12b4629a ID	489	u64 supported_features,
12b4629a ID	490	u64 required_features)
3d14c5d2 YS	491	{
3d14c5d2 YS	492	struct ceph_client *client;
6ab00d46	493	struct ceph_entity_addr *myaddr = NULL;
3d14c5d2 YS	494	int err = -ENOMEM;
	495
	496	client = kzalloc(sizeof(*client), GFP_KERNEL);
	497	if (client == NULL)
	498	return ERR_PTR(-ENOMEM);
	499
	500	client->private = private;
	501	client->options = opt;
	502
	503	mutex_init(&client->mount_mutex);
	504	init_waitqueue_head(&client->auth_wq);
	505	client->auth_err = 0;
	506
	507	client->extra_mon_dispatch = NULL;
1fe60e51	508	client->supported_features = CEPH_FEATURES_SUPPORTED_DEFAULT \|
6ab00d46	509	supported_features;
1fe60e51	510	client->required_features = CEPH_FEATURES_REQUIRED_DEFAULT \|
6ab00d46 SW	511	required_features;
	512
	513	/* msgr */
	514	if (ceph_test_opt(client, MYIP))
	515	myaddr = &client->options->my_addr;
15d9882c AE	516	ceph_messenger_init(&client->msgr, myaddr,
	517	client->supported_features,
	518	client->required_features,
	519	ceph_test_opt(client, NOCRC));
3d14c5d2 YS	520
	521	/* subsystems */
	522	err = ceph_monc_init(&client->monc, client);
	523	if (err < 0)
15d9882c	524	goto fail;
3d14c5d2 YS	525	err = ceph_osdc_init(&client->osdc, client);
	526	if (err < 0)
	527	goto fail_monc;
	528
	529	return client;
	530
	531	fail_monc:
	532	ceph_monc_stop(&client->monc);
	533	fail:
	534	kfree(client);
	535	return ERR_PTR(err);
	536	}
	537	EXPORT_SYMBOL(ceph_create_client);
	538
	539	void ceph_destroy_client(struct ceph_client *client)
	540	{
	541	dout("destroy_client %p\n", client);
	542
a2a32584 GH	543	atomic_set(&client->msgr.stopping, 1);
a2a32584 GH	544
3d14c5d2 YS	545	/* unmount */
	546	ceph_osdc_stop(&client->osdc);
	547
3d14c5d2 YS	548	ceph_monc_stop(&client->monc);
	549
	550	ceph_debugfs_client_cleanup(client);
	551
3d14c5d2 YS	552	ceph_destroy_options(client->options);
	553
	554	kfree(client);
	555	dout("destroy_client %p done\n", client);
	556	}
	557	EXPORT_SYMBOL(ceph_destroy_client);
	558
	559	/*
	560	* true if we have the mon map (and have thus joined the cluster)
	561	*/
	562	static int have_mon_and_osd_map(struct ceph_client *client)
	563	{
	564	return client->monc.monmap && client->monc.monmap->epoch &&
	565	client->osdc.osdmap && client->osdc.osdmap->epoch;
	566	}
	567
	568	/*
	569	* mount: join the ceph cluster, and open root directory.
	570	*/
	571	int __ceph_open_session(struct ceph_client *client, unsigned long started)
	572	{
3d14c5d2 YS	573	int err;
	574	unsigned long timeout = client->options->mount_timeout * HZ;
	575
3d14c5d2 YS	576	/* open session, and wait for mon and osd maps */
	577	err = ceph_monc_open_session(&client->monc);
	578	if (err < 0)
	579	return err;
	580
	581	while (!have_mon_and_osd_map(client)) {
	582	err = -EIO;
	583	if (timeout && time_after_eq(jiffies, started + timeout))
	584	return err;
	585
	586	/* wait */
	587	dout("mount waiting for mon_map\n");
	588	err = wait_event_interruptible_timeout(client->auth_wq,
	589	have_mon_and_osd_map(client) \|\| (client->auth_err < 0),
	590	timeout);
	591	if (err == -EINTR \|\| err == -ERESTARTSYS)
	592	return err;
	593	if (client->auth_err < 0)
	594	return client->auth_err;
	595	}
	596
	597	return 0;
	598	}
	599	EXPORT_SYMBOL(__ceph_open_session);
	600
	601
	602	int ceph_open_session(struct ceph_client *client)
	603	{
	604	int ret;
	605	unsigned long started = jiffies; /* note the start time */
	606
	607	dout("open_session start\n");
	608	mutex_lock(&client->mount_mutex);
	609
	610	ret = __ceph_open_session(client, started);
	611
	612	mutex_unlock(&client->mount_mutex);
	613	return ret;
	614	}
	615	EXPORT_SYMBOL(ceph_open_session);
	616
	617
	618	static int __init init_ceph_lib(void)
	619	{
	620	int ret = 0;
	621
	622	ret = ceph_debugfs_init();
	623	if (ret < 0)
	624	goto out;
	625
4b2a58ab	626	ret = ceph_crypto_init();
3d14c5d2 YS	627	if (ret < 0)
	628	goto out_debugfs;
	629
4b2a58ab TV	630	ret = ceph_msgr_init();
	631	if (ret < 0)
	632	goto out_crypto;
	633
5522ae0b AE	634	ret = ceph_osdc_setup();
	635	if (ret < 0)
	636	goto out_msgr;
	637
4f6a7e5e SW	638	pr_info("loaded (mon/osd proto %d/%d)\n",
4f6a7e5e SW	639	CEPH_MONC_PROTOCOL, CEPH_OSDC_PROTOCOL);
3d14c5d2 YS	640
	641	return 0;
	642
5522ae0b AE	643	out_msgr:
5522ae0b AE	644	ceph_msgr_exit();
4b2a58ab TV	645	out_crypto:
4b2a58ab TV	646	ceph_crypto_shutdown();
3d14c5d2 YS	647	out_debugfs:
	648	ceph_debugfs_cleanup();
	649	out:
	650	return ret;
	651	}
	652
	653	static void __exit exit_ceph_lib(void)
	654	{
	655	dout("exit_ceph_lib\n");
5522ae0b	656	ceph_osdc_cleanup();
3d14c5d2	657	ceph_msgr_exit();
4b2a58ab	658	ceph_crypto_shutdown();
3d14c5d2 YS	659	ceph_debugfs_cleanup();
	660	}
	661
	662	module_init(init_ceph_lib);
	663	module_exit(exit_ceph_lib);
	664
	665	MODULE_AUTHOR("Sage Weil <sage@newdream.net>");
	666	MODULE_AUTHOR("Yehuda Sadeh <yehuda@hq.newdream.net>");
	667	MODULE_AUTHOR("Patience Warnick <patience@newdream.net>");
	668	MODULE_DESCRIPTION("Ceph filesystem for Linux");
	669	MODULE_LICENSE("GPL");