mon.c 14.0 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11
/*
 * linux/fs/lockd/mon.c
 *
 * The kernel statd client.
 *
 * Copyright (C) 1996, Olaf Kirch <okir@monad.swb.de>
 */

#include <linux/types.h>
#include <linux/utsname.h>
#include <linux/kernel.h>
12
#include <linux/ktime.h>
13
#include <linux/slab.h>
14

L
Linus Torvalds 已提交
15
#include <linux/sunrpc/clnt.h>
16
#include <linux/sunrpc/xprtsock.h>
L
Linus Torvalds 已提交
17 18 19
#include <linux/sunrpc/svc.h>
#include <linux/lockd/lockd.h>

20 21
#include <asm/unaligned.h>

L
Linus Torvalds 已提交
22
#define NLMDBG_FACILITY		NLMDBG_MONITOR
23 24 25 26 27 28 29 30 31 32 33 34
#define NSM_PROGRAM		100024
#define NSM_VERSION		1

enum {
	NSMPROC_NULL,
	NSMPROC_STAT,
	NSMPROC_MON,
	NSMPROC_UNMON,
	NSMPROC_UNMON_ALL,
	NSMPROC_SIMU_CRASH,
	NSMPROC_NOTIFY,
};
L
Linus Torvalds 已提交
35

36
struct nsm_args {
37
	struct nsm_private	*priv;
38 39 40 41 42 43 44 45 46 47 48 49
	u32			prog;		/* RPC callback info */
	u32			vers;
	u32			proc;

	char			*mon_name;
};

struct nsm_res {
	u32			status;
	u32			state;
};

L
Linus Torvalds 已提交
50
static struct rpc_program	nsm_program;
51 52
static				LIST_HEAD(nsm_handles);
static				DEFINE_SPINLOCK(nsm_lock);
L
Linus Torvalds 已提交
53 54 55 56

/*
 * Local NSM state
 */
57
u32	__read_mostly		nsm_local_state;
58
int	__read_mostly		nsm_use_hostnames;
L
Linus Torvalds 已提交
59

60 61 62 63 64
static inline struct sockaddr *nsm_addr(const struct nsm_handle *nsm)
{
	return (struct sockaddr *)&nsm->sm_addr;
}

C
Chuck Lever 已提交
65 66 67 68 69 70 71
static struct rpc_clnt *nsm_create(void)
{
	struct sockaddr_in sin = {
		.sin_family		= AF_INET,
		.sin_addr.s_addr	= htonl(INADDR_LOOPBACK),
	};
	struct rpc_create_args args = {
72
		.net			= &init_net,
C
Chuck Lever 已提交
73 74 75 76 77 78 79
		.protocol		= XPRT_TRANSPORT_UDP,
		.address		= (struct sockaddr *)&sin,
		.addrsize		= sizeof(sin),
		.servername		= "rpc.statd",
		.program		= &nsm_program,
		.version		= NSM_VERSION,
		.authflavor		= RPC_AUTH_NULL,
80
		.flags			= RPC_CLNT_CREATE_NOPING,
C
Chuck Lever 已提交
81 82 83 84 85 86
	};

	return rpc_create(&args);
}

static int nsm_mon_unmon(struct nsm_handle *nsm, u32 proc, struct nsm_res *res)
L
Linus Torvalds 已提交
87 88 89
{
	struct rpc_clnt	*clnt;
	int		status;
90
	struct nsm_args args = {
91
		.priv		= &nsm->sm_priv,
92 93 94
		.prog		= NLM_PROGRAM,
		.vers		= 3,
		.proc		= NLMPROC_NSM_NOTIFY,
95
		.mon_name	= nsm->sm_mon_name,
96
	};
97 98 99 100
	struct rpc_message msg = {
		.rpc_argp	= &args,
		.rpc_resp	= res,
	};
L
Linus Torvalds 已提交
101 102 103 104

	clnt = nsm_create();
	if (IS_ERR(clnt)) {
		status = PTR_ERR(clnt);
105 106
		dprintk("lockd: failed to create NSM upcall transport, "
				"status=%d\n", status);
L
Linus Torvalds 已提交
107 108 109 110 111
		goto out;
	}

	memset(res, 0, sizeof(*res));

112 113
	msg.rpc_proc = &clnt->cl_procinfo[proc];
	status = rpc_call_sync(clnt, &msg, 0);
L
Linus Torvalds 已提交
114
	if (status < 0)
115 116
		dprintk("lockd: NSM upcall RPC failed, status=%d\n",
				status);
L
Linus Torvalds 已提交
117 118
	else
		status = 0;
119
	rpc_shutdown_client(clnt);
L
Linus Torvalds 已提交
120 121 122 123
 out:
	return status;
}

124 125 126 127 128 129 130 131 132 133
/**
 * nsm_monitor - Notify a peer in case we reboot
 * @host: pointer to nlm_host of peer to notify
 *
 * If this peer is not already monitored, this function sends an
 * upcall to the local rpc.statd to record the name/address of
 * the peer to notify in case we reboot.
 *
 * Returns zero if the peer is monitored by the local rpc.statd;
 * otherwise a negative errno value is returned.
L
Linus Torvalds 已提交
134
 */
135
int nsm_monitor(const struct nlm_host *host)
L
Linus Torvalds 已提交
136
{
137
	struct nsm_handle *nsm = host->h_nsmhandle;
L
Linus Torvalds 已提交
138 139 140
	struct nsm_res	res;
	int		status;

141
	dprintk("lockd: nsm_monitor(%s)\n", nsm->sm_name);
142 143

	if (nsm->sm_monitored)
144
		return 0;
L
Linus Torvalds 已提交
145

146 147 148 149 150 151
	/*
	 * Choose whether to record the caller_name or IP address of
	 * this peer in the local rpc.statd's database.
	 */
	nsm->sm_mon_name = nsm_use_hostnames ? nsm->sm_name : nsm->sm_addrbuf;

152
	status = nsm_mon_unmon(nsm, NSMPROC_MON, &res);
153
	if (unlikely(res.status != 0))
154
		status = -EIO;
155
	if (unlikely(status < 0)) {
156
		printk(KERN_NOTICE "lockd: cannot monitor %s\n", nsm->sm_name);
157 158 159 160 161 162 163 164 165
		return status;
	}

	nsm->sm_monitored = 1;
	if (unlikely(nsm_local_state != res.state)) {
		nsm_local_state = res.state;
		dprintk("lockd: NSM state changed to %d\n", nsm_local_state);
	}
	return 0;
L
Linus Torvalds 已提交
166 167
}

168 169 170 171 172 173 174
/**
 * nsm_unmonitor - Unregister peer notification
 * @host: pointer to nlm_host of peer to stop monitoring
 *
 * If this peer is monitored, this function sends an upcall to
 * tell the local rpc.statd not to send this peer a notification
 * when we reboot.
L
Linus Torvalds 已提交
175
 */
176
void nsm_unmonitor(const struct nlm_host *host)
L
Linus Torvalds 已提交
177
{
178
	struct nsm_handle *nsm = host->h_nsmhandle;
L
Linus Torvalds 已提交
179
	struct nsm_res	res;
180
	int status;
L
Linus Torvalds 已提交
181

182 183
	if (atomic_read(&nsm->sm_count) == 1
	 && nsm->sm_monitored && !nsm->sm_sticky) {
184
		dprintk("lockd: nsm_unmonitor(%s)\n", nsm->sm_name);
185

186
		status = nsm_mon_unmon(nsm, NSMPROC_UNMON, &res);
187 188
		if (res.status != 0)
			status = -EIO;
189
		if (status < 0)
190
			printk(KERN_NOTICE "lockd: cannot unmonitor %s\n",
191
					nsm->sm_name);
192 193
		else
			nsm->sm_monitored = 0;
194
	}
L
Linus Torvalds 已提交
195 196
}

197 198 199 200 201 202 203 204 205 206 207 208
static struct nsm_handle *nsm_lookup_hostname(const char *hostname,
					      const size_t len)
{
	struct nsm_handle *nsm;

	list_for_each_entry(nsm, &nsm_handles, sm_link)
		if (strlen(nsm->sm_name) == len &&
		    memcmp(nsm->sm_name, hostname, len) == 0)
			return nsm;
	return NULL;
}

209 210 211 212 213
static struct nsm_handle *nsm_lookup_addr(const struct sockaddr *sap)
{
	struct nsm_handle *nsm;

	list_for_each_entry(nsm, &nsm_handles, sm_link)
214
		if (rpc_cmp_addr(nsm_addr(nsm), sap))
215 216 217 218
			return nsm;
	return NULL;
}

219 220 221 222 223 224 225 226 227 228 229
static struct nsm_handle *nsm_lookup_priv(const struct nsm_private *priv)
{
	struct nsm_handle *nsm;

	list_for_each_entry(nsm, &nsm_handles, sm_link)
		if (memcmp(nsm->sm_priv.data, priv->data,
					sizeof(priv->data)) == 0)
			return nsm;
	return NULL;
}

230 231 232 233 234 235
/*
 * Construct a unique cookie to match this nsm_handle to this monitored
 * host.  It is passed to the local rpc.statd via NSMPROC_MON, and
 * returned via NLMPROC_SM_NOTIFY, in the "priv" field of these
 * requests.
 *
236 237 238 239 240 241 242 243 244 245
 * The NSM protocol requires that these cookies be unique while the
 * system is running.  We prefer a stronger requirement of making them
 * unique across reboots.  If user space bugs cause a stale cookie to
 * be sent to the kernel, it could cause the wrong host to lose its
 * lock state if cookies were not unique across reboots.
 *
 * The cookies are exposed only to local user space via loopback.  They
 * do not appear on the physical network.  If we want greater security
 * for some reason, nsm_init_private() could perform a one-way hash to
 * obscure the contents of the cookie.
246 247 248
 */
static void nsm_init_private(struct nsm_handle *nsm)
{
249 250
	u64 *p = (u64 *)&nsm->sm_priv.data;
	struct timespec ts;
251
	s64 ns;
252 253

	ktime_get_ts(&ts);
254 255 256
	ns = timespec_to_ns(&ts);
	put_unaligned(ns, p);
	put_unaligned((unsigned long)nsm, p + 1);
257 258
}

259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274
static struct nsm_handle *nsm_create_handle(const struct sockaddr *sap,
					    const size_t salen,
					    const char *hostname,
					    const size_t hostname_len)
{
	struct nsm_handle *new;

	new = kzalloc(sizeof(*new) + hostname_len + 1, GFP_KERNEL);
	if (unlikely(new == NULL))
		return NULL;

	atomic_set(&new->sm_count, 1);
	new->sm_name = (char *)(new + 1);
	memcpy(nsm_addr(new), sap, salen);
	new->sm_addrlen = salen;
	nsm_init_private(new);
275 276 277 278 279

	if (rpc_ntop(nsm_addr(new), new->sm_addrbuf,
					sizeof(new->sm_addrbuf)) == 0)
		(void)snprintf(new->sm_addrbuf, sizeof(new->sm_addrbuf),
				"unsupported address family");
280 281 282 283 284 285
	memcpy(new->sm_name, hostname, hostname_len);
	new->sm_name[hostname_len] = '\0';

	return new;
}

286
/**
287
 * nsm_get_handle - Find or create a cached nsm_handle
288 289 290 291 292
 * @sap: pointer to socket address of handle to find
 * @salen: length of socket address
 * @hostname: pointer to C string containing hostname to find
 * @hostname_len: length of C string
 *
293
 * Behavior is modulated by the global nsm_use_hostnames variable.
294
 *
295 296 297 298
 * Returns a cached nsm_handle after bumping its ref count, or
 * returns a fresh nsm_handle if a handle that matches @sap and/or
 * @hostname cannot be found in the handle cache.  Returns NULL if
 * an error occurs.
299
 */
300 301 302
struct nsm_handle *nsm_get_handle(const struct sockaddr *sap,
				  const size_t salen, const char *hostname,
				  const size_t hostname_len)
303
{
304
	struct nsm_handle *cached, *new = NULL;
305 306 307 308 309 310 311 312 313 314 315 316

	if (hostname && memchr(hostname, '/', hostname_len) != NULL) {
		if (printk_ratelimit()) {
			printk(KERN_WARNING "Invalid hostname \"%.*s\" "
					    "in NFS lock request\n",
				(int)hostname_len, hostname);
		}
		return NULL;
	}

retry:
	spin_lock(&nsm_lock);
317 318 319 320 321 322 323 324 325 326 327 328 329 330 331

	if (nsm_use_hostnames && hostname != NULL)
		cached = nsm_lookup_hostname(hostname, hostname_len);
	else
		cached = nsm_lookup_addr(sap);

	if (cached != NULL) {
		atomic_inc(&cached->sm_count);
		spin_unlock(&nsm_lock);
		kfree(new);
		dprintk("lockd: found nsm_handle for %s (%s), "
				"cnt %d\n", cached->sm_name,
				cached->sm_addrbuf,
				atomic_read(&cached->sm_count));
		return cached;
332
	}
333 334 335 336

	if (new != NULL) {
		list_add(&new->sm_link, &nsm_handles);
		spin_unlock(&nsm_lock);
337
		dprintk("lockd: created nsm_handle for %s (%s)\n",
338 339
				new->sm_name, new->sm_addrbuf);
		return new;
340
	}
341

342 343
	spin_unlock(&nsm_lock);

344 345
	new = nsm_create_handle(sap, salen, hostname, hostname_len);
	if (unlikely(new == NULL))
346 347 348 349
		return NULL;
	goto retry;
}

350 351 352 353
/**
 * nsm_reboot_lookup - match NLMPROC_SM_NOTIFY arguments to an nsm_handle
 * @info: pointer to NLMPROC_SM_NOTIFY arguments
 *
354 355 356
 * Returns a matching nsm_handle if found in the nsm cache. The returned
 * nsm_handle's reference count is bumped. Otherwise returns NULL if some
 * error occurred.
357 358 359 360 361 362 363
 */
struct nsm_handle *nsm_reboot_lookup(const struct nlm_reboot *info)
{
	struct nsm_handle *cached;

	spin_lock(&nsm_lock);

364
	cached = nsm_lookup_priv(&info->priv);
365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380
	if (unlikely(cached == NULL)) {
		spin_unlock(&nsm_lock);
		dprintk("lockd: never saw rebooted peer '%.*s' before\n",
				info->len, info->mon);
		return cached;
	}

	atomic_inc(&cached->sm_count);
	spin_unlock(&nsm_lock);

	dprintk("lockd: host %s (%s) rebooted, cnt %d\n",
			cached->sm_name, cached->sm_addrbuf,
			atomic_read(&cached->sm_count));
	return cached;
}

381 382 383 384 385 386 387 388 389 390
/**
 * nsm_release - Release an NSM handle
 * @nsm: pointer to handle to be released
 *
 */
void nsm_release(struct nsm_handle *nsm)
{
	if (atomic_dec_and_lock(&nsm->sm_count, &nsm_lock)) {
		list_del(&nsm->sm_link);
		spin_unlock(&nsm_lock);
391 392
		dprintk("lockd: destroyed nsm_handle for %s (%s)\n",
				nsm->sm_name, nsm->sm_addrbuf);
393 394 395 396
		kfree(nsm);
	}
}

L
Linus Torvalds 已提交
397 398
/*
 * XDR functions for NSM.
399 400 401
 *
 * See http://www.opengroup.org/ for details on the Network
 * Status Monitor wire protocol.
L
Linus Torvalds 已提交
402 403
 */

404
static int encode_nsm_string(struct xdr_stream *xdr, const char *string)
405
{
406 407 408 409 410 411 412 413 414 415
	const u32 len = strlen(string);
	__be32 *p;

	if (unlikely(len > SM_MAXSTRLEN))
		return -EIO;
	p = xdr_reserve_space(xdr, sizeof(u32) + len);
	if (unlikely(p == NULL))
		return -EIO;
	xdr_encode_opaque(p, string, len);
	return 0;
416 417
}

418 419 420
/*
 * "mon_name" specifies the host to be monitored.
 */
421
static int encode_mon_name(struct xdr_stream *xdr, const struct nsm_args *argp)
422
{
423
	return encode_nsm_string(xdr, argp->mon_name);
424 425
}

426 427 428
/*
 * The "my_id" argument specifies the hostname and RPC procedure
 * to be called when the status manager receives notification
429
 * (via the NLMPROC_SM_NOTIFY call) that the state of host "mon_name"
430 431
 * has changed.
 */
432
static int encode_my_id(struct xdr_stream *xdr, const struct nsm_args *argp)
433
{
434 435 436 437 438 439 440 441 442
	int status;
	__be32 *p;

	status = encode_nsm_string(xdr, utsname()->nodename);
	if (unlikely(status != 0))
		return status;
	p = xdr_reserve_space(xdr, 3 * sizeof(u32));
	if (unlikely(p == NULL))
		return -EIO;
443 444 445
	*p++ = htonl(argp->prog);
	*p++ = htonl(argp->vers);
	*p++ = htonl(argp->proc);
446
	return 0;
447 448
}

449 450
/*
 * The "mon_id" argument specifies the non-private arguments
451
 * of an NSMPROC_MON or NSMPROC_UNMON call.
452
 */
453
static int encode_mon_id(struct xdr_stream *xdr, const struct nsm_args *argp)
454
{
455
	int status;
456

457 458 459 460
	status = encode_mon_name(xdr, argp);
	if (unlikely(status != 0))
		return status;
	return encode_my_id(xdr, argp);
461 462
}

463 464
/*
 * The "priv" argument may contain private information required
465 466
 * by the NSMPROC_MON call. This information will be supplied in the
 * NLMPROC_SM_NOTIFY call.
467
 */
468
static int encode_priv(struct xdr_stream *xdr, const struct nsm_args *argp)
469
{
470 471 472 473 474
	__be32 *p;

	p = xdr_reserve_space(xdr, SM_PRIV_SIZE);
	if (unlikely(p == NULL))
		return -EIO;
475
	xdr_encode_opaque_fixed(p, argp->priv->data, SM_PRIV_SIZE);
476
	return 0;
477 478
}

479 480
static int xdr_enc_mon(struct rpc_rqst *req, __be32 *p,
		       const struct nsm_args *argp)
L
Linus Torvalds 已提交
481
{
482 483
	struct xdr_stream xdr;
	int status;
484

485 486 487 488 489
	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
	status = encode_mon_id(&xdr, argp);
	if (unlikely(status))
		return status;
	return encode_priv(&xdr, argp);
L
Linus Torvalds 已提交
490 491
}

492 493
static int xdr_enc_unmon(struct rpc_rqst *req, __be32 *p,
			 const struct nsm_args *argp)
L
Linus Torvalds 已提交
494
{
495 496 497 498
	struct xdr_stream xdr;

	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
	return encode_mon_id(&xdr, argp);
L
Linus Torvalds 已提交
499 500
}

501 502
static int xdr_dec_stat_res(struct rpc_rqst *rqstp, __be32 *p,
			    struct nsm_res *resp)
L
Linus Torvalds 已提交
503
{
504 505 506 507 508 509
	struct xdr_stream xdr;

	xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
	p = xdr_inline_decode(&xdr, 2 * sizeof(u32));
	if (unlikely(p == NULL))
		return -EIO;
L
Linus Torvalds 已提交
510
	resp->status = ntohl(*p++);
511 512 513
	resp->state = ntohl(*p);

	dprintk("lockd: xdr_dec_stat_res status %d state %d\n",
L
Linus Torvalds 已提交
514 515 516 517
			resp->status, resp->state);
	return 0;
}

518 519
static int xdr_dec_stat(struct rpc_rqst *rqstp, __be32 *p,
			struct nsm_res *resp)
L
Linus Torvalds 已提交
520
{
521 522 523 524 525 526 527 528 529
	struct xdr_stream xdr;

	xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
	p = xdr_inline_decode(&xdr, sizeof(u32));
	if (unlikely(p == NULL))
		return -EIO;
	resp->state = ntohl(*p);

	dprintk("lockd: xdr_dec_stat state %d\n", resp->state);
L
Linus Torvalds 已提交
530 531 532 533
	return 0;
}

#define SM_my_name_sz	(1+XDR_QUADLEN(SM_MAXSTRLEN))
534 535 536
#define SM_my_id_sz	(SM_my_name_sz+3)
#define SM_mon_name_sz	(1+XDR_QUADLEN(SM_MAXSTRLEN))
#define SM_mon_id_sz	(SM_mon_name_sz+SM_my_id_sz)
537 538
#define SM_priv_sz	(XDR_QUADLEN(SM_PRIV_SIZE))
#define SM_mon_sz	(SM_mon_id_sz+SM_priv_sz)
L
Linus Torvalds 已提交
539 540 541 542
#define SM_monres_sz	2
#define SM_unmonres_sz	1

static struct rpc_procinfo	nsm_procedures[] = {
543 544
[NSMPROC_MON] = {
		.p_proc		= NSMPROC_MON,
545 546
		.p_encode	= (kxdrproc_t)xdr_enc_mon,
		.p_decode	= (kxdrproc_t)xdr_dec_stat_res,
547 548
		.p_arglen	= SM_mon_sz,
		.p_replen	= SM_monres_sz,
549
		.p_statidx	= NSMPROC_MON,
550
		.p_name		= "MONITOR",
L
Linus Torvalds 已提交
551
	},
552 553
[NSMPROC_UNMON] = {
		.p_proc		= NSMPROC_UNMON,
554 555
		.p_encode	= (kxdrproc_t)xdr_enc_unmon,
		.p_decode	= (kxdrproc_t)xdr_dec_stat,
556 557
		.p_arglen	= SM_mon_id_sz,
		.p_replen	= SM_unmonres_sz,
558
		.p_statidx	= NSMPROC_UNMON,
559
		.p_name		= "UNMONITOR",
L
Linus Torvalds 已提交
560 561 562 563
	},
};

static struct rpc_version	nsm_version1 = {
564 565
		.number		= 1,
		.nrprocs	= ARRAY_SIZE(nsm_procedures),
L
Linus Torvalds 已提交
566 567 568 569 570 571 572 573 574 575 576
		.procs		= nsm_procedures
};

static struct rpc_version *	nsm_version[] = {
	[1] = &nsm_version1,
};

static struct rpc_stat		nsm_stats;

static struct rpc_program	nsm_program = {
		.name		= "statd",
577
		.number		= NSM_PROGRAM,
578
		.nrvers		= ARRAY_SIZE(nsm_version),
L
Linus Torvalds 已提交
579 580 581
		.version	= nsm_version,
		.stats		= &nsm_stats
};
新手
引导
客服 返回
顶部