2 * linux/net/sunrpc/svc_xprt.c
4 * Author: Tom Tucker <tom@opengridcomputing.com>
7 #include <linux/sched.h>
8 #include <linux/smp_lock.h>
9 #include <linux/errno.h>
10 #include <linux/freezer.h>
11 #include <linux/kthread.h>
12 #include <linux/slab.h>
14 #include <linux/sunrpc/stats.h>
15 #include <linux/sunrpc/svc_xprt.h>
16 #include <linux/sunrpc/svcsock.h>
18 #define RPCDBG_FACILITY RPCDBG_SVCXPRT
20 static struct svc_deferred_req
*svc_deferred_dequeue(struct svc_xprt
*xprt
);
21 static int svc_deferred_recv(struct svc_rqst
*rqstp
);
22 static struct cache_deferred_req
*svc_defer(struct cache_req
*req
);
23 static void svc_age_temp_xprts(unsigned long closure
);
25 /* apparently the "standard" is that clients close
26 * idle connections after 5 minutes, servers after
28 * http://www.connectathon.org/talks96/nfstcp.pdf
30 static int svc_conn_age_period
= 6*60;
32 /* List of registered transport classes */
33 static DEFINE_SPINLOCK(svc_xprt_class_lock
);
34 static LIST_HEAD(svc_xprt_class_list
);
36 /* SMP locking strategy:
38 * svc_pool->sp_lock protects most of the fields of that pool.
39 * svc_serv->sv_lock protects sv_tempsocks, sv_permsocks, sv_tmpcnt.
40 * when both need to be taken (rare), svc_serv->sv_lock is first.
41 * BKL protects svc_serv->sv_nrthread.
42 * svc_sock->sk_lock protects the svc_sock->sk_deferred list
43 * and the ->sk_info_authunix cache.
45 * The XPT_BUSY bit in xprt->xpt_flags prevents a transport being
46 * enqueued multiply. During normal transport processing this bit
47 * is set by svc_xprt_enqueue and cleared by svc_xprt_received.
48 * Providers should not manipulate this bit directly.
50 * Some flags can be set to certain values at any time
51 * providing that certain rules are followed:
54 * - Can be set or cleared at any time.
55 * - After a set, svc_xprt_enqueue must be called to enqueue
56 * the transport for processing.
57 * - After a clear, the transport must be read/accepted.
58 * If this succeeds, it must be set again.
60 * - Can set at any time. It is never cleared.
62 * - Can only be set while XPT_BUSY is held which ensures
63 * that no other thread will be using the transport or will
64 * try to set XPT_DEAD.
67 int svc_reg_xprt_class(struct svc_xprt_class
*xcl
)
69 struct svc_xprt_class
*cl
;
72 dprintk("svc: Adding svc transport class '%s'\n", xcl
->xcl_name
);
74 INIT_LIST_HEAD(&xcl
->xcl_list
);
75 spin_lock(&svc_xprt_class_lock
);
76 /* Make sure there isn't already a class with the same name */
77 list_for_each_entry(cl
, &svc_xprt_class_list
, xcl_list
) {
78 if (strcmp(xcl
->xcl_name
, cl
->xcl_name
) == 0)
81 list_add_tail(&xcl
->xcl_list
, &svc_xprt_class_list
);
84 spin_unlock(&svc_xprt_class_lock
);
87 EXPORT_SYMBOL_GPL(svc_reg_xprt_class
);
89 void svc_unreg_xprt_class(struct svc_xprt_class
*xcl
)
91 dprintk("svc: Removing svc transport class '%s'\n", xcl
->xcl_name
);
92 spin_lock(&svc_xprt_class_lock
);
93 list_del_init(&xcl
->xcl_list
);
94 spin_unlock(&svc_xprt_class_lock
);
96 EXPORT_SYMBOL_GPL(svc_unreg_xprt_class
);
99 * Format the transport list for printing
101 int svc_print_xprts(char *buf
, int maxlen
)
103 struct list_head
*le
;
108 spin_lock(&svc_xprt_class_lock
);
109 list_for_each(le
, &svc_xprt_class_list
) {
111 struct svc_xprt_class
*xcl
=
112 list_entry(le
, struct svc_xprt_class
, xcl_list
);
114 sprintf(tmpstr
, "%s %d\n", xcl
->xcl_name
, xcl
->xcl_max_payload
);
115 slen
= strlen(tmpstr
);
116 if (len
+ slen
> maxlen
)
121 spin_unlock(&svc_xprt_class_lock
);
126 static void svc_xprt_free(struct kref
*kref
)
128 struct svc_xprt
*xprt
=
129 container_of(kref
, struct svc_xprt
, xpt_ref
);
130 struct module
*owner
= xprt
->xpt_class
->xcl_owner
;
131 if (test_bit(XPT_CACHE_AUTH
, &xprt
->xpt_flags
) &&
132 xprt
->xpt_auth_cache
!= NULL
)
133 svcauth_unix_info_release(xprt
->xpt_auth_cache
);
134 xprt
->xpt_ops
->xpo_free(xprt
);
138 void svc_xprt_put(struct svc_xprt
*xprt
)
140 kref_put(&xprt
->xpt_ref
, svc_xprt_free
);
142 EXPORT_SYMBOL_GPL(svc_xprt_put
);
145 * Called by transport drivers to initialize the transport independent
146 * portion of the transport instance.
148 void svc_xprt_init(struct svc_xprt_class
*xcl
, struct svc_xprt
*xprt
,
149 struct svc_serv
*serv
)
151 memset(xprt
, 0, sizeof(*xprt
));
152 xprt
->xpt_class
= xcl
;
153 xprt
->xpt_ops
= xcl
->xcl_ops
;
154 kref_init(&xprt
->xpt_ref
);
155 xprt
->xpt_server
= serv
;
156 INIT_LIST_HEAD(&xprt
->xpt_list
);
157 INIT_LIST_HEAD(&xprt
->xpt_ready
);
158 INIT_LIST_HEAD(&xprt
->xpt_deferred
);
159 mutex_init(&xprt
->xpt_mutex
);
160 spin_lock_init(&xprt
->xpt_lock
);
161 set_bit(XPT_BUSY
, &xprt
->xpt_flags
);
162 rpc_init_wait_queue(&xprt
->xpt_bc_pending
, "xpt_bc_pending");
164 EXPORT_SYMBOL_GPL(svc_xprt_init
);
166 static struct svc_xprt
*__svc_xpo_create(struct svc_xprt_class
*xcl
,
167 struct svc_serv
*serv
,
169 const unsigned short port
,
172 struct sockaddr_in sin
= {
173 .sin_family
= AF_INET
,
174 .sin_addr
.s_addr
= htonl(INADDR_ANY
),
175 .sin_port
= htons(port
),
177 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
178 struct sockaddr_in6 sin6
= {
179 .sin6_family
= AF_INET6
,
180 .sin6_addr
= IN6ADDR_ANY_INIT
,
181 .sin6_port
= htons(port
),
183 #endif /* defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) */
184 struct sockaddr
*sap
;
189 sap
= (struct sockaddr
*)&sin
;
192 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
194 sap
= (struct sockaddr
*)&sin6
;
197 #endif /* defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) */
199 return ERR_PTR(-EAFNOSUPPORT
);
202 return xcl
->xcl_ops
->xpo_create(serv
, sap
, len
, flags
);
205 int svc_create_xprt(struct svc_serv
*serv
, const char *xprt_name
,
206 const int family
, const unsigned short port
,
209 struct svc_xprt_class
*xcl
;
211 dprintk("svc: creating transport %s[%d]\n", xprt_name
, port
);
212 spin_lock(&svc_xprt_class_lock
);
213 list_for_each_entry(xcl
, &svc_xprt_class_list
, xcl_list
) {
214 struct svc_xprt
*newxprt
;
216 if (strcmp(xprt_name
, xcl
->xcl_name
))
219 if (!try_module_get(xcl
->xcl_owner
))
222 spin_unlock(&svc_xprt_class_lock
);
223 newxprt
= __svc_xpo_create(xcl
, serv
, family
, port
, flags
);
224 if (IS_ERR(newxprt
)) {
225 module_put(xcl
->xcl_owner
);
226 return PTR_ERR(newxprt
);
229 clear_bit(XPT_TEMP
, &newxprt
->xpt_flags
);
230 spin_lock_bh(&serv
->sv_lock
);
231 list_add(&newxprt
->xpt_list
, &serv
->sv_permsocks
);
232 spin_unlock_bh(&serv
->sv_lock
);
233 clear_bit(XPT_BUSY
, &newxprt
->xpt_flags
);
234 return svc_xprt_local_port(newxprt
);
237 spin_unlock(&svc_xprt_class_lock
);
238 dprintk("svc: transport %s not found\n", xprt_name
);
240 /* This errno is exposed to user space. Provide a reasonable
241 * perror msg for a bad transport. */
242 return -EPROTONOSUPPORT
;
244 EXPORT_SYMBOL_GPL(svc_create_xprt
);
247 * Copy the local and remote xprt addresses to the rqstp structure
249 void svc_xprt_copy_addrs(struct svc_rqst
*rqstp
, struct svc_xprt
*xprt
)
251 struct sockaddr
*sin
;
253 memcpy(&rqstp
->rq_addr
, &xprt
->xpt_remote
, xprt
->xpt_remotelen
);
254 rqstp
->rq_addrlen
= xprt
->xpt_remotelen
;
257 * Destination address in request is needed for binding the
258 * source address in RPC replies/callbacks later.
260 sin
= (struct sockaddr
*)&xprt
->xpt_local
;
261 switch (sin
->sa_family
) {
263 rqstp
->rq_daddr
.addr
= ((struct sockaddr_in
*)sin
)->sin_addr
;
266 rqstp
->rq_daddr
.addr6
= ((struct sockaddr_in6
*)sin
)->sin6_addr
;
270 EXPORT_SYMBOL_GPL(svc_xprt_copy_addrs
);
273 * svc_print_addr - Format rq_addr field for printing
274 * @rqstp: svc_rqst struct containing address to print
275 * @buf: target buffer for formatted address
276 * @len: length of target buffer
279 char *svc_print_addr(struct svc_rqst
*rqstp
, char *buf
, size_t len
)
281 return __svc_print_addr(svc_addr(rqstp
), buf
, len
);
283 EXPORT_SYMBOL_GPL(svc_print_addr
);
286 * Queue up an idle server thread. Must have pool->sp_lock held.
287 * Note: this is really a stack rather than a queue, so that we only
288 * use as many different threads as we need, and the rest don't pollute
291 static void svc_thread_enqueue(struct svc_pool
*pool
, struct svc_rqst
*rqstp
)
293 list_add(&rqstp
->rq_list
, &pool
->sp_threads
);
297 * Dequeue an nfsd thread. Must have pool->sp_lock held.
299 static void svc_thread_dequeue(struct svc_pool
*pool
, struct svc_rqst
*rqstp
)
301 list_del(&rqstp
->rq_list
);
305 * Queue up a transport with data pending. If there are idle nfsd
306 * processes, wake 'em up.
309 void svc_xprt_enqueue(struct svc_xprt
*xprt
)
311 struct svc_serv
*serv
= xprt
->xpt_server
;
312 struct svc_pool
*pool
;
313 struct svc_rqst
*rqstp
;
316 if (!(xprt
->xpt_flags
&
317 ((1<<XPT_CONN
)|(1<<XPT_DATA
)|(1<<XPT_CLOSE
)|(1<<XPT_DEFERRED
))))
321 pool
= svc_pool_for_cpu(xprt
->xpt_server
, cpu
);
324 spin_lock_bh(&pool
->sp_lock
);
326 if (!list_empty(&pool
->sp_threads
) &&
327 !list_empty(&pool
->sp_sockets
))
330 "threads and transports both waiting??\n");
332 if (test_bit(XPT_DEAD
, &xprt
->xpt_flags
)) {
333 /* Don't enqueue dead transports */
334 dprintk("svc: transport %p is dead, not enqueued\n", xprt
);
338 pool
->sp_stats
.packets
++;
340 /* Mark transport as busy. It will remain in this state until
341 * the provider calls svc_xprt_received. We update XPT_BUSY
342 * atomically because it also guards against trying to enqueue
343 * the transport twice.
345 if (test_and_set_bit(XPT_BUSY
, &xprt
->xpt_flags
)) {
346 /* Don't enqueue transport while already enqueued */
347 dprintk("svc: transport %p busy, not enqueued\n", xprt
);
350 BUG_ON(xprt
->xpt_pool
!= NULL
);
351 xprt
->xpt_pool
= pool
;
353 /* Handle pending connection */
354 if (test_bit(XPT_CONN
, &xprt
->xpt_flags
))
357 /* Handle close in-progress */
358 if (test_bit(XPT_CLOSE
, &xprt
->xpt_flags
))
361 /* Check if we have space to reply to a request */
362 if (!xprt
->xpt_ops
->xpo_has_wspace(xprt
)) {
363 /* Don't enqueue while not enough space for reply */
364 dprintk("svc: no write space, transport %p not enqueued\n",
366 xprt
->xpt_pool
= NULL
;
367 clear_bit(XPT_BUSY
, &xprt
->xpt_flags
);
372 if (!list_empty(&pool
->sp_threads
)) {
373 rqstp
= list_entry(pool
->sp_threads
.next
,
376 dprintk("svc: transport %p served by daemon %p\n",
378 svc_thread_dequeue(pool
, rqstp
);
381 "svc_xprt_enqueue: server %p, rq_xprt=%p!\n",
382 rqstp
, rqstp
->rq_xprt
);
383 rqstp
->rq_xprt
= xprt
;
385 rqstp
->rq_reserved
= serv
->sv_max_mesg
;
386 atomic_add(rqstp
->rq_reserved
, &xprt
->xpt_reserved
);
387 pool
->sp_stats
.threads_woken
++;
388 BUG_ON(xprt
->xpt_pool
!= pool
);
389 wake_up(&rqstp
->rq_wait
);
391 dprintk("svc: transport %p put into queue\n", xprt
);
392 list_add_tail(&xprt
->xpt_ready
, &pool
->sp_sockets
);
393 pool
->sp_stats
.sockets_queued
++;
394 BUG_ON(xprt
->xpt_pool
!= pool
);
398 spin_unlock_bh(&pool
->sp_lock
);
400 EXPORT_SYMBOL_GPL(svc_xprt_enqueue
);
403 * Dequeue the first transport. Must be called with the pool->sp_lock held.
405 static struct svc_xprt
*svc_xprt_dequeue(struct svc_pool
*pool
)
407 struct svc_xprt
*xprt
;
409 if (list_empty(&pool
->sp_sockets
))
412 xprt
= list_entry(pool
->sp_sockets
.next
,
413 struct svc_xprt
, xpt_ready
);
414 list_del_init(&xprt
->xpt_ready
);
416 dprintk("svc: transport %p dequeued, inuse=%d\n",
417 xprt
, atomic_read(&xprt
->xpt_ref
.refcount
));
423 * svc_xprt_received conditionally queues the transport for processing
424 * by another thread. The caller must hold the XPT_BUSY bit and must
425 * not thereafter touch transport data.
427 * Note: XPT_DATA only gets cleared when a read-attempt finds no (or
428 * insufficient) data.
430 void svc_xprt_received(struct svc_xprt
*xprt
)
432 BUG_ON(!test_bit(XPT_BUSY
, &xprt
->xpt_flags
));
433 xprt
->xpt_pool
= NULL
;
434 clear_bit(XPT_BUSY
, &xprt
->xpt_flags
);
435 svc_xprt_enqueue(xprt
);
437 EXPORT_SYMBOL_GPL(svc_xprt_received
);
440 * svc_reserve - change the space reserved for the reply to a request.
441 * @rqstp: The request in question
442 * @space: new max space to reserve
444 * Each request reserves some space on the output queue of the transport
445 * to make sure the reply fits. This function reduces that reserved
446 * space to be the amount of space used already, plus @space.
449 void svc_reserve(struct svc_rqst
*rqstp
, int space
)
451 space
+= rqstp
->rq_res
.head
[0].iov_len
;
453 if (space
< rqstp
->rq_reserved
) {
454 struct svc_xprt
*xprt
= rqstp
->rq_xprt
;
455 atomic_sub((rqstp
->rq_reserved
- space
), &xprt
->xpt_reserved
);
456 rqstp
->rq_reserved
= space
;
458 svc_xprt_enqueue(xprt
);
461 EXPORT_SYMBOL_GPL(svc_reserve
);
463 static void svc_xprt_release(struct svc_rqst
*rqstp
)
465 struct svc_xprt
*xprt
= rqstp
->rq_xprt
;
467 rqstp
->rq_xprt
->xpt_ops
->xpo_release_rqst(rqstp
);
469 kfree(rqstp
->rq_deferred
);
470 rqstp
->rq_deferred
= NULL
;
472 svc_free_res_pages(rqstp
);
473 rqstp
->rq_res
.page_len
= 0;
474 rqstp
->rq_res
.page_base
= 0;
476 /* Reset response buffer and release
478 * But first, check that enough space was reserved
479 * for the reply, otherwise we have a bug!
481 if ((rqstp
->rq_res
.len
) > rqstp
->rq_reserved
)
482 printk(KERN_ERR
"RPC request reserved %d but used %d\n",
486 rqstp
->rq_res
.head
[0].iov_len
= 0;
487 svc_reserve(rqstp
, 0);
488 rqstp
->rq_xprt
= NULL
;
494 * External function to wake up a server waiting for data
495 * This really only makes sense for services like lockd
496 * which have exactly one thread anyway.
498 void svc_wake_up(struct svc_serv
*serv
)
500 struct svc_rqst
*rqstp
;
502 struct svc_pool
*pool
;
504 for (i
= 0; i
< serv
->sv_nrpools
; i
++) {
505 pool
= &serv
->sv_pools
[i
];
507 spin_lock_bh(&pool
->sp_lock
);
508 if (!list_empty(&pool
->sp_threads
)) {
509 rqstp
= list_entry(pool
->sp_threads
.next
,
512 dprintk("svc: daemon %p woken up.\n", rqstp
);
514 svc_thread_dequeue(pool, rqstp);
515 rqstp->rq_xprt = NULL;
517 wake_up(&rqstp
->rq_wait
);
519 spin_unlock_bh(&pool
->sp_lock
);
522 EXPORT_SYMBOL_GPL(svc_wake_up
);
524 int svc_port_is_privileged(struct sockaddr
*sin
)
526 switch (sin
->sa_family
) {
528 return ntohs(((struct sockaddr_in
*)sin
)->sin_port
)
531 return ntohs(((struct sockaddr_in6
*)sin
)->sin6_port
)
539 * Make sure that we don't have too many active connections. If we have,
540 * something must be dropped. It's not clear what will happen if we allow
541 * "too many" connections, but when dealing with network-facing software,
542 * we have to code defensively. Here we do that by imposing hard limits.
544 * There's no point in trying to do random drop here for DoS
545 * prevention. The NFS clients does 1 reconnect in 15 seconds. An
546 * attacker can easily beat that.
548 * The only somewhat efficient mechanism would be if drop old
549 * connections from the same IP first. But right now we don't even
550 * record the client IP in svc_sock.
552 * single-threaded services that expect a lot of clients will probably
553 * need to set sv_maxconn to override the default value which is based
554 * on the number of threads
556 static void svc_check_conn_limits(struct svc_serv
*serv
)
558 unsigned int limit
= serv
->sv_maxconn
? serv
->sv_maxconn
:
559 (serv
->sv_nrthreads
+3) * 20;
561 if (serv
->sv_tmpcnt
> limit
) {
562 struct svc_xprt
*xprt
= NULL
;
563 spin_lock_bh(&serv
->sv_lock
);
564 if (!list_empty(&serv
->sv_tempsocks
)) {
565 if (net_ratelimit()) {
566 /* Try to help the admin */
567 printk(KERN_NOTICE
"%s: too many open "
568 "connections, consider increasing %s\n",
569 serv
->sv_name
, serv
->sv_maxconn
?
570 "the max number of connections." :
571 "the number of threads.");
574 * Always select the oldest connection. It's not fair,
577 xprt
= list_entry(serv
->sv_tempsocks
.prev
,
580 set_bit(XPT_CLOSE
, &xprt
->xpt_flags
);
583 spin_unlock_bh(&serv
->sv_lock
);
586 svc_xprt_enqueue(xprt
);
593 * Receive the next request on any transport. This code is carefully
594 * organised not to touch any cachelines in the shared svc_serv
595 * structure, only cachelines in the local svc_pool.
597 int svc_recv(struct svc_rqst
*rqstp
, long timeout
)
599 struct svc_xprt
*xprt
= NULL
;
600 struct svc_serv
*serv
= rqstp
->rq_server
;
601 struct svc_pool
*pool
= rqstp
->rq_pool
;
605 DECLARE_WAITQUEUE(wait
, current
);
608 dprintk("svc: server %p waiting for data (to = %ld)\n",
613 "svc_recv: service %p, transport not NULL!\n",
615 if (waitqueue_active(&rqstp
->rq_wait
))
617 "svc_recv: service %p, wait queue active!\n",
620 /* now allocate needed pages. If we get a failure, sleep briefly */
621 pages
= (serv
->sv_max_mesg
+ PAGE_SIZE
) / PAGE_SIZE
;
622 for (i
= 0; i
< pages
; i
++)
623 while (rqstp
->rq_pages
[i
] == NULL
) {
624 struct page
*p
= alloc_page(GFP_KERNEL
);
626 set_current_state(TASK_INTERRUPTIBLE
);
627 if (signalled() || kthread_should_stop()) {
628 set_current_state(TASK_RUNNING
);
631 schedule_timeout(msecs_to_jiffies(500));
633 rqstp
->rq_pages
[i
] = p
;
635 rqstp
->rq_pages
[i
++] = NULL
; /* this might be seen in nfs_read_actor */
636 BUG_ON(pages
>= RPCSVC_MAXPAGES
);
638 /* Make arg->head point to first page and arg->pages point to rest */
639 arg
= &rqstp
->rq_arg
;
640 arg
->head
[0].iov_base
= page_address(rqstp
->rq_pages
[0]);
641 arg
->head
[0].iov_len
= PAGE_SIZE
;
642 arg
->pages
= rqstp
->rq_pages
+ 1;
644 /* save at least one page for response */
645 arg
->page_len
= (pages
-2)*PAGE_SIZE
;
646 arg
->len
= (pages
-1)*PAGE_SIZE
;
647 arg
->tail
[0].iov_len
= 0;
651 if (signalled() || kthread_should_stop())
654 /* Normally we will wait up to 5 seconds for any required
655 * cache information to be provided.
657 rqstp
->rq_chandle
.thread_wait
= 5*HZ
;
659 spin_lock_bh(&pool
->sp_lock
);
660 xprt
= svc_xprt_dequeue(pool
);
662 rqstp
->rq_xprt
= xprt
;
664 rqstp
->rq_reserved
= serv
->sv_max_mesg
;
665 atomic_add(rqstp
->rq_reserved
, &xprt
->xpt_reserved
);
667 /* As there is a shortage of threads and this request
668 * had to be queue, don't allow the thread to wait so
669 * long for cache updates.
671 rqstp
->rq_chandle
.thread_wait
= 1*HZ
;
673 /* No data pending. Go to sleep */
674 svc_thread_enqueue(pool
, rqstp
);
677 * We have to be able to interrupt this wait
678 * to bring down the daemons ...
680 set_current_state(TASK_INTERRUPTIBLE
);
683 * checking kthread_should_stop() here allows us to avoid
684 * locking and signalling when stopping kthreads that call
685 * svc_recv. If the thread has already been woken up, then
686 * we can exit here without sleeping. If not, then it
687 * it'll be woken up quickly during the schedule_timeout
689 if (kthread_should_stop()) {
690 set_current_state(TASK_RUNNING
);
691 spin_unlock_bh(&pool
->sp_lock
);
695 add_wait_queue(&rqstp
->rq_wait
, &wait
);
696 spin_unlock_bh(&pool
->sp_lock
);
698 time_left
= schedule_timeout(timeout
);
702 spin_lock_bh(&pool
->sp_lock
);
703 remove_wait_queue(&rqstp
->rq_wait
, &wait
);
705 pool
->sp_stats
.threads_timedout
++;
707 xprt
= rqstp
->rq_xprt
;
709 svc_thread_dequeue(pool
, rqstp
);
710 spin_unlock_bh(&pool
->sp_lock
);
711 dprintk("svc: server %p, no data yet\n", rqstp
);
712 if (signalled() || kthread_should_stop())
718 spin_unlock_bh(&pool
->sp_lock
);
721 if (test_bit(XPT_CLOSE
, &xprt
->xpt_flags
)) {
722 dprintk("svc_recv: found XPT_CLOSE\n");
723 svc_delete_xprt(xprt
);
724 } else if (test_bit(XPT_LISTENER
, &xprt
->xpt_flags
)) {
725 struct svc_xprt
*newxpt
;
726 newxpt
= xprt
->xpt_ops
->xpo_accept(xprt
);
729 * We know this module_get will succeed because the
730 * listener holds a reference too
732 __module_get(newxpt
->xpt_class
->xcl_owner
);
733 svc_check_conn_limits(xprt
->xpt_server
);
734 spin_lock_bh(&serv
->sv_lock
);
735 set_bit(XPT_TEMP
, &newxpt
->xpt_flags
);
736 list_add(&newxpt
->xpt_list
, &serv
->sv_tempsocks
);
738 if (serv
->sv_temptimer
.function
== NULL
) {
739 /* setup timer to age temp transports */
740 setup_timer(&serv
->sv_temptimer
,
742 (unsigned long)serv
);
743 mod_timer(&serv
->sv_temptimer
,
744 jiffies
+ svc_conn_age_period
* HZ
);
746 spin_unlock_bh(&serv
->sv_lock
);
747 svc_xprt_received(newxpt
);
749 svc_xprt_received(xprt
);
751 dprintk("svc: server %p, pool %u, transport %p, inuse=%d\n",
752 rqstp
, pool
->sp_id
, xprt
,
753 atomic_read(&xprt
->xpt_ref
.refcount
));
754 rqstp
->rq_deferred
= svc_deferred_dequeue(xprt
);
755 if (rqstp
->rq_deferred
) {
756 svc_xprt_received(xprt
);
757 len
= svc_deferred_recv(rqstp
);
759 len
= xprt
->xpt_ops
->xpo_recvfrom(rqstp
);
760 svc_xprt_received(xprt
);
762 dprintk("svc: got len=%d\n", len
);
765 /* No data, incomplete (TCP) read, or accept() */
766 if (len
== 0 || len
== -EAGAIN
) {
767 rqstp
->rq_res
.len
= 0;
768 svc_xprt_release(rqstp
);
771 clear_bit(XPT_OLD
, &xprt
->xpt_flags
);
773 rqstp
->rq_secure
= svc_port_is_privileged(svc_addr(rqstp
));
774 rqstp
->rq_chandle
.defer
= svc_defer
;
777 serv
->sv_stats
->netcnt
++;
780 EXPORT_SYMBOL_GPL(svc_recv
);
785 void svc_drop(struct svc_rqst
*rqstp
)
787 dprintk("svc: xprt %p dropped request\n", rqstp
->rq_xprt
);
788 svc_xprt_release(rqstp
);
790 EXPORT_SYMBOL_GPL(svc_drop
);
793 * Return reply to client.
795 int svc_send(struct svc_rqst
*rqstp
)
797 struct svc_xprt
*xprt
;
801 xprt
= rqstp
->rq_xprt
;
805 /* release the receive skb before sending the reply */
806 rqstp
->rq_xprt
->xpt_ops
->xpo_release_rqst(rqstp
);
808 /* calculate over-all length */
810 xb
->len
= xb
->head
[0].iov_len
+
814 /* Grab mutex to serialize outgoing data. */
815 mutex_lock(&xprt
->xpt_mutex
);
816 if (test_bit(XPT_DEAD
, &xprt
->xpt_flags
))
819 len
= xprt
->xpt_ops
->xpo_sendto(rqstp
);
820 mutex_unlock(&xprt
->xpt_mutex
);
821 rpc_wake_up(&xprt
->xpt_bc_pending
);
822 svc_xprt_release(rqstp
);
824 if (len
== -ECONNREFUSED
|| len
== -ENOTCONN
|| len
== -EAGAIN
)
830 * Timer function to close old temporary transports, using
831 * a mark-and-sweep algorithm.
833 static void svc_age_temp_xprts(unsigned long closure
)
835 struct svc_serv
*serv
= (struct svc_serv
*)closure
;
836 struct svc_xprt
*xprt
;
837 struct list_head
*le
, *next
;
838 LIST_HEAD(to_be_aged
);
840 dprintk("svc_age_temp_xprts\n");
842 if (!spin_trylock_bh(&serv
->sv_lock
)) {
843 /* busy, try again 1 sec later */
844 dprintk("svc_age_temp_xprts: busy\n");
845 mod_timer(&serv
->sv_temptimer
, jiffies
+ HZ
);
849 list_for_each_safe(le
, next
, &serv
->sv_tempsocks
) {
850 xprt
= list_entry(le
, struct svc_xprt
, xpt_list
);
852 /* First time through, just mark it OLD. Second time
853 * through, close it. */
854 if (!test_and_set_bit(XPT_OLD
, &xprt
->xpt_flags
))
856 if (atomic_read(&xprt
->xpt_ref
.refcount
) > 1 ||
857 test_bit(XPT_BUSY
, &xprt
->xpt_flags
))
860 list_move(le
, &to_be_aged
);
861 set_bit(XPT_CLOSE
, &xprt
->xpt_flags
);
862 set_bit(XPT_DETACHED
, &xprt
->xpt_flags
);
864 spin_unlock_bh(&serv
->sv_lock
);
866 while (!list_empty(&to_be_aged
)) {
867 le
= to_be_aged
.next
;
868 /* fiddling the xpt_list node is safe 'cos we're XPT_DETACHED */
870 xprt
= list_entry(le
, struct svc_xprt
, xpt_list
);
872 dprintk("queuing xprt %p for closing\n", xprt
);
874 /* a thread will dequeue and close it soon */
875 svc_xprt_enqueue(xprt
);
879 mod_timer(&serv
->sv_temptimer
, jiffies
+ svc_conn_age_period
* HZ
);
883 * Remove a dead transport
885 void svc_delete_xprt(struct svc_xprt
*xprt
)
887 struct svc_serv
*serv
= xprt
->xpt_server
;
888 struct svc_deferred_req
*dr
;
890 /* Only do this once */
891 if (test_and_set_bit(XPT_DEAD
, &xprt
->xpt_flags
))
894 dprintk("svc: svc_delete_xprt(%p)\n", xprt
);
895 xprt
->xpt_ops
->xpo_detach(xprt
);
897 spin_lock_bh(&serv
->sv_lock
);
898 if (!test_and_set_bit(XPT_DETACHED
, &xprt
->xpt_flags
))
899 list_del_init(&xprt
->xpt_list
);
901 * We used to delete the transport from whichever list
902 * it's sk_xprt.xpt_ready node was on, but we don't actually
903 * need to. This is because the only time we're called
904 * while still attached to a queue, the queue itself
905 * is about to be destroyed (in svc_destroy).
907 if (test_bit(XPT_TEMP
, &xprt
->xpt_flags
))
909 spin_unlock_bh(&serv
->sv_lock
);
911 while ((dr
= svc_deferred_dequeue(xprt
)) != NULL
)
917 void svc_close_xprt(struct svc_xprt
*xprt
)
919 set_bit(XPT_CLOSE
, &xprt
->xpt_flags
);
920 if (test_and_set_bit(XPT_BUSY
, &xprt
->xpt_flags
))
921 /* someone else will have to effect the close */
925 svc_delete_xprt(xprt
);
926 clear_bit(XPT_BUSY
, &xprt
->xpt_flags
);
929 EXPORT_SYMBOL_GPL(svc_close_xprt
);
931 void svc_close_all(struct list_head
*xprt_list
)
933 struct svc_xprt
*xprt
;
934 struct svc_xprt
*tmp
;
936 list_for_each_entry_safe(xprt
, tmp
, xprt_list
, xpt_list
) {
937 set_bit(XPT_CLOSE
, &xprt
->xpt_flags
);
938 if (test_bit(XPT_BUSY
, &xprt
->xpt_flags
)) {
939 /* Waiting to be processed, but no threads left,
940 * So just remove it from the waiting list
942 list_del_init(&xprt
->xpt_ready
);
943 clear_bit(XPT_BUSY
, &xprt
->xpt_flags
);
945 svc_close_xprt(xprt
);
950 * Handle defer and revisit of requests
953 static void svc_revisit(struct cache_deferred_req
*dreq
, int too_many
)
955 struct svc_deferred_req
*dr
=
956 container_of(dreq
, struct svc_deferred_req
, handle
);
957 struct svc_xprt
*xprt
= dr
->xprt
;
959 spin_lock(&xprt
->xpt_lock
);
960 set_bit(XPT_DEFERRED
, &xprt
->xpt_flags
);
961 if (too_many
|| test_bit(XPT_DEAD
, &xprt
->xpt_flags
)) {
962 spin_unlock(&xprt
->xpt_lock
);
963 dprintk("revisit canceled\n");
968 dprintk("revisit queued\n");
970 list_add(&dr
->handle
.recent
, &xprt
->xpt_deferred
);
971 spin_unlock(&xprt
->xpt_lock
);
972 svc_xprt_enqueue(xprt
);
977 * Save the request off for later processing. The request buffer looks
980 * <xprt-header><rpc-header><rpc-pagelist><rpc-tail>
982 * This code can only handle requests that consist of an xprt-header
985 static struct cache_deferred_req
*svc_defer(struct cache_req
*req
)
987 struct svc_rqst
*rqstp
= container_of(req
, struct svc_rqst
, rq_chandle
);
988 struct svc_deferred_req
*dr
;
990 if (rqstp
->rq_arg
.page_len
|| !rqstp
->rq_usedeferral
)
991 return NULL
; /* if more than a page, give up FIXME */
992 if (rqstp
->rq_deferred
) {
993 dr
= rqstp
->rq_deferred
;
994 rqstp
->rq_deferred
= NULL
;
998 /* FIXME maybe discard if size too large */
999 size
= sizeof(struct svc_deferred_req
) + rqstp
->rq_arg
.len
;
1000 dr
= kmalloc(size
, GFP_KERNEL
);
1004 dr
->handle
.owner
= rqstp
->rq_server
;
1005 dr
->prot
= rqstp
->rq_prot
;
1006 memcpy(&dr
->addr
, &rqstp
->rq_addr
, rqstp
->rq_addrlen
);
1007 dr
->addrlen
= rqstp
->rq_addrlen
;
1008 dr
->daddr
= rqstp
->rq_daddr
;
1009 dr
->argslen
= rqstp
->rq_arg
.len
>> 2;
1010 dr
->xprt_hlen
= rqstp
->rq_xprt_hlen
;
1012 /* back up head to the start of the buffer and copy */
1013 skip
= rqstp
->rq_arg
.len
- rqstp
->rq_arg
.head
[0].iov_len
;
1014 memcpy(dr
->args
, rqstp
->rq_arg
.head
[0].iov_base
- skip
,
1017 svc_xprt_get(rqstp
->rq_xprt
);
1018 dr
->xprt
= rqstp
->rq_xprt
;
1020 dr
->handle
.revisit
= svc_revisit
;
1025 * recv data from a deferred request into an active one
1027 static int svc_deferred_recv(struct svc_rqst
*rqstp
)
1029 struct svc_deferred_req
*dr
= rqstp
->rq_deferred
;
1031 /* setup iov_base past transport header */
1032 rqstp
->rq_arg
.head
[0].iov_base
= dr
->args
+ (dr
->xprt_hlen
>>2);
1033 /* The iov_len does not include the transport header bytes */
1034 rqstp
->rq_arg
.head
[0].iov_len
= (dr
->argslen
<<2) - dr
->xprt_hlen
;
1035 rqstp
->rq_arg
.page_len
= 0;
1036 /* The rq_arg.len includes the transport header bytes */
1037 rqstp
->rq_arg
.len
= dr
->argslen
<<2;
1038 rqstp
->rq_prot
= dr
->prot
;
1039 memcpy(&rqstp
->rq_addr
, &dr
->addr
, dr
->addrlen
);
1040 rqstp
->rq_addrlen
= dr
->addrlen
;
1041 /* Save off transport header len in case we get deferred again */
1042 rqstp
->rq_xprt_hlen
= dr
->xprt_hlen
;
1043 rqstp
->rq_daddr
= dr
->daddr
;
1044 rqstp
->rq_respages
= rqstp
->rq_pages
;
1045 return (dr
->argslen
<<2) - dr
->xprt_hlen
;
1049 static struct svc_deferred_req
*svc_deferred_dequeue(struct svc_xprt
*xprt
)
1051 struct svc_deferred_req
*dr
= NULL
;
1053 if (!test_bit(XPT_DEFERRED
, &xprt
->xpt_flags
))
1055 spin_lock(&xprt
->xpt_lock
);
1056 clear_bit(XPT_DEFERRED
, &xprt
->xpt_flags
);
1057 if (!list_empty(&xprt
->xpt_deferred
)) {
1058 dr
= list_entry(xprt
->xpt_deferred
.next
,
1059 struct svc_deferred_req
,
1061 list_del_init(&dr
->handle
.recent
);
1062 set_bit(XPT_DEFERRED
, &xprt
->xpt_flags
);
1064 spin_unlock(&xprt
->xpt_lock
);
1069 * svc_find_xprt - find an RPC transport instance
1070 * @serv: pointer to svc_serv to search
1071 * @xcl_name: C string containing transport's class name
1072 * @af: Address family of transport's local address
1073 * @port: transport's IP port number
1075 * Return the transport instance pointer for the endpoint accepting
1076 * connections/peer traffic from the specified transport class,
1077 * address family and port.
1079 * Specifying 0 for the address family or port is effectively a
1080 * wild-card, and will result in matching the first transport in the
1081 * service's list that has a matching class name.
1083 struct svc_xprt
*svc_find_xprt(struct svc_serv
*serv
, const char *xcl_name
,
1084 const sa_family_t af
, const unsigned short port
)
1086 struct svc_xprt
*xprt
;
1087 struct svc_xprt
*found
= NULL
;
1089 /* Sanity check the args */
1090 if (serv
== NULL
|| xcl_name
== NULL
)
1093 spin_lock_bh(&serv
->sv_lock
);
1094 list_for_each_entry(xprt
, &serv
->sv_permsocks
, xpt_list
) {
1095 if (strcmp(xprt
->xpt_class
->xcl_name
, xcl_name
))
1097 if (af
!= AF_UNSPEC
&& af
!= xprt
->xpt_local
.ss_family
)
1099 if (port
!= 0 && port
!= svc_xprt_local_port(xprt
))
1105 spin_unlock_bh(&serv
->sv_lock
);
1108 EXPORT_SYMBOL_GPL(svc_find_xprt
);
1110 static int svc_one_xprt_name(const struct svc_xprt
*xprt
,
1111 char *pos
, int remaining
)
1115 len
= snprintf(pos
, remaining
, "%s %u\n",
1116 xprt
->xpt_class
->xcl_name
,
1117 svc_xprt_local_port(xprt
));
1118 if (len
>= remaining
)
1119 return -ENAMETOOLONG
;
1124 * svc_xprt_names - format a buffer with a list of transport names
1125 * @serv: pointer to an RPC service
1126 * @buf: pointer to a buffer to be filled in
1127 * @buflen: length of buffer to be filled in
1129 * Fills in @buf with a string containing a list of transport names,
1130 * each name terminated with '\n'.
1132 * Returns positive length of the filled-in string on success; otherwise
1133 * a negative errno value is returned if an error occurs.
1135 int svc_xprt_names(struct svc_serv
*serv
, char *buf
, const int buflen
)
1137 struct svc_xprt
*xprt
;
1141 /* Sanity check args */
1145 spin_lock_bh(&serv
->sv_lock
);
1149 list_for_each_entry(xprt
, &serv
->sv_permsocks
, xpt_list
) {
1150 len
= svc_one_xprt_name(xprt
, pos
, buflen
- totlen
);
1162 spin_unlock_bh(&serv
->sv_lock
);
1165 EXPORT_SYMBOL_GPL(svc_xprt_names
);
1168 /*----------------------------------------------------------------------------*/
1170 static void *svc_pool_stats_start(struct seq_file
*m
, loff_t
*pos
)
1172 unsigned int pidx
= (unsigned int)*pos
;
1173 struct svc_serv
*serv
= m
->private;
1175 dprintk("svc_pool_stats_start, *pidx=%u\n", pidx
);
1178 return SEQ_START_TOKEN
;
1179 return (pidx
> serv
->sv_nrpools
? NULL
: &serv
->sv_pools
[pidx
-1]);
1182 static void *svc_pool_stats_next(struct seq_file
*m
, void *p
, loff_t
*pos
)
1184 struct svc_pool
*pool
= p
;
1185 struct svc_serv
*serv
= m
->private;
1187 dprintk("svc_pool_stats_next, *pos=%llu\n", *pos
);
1189 if (p
== SEQ_START_TOKEN
) {
1190 pool
= &serv
->sv_pools
[0];
1192 unsigned int pidx
= (pool
- &serv
->sv_pools
[0]);
1193 if (pidx
< serv
->sv_nrpools
-1)
1194 pool
= &serv
->sv_pools
[pidx
+1];
1202 static void svc_pool_stats_stop(struct seq_file
*m
, void *p
)
1206 static int svc_pool_stats_show(struct seq_file
*m
, void *p
)
1208 struct svc_pool
*pool
= p
;
1210 if (p
== SEQ_START_TOKEN
) {
1211 seq_puts(m
, "# pool packets-arrived sockets-enqueued threads-woken threads-timedout\n");
1215 seq_printf(m
, "%u %lu %lu %lu %lu\n",
1217 pool
->sp_stats
.packets
,
1218 pool
->sp_stats
.sockets_queued
,
1219 pool
->sp_stats
.threads_woken
,
1220 pool
->sp_stats
.threads_timedout
);
1225 static const struct seq_operations svc_pool_stats_seq_ops
= {
1226 .start
= svc_pool_stats_start
,
1227 .next
= svc_pool_stats_next
,
1228 .stop
= svc_pool_stats_stop
,
1229 .show
= svc_pool_stats_show
,
1232 int svc_pool_stats_open(struct svc_serv
*serv
, struct file
*file
)
1236 err
= seq_open(file
, &svc_pool_stats_seq_ops
);
1238 ((struct seq_file
*) file
->private_data
)->private = serv
;
1241 EXPORT_SYMBOL(svc_pool_stats_open
);
1243 /*----------------------------------------------------------------------------*/