svc_xprt.c 37.9 KB
Newer Older
T
Tom Tucker 已提交
1 2 3 4 5 6 7 8 9
/*
 * linux/net/sunrpc/svc_xprt.c
 *
 * Author: Tom Tucker <tom@opengridcomputing.com>
 */

#include <linux/sched.h>
#include <linux/errno.h>
#include <linux/freezer.h>
10
#include <linux/kthread.h>
11
#include <linux/slab.h>
T
Tom Tucker 已提交
12
#include <net/sock.h>
13
#include <linux/sunrpc/addr.h>
T
Tom Tucker 已提交
14 15
#include <linux/sunrpc/stats.h>
#include <linux/sunrpc/svc_xprt.h>
16
#include <linux/sunrpc/svcsock.h>
17
#include <linux/sunrpc/xprt.h>
18
#include <linux/module.h>
19
#include <linux/netdevice.h>
20
#include <trace/events/sunrpc.h>
T
Tom Tucker 已提交
21 22 23

#define RPCDBG_FACILITY	RPCDBG_SVCXPRT

24 25 26 27
static struct svc_deferred_req *svc_deferred_dequeue(struct svc_xprt *xprt);
static int svc_deferred_recv(struct svc_rqst *rqstp);
static struct cache_deferred_req *svc_defer(struct cache_req *req);
static void svc_age_temp_xprts(unsigned long closure);
28
static void svc_delete_xprt(struct svc_xprt *xprt);
29 30 31 32 33 34 35 36

/* apparently the "standard" is that clients close
 * idle connections after 5 minutes, servers after
 * 6 minutes
 *   http://www.connectathon.org/talks96/nfstcp.pdf
 */
static int svc_conn_age_period = 6*60;

T
Tom Tucker 已提交
37 38 39 40
/* List of registered transport classes */
static DEFINE_SPINLOCK(svc_xprt_class_lock);
static LIST_HEAD(svc_xprt_class_list);

41 42 43 44 45
/* SMP locking strategy:
 *
 *	svc_pool->sp_lock protects most of the fields of that pool.
 *	svc_serv->sv_lock protects sv_tempsocks, sv_permsocks, sv_tmpcnt.
 *	when both need to be taken (rare), svc_serv->sv_lock is first.
46
 *	The "service mutex" protects svc_serv->sv_nrthread.
47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70
 *	svc_sock->sk_lock protects the svc_sock->sk_deferred list
 *             and the ->sk_info_authunix cache.
 *
 *	The XPT_BUSY bit in xprt->xpt_flags prevents a transport being
 *	enqueued multiply. During normal transport processing this bit
 *	is set by svc_xprt_enqueue and cleared by svc_xprt_received.
 *	Providers should not manipulate this bit directly.
 *
 *	Some flags can be set to certain values at any time
 *	providing that certain rules are followed:
 *
 *	XPT_CONN, XPT_DATA:
 *		- Can be set or cleared at any time.
 *		- After a set, svc_xprt_enqueue must be called to enqueue
 *		  the transport for processing.
 *		- After a clear, the transport must be read/accepted.
 *		  If this succeeds, it must be set again.
 *	XPT_CLOSE:
 *		- Can set at any time. It is never cleared.
 *      XPT_DEAD:
 *		- Can only be set while XPT_BUSY is held which ensures
 *		  that no other thread will be using the transport or will
 *		  try to set XPT_DEAD.
 */
T
Tom Tucker 已提交
71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101
int svc_reg_xprt_class(struct svc_xprt_class *xcl)
{
	struct svc_xprt_class *cl;
	int res = -EEXIST;

	dprintk("svc: Adding svc transport class '%s'\n", xcl->xcl_name);

	INIT_LIST_HEAD(&xcl->xcl_list);
	spin_lock(&svc_xprt_class_lock);
	/* Make sure there isn't already a class with the same name */
	list_for_each_entry(cl, &svc_xprt_class_list, xcl_list) {
		if (strcmp(xcl->xcl_name, cl->xcl_name) == 0)
			goto out;
	}
	list_add_tail(&xcl->xcl_list, &svc_xprt_class_list);
	res = 0;
out:
	spin_unlock(&svc_xprt_class_lock);
	return res;
}
EXPORT_SYMBOL_GPL(svc_reg_xprt_class);

void svc_unreg_xprt_class(struct svc_xprt_class *xcl)
{
	dprintk("svc: Removing svc transport class '%s'\n", xcl->xcl_name);
	spin_lock(&svc_xprt_class_lock);
	list_del_init(&xcl->xcl_list);
	spin_unlock(&svc_xprt_class_lock);
}
EXPORT_SYMBOL_GPL(svc_unreg_xprt_class);

102 103 104 105 106
/*
 * Format the transport list for printing
 */
int svc_print_xprts(char *buf, int maxlen)
{
107
	struct svc_xprt_class *xcl;
108 109 110 111 112
	char tmpstr[80];
	int len = 0;
	buf[0] = '\0';

	spin_lock(&svc_xprt_class_lock);
113
	list_for_each_entry(xcl, &svc_xprt_class_list, xcl_list) {
114 115 116 117 118 119 120 121 122 123 124 125 126 127
		int slen;

		sprintf(tmpstr, "%s %d\n", xcl->xcl_name, xcl->xcl_max_payload);
		slen = strlen(tmpstr);
		if (len + slen > maxlen)
			break;
		len += slen;
		strcat(buf, tmpstr);
	}
	spin_unlock(&svc_xprt_class_lock);

	return len;
}

T
Tom Tucker 已提交
128 129 130 131 132
static void svc_xprt_free(struct kref *kref)
{
	struct svc_xprt *xprt =
		container_of(kref, struct svc_xprt, xpt_ref);
	struct module *owner = xprt->xpt_class->xcl_owner;
133 134
	if (test_bit(XPT_CACHE_AUTH, &xprt->xpt_flags))
		svcauth_unix_info_release(xprt);
P
Pavel Emelyanov 已提交
135
	put_net(xprt->xpt_net);
136 137 138
	/* See comment on corresponding get in xs_setup_bc_tcp(): */
	if (xprt->xpt_bc_xprt)
		xprt_put(xprt->xpt_bc_xprt);
139 140
	if (xprt->xpt_bc_xps)
		xprt_switch_put(xprt->xpt_bc_xps);
T
Tom Tucker 已提交
141 142 143 144 145 146 147 148 149 150
	xprt->xpt_ops->xpo_free(xprt);
	module_put(owner);
}

void svc_xprt_put(struct svc_xprt *xprt)
{
	kref_put(&xprt->xpt_ref, svc_xprt_free);
}
EXPORT_SYMBOL_GPL(svc_xprt_put);

T
Tom Tucker 已提交
151 152 153 154
/*
 * Called by transport drivers to initialize the transport independent
 * portion of the transport instance.
 */
155 156
void svc_xprt_init(struct net *net, struct svc_xprt_class *xcl,
		   struct svc_xprt *xprt, struct svc_serv *serv)
T
Tom Tucker 已提交
157 158 159 160
{
	memset(xprt, 0, sizeof(*xprt));
	xprt->xpt_class = xcl;
	xprt->xpt_ops = xcl->xcl_ops;
T
Tom Tucker 已提交
161
	kref_init(&xprt->xpt_ref);
162
	xprt->xpt_server = serv;
163 164
	INIT_LIST_HEAD(&xprt->xpt_list);
	INIT_LIST_HEAD(&xprt->xpt_ready);
165
	INIT_LIST_HEAD(&xprt->xpt_deferred);
166
	INIT_LIST_HEAD(&xprt->xpt_users);
167
	mutex_init(&xprt->xpt_mutex);
168
	spin_lock_init(&xprt->xpt_lock);
169
	set_bit(XPT_BUSY, &xprt->xpt_flags);
170
	rpc_init_wait_queue(&xprt->xpt_bc_pending, "xpt_bc_pending");
171
	xprt->xpt_net = get_net(net);
T
Tom Tucker 已提交
172 173
}
EXPORT_SYMBOL_GPL(svc_xprt_init);
174

175 176
static struct svc_xprt *__svc_xpo_create(struct svc_xprt_class *xcl,
					 struct svc_serv *serv,
177
					 struct net *net,
178 179 180
					 const int family,
					 const unsigned short port,
					 int flags)
181 182 183
{
	struct sockaddr_in sin = {
		.sin_family		= AF_INET,
A
Al Viro 已提交
184
		.sin_addr.s_addr	= htonl(INADDR_ANY),
185 186
		.sin_port		= htons(port),
	};
E
Eric Dumazet 已提交
187
#if IS_ENABLED(CONFIG_IPV6)
188 189 190 191 192
	struct sockaddr_in6 sin6 = {
		.sin6_family		= AF_INET6,
		.sin6_addr		= IN6ADDR_ANY_INIT,
		.sin6_port		= htons(port),
	};
E
Eric Dumazet 已提交
193
#endif
194 195 196
	struct sockaddr *sap;
	size_t len;

197 198
	switch (family) {
	case PF_INET:
199 200 201
		sap = (struct sockaddr *)&sin;
		len = sizeof(sin);
		break;
E
Eric Dumazet 已提交
202
#if IS_ENABLED(CONFIG_IPV6)
203
	case PF_INET6:
204 205 206
		sap = (struct sockaddr *)&sin6;
		len = sizeof(sin6);
		break;
E
Eric Dumazet 已提交
207
#endif
208 209 210 211
	default:
		return ERR_PTR(-EAFNOSUPPORT);
	}

212
	return xcl->xcl_ops->xpo_create(serv, net, sap, len, flags);
213 214
}

215 216 217 218 219 220 221 222 223 224
/*
 * svc_xprt_received conditionally queues the transport for processing
 * by another thread. The caller must hold the XPT_BUSY bit and must
 * not thereafter touch transport data.
 *
 * Note: XPT_DATA only gets cleared when a read-attempt finds no (or
 * insufficient) data.
 */
static void svc_xprt_received(struct svc_xprt *xprt)
{
225 226
	if (!test_bit(XPT_BUSY, &xprt->xpt_flags)) {
		WARN_ONCE(1, "xprt=0x%p already busy!", xprt);
227
		return;
228 229
	}

230
	/* As soon as we clear busy, the xprt could be closed and
231
	 * 'put', so we need a reference to call svc_enqueue_xprt with:
232 233
	 */
	svc_xprt_get(xprt);
234
	smp_mb__before_atomic();
235
	clear_bit(XPT_BUSY, &xprt->xpt_flags);
236
	xprt->xpt_server->sv_ops->svo_enqueue_xprt(xprt);
237 238 239
	svc_xprt_put(xprt);
}

240 241 242 243 244 245 246 247 248
void svc_add_new_perm_xprt(struct svc_serv *serv, struct svc_xprt *new)
{
	clear_bit(XPT_TEMP, &new->xpt_flags);
	spin_lock_bh(&serv->sv_lock);
	list_add(&new->xpt_list, &serv->sv_permsocks);
	spin_unlock_bh(&serv->sv_lock);
	svc_xprt_received(new);
}

J
J. Bruce Fields 已提交
249
int _svc_create_xprt(struct svc_serv *serv, const char *xprt_name,
250 251
		    struct net *net, const int family,
		    const unsigned short port, int flags)
252 253 254
{
	struct svc_xprt_class *xcl;

255 256
	spin_lock(&svc_xprt_class_lock);
	list_for_each_entry(xcl, &svc_xprt_class_list, xcl_list) {
257
		struct svc_xprt *newxprt;
258
		unsigned short newport;
259 260 261 262 263 264 265 266

		if (strcmp(xprt_name, xcl->xcl_name))
			continue;

		if (!try_module_get(xcl->xcl_owner))
			goto err;

		spin_unlock(&svc_xprt_class_lock);
267
		newxprt = __svc_xpo_create(xcl, serv, net, family, port, flags);
268 269 270
		if (IS_ERR(newxprt)) {
			module_put(xcl->xcl_owner);
			return PTR_ERR(newxprt);
271
		}
272
		svc_add_new_perm_xprt(serv, newxprt);
273 274
		newport = svc_xprt_local_port(newxprt);
		return newport;
275
	}
276
 err:
277
	spin_unlock(&svc_xprt_class_lock);
278 279 280
	/* This errno is exposed to user space.  Provide a reasonable
	 * perror msg for a bad transport. */
	return -EPROTONOSUPPORT;
281
}
J
J. Bruce Fields 已提交
282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299

int svc_create_xprt(struct svc_serv *serv, const char *xprt_name,
		    struct net *net, const int family,
		    const unsigned short port, int flags)
{
	int err;

	dprintk("svc: creating transport %s[%d]\n", xprt_name, port);
	err = _svc_create_xprt(serv, xprt_name, net, family, port, flags);
	if (err == -EPROTONOSUPPORT) {
		request_module("svc%s", xprt_name);
		err = _svc_create_xprt(serv, xprt_name, net, family, port, flags);
	}
	if (err)
		dprintk("svc: transport %s not found, err %d\n",
			xprt_name, err);
	return err;
}
300
EXPORT_SYMBOL_GPL(svc_create_xprt);
301 302 303 304 305 306 307 308 309 310 311 312 313

/*
 * Copy the local and remote xprt addresses to the rqstp structure
 */
void svc_xprt_copy_addrs(struct svc_rqst *rqstp, struct svc_xprt *xprt)
{
	memcpy(&rqstp->rq_addr, &xprt->xpt_remote, xprt->xpt_remotelen);
	rqstp->rq_addrlen = xprt->xpt_remotelen;

	/*
	 * Destination address in request is needed for binding the
	 * source address in RPC replies/callbacks later.
	 */
314 315
	memcpy(&rqstp->rq_daddr, &xprt->xpt_local, xprt->xpt_locallen);
	rqstp->rq_daddrlen = xprt->xpt_locallen;
316 317 318
}
EXPORT_SYMBOL_GPL(svc_xprt_copy_addrs);

319 320 321 322 323 324 325 326 327 328 329 330 331
/**
 * svc_print_addr - Format rq_addr field for printing
 * @rqstp: svc_rqst struct containing address to print
 * @buf: target buffer for formatted address
 * @len: length of target buffer
 *
 */
char *svc_print_addr(struct svc_rqst *rqstp, char *buf, size_t len)
{
	return __svc_print_addr(svc_addr(rqstp), buf, len);
}
EXPORT_SYMBOL_GPL(svc_print_addr);

J
J. Bruce Fields 已提交
332 333 334 335
static bool svc_xprt_has_something_to_do(struct svc_xprt *xprt)
{
	if (xprt->xpt_flags & ((1<<XPT_CONN)|(1<<XPT_CLOSE)))
		return true;
336 337 338 339 340 341
	if (xprt->xpt_flags & ((1<<XPT_DATA)|(1<<XPT_DEFERRED))) {
		if (xprt->xpt_ops->xpo_has_wspace(xprt))
			return true;
		trace_svc_xprt_no_write_space(xprt);
		return false;
	}
J
J. Bruce Fields 已提交
342 343 344
	return false;
}

345
void svc_xprt_do_enqueue(struct svc_xprt *xprt)
346 347
{
	struct svc_pool *pool;
348
	struct svc_rqst	*rqstp = NULL;
349
	int cpu;
350
	bool queued = false;
351

J
J. Bruce Fields 已提交
352
	if (!svc_xprt_has_something_to_do(xprt))
353
		goto out;
354 355 356 357 358 359 360 361 362

	/* Mark transport as busy. It will remain in this state until
	 * the provider calls svc_xprt_received. We update XPT_BUSY
	 * atomically because it also guards against trying to enqueue
	 * the transport twice.
	 */
	if (test_and_set_bit(XPT_BUSY, &xprt->xpt_flags)) {
		/* Don't enqueue transport while already enqueued */
		dprintk("svc: transport %p busy, not enqueued\n", xprt);
363
		goto out;
364 365
	}

366 367 368
	cpu = get_cpu();
	pool = svc_pool_for_cpu(xprt->xpt_server, cpu);

369
	atomic_long_inc(&pool->sp_stats.packets);
370

371 372 373 374 375 376 377 378 379 380 381 382 383
redo_search:
	/* find a thread for this xprt */
	rcu_read_lock();
	list_for_each_entry_rcu(rqstp, &pool->sp_all_threads, rq_all) {
		/* Do a lockless check first */
		if (test_bit(RQ_BUSY, &rqstp->rq_flags))
			continue;

		/*
		 * Once the xprt has been queued, it can only be dequeued by
		 * the task that intends to service it. All we can do at that
		 * point is to try to wake this thread back up so that it can
		 * do so.
384
		 */
385 386 387 388 389 390 391 392 393 394 395 396 397 398 399
		if (!queued) {
			spin_lock_bh(&rqstp->rq_lock);
			if (test_and_set_bit(RQ_BUSY, &rqstp->rq_flags)) {
				/* already busy, move on... */
				spin_unlock_bh(&rqstp->rq_lock);
				continue;
			}

			/* this one will do */
			rqstp->rq_xprt = xprt;
			svc_xprt_get(xprt);
			spin_unlock_bh(&rqstp->rq_lock);
		}
		rcu_read_unlock();

400
		atomic_long_inc(&pool->sp_stats.threads_woken);
401 402
		wake_up_process(rqstp->rq_task);
		put_cpu();
403
		goto out;
404 405 406 407 408 409 410 411 412 413 414
	}
	rcu_read_unlock();

	/*
	 * We didn't find an idle thread to use, so we need to queue the xprt.
	 * Do so and then search again. If we find one, we can't hook this one
	 * up to it directly but we can wake the thread up in the hopes that it
	 * will pick it up once it searches for a xprt to service.
	 */
	if (!queued) {
		queued = true;
415
		dprintk("svc: transport %p put into queue\n", xprt);
416
		spin_lock_bh(&pool->sp_lock);
417
		list_add_tail(&xprt->xpt_ready, &pool->sp_sockets);
418
		pool->sp_stats.sockets_queued++;
419 420
		spin_unlock_bh(&pool->sp_lock);
		goto redo_search;
421
	}
422
	rqstp = NULL;
423
	put_cpu();
424 425
out:
	trace_svc_xprt_do_enqueue(xprt, rqstp);
426
}
427
EXPORT_SYMBOL_GPL(svc_xprt_do_enqueue);
428 429 430 431 432 433 434 435 436 437

/*
 * Queue up a transport with data pending. If there are idle nfsd
 * processes, wake 'em up.
 *
 */
void svc_xprt_enqueue(struct svc_xprt *xprt)
{
	if (test_bit(XPT_BUSY, &xprt->xpt_flags))
		return;
438
	xprt->xpt_server->sv_ops->svo_enqueue_xprt(xprt);
439
}
440 441 442
EXPORT_SYMBOL_GPL(svc_xprt_enqueue);

/*
443
 * Dequeue the first transport, if there is one.
444 445 446
 */
static struct svc_xprt *svc_xprt_dequeue(struct svc_pool *pool)
{
447
	struct svc_xprt	*xprt = NULL;
448 449

	if (list_empty(&pool->sp_sockets))
450
		goto out;
451

452 453 454 455 456 457
	spin_lock_bh(&pool->sp_lock);
	if (likely(!list_empty(&pool->sp_sockets))) {
		xprt = list_first_entry(&pool->sp_sockets,
					struct svc_xprt, xpt_ready);
		list_del_init(&xprt->xpt_ready);
		svc_xprt_get(xprt);
458

459 460 461 462
		dprintk("svc: transport %p dequeued, inuse=%d\n",
			xprt, atomic_read(&xprt->xpt_ref.refcount));
	}
	spin_unlock_bh(&pool->sp_lock);
463 464
out:
	trace_svc_xprt_dequeue(xprt);
465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486
	return xprt;
}

/**
 * svc_reserve - change the space reserved for the reply to a request.
 * @rqstp:  The request in question
 * @space: new max space to reserve
 *
 * Each request reserves some space on the output queue of the transport
 * to make sure the reply fits.  This function reduces that reserved
 * space to be the amount of space used already, plus @space.
 *
 */
void svc_reserve(struct svc_rqst *rqstp, int space)
{
	space += rqstp->rq_res.head[0].iov_len;

	if (space < rqstp->rq_reserved) {
		struct svc_xprt *xprt = rqstp->rq_xprt;
		atomic_sub((rqstp->rq_reserved - space), &xprt->xpt_reserved);
		rqstp->rq_reserved = space;

487 488
		if (xprt->xpt_ops->xpo_adjust_wspace)
			xprt->xpt_ops->xpo_adjust_wspace(xprt);
489 490 491
		svc_xprt_enqueue(xprt);
	}
}
492
EXPORT_SYMBOL_GPL(svc_reserve);
493 494 495 496 497 498 499

static void svc_xprt_release(struct svc_rqst *rqstp)
{
	struct svc_xprt	*xprt = rqstp->rq_xprt;

	rqstp->rq_xprt->xpt_ops->xpo_release_rqst(rqstp);

500 501 502
	kfree(rqstp->rq_deferred);
	rqstp->rq_deferred = NULL;

503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524
	svc_free_res_pages(rqstp);
	rqstp->rq_res.page_len = 0;
	rqstp->rq_res.page_base = 0;

	/* Reset response buffer and release
	 * the reservation.
	 * But first, check that enough space was reserved
	 * for the reply, otherwise we have a bug!
	 */
	if ((rqstp->rq_res.len) >  rqstp->rq_reserved)
		printk(KERN_ERR "RPC request reserved %d but used %d\n",
		       rqstp->rq_reserved,
		       rqstp->rq_res.len);

	rqstp->rq_res.head[0].iov_len = 0;
	svc_reserve(rqstp, 0);
	rqstp->rq_xprt = NULL;

	svc_xprt_put(xprt);
}

/*
525 526 527 528 529
 * Some svc_serv's will have occasional work to do, even when a xprt is not
 * waiting to be serviced. This function is there to "kick" a task in one of
 * those services so that it can wake up and do that work. Note that we only
 * bother with pool 0 as we don't need to wake up more than one thread for
 * this purpose.
530 531 532 533 534 535
 */
void svc_wake_up(struct svc_serv *serv)
{
	struct svc_rqst	*rqstp;
	struct svc_pool *pool;

536 537
	pool = &serv->sv_pools[0];

538 539 540 541 542 543
	rcu_read_lock();
	list_for_each_entry_rcu(rqstp, &pool->sp_all_threads, rq_all) {
		/* skip any that aren't queued */
		if (test_bit(RQ_BUSY, &rqstp->rq_flags))
			continue;
		rcu_read_unlock();
544 545
		dprintk("svc: daemon %p woken up.\n", rqstp);
		wake_up_process(rqstp->rq_task);
546
		trace_svc_wake_up(rqstp->rq_task->pid);
547 548 549 550 551 552 553
		return;
	}
	rcu_read_unlock();

	/* No free entries available */
	set_bit(SP_TASK_PENDING, &pool->sp_flags);
	smp_wmb();
554
	trace_svc_wake_up(0);
555
}
556
EXPORT_SYMBOL_GPL(svc_wake_up);
557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572

int svc_port_is_privileged(struct sockaddr *sin)
{
	switch (sin->sa_family) {
	case AF_INET:
		return ntohs(((struct sockaddr_in *)sin)->sin_port)
			< PROT_SOCK;
	case AF_INET6:
		return ntohs(((struct sockaddr_in6 *)sin)->sin6_port)
			< PROT_SOCK;
	default:
		return 0;
	}
}

/*
573 574 575 576
 * Make sure that we don't have too many active connections. If we have,
 * something must be dropped. It's not clear what will happen if we allow
 * "too many" connections, but when dealing with network-facing software,
 * we have to code defensively. Here we do that by imposing hard limits.
577 578 579 580 581 582 583 584
 *
 * There's no point in trying to do random drop here for DoS
 * prevention. The NFS clients does 1 reconnect in 15 seconds. An
 * attacker can easily beat that.
 *
 * The only somewhat efficient mechanism would be if drop old
 * connections from the same IP first. But right now we don't even
 * record the client IP in svc_sock.
585 586 587 588
 *
 * single-threaded services that expect a lot of clients will probably
 * need to set sv_maxconn to override the default value which is based
 * on the number of threads
589 590 591
 */
static void svc_check_conn_limits(struct svc_serv *serv)
{
592 593 594 595
	unsigned int limit = serv->sv_maxconn ? serv->sv_maxconn :
				(serv->sv_nrthreads+3) * 20;

	if (serv->sv_tmpcnt > limit) {
596 597 598
		struct svc_xprt *xprt = NULL;
		spin_lock_bh(&serv->sv_lock);
		if (!list_empty(&serv->sv_tempsocks)) {
599 600 601 602 603
			/* Try to help the admin */
			net_notice_ratelimited("%s: too many open connections, consider increasing the %s\n",
					       serv->sv_name, serv->sv_maxconn ?
					       "max number of connections" :
					       "number of threads");
604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622
			/*
			 * Always select the oldest connection. It's not fair,
			 * but so is life
			 */
			xprt = list_entry(serv->sv_tempsocks.prev,
					  struct svc_xprt,
					  xpt_list);
			set_bit(XPT_CLOSE, &xprt->xpt_flags);
			svc_xprt_get(xprt);
		}
		spin_unlock_bh(&serv->sv_lock);

		if (xprt) {
			svc_xprt_enqueue(xprt);
			svc_xprt_put(xprt);
		}
	}
}

623
static int svc_alloc_arg(struct svc_rqst *rqstp)
624
{
J
J. Bruce Fields 已提交
625 626 627 628
	struct svc_serv *serv = rqstp->rq_server;
	struct xdr_buf *arg;
	int pages;
	int i;
629 630 631

	/* now allocate needed pages.  If we get a failure, sleep briefly */
	pages = (serv->sv_max_mesg + PAGE_SIZE) / PAGE_SIZE;
632 633 634 635
	WARN_ON_ONCE(pages >= RPCSVC_MAXPAGES);
	if (pages >= RPCSVC_MAXPAGES)
		/* use as many pages as possible */
		pages = RPCSVC_MAXPAGES - 1;
636 637 638 639
	for (i = 0; i < pages ; i++)
		while (rqstp->rq_pages[i] == NULL) {
			struct page *p = alloc_page(GFP_KERNEL);
			if (!p) {
640 641 642
				set_current_state(TASK_INTERRUPTIBLE);
				if (signalled() || kthread_should_stop()) {
					set_current_state(TASK_RUNNING);
643
					return -EINTR;
644 645
				}
				schedule_timeout(msecs_to_jiffies(500));
646 647 648
			}
			rqstp->rq_pages[i] = p;
		}
649
	rqstp->rq_page_end = &rqstp->rq_pages[i];
650 651 652 653 654 655 656 657 658 659 660 661
	rqstp->rq_pages[i++] = NULL; /* this might be seen in nfs_read_actor */

	/* Make arg->head point to first page and arg->pages point to rest */
	arg = &rqstp->rq_arg;
	arg->head[0].iov_base = page_address(rqstp->rq_pages[0]);
	arg->head[0].iov_len = PAGE_SIZE;
	arg->pages = rqstp->rq_pages + 1;
	arg->page_base = 0;
	/* save at least one page for response */
	arg->page_len = (pages-2)*PAGE_SIZE;
	arg->len = (pages-1)*PAGE_SIZE;
	arg->tail[0].iov_len = 0;
J
J. Bruce Fields 已提交
662 663
	return 0;
}
664

665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688
static bool
rqst_should_sleep(struct svc_rqst *rqstp)
{
	struct svc_pool		*pool = rqstp->rq_pool;

	/* did someone call svc_wake_up? */
	if (test_and_clear_bit(SP_TASK_PENDING, &pool->sp_flags))
		return false;

	/* was a socket queued? */
	if (!list_empty(&pool->sp_sockets))
		return false;

	/* are we shutting down? */
	if (signalled() || kthread_should_stop())
		return false;

	/* are we freezing? */
	if (freezing(current))
		return false;

	return true;
}

689
static struct svc_xprt *svc_get_next_xprt(struct svc_rqst *rqstp, long timeout)
J
J. Bruce Fields 已提交
690 691 692
{
	struct svc_xprt *xprt;
	struct svc_pool		*pool = rqstp->rq_pool;
693
	long			time_left = 0;
694

695 696 697
	/* rq_xprt should be clear on entry */
	WARN_ON_ONCE(rqstp->rq_xprt);

698 699 700 701 702
	/* Normally we will wait up to 5 seconds for any required
	 * cache information to be provided.
	 */
	rqstp->rq_chandle.thread_wait = 5*HZ;

703 704 705
	xprt = svc_xprt_dequeue(pool);
	if (xprt) {
		rqstp->rq_xprt = xprt;
706 707

		/* As there is a shortage of threads and this request
J
J. Bruce Fields 已提交
708
		 * had to be queued, don't allow the thread to wait so
709 710 711
		 * long for cache updates.
		 */
		rqstp->rq_chandle.thread_wait = 1*HZ;
712
		clear_bit(SP_TASK_PENDING, &pool->sp_flags);
713 714
		return xprt;
	}
715

716 717 718 719 720 721 722
	/*
	 * We have to be able to interrupt this wait
	 * to bring down the daemons ...
	 */
	set_current_state(TASK_INTERRUPTIBLE);
	clear_bit(RQ_BUSY, &rqstp->rq_flags);
	smp_mb();
723

724 725 726 727
	if (likely(rqst_should_sleep(rqstp)))
		time_left = schedule_timeout(timeout);
	else
		__set_current_state(TASK_RUNNING);
728

729
	try_to_freeze();
730

731 732 733
	spin_lock_bh(&rqstp->rq_lock);
	set_bit(RQ_BUSY, &rqstp->rq_flags);
	spin_unlock_bh(&rqstp->rq_lock);
734

735 736 737 738 739 740 741 742 743 744
	xprt = rqstp->rq_xprt;
	if (xprt != NULL)
		return xprt;

	if (!time_left)
		atomic_long_inc(&pool->sp_stats.threads_timedout);

	if (signalled() || kthread_should_stop())
		return ERR_PTR(-EINTR);
	return ERR_PTR(-EAGAIN);
J
J. Bruce Fields 已提交
745 746
}

747
static void svc_add_new_temp_xprt(struct svc_serv *serv, struct svc_xprt *newxpt)
748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763
{
	spin_lock_bh(&serv->sv_lock);
	set_bit(XPT_TEMP, &newxpt->xpt_flags);
	list_add(&newxpt->xpt_list, &serv->sv_tempsocks);
	serv->sv_tmpcnt++;
	if (serv->sv_temptimer.function == NULL) {
		/* setup timer to age temp transports */
		setup_timer(&serv->sv_temptimer, svc_age_temp_xprts,
			    (unsigned long)serv);
		mod_timer(&serv->sv_temptimer,
			  jiffies + svc_conn_age_period * HZ);
	}
	spin_unlock_bh(&serv->sv_lock);
	svc_xprt_received(newxpt);
}

J
J. Bruce Fields 已提交
764 765 766 767
static int svc_handle_xprt(struct svc_rqst *rqstp, struct svc_xprt *xprt)
{
	struct svc_serv *serv = rqstp->rq_server;
	int len = 0;
768

769 770 771
	if (test_bit(XPT_CLOSE, &xprt->xpt_flags)) {
		dprintk("svc_recv: found XPT_CLOSE\n");
		svc_delete_xprt(xprt);
772
		/* Leave XPT_BUSY set on the dead xprt: */
773
		goto out;
774 775
	}
	if (test_bit(XPT_LISTENER, &xprt->xpt_flags)) {
776
		struct svc_xprt *newxpt;
777 778 779 780 781 782
		/*
		 * We know this module_get will succeed because the
		 * listener holds a reference too
		 */
		__module_get(xprt->xpt_class->xcl_owner);
		svc_check_conn_limits(xprt->xpt_server);
783
		newxpt = xprt->xpt_ops->xpo_accept(xprt);
784 785
		if (newxpt)
			svc_add_new_temp_xprt(serv, newxpt);
786 787
		else
			module_put(xprt->xpt_class->xcl_owner);
788
	} else {
J
J. Bruce Fields 已提交
789
		/* XPT_DATA|XPT_DEFERRED case: */
790
		dprintk("svc: server %p, pool %u, transport %p, inuse=%d\n",
J
J. Bruce Fields 已提交
791
			rqstp, rqstp->rq_pool->sp_id, xprt,
792 793
			atomic_read(&xprt->xpt_ref.refcount));
		rqstp->rq_deferred = svc_deferred_dequeue(xprt);
794
		if (rqstp->rq_deferred)
795
			len = svc_deferred_recv(rqstp);
796
		else
797 798
			len = xprt->xpt_ops->xpo_recvfrom(rqstp);
		dprintk("svc: got len=%d\n", len);
799 800
		rqstp->rq_reserved = serv->sv_max_mesg;
		atomic_add(rqstp->rq_reserved, &xprt->xpt_reserved);
801
	}
J
J. Bruce Fields 已提交
802
	/* clear XPT_BUSY: */
803
	svc_xprt_received(xprt);
804 805
out:
	trace_svc_handle_xprt(xprt, len);
J
J. Bruce Fields 已提交
806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826
	return len;
}

/*
 * Receive the next request on any transport.  This code is carefully
 * organised not to touch any cachelines in the shared svc_serv
 * structure, only cachelines in the local svc_pool.
 */
int svc_recv(struct svc_rqst *rqstp, long timeout)
{
	struct svc_xprt		*xprt = NULL;
	struct svc_serv		*serv = rqstp->rq_server;
	int			len, err;

	dprintk("svc: server %p waiting for data (to = %ld)\n",
		rqstp, timeout);

	if (rqstp->rq_xprt)
		printk(KERN_ERR
			"svc_recv: service %p, transport not NULL!\n",
			 rqstp);
827

J
J. Bruce Fields 已提交
828 829
	err = svc_alloc_arg(rqstp);
	if (err)
830
		goto out;
J
J. Bruce Fields 已提交
831 832 833

	try_to_freeze();
	cond_resched();
834
	err = -EINTR;
J
J. Bruce Fields 已提交
835
	if (signalled() || kthread_should_stop())
836
		goto out;
J
J. Bruce Fields 已提交
837 838

	xprt = svc_get_next_xprt(rqstp, timeout);
839 840 841 842
	if (IS_ERR(xprt)) {
		err = PTR_ERR(xprt);
		goto out;
	}
J
J. Bruce Fields 已提交
843 844

	len = svc_handle_xprt(rqstp, xprt);
845 846

	/* No data, incomplete (TCP) read, or accept() */
847
	err = -EAGAIN;
848
	if (len <= 0)
849
		goto out_release;
850

851 852
	clear_bit(XPT_OLD, &xprt->xpt_flags);

853 854 855 856
	if (xprt->xpt_ops->xpo_secure_port(rqstp))
		set_bit(RQ_SECURE, &rqstp->rq_flags);
	else
		clear_bit(RQ_SECURE, &rqstp->rq_flags);
857
	rqstp->rq_chandle.defer = svc_defer;
858
	rqstp->rq_xid = svc_getu32(&rqstp->rq_arg.head[0]);
859 860 861

	if (serv->sv_stats)
		serv->sv_stats->netcnt++;
862
	trace_svc_recv(rqstp, len);
863
	return len;
864
out_release:
865 866
	rqstp->rq_res.len = 0;
	svc_xprt_release(rqstp);
867 868 869
out:
	trace_svc_recv(rqstp, err);
	return err;
870
}
871
EXPORT_SYMBOL_GPL(svc_recv);
872 873 874 875 876 877

/*
 * Drop request
 */
void svc_drop(struct svc_rqst *rqstp)
{
878
	trace_svc_drop(rqstp);
879 880 881
	dprintk("svc: xprt %p dropped request\n", rqstp->rq_xprt);
	svc_xprt_release(rqstp);
}
882
EXPORT_SYMBOL_GPL(svc_drop);
883 884 885 886 887 888 889

/*
 * Return reply to client.
 */
int svc_send(struct svc_rqst *rqstp)
{
	struct svc_xprt	*xprt;
890
	int		len = -EFAULT;
891 892 893 894
	struct xdr_buf	*xb;

	xprt = rqstp->rq_xprt;
	if (!xprt)
895
		goto out;
896 897 898 899 900 901 902 903 904 905 906 907

	/* release the receive skb before sending the reply */
	rqstp->rq_xprt->xpt_ops->xpo_release_rqst(rqstp);

	/* calculate over-all length */
	xb = &rqstp->rq_res;
	xb->len = xb->head[0].iov_len +
		xb->page_len +
		xb->tail[0].iov_len;

	/* Grab mutex to serialize outgoing data. */
	mutex_lock(&xprt->xpt_mutex);
908 909
	if (test_bit(XPT_DEAD, &xprt->xpt_flags)
			|| test_bit(XPT_CLOSE, &xprt->xpt_flags))
910 911 912 913
		len = -ENOTCONN;
	else
		len = xprt->xpt_ops->xpo_sendto(rqstp);
	mutex_unlock(&xprt->xpt_mutex);
914
	rpc_wake_up(&xprt->xpt_bc_pending);
915 916 917
	svc_xprt_release(rqstp);

	if (len == -ECONNREFUSED || len == -ENOTCONN || len == -EAGAIN)
918 919 920
		len = 0;
out:
	trace_svc_send(rqstp, len);
921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949
	return len;
}

/*
 * Timer function to close old temporary transports, using
 * a mark-and-sweep algorithm.
 */
static void svc_age_temp_xprts(unsigned long closure)
{
	struct svc_serv *serv = (struct svc_serv *)closure;
	struct svc_xprt *xprt;
	struct list_head *le, *next;

	dprintk("svc_age_temp_xprts\n");

	if (!spin_trylock_bh(&serv->sv_lock)) {
		/* busy, try again 1 sec later */
		dprintk("svc_age_temp_xprts: busy\n");
		mod_timer(&serv->sv_temptimer, jiffies + HZ);
		return;
	}

	list_for_each_safe(le, next, &serv->sv_tempsocks) {
		xprt = list_entry(le, struct svc_xprt, xpt_list);

		/* First time through, just mark it OLD. Second time
		 * through, close it. */
		if (!test_and_set_bit(XPT_OLD, &xprt->xpt_flags))
			continue;
950 951
		if (atomic_read(&xprt->xpt_ref.refcount) > 1 ||
		    test_bit(XPT_BUSY, &xprt->xpt_flags))
952
			continue;
953
		list_del_init(le);
954 955 956 957 958 959
		set_bit(XPT_CLOSE, &xprt->xpt_flags);
		dprintk("queuing xprt %p for closing\n", xprt);

		/* a thread will dequeue and close it soon */
		svc_xprt_enqueue(xprt);
	}
960
	spin_unlock_bh(&serv->sv_lock);
961 962 963 964

	mod_timer(&serv->sv_temptimer, jiffies + svc_conn_age_period * HZ);
}

965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007
/* Close temporary transports whose xpt_local matches server_addr immediately
 * instead of waiting for them to be picked up by the timer.
 *
 * This is meant to be called from a notifier_block that runs when an ip
 * address is deleted.
 */
void svc_age_temp_xprts_now(struct svc_serv *serv, struct sockaddr *server_addr)
{
	struct svc_xprt *xprt;
	struct svc_sock *svsk;
	struct socket *sock;
	struct list_head *le, *next;
	LIST_HEAD(to_be_closed);
	struct linger no_linger = {
		.l_onoff = 1,
		.l_linger = 0,
	};

	spin_lock_bh(&serv->sv_lock);
	list_for_each_safe(le, next, &serv->sv_tempsocks) {
		xprt = list_entry(le, struct svc_xprt, xpt_list);
		if (rpc_cmp_addr(server_addr, (struct sockaddr *)
				&xprt->xpt_local)) {
			dprintk("svc_age_temp_xprts_now: found %p\n", xprt);
			list_move(le, &to_be_closed);
		}
	}
	spin_unlock_bh(&serv->sv_lock);

	while (!list_empty(&to_be_closed)) {
		le = to_be_closed.next;
		list_del_init(le);
		xprt = list_entry(le, struct svc_xprt, xpt_list);
		dprintk("svc_age_temp_xprts_now: closing %p\n", xprt);
		svsk = container_of(xprt, struct svc_sock, sk_xprt);
		sock = svsk->sk_sock;
		kernel_setsockopt(sock, SOL_SOCKET, SO_LINGER,
				  (char *)&no_linger, sizeof(no_linger));
		svc_close_xprt(xprt);
	}
}
EXPORT_SYMBOL_GPL(svc_age_temp_xprts_now);

1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020
static void call_xpt_users(struct svc_xprt *xprt)
{
	struct svc_xpt_user *u;

	spin_lock(&xprt->xpt_lock);
	while (!list_empty(&xprt->xpt_users)) {
		u = list_first_entry(&xprt->xpt_users, struct svc_xpt_user, list);
		list_del(&u->list);
		u->callback(u);
	}
	spin_unlock(&xprt->xpt_lock);
}

1021 1022 1023
/*
 * Remove a dead transport
 */
1024
static void svc_delete_xprt(struct svc_xprt *xprt)
1025 1026
{
	struct svc_serv	*serv = xprt->xpt_server;
1027 1028 1029 1030
	struct svc_deferred_req *dr;

	/* Only do this once */
	if (test_and_set_bit(XPT_DEAD, &xprt->xpt_flags))
1031
		BUG();
1032 1033 1034 1035 1036

	dprintk("svc: svc_delete_xprt(%p)\n", xprt);
	xprt->xpt_ops->xpo_detach(xprt);

	spin_lock_bh(&serv->sv_lock);
1037
	list_del_init(&xprt->xpt_list);
1038
	WARN_ON_ONCE(!list_empty(&xprt->xpt_ready));
1039 1040
	if (test_bit(XPT_TEMP, &xprt->xpt_flags))
		serv->sv_tmpcnt--;
1041
	spin_unlock_bh(&serv->sv_lock);
1042

1043
	while ((dr = svc_deferred_dequeue(xprt)) != NULL)
1044 1045
		kfree(dr);

1046
	call_xpt_users(xprt);
1047
	svc_xprt_put(xprt);
1048 1049 1050 1051 1052 1053 1054 1055
}

void svc_close_xprt(struct svc_xprt *xprt)
{
	set_bit(XPT_CLOSE, &xprt->xpt_flags);
	if (test_and_set_bit(XPT_BUSY, &xprt->xpt_flags))
		/* someone else will have to effect the close */
		return;
J
J. Bruce Fields 已提交
1056 1057 1058 1059 1060 1061
	/*
	 * We expect svc_close_xprt() to work even when no threads are
	 * running (e.g., while configuring the server before starting
	 * any threads), so if the transport isn't busy, we delete
	 * it ourself:
	 */
1062 1063
	svc_delete_xprt(xprt);
}
1064
EXPORT_SYMBOL_GPL(svc_close_xprt);
1065

1066
static int svc_close_list(struct svc_serv *serv, struct list_head *xprt_list, struct net *net)
1067 1068
{
	struct svc_xprt *xprt;
1069
	int ret = 0;
1070

1071
	spin_lock(&serv->sv_lock);
1072
	list_for_each_entry(xprt, xprt_list, xpt_list) {
1073 1074
		if (xprt->xpt_net != net)
			continue;
1075
		ret++;
1076
		set_bit(XPT_CLOSE, &xprt->xpt_flags);
1077
		svc_xprt_enqueue(xprt);
1078
	}
1079
	spin_unlock(&serv->sv_lock);
1080
	return ret;
1081 1082
}

1083
static struct svc_xprt *svc_dequeue_net(struct svc_serv *serv, struct net *net)
1084
{
1085
	struct svc_pool *pool;
1086 1087
	struct svc_xprt *xprt;
	struct svc_xprt *tmp;
1088 1089 1090 1091 1092 1093
	int i;

	for (i = 0; i < serv->sv_nrpools; i++) {
		pool = &serv->sv_pools[i];

		spin_lock_bh(&pool->sp_lock);
1094
		list_for_each_entry_safe(xprt, tmp, &pool->sp_sockets, xpt_ready) {
1095 1096
			if (xprt->xpt_net != net)
				continue;
1097
			list_del_init(&xprt->xpt_ready);
1098 1099
			spin_unlock_bh(&pool->sp_lock);
			return xprt;
1100 1101 1102
		}
		spin_unlock_bh(&pool->sp_lock);
	}
1103
	return NULL;
1104 1105
}

1106
static void svc_clean_up_xprts(struct svc_serv *serv, struct net *net)
1107 1108
{
	struct svc_xprt *xprt;
1109

1110 1111
	while ((xprt = svc_dequeue_net(serv, net))) {
		set_bit(XPT_CLOSE, &xprt->xpt_flags);
1112
		svc_delete_xprt(xprt);
1113
	}
1114 1115
}

1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127
/*
 * Server threads may still be running (especially in the case where the
 * service is still running in other network namespaces).
 *
 * So we shut down sockets the same way we would on a running server, by
 * setting XPT_CLOSE, enqueuing, and letting a thread pick it up to do
 * the close.  In the case there are no such other threads,
 * threads running, svc_clean_up_xprts() does a simple version of a
 * server's main event loop, and in the case where there are other
 * threads, we may need to wait a little while and then check again to
 * see if they're done.
 */
1128
void svc_close_net(struct svc_serv *serv, struct net *net)
1129
{
1130
	int delay = 0;
1131

1132 1133 1134 1135 1136 1137
	while (svc_close_list(serv, &serv->sv_permsocks, net) +
	       svc_close_list(serv, &serv->sv_tempsocks, net)) {

		svc_clean_up_xprts(serv, net);
		msleep(delay++);
	}
1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149
}

/*
 * Handle defer and revisit of requests
 */

static void svc_revisit(struct cache_deferred_req *dreq, int too_many)
{
	struct svc_deferred_req *dr =
		container_of(dreq, struct svc_deferred_req, handle);
	struct svc_xprt *xprt = dr->xprt;

1150 1151 1152 1153 1154
	spin_lock(&xprt->xpt_lock);
	set_bit(XPT_DEFERRED, &xprt->xpt_flags);
	if (too_many || test_bit(XPT_DEAD, &xprt->xpt_flags)) {
		spin_unlock(&xprt->xpt_lock);
		dprintk("revisit canceled\n");
1155
		svc_xprt_put(xprt);
1156
		trace_svc_drop_deferred(dr);
1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167
		kfree(dr);
		return;
	}
	dprintk("revisit queued\n");
	dr->xprt = NULL;
	list_add(&dr->handle.recent, &xprt->xpt_deferred);
	spin_unlock(&xprt->xpt_lock);
	svc_xprt_enqueue(xprt);
	svc_xprt_put(xprt);
}

1168 1169 1170 1171 1172 1173 1174 1175 1176
/*
 * Save the request off for later processing. The request buffer looks
 * like this:
 *
 * <xprt-header><rpc-header><rpc-pagelist><rpc-tail>
 *
 * This code can only handle requests that consist of an xprt-header
 * and rpc-header.
 */
1177 1178 1179 1180 1181
static struct cache_deferred_req *svc_defer(struct cache_req *req)
{
	struct svc_rqst *rqstp = container_of(req, struct svc_rqst, rq_chandle);
	struct svc_deferred_req *dr;

1182
	if (rqstp->rq_arg.page_len || !test_bit(RQ_USEDEFERRAL, &rqstp->rq_flags))
1183 1184 1185 1186 1187
		return NULL; /* if more than a page, give up FIXME */
	if (rqstp->rq_deferred) {
		dr = rqstp->rq_deferred;
		rqstp->rq_deferred = NULL;
	} else {
1188 1189
		size_t skip;
		size_t size;
1190
		/* FIXME maybe discard if size too large */
1191
		size = sizeof(struct svc_deferred_req) + rqstp->rq_arg.len;
1192 1193 1194 1195 1196 1197 1198 1199 1200 1201
		dr = kmalloc(size, GFP_KERNEL);
		if (dr == NULL)
			return NULL;

		dr->handle.owner = rqstp->rq_server;
		dr->prot = rqstp->rq_prot;
		memcpy(&dr->addr, &rqstp->rq_addr, rqstp->rq_addrlen);
		dr->addrlen = rqstp->rq_addrlen;
		dr->daddr = rqstp->rq_daddr;
		dr->argslen = rqstp->rq_arg.len >> 2;
1202 1203 1204 1205 1206 1207
		dr->xprt_hlen = rqstp->rq_xprt_hlen;

		/* back up head to the start of the buffer and copy */
		skip = rqstp->rq_arg.len - rqstp->rq_arg.head[0].iov_len;
		memcpy(dr->args, rqstp->rq_arg.head[0].iov_base - skip,
		       dr->argslen << 2);
1208 1209 1210
	}
	svc_xprt_get(rqstp->rq_xprt);
	dr->xprt = rqstp->rq_xprt;
1211
	set_bit(RQ_DROPME, &rqstp->rq_flags);
1212 1213

	dr->handle.revisit = svc_revisit;
1214
	trace_svc_defer(rqstp);
1215 1216 1217 1218 1219 1220 1221 1222 1223 1224
	return &dr->handle;
}

/*
 * recv data from a deferred request into an active one
 */
static int svc_deferred_recv(struct svc_rqst *rqstp)
{
	struct svc_deferred_req *dr = rqstp->rq_deferred;

1225 1226 1227 1228
	/* setup iov_base past transport header */
	rqstp->rq_arg.head[0].iov_base = dr->args + (dr->xprt_hlen>>2);
	/* The iov_len does not include the transport header bytes */
	rqstp->rq_arg.head[0].iov_len = (dr->argslen<<2) - dr->xprt_hlen;
1229
	rqstp->rq_arg.page_len = 0;
1230 1231
	/* The rq_arg.len includes the transport header bytes */
	rqstp->rq_arg.len     = dr->argslen<<2;
1232 1233 1234
	rqstp->rq_prot        = dr->prot;
	memcpy(&rqstp->rq_addr, &dr->addr, dr->addrlen);
	rqstp->rq_addrlen     = dr->addrlen;
1235 1236
	/* Save off transport header len in case we get deferred again */
	rqstp->rq_xprt_hlen   = dr->xprt_hlen;
1237 1238
	rqstp->rq_daddr       = dr->daddr;
	rqstp->rq_respages    = rqstp->rq_pages;
1239
	return (dr->argslen<<2) - dr->xprt_hlen;
1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254
}


static struct svc_deferred_req *svc_deferred_dequeue(struct svc_xprt *xprt)
{
	struct svc_deferred_req *dr = NULL;

	if (!test_bit(XPT_DEFERRED, &xprt->xpt_flags))
		return NULL;
	spin_lock(&xprt->xpt_lock);
	if (!list_empty(&xprt->xpt_deferred)) {
		dr = list_entry(xprt->xpt_deferred.next,
				struct svc_deferred_req,
				handle.recent);
		list_del_init(&dr->handle.recent);
1255
		trace_svc_revisit_deferred(dr);
1256 1257
	} else
		clear_bit(XPT_DEFERRED, &xprt->xpt_flags);
1258 1259 1260
	spin_unlock(&xprt->xpt_lock);
	return dr;
}
1261

1262 1263 1264 1265
/**
 * svc_find_xprt - find an RPC transport instance
 * @serv: pointer to svc_serv to search
 * @xcl_name: C string containing transport's class name
1266
 * @net: owner net pointer
1267 1268 1269
 * @af: Address family of transport's local address
 * @port: transport's IP port number
 *
1270 1271 1272 1273 1274 1275 1276 1277
 * Return the transport instance pointer for the endpoint accepting
 * connections/peer traffic from the specified transport class,
 * address family and port.
 *
 * Specifying 0 for the address family or port is effectively a
 * wild-card, and will result in matching the first transport in the
 * service's list that has a matching class name.
 */
1278
struct svc_xprt *svc_find_xprt(struct svc_serv *serv, const char *xcl_name,
1279 1280
			       struct net *net, const sa_family_t af,
			       const unsigned short port)
1281 1282 1283 1284 1285
{
	struct svc_xprt *xprt;
	struct svc_xprt *found = NULL;

	/* Sanity check the args */
1286
	if (serv == NULL || xcl_name == NULL)
1287 1288 1289 1290
		return found;

	spin_lock_bh(&serv->sv_lock);
	list_for_each_entry(xprt, &serv->sv_permsocks, xpt_list) {
1291 1292
		if (xprt->xpt_net != net)
			continue;
1293 1294 1295 1296
		if (strcmp(xprt->xpt_class->xcl_name, xcl_name))
			continue;
		if (af != AF_UNSPEC && af != xprt->xpt_local.ss_family)
			continue;
1297
		if (port != 0 && port != svc_xprt_local_port(xprt))
1298 1299
			continue;
		found = xprt;
1300
		svc_xprt_get(xprt);
1301 1302 1303 1304 1305 1306
		break;
	}
	spin_unlock_bh(&serv->sv_lock);
	return found;
}
EXPORT_SYMBOL_GPL(svc_find_xprt);
1307

1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331
static int svc_one_xprt_name(const struct svc_xprt *xprt,
			     char *pos, int remaining)
{
	int len;

	len = snprintf(pos, remaining, "%s %u\n",
			xprt->xpt_class->xcl_name,
			svc_xprt_local_port(xprt));
	if (len >= remaining)
		return -ENAMETOOLONG;
	return len;
}

/**
 * svc_xprt_names - format a buffer with a list of transport names
 * @serv: pointer to an RPC service
 * @buf: pointer to a buffer to be filled in
 * @buflen: length of buffer to be filled in
 *
 * Fills in @buf with a string containing a list of transport names,
 * each name terminated with '\n'.
 *
 * Returns positive length of the filled-in string on success; otherwise
 * a negative errno value is returned if an error occurs.
1332
 */
1333
int svc_xprt_names(struct svc_serv *serv, char *buf, const int buflen)
1334 1335
{
	struct svc_xprt *xprt;
1336 1337
	int len, totlen;
	char *pos;
1338 1339 1340 1341 1342 1343

	/* Sanity check args */
	if (!serv)
		return 0;

	spin_lock_bh(&serv->sv_lock);
1344 1345 1346

	pos = buf;
	totlen = 0;
1347
	list_for_each_entry(xprt, &serv->sv_permsocks, xpt_list) {
1348 1349 1350 1351 1352 1353
		len = svc_one_xprt_name(xprt, pos, buflen - totlen);
		if (len < 0) {
			*buf = '\0';
			totlen = len;
		}
		if (len <= 0)
1354
			break;
1355 1356

		pos += len;
1357 1358
		totlen += len;
	}
1359

1360 1361 1362 1363
	spin_unlock_bh(&serv->sv_lock);
	return totlen;
}
EXPORT_SYMBOL_GPL(svc_xprt_names);
1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408


/*----------------------------------------------------------------------------*/

static void *svc_pool_stats_start(struct seq_file *m, loff_t *pos)
{
	unsigned int pidx = (unsigned int)*pos;
	struct svc_serv *serv = m->private;

	dprintk("svc_pool_stats_start, *pidx=%u\n", pidx);

	if (!pidx)
		return SEQ_START_TOKEN;
	return (pidx > serv->sv_nrpools ? NULL : &serv->sv_pools[pidx-1]);
}

static void *svc_pool_stats_next(struct seq_file *m, void *p, loff_t *pos)
{
	struct svc_pool *pool = p;
	struct svc_serv *serv = m->private;

	dprintk("svc_pool_stats_next, *pos=%llu\n", *pos);

	if (p == SEQ_START_TOKEN) {
		pool = &serv->sv_pools[0];
	} else {
		unsigned int pidx = (pool - &serv->sv_pools[0]);
		if (pidx < serv->sv_nrpools-1)
			pool = &serv->sv_pools[pidx+1];
		else
			pool = NULL;
	}
	++*pos;
	return pool;
}

static void svc_pool_stats_stop(struct seq_file *m, void *p)
{
}

static int svc_pool_stats_show(struct seq_file *m, void *p)
{
	struct svc_pool *pool = p;

	if (p == SEQ_START_TOKEN) {
1409
		seq_puts(m, "# pool packets-arrived sockets-enqueued threads-woken threads-timedout\n");
1410 1411 1412
		return 0;
	}

1413
	seq_printf(m, "%u %lu %lu %lu %lu\n",
1414
		pool->sp_id,
1415
		(unsigned long)atomic_long_read(&pool->sp_stats.packets),
1416
		pool->sp_stats.sockets_queued,
1417 1418
		(unsigned long)atomic_long_read(&pool->sp_stats.threads_woken),
		(unsigned long)atomic_long_read(&pool->sp_stats.threads_timedout));
1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441

	return 0;
}

static const struct seq_operations svc_pool_stats_seq_ops = {
	.start	= svc_pool_stats_start,
	.next	= svc_pool_stats_next,
	.stop	= svc_pool_stats_stop,
	.show	= svc_pool_stats_show,
};

int svc_pool_stats_open(struct svc_serv *serv, struct file *file)
{
	int err;

	err = seq_open(file, &svc_pool_stats_seq_ops);
	if (!err)
		((struct seq_file *) file->private_data)->private = serv;
	return err;
}
EXPORT_SYMBOL(svc_pool_stats_open);

/*----------------------------------------------------------------------------*/