xref: /openbmc/linux/net/sunrpc/svc.c (revision f5c27da4)
1 // SPDX-License-Identifier: GPL-2.0-only
2 /*
3  * linux/net/sunrpc/svc.c
4  *
5  * High-level RPC service routines
6  *
7  * Copyright (C) 1995, 1996 Olaf Kirch <okir@monad.swb.de>
8  *
9  * Multiple threads pools and NUMAisation
10  * Copyright (c) 2006 Silicon Graphics, Inc.
11  * by Greg Banks <gnb@melbourne.sgi.com>
12  */
13 
14 #include <linux/linkage.h>
15 #include <linux/sched/signal.h>
16 #include <linux/errno.h>
17 #include <linux/net.h>
18 #include <linux/in.h>
19 #include <linux/mm.h>
20 #include <linux/interrupt.h>
21 #include <linux/module.h>
22 #include <linux/kthread.h>
23 #include <linux/slab.h>
24 
25 #include <linux/sunrpc/types.h>
26 #include <linux/sunrpc/xdr.h>
27 #include <linux/sunrpc/stats.h>
28 #include <linux/sunrpc/svcsock.h>
29 #include <linux/sunrpc/clnt.h>
30 #include <linux/sunrpc/bc_xprt.h>
31 
32 #include <trace/events/sunrpc.h>
33 
34 #include "fail.h"
35 
36 #define RPCDBG_FACILITY	RPCDBG_SVCDSP
37 
38 static void svc_unregister(const struct svc_serv *serv, struct net *net);
39 
40 #define SVC_POOL_DEFAULT	SVC_POOL_GLOBAL
41 
42 /*
43  * Mode for mapping cpus to pools.
44  */
45 enum {
46 	SVC_POOL_AUTO = -1,	/* choose one of the others */
47 	SVC_POOL_GLOBAL,	/* no mapping, just a single global pool
48 				 * (legacy & UP mode) */
49 	SVC_POOL_PERCPU,	/* one pool per cpu */
50 	SVC_POOL_PERNODE	/* one pool per numa node */
51 };
52 
53 /*
54  * Structure for mapping cpus to pools and vice versa.
55  * Setup once during sunrpc initialisation.
56  */
57 
58 struct svc_pool_map {
59 	int count;			/* How many svc_servs use us */
60 	int mode;			/* Note: int not enum to avoid
61 					 * warnings about "enumeration value
62 					 * not handled in switch" */
63 	unsigned int npools;
64 	unsigned int *pool_to;		/* maps pool id to cpu or node */
65 	unsigned int *to_pool;		/* maps cpu or node to pool id */
66 };
67 
68 static struct svc_pool_map svc_pool_map = {
69 	.mode = SVC_POOL_DEFAULT
70 };
71 
72 static DEFINE_MUTEX(svc_pool_map_mutex);/* protects svc_pool_map.count only */
73 
74 static int
75 param_set_pool_mode(const char *val, const struct kernel_param *kp)
76 {
77 	int *ip = (int *)kp->arg;
78 	struct svc_pool_map *m = &svc_pool_map;
79 	int err;
80 
81 	mutex_lock(&svc_pool_map_mutex);
82 
83 	err = -EBUSY;
84 	if (m->count)
85 		goto out;
86 
87 	err = 0;
88 	if (!strncmp(val, "auto", 4))
89 		*ip = SVC_POOL_AUTO;
90 	else if (!strncmp(val, "global", 6))
91 		*ip = SVC_POOL_GLOBAL;
92 	else if (!strncmp(val, "percpu", 6))
93 		*ip = SVC_POOL_PERCPU;
94 	else if (!strncmp(val, "pernode", 7))
95 		*ip = SVC_POOL_PERNODE;
96 	else
97 		err = -EINVAL;
98 
99 out:
100 	mutex_unlock(&svc_pool_map_mutex);
101 	return err;
102 }
103 
104 static int
105 param_get_pool_mode(char *buf, const struct kernel_param *kp)
106 {
107 	int *ip = (int *)kp->arg;
108 
109 	switch (*ip)
110 	{
111 	case SVC_POOL_AUTO:
112 		return strlcpy(buf, "auto\n", 20);
113 	case SVC_POOL_GLOBAL:
114 		return strlcpy(buf, "global\n", 20);
115 	case SVC_POOL_PERCPU:
116 		return strlcpy(buf, "percpu\n", 20);
117 	case SVC_POOL_PERNODE:
118 		return strlcpy(buf, "pernode\n", 20);
119 	default:
120 		return sprintf(buf, "%d\n", *ip);
121 	}
122 }
123 
124 module_param_call(pool_mode, param_set_pool_mode, param_get_pool_mode,
125 		 &svc_pool_map.mode, 0644);
126 
127 /*
128  * Detect best pool mapping mode heuristically,
129  * according to the machine's topology.
130  */
131 static int
132 svc_pool_map_choose_mode(void)
133 {
134 	unsigned int node;
135 
136 	if (nr_online_nodes > 1) {
137 		/*
138 		 * Actually have multiple NUMA nodes,
139 		 * so split pools on NUMA node boundaries
140 		 */
141 		return SVC_POOL_PERNODE;
142 	}
143 
144 	node = first_online_node;
145 	if (nr_cpus_node(node) > 2) {
146 		/*
147 		 * Non-trivial SMP, or CONFIG_NUMA on
148 		 * non-NUMA hardware, e.g. with a generic
149 		 * x86_64 kernel on Xeons.  In this case we
150 		 * want to divide the pools on cpu boundaries.
151 		 */
152 		return SVC_POOL_PERCPU;
153 	}
154 
155 	/* default: one global pool */
156 	return SVC_POOL_GLOBAL;
157 }
158 
159 /*
160  * Allocate the to_pool[] and pool_to[] arrays.
161  * Returns 0 on success or an errno.
162  */
163 static int
164 svc_pool_map_alloc_arrays(struct svc_pool_map *m, unsigned int maxpools)
165 {
166 	m->to_pool = kcalloc(maxpools, sizeof(unsigned int), GFP_KERNEL);
167 	if (!m->to_pool)
168 		goto fail;
169 	m->pool_to = kcalloc(maxpools, sizeof(unsigned int), GFP_KERNEL);
170 	if (!m->pool_to)
171 		goto fail_free;
172 
173 	return 0;
174 
175 fail_free:
176 	kfree(m->to_pool);
177 	m->to_pool = NULL;
178 fail:
179 	return -ENOMEM;
180 }
181 
182 /*
183  * Initialise the pool map for SVC_POOL_PERCPU mode.
184  * Returns number of pools or <0 on error.
185  */
186 static int
187 svc_pool_map_init_percpu(struct svc_pool_map *m)
188 {
189 	unsigned int maxpools = nr_cpu_ids;
190 	unsigned int pidx = 0;
191 	unsigned int cpu;
192 	int err;
193 
194 	err = svc_pool_map_alloc_arrays(m, maxpools);
195 	if (err)
196 		return err;
197 
198 	for_each_online_cpu(cpu) {
199 		BUG_ON(pidx >= maxpools);
200 		m->to_pool[cpu] = pidx;
201 		m->pool_to[pidx] = cpu;
202 		pidx++;
203 	}
204 	/* cpus brought online later all get mapped to pool0, sorry */
205 
206 	return pidx;
207 };
208 
209 
210 /*
211  * Initialise the pool map for SVC_POOL_PERNODE mode.
212  * Returns number of pools or <0 on error.
213  */
214 static int
215 svc_pool_map_init_pernode(struct svc_pool_map *m)
216 {
217 	unsigned int maxpools = nr_node_ids;
218 	unsigned int pidx = 0;
219 	unsigned int node;
220 	int err;
221 
222 	err = svc_pool_map_alloc_arrays(m, maxpools);
223 	if (err)
224 		return err;
225 
226 	for_each_node_with_cpus(node) {
227 		/* some architectures (e.g. SN2) have cpuless nodes */
228 		BUG_ON(pidx > maxpools);
229 		m->to_pool[node] = pidx;
230 		m->pool_to[pidx] = node;
231 		pidx++;
232 	}
233 	/* nodes brought online later all get mapped to pool0, sorry */
234 
235 	return pidx;
236 }
237 
238 
239 /*
240  * Add a reference to the global map of cpus to pools (and
241  * vice versa) if pools are in use.
242  * Initialise the map if we're the first user.
243  * Returns the number of pools. If this is '1', no reference
244  * was taken.
245  */
246 static unsigned int
247 svc_pool_map_get(void)
248 {
249 	struct svc_pool_map *m = &svc_pool_map;
250 	int npools = -1;
251 
252 	mutex_lock(&svc_pool_map_mutex);
253 
254 	if (m->count++) {
255 		mutex_unlock(&svc_pool_map_mutex);
256 		WARN_ON_ONCE(m->npools <= 1);
257 		return m->npools;
258 	}
259 
260 	if (m->mode == SVC_POOL_AUTO)
261 		m->mode = svc_pool_map_choose_mode();
262 
263 	switch (m->mode) {
264 	case SVC_POOL_PERCPU:
265 		npools = svc_pool_map_init_percpu(m);
266 		break;
267 	case SVC_POOL_PERNODE:
268 		npools = svc_pool_map_init_pernode(m);
269 		break;
270 	}
271 
272 	if (npools <= 0) {
273 		/* default, or memory allocation failure */
274 		npools = 1;
275 		m->mode = SVC_POOL_GLOBAL;
276 	}
277 	m->npools = npools;
278 
279 	if (npools == 1)
280 		/* service is unpooled, so doesn't hold a reference */
281 		m->count--;
282 
283 	mutex_unlock(&svc_pool_map_mutex);
284 	return npools;
285 }
286 
287 /*
288  * Drop a reference to the global map of cpus to pools, if
289  * pools were in use, i.e. if npools > 1.
290  * When the last reference is dropped, the map data is
291  * freed; this allows the sysadmin to change the pool
292  * mode using the pool_mode module option without
293  * rebooting or re-loading sunrpc.ko.
294  */
295 static void
296 svc_pool_map_put(int npools)
297 {
298 	struct svc_pool_map *m = &svc_pool_map;
299 
300 	if (npools <= 1)
301 		return;
302 	mutex_lock(&svc_pool_map_mutex);
303 
304 	if (!--m->count) {
305 		kfree(m->to_pool);
306 		m->to_pool = NULL;
307 		kfree(m->pool_to);
308 		m->pool_to = NULL;
309 		m->npools = 0;
310 	}
311 
312 	mutex_unlock(&svc_pool_map_mutex);
313 }
314 
315 static int svc_pool_map_get_node(unsigned int pidx)
316 {
317 	const struct svc_pool_map *m = &svc_pool_map;
318 
319 	if (m->count) {
320 		if (m->mode == SVC_POOL_PERCPU)
321 			return cpu_to_node(m->pool_to[pidx]);
322 		if (m->mode == SVC_POOL_PERNODE)
323 			return m->pool_to[pidx];
324 	}
325 	return NUMA_NO_NODE;
326 }
327 /*
328  * Set the given thread's cpus_allowed mask so that it
329  * will only run on cpus in the given pool.
330  */
331 static inline void
332 svc_pool_map_set_cpumask(struct task_struct *task, unsigned int pidx)
333 {
334 	struct svc_pool_map *m = &svc_pool_map;
335 	unsigned int node = m->pool_to[pidx];
336 
337 	/*
338 	 * The caller checks for sv_nrpools > 1, which
339 	 * implies that we've been initialized.
340 	 */
341 	WARN_ON_ONCE(m->count == 0);
342 	if (m->count == 0)
343 		return;
344 
345 	switch (m->mode) {
346 	case SVC_POOL_PERCPU:
347 	{
348 		set_cpus_allowed_ptr(task, cpumask_of(node));
349 		break;
350 	}
351 	case SVC_POOL_PERNODE:
352 	{
353 		set_cpus_allowed_ptr(task, cpumask_of_node(node));
354 		break;
355 	}
356 	}
357 }
358 
359 /**
360  * svc_pool_for_cpu - Select pool to run a thread on this cpu
361  * @serv: An RPC service
362  *
363  * Use the active CPU and the svc_pool_map's mode setting to
364  * select the svc thread pool to use. Once initialized, the
365  * svc_pool_map does not change.
366  *
367  * Return value:
368  *   A pointer to an svc_pool
369  */
370 struct svc_pool *svc_pool_for_cpu(struct svc_serv *serv)
371 {
372 	struct svc_pool_map *m = &svc_pool_map;
373 	int cpu = raw_smp_processor_id();
374 	unsigned int pidx = 0;
375 
376 	if (serv->sv_nrpools <= 1)
377 		return serv->sv_pools;
378 
379 	switch (m->mode) {
380 	case SVC_POOL_PERCPU:
381 		pidx = m->to_pool[cpu];
382 		break;
383 	case SVC_POOL_PERNODE:
384 		pidx = m->to_pool[cpu_to_node(cpu)];
385 		break;
386 	}
387 
388 	return &serv->sv_pools[pidx % serv->sv_nrpools];
389 }
390 
391 int svc_rpcb_setup(struct svc_serv *serv, struct net *net)
392 {
393 	int err;
394 
395 	err = rpcb_create_local(net);
396 	if (err)
397 		return err;
398 
399 	/* Remove any stale portmap registrations */
400 	svc_unregister(serv, net);
401 	return 0;
402 }
403 EXPORT_SYMBOL_GPL(svc_rpcb_setup);
404 
405 void svc_rpcb_cleanup(struct svc_serv *serv, struct net *net)
406 {
407 	svc_unregister(serv, net);
408 	rpcb_put_local(net);
409 }
410 EXPORT_SYMBOL_GPL(svc_rpcb_cleanup);
411 
412 static int svc_uses_rpcbind(struct svc_serv *serv)
413 {
414 	struct svc_program	*progp;
415 	unsigned int		i;
416 
417 	for (progp = serv->sv_program; progp; progp = progp->pg_next) {
418 		for (i = 0; i < progp->pg_nvers; i++) {
419 			if (progp->pg_vers[i] == NULL)
420 				continue;
421 			if (!progp->pg_vers[i]->vs_hidden)
422 				return 1;
423 		}
424 	}
425 
426 	return 0;
427 }
428 
429 int svc_bind(struct svc_serv *serv, struct net *net)
430 {
431 	if (!svc_uses_rpcbind(serv))
432 		return 0;
433 	return svc_rpcb_setup(serv, net);
434 }
435 EXPORT_SYMBOL_GPL(svc_bind);
436 
437 #if defined(CONFIG_SUNRPC_BACKCHANNEL)
438 static void
439 __svc_init_bc(struct svc_serv *serv)
440 {
441 	INIT_LIST_HEAD(&serv->sv_cb_list);
442 	spin_lock_init(&serv->sv_cb_lock);
443 	init_waitqueue_head(&serv->sv_cb_waitq);
444 }
445 #else
446 static void
447 __svc_init_bc(struct svc_serv *serv)
448 {
449 }
450 #endif
451 
452 /*
453  * Create an RPC service
454  */
455 static struct svc_serv *
456 __svc_create(struct svc_program *prog, unsigned int bufsize, int npools,
457 	     int (*threadfn)(void *data))
458 {
459 	struct svc_serv	*serv;
460 	unsigned int vers;
461 	unsigned int xdrsize;
462 	unsigned int i;
463 
464 	if (!(serv = kzalloc(sizeof(*serv), GFP_KERNEL)))
465 		return NULL;
466 	serv->sv_name      = prog->pg_name;
467 	serv->sv_program   = prog;
468 	kref_init(&serv->sv_refcnt);
469 	serv->sv_stats     = prog->pg_stats;
470 	if (bufsize > RPCSVC_MAXPAYLOAD)
471 		bufsize = RPCSVC_MAXPAYLOAD;
472 	serv->sv_max_payload = bufsize? bufsize : 4096;
473 	serv->sv_max_mesg  = roundup(serv->sv_max_payload + PAGE_SIZE, PAGE_SIZE);
474 	serv->sv_threadfn = threadfn;
475 	xdrsize = 0;
476 	while (prog) {
477 		prog->pg_lovers = prog->pg_nvers-1;
478 		for (vers=0; vers<prog->pg_nvers ; vers++)
479 			if (prog->pg_vers[vers]) {
480 				prog->pg_hivers = vers;
481 				if (prog->pg_lovers > vers)
482 					prog->pg_lovers = vers;
483 				if (prog->pg_vers[vers]->vs_xdrsize > xdrsize)
484 					xdrsize = prog->pg_vers[vers]->vs_xdrsize;
485 			}
486 		prog = prog->pg_next;
487 	}
488 	serv->sv_xdrsize   = xdrsize;
489 	INIT_LIST_HEAD(&serv->sv_tempsocks);
490 	INIT_LIST_HEAD(&serv->sv_permsocks);
491 	timer_setup(&serv->sv_temptimer, NULL, 0);
492 	spin_lock_init(&serv->sv_lock);
493 
494 	__svc_init_bc(serv);
495 
496 	serv->sv_nrpools = npools;
497 	serv->sv_pools =
498 		kcalloc(serv->sv_nrpools, sizeof(struct svc_pool),
499 			GFP_KERNEL);
500 	if (!serv->sv_pools) {
501 		kfree(serv);
502 		return NULL;
503 	}
504 
505 	for (i = 0; i < serv->sv_nrpools; i++) {
506 		struct svc_pool *pool = &serv->sv_pools[i];
507 
508 		dprintk("svc: initialising pool %u for %s\n",
509 				i, serv->sv_name);
510 
511 		pool->sp_id = i;
512 		INIT_LIST_HEAD(&pool->sp_sockets);
513 		INIT_LIST_HEAD(&pool->sp_all_threads);
514 		spin_lock_init(&pool->sp_lock);
515 	}
516 
517 	return serv;
518 }
519 
520 /**
521  * svc_create - Create an RPC service
522  * @prog: the RPC program the new service will handle
523  * @bufsize: maximum message size for @prog
524  * @threadfn: a function to service RPC requests for @prog
525  *
526  * Returns an instantiated struct svc_serv object or NULL.
527  */
528 struct svc_serv *svc_create(struct svc_program *prog, unsigned int bufsize,
529 			    int (*threadfn)(void *data))
530 {
531 	return __svc_create(prog, bufsize, 1, threadfn);
532 }
533 EXPORT_SYMBOL_GPL(svc_create);
534 
535 /**
536  * svc_create_pooled - Create an RPC service with pooled threads
537  * @prog: the RPC program the new service will handle
538  * @bufsize: maximum message size for @prog
539  * @threadfn: a function to service RPC requests for @prog
540  *
541  * Returns an instantiated struct svc_serv object or NULL.
542  */
543 struct svc_serv *svc_create_pooled(struct svc_program *prog,
544 				   unsigned int bufsize,
545 				   int (*threadfn)(void *data))
546 {
547 	struct svc_serv *serv;
548 	unsigned int npools = svc_pool_map_get();
549 
550 	serv = __svc_create(prog, bufsize, npools, threadfn);
551 	if (!serv)
552 		goto out_err;
553 	return serv;
554 out_err:
555 	svc_pool_map_put(npools);
556 	return NULL;
557 }
558 EXPORT_SYMBOL_GPL(svc_create_pooled);
559 
560 /*
561  * Destroy an RPC service. Should be called with appropriate locking to
562  * protect sv_permsocks and sv_tempsocks.
563  */
564 void
565 svc_destroy(struct kref *ref)
566 {
567 	struct svc_serv *serv = container_of(ref, struct svc_serv, sv_refcnt);
568 
569 	dprintk("svc: svc_destroy(%s)\n", serv->sv_program->pg_name);
570 	del_timer_sync(&serv->sv_temptimer);
571 
572 	/*
573 	 * The last user is gone and thus all sockets have to be destroyed to
574 	 * the point. Check this.
575 	 */
576 	BUG_ON(!list_empty(&serv->sv_permsocks));
577 	BUG_ON(!list_empty(&serv->sv_tempsocks));
578 
579 	cache_clean_deferred(serv);
580 
581 	svc_pool_map_put(serv->sv_nrpools);
582 
583 	kfree(serv->sv_pools);
584 	kfree(serv);
585 }
586 EXPORT_SYMBOL_GPL(svc_destroy);
587 
588 /*
589  * Allocate an RPC server's buffer space.
590  * We allocate pages and place them in rq_pages.
591  */
592 static int
593 svc_init_buffer(struct svc_rqst *rqstp, unsigned int size, int node)
594 {
595 	unsigned int pages, arghi;
596 
597 	/* bc_xprt uses fore channel allocated buffers */
598 	if (svc_is_backchannel(rqstp))
599 		return 1;
600 
601 	pages = size / PAGE_SIZE + 1; /* extra page as we hold both request and reply.
602 				       * We assume one is at most one page
603 				       */
604 	arghi = 0;
605 	WARN_ON_ONCE(pages > RPCSVC_MAXPAGES);
606 	if (pages > RPCSVC_MAXPAGES)
607 		pages = RPCSVC_MAXPAGES;
608 	while (pages) {
609 		struct page *p = alloc_pages_node(node, GFP_KERNEL, 0);
610 		if (!p)
611 			break;
612 		rqstp->rq_pages[arghi++] = p;
613 		pages--;
614 	}
615 	return pages == 0;
616 }
617 
618 /*
619  * Release an RPC server buffer
620  */
621 static void
622 svc_release_buffer(struct svc_rqst *rqstp)
623 {
624 	unsigned int i;
625 
626 	for (i = 0; i < ARRAY_SIZE(rqstp->rq_pages); i++)
627 		if (rqstp->rq_pages[i])
628 			put_page(rqstp->rq_pages[i]);
629 }
630 
631 struct svc_rqst *
632 svc_rqst_alloc(struct svc_serv *serv, struct svc_pool *pool, int node)
633 {
634 	struct svc_rqst	*rqstp;
635 
636 	rqstp = kzalloc_node(sizeof(*rqstp), GFP_KERNEL, node);
637 	if (!rqstp)
638 		return rqstp;
639 
640 	__set_bit(RQ_BUSY, &rqstp->rq_flags);
641 	spin_lock_init(&rqstp->rq_lock);
642 	rqstp->rq_server = serv;
643 	rqstp->rq_pool = pool;
644 
645 	rqstp->rq_scratch_page = alloc_pages_node(node, GFP_KERNEL, 0);
646 	if (!rqstp->rq_scratch_page)
647 		goto out_enomem;
648 
649 	rqstp->rq_argp = kmalloc_node(serv->sv_xdrsize, GFP_KERNEL, node);
650 	if (!rqstp->rq_argp)
651 		goto out_enomem;
652 
653 	rqstp->rq_resp = kmalloc_node(serv->sv_xdrsize, GFP_KERNEL, node);
654 	if (!rqstp->rq_resp)
655 		goto out_enomem;
656 
657 	if (!svc_init_buffer(rqstp, serv->sv_max_mesg, node))
658 		goto out_enomem;
659 
660 	return rqstp;
661 out_enomem:
662 	svc_rqst_free(rqstp);
663 	return NULL;
664 }
665 EXPORT_SYMBOL_GPL(svc_rqst_alloc);
666 
667 static struct svc_rqst *
668 svc_prepare_thread(struct svc_serv *serv, struct svc_pool *pool, int node)
669 {
670 	struct svc_rqst	*rqstp;
671 
672 	rqstp = svc_rqst_alloc(serv, pool, node);
673 	if (!rqstp)
674 		return ERR_PTR(-ENOMEM);
675 
676 	svc_get(serv);
677 	spin_lock_bh(&serv->sv_lock);
678 	serv->sv_nrthreads += 1;
679 	spin_unlock_bh(&serv->sv_lock);
680 
681 	spin_lock_bh(&pool->sp_lock);
682 	pool->sp_nrthreads++;
683 	list_add_rcu(&rqstp->rq_all, &pool->sp_all_threads);
684 	spin_unlock_bh(&pool->sp_lock);
685 	return rqstp;
686 }
687 
688 /*
689  * Choose a pool in which to create a new thread, for svc_set_num_threads
690  */
691 static inline struct svc_pool *
692 choose_pool(struct svc_serv *serv, struct svc_pool *pool, unsigned int *state)
693 {
694 	if (pool != NULL)
695 		return pool;
696 
697 	return &serv->sv_pools[(*state)++ % serv->sv_nrpools];
698 }
699 
700 /*
701  * Choose a thread to kill, for svc_set_num_threads
702  */
703 static inline struct task_struct *
704 choose_victim(struct svc_serv *serv, struct svc_pool *pool, unsigned int *state)
705 {
706 	unsigned int i;
707 	struct task_struct *task = NULL;
708 
709 	if (pool != NULL) {
710 		spin_lock_bh(&pool->sp_lock);
711 	} else {
712 		/* choose a pool in round-robin fashion */
713 		for (i = 0; i < serv->sv_nrpools; i++) {
714 			pool = &serv->sv_pools[--(*state) % serv->sv_nrpools];
715 			spin_lock_bh(&pool->sp_lock);
716 			if (!list_empty(&pool->sp_all_threads))
717 				goto found_pool;
718 			spin_unlock_bh(&pool->sp_lock);
719 		}
720 		return NULL;
721 	}
722 
723 found_pool:
724 	if (!list_empty(&pool->sp_all_threads)) {
725 		struct svc_rqst *rqstp;
726 
727 		/*
728 		 * Remove from the pool->sp_all_threads list
729 		 * so we don't try to kill it again.
730 		 */
731 		rqstp = list_entry(pool->sp_all_threads.next, struct svc_rqst, rq_all);
732 		set_bit(RQ_VICTIM, &rqstp->rq_flags);
733 		list_del_rcu(&rqstp->rq_all);
734 		task = rqstp->rq_task;
735 	}
736 	spin_unlock_bh(&pool->sp_lock);
737 
738 	return task;
739 }
740 
741 /* create new threads */
742 static int
743 svc_start_kthreads(struct svc_serv *serv, struct svc_pool *pool, int nrservs)
744 {
745 	struct svc_rqst	*rqstp;
746 	struct task_struct *task;
747 	struct svc_pool *chosen_pool;
748 	unsigned int state = serv->sv_nrthreads-1;
749 	int node;
750 
751 	do {
752 		nrservs--;
753 		chosen_pool = choose_pool(serv, pool, &state);
754 
755 		node = svc_pool_map_get_node(chosen_pool->sp_id);
756 		rqstp = svc_prepare_thread(serv, chosen_pool, node);
757 		if (IS_ERR(rqstp))
758 			return PTR_ERR(rqstp);
759 
760 		task = kthread_create_on_node(serv->sv_threadfn, rqstp,
761 					      node, "%s", serv->sv_name);
762 		if (IS_ERR(task)) {
763 			svc_exit_thread(rqstp);
764 			return PTR_ERR(task);
765 		}
766 
767 		rqstp->rq_task = task;
768 		if (serv->sv_nrpools > 1)
769 			svc_pool_map_set_cpumask(task, chosen_pool->sp_id);
770 
771 		svc_sock_update_bufs(serv);
772 		wake_up_process(task);
773 	} while (nrservs > 0);
774 
775 	return 0;
776 }
777 
778 /*
779  * Create or destroy enough new threads to make the number
780  * of threads the given number.  If `pool' is non-NULL, applies
781  * only to threads in that pool, otherwise round-robins between
782  * all pools.  Caller must ensure that mutual exclusion between this and
783  * server startup or shutdown.
784  */
785 
786 /* destroy old threads */
787 static int
788 svc_stop_kthreads(struct svc_serv *serv, struct svc_pool *pool, int nrservs)
789 {
790 	struct task_struct *task;
791 	unsigned int state = serv->sv_nrthreads-1;
792 
793 	/* destroy old threads */
794 	do {
795 		task = choose_victim(serv, pool, &state);
796 		if (task == NULL)
797 			break;
798 		kthread_stop(task);
799 		nrservs++;
800 	} while (nrservs < 0);
801 	return 0;
802 }
803 
804 int
805 svc_set_num_threads(struct svc_serv *serv, struct svc_pool *pool, int nrservs)
806 {
807 	if (pool == NULL) {
808 		nrservs -= serv->sv_nrthreads;
809 	} else {
810 		spin_lock_bh(&pool->sp_lock);
811 		nrservs -= pool->sp_nrthreads;
812 		spin_unlock_bh(&pool->sp_lock);
813 	}
814 
815 	if (nrservs > 0)
816 		return svc_start_kthreads(serv, pool, nrservs);
817 	if (nrservs < 0)
818 		return svc_stop_kthreads(serv, pool, nrservs);
819 	return 0;
820 }
821 EXPORT_SYMBOL_GPL(svc_set_num_threads);
822 
823 /**
824  * svc_rqst_replace_page - Replace one page in rq_pages[]
825  * @rqstp: svc_rqst with pages to replace
826  * @page: replacement page
827  *
828  * When replacing a page in rq_pages, batch the release of the
829  * replaced pages to avoid hammering the page allocator.
830  */
831 void svc_rqst_replace_page(struct svc_rqst *rqstp, struct page *page)
832 {
833 	if (*rqstp->rq_next_page) {
834 		if (!pagevec_space(&rqstp->rq_pvec))
835 			__pagevec_release(&rqstp->rq_pvec);
836 		pagevec_add(&rqstp->rq_pvec, *rqstp->rq_next_page);
837 	}
838 
839 	get_page(page);
840 	*(rqstp->rq_next_page++) = page;
841 }
842 EXPORT_SYMBOL_GPL(svc_rqst_replace_page);
843 
844 /*
845  * Called from a server thread as it's exiting. Caller must hold the "service
846  * mutex" for the service.
847  */
848 void
849 svc_rqst_free(struct svc_rqst *rqstp)
850 {
851 	svc_release_buffer(rqstp);
852 	if (rqstp->rq_scratch_page)
853 		put_page(rqstp->rq_scratch_page);
854 	kfree(rqstp->rq_resp);
855 	kfree(rqstp->rq_argp);
856 	kfree(rqstp->rq_auth_data);
857 	kfree_rcu(rqstp, rq_rcu_head);
858 }
859 EXPORT_SYMBOL_GPL(svc_rqst_free);
860 
861 void
862 svc_exit_thread(struct svc_rqst *rqstp)
863 {
864 	struct svc_serv	*serv = rqstp->rq_server;
865 	struct svc_pool	*pool = rqstp->rq_pool;
866 
867 	spin_lock_bh(&pool->sp_lock);
868 	pool->sp_nrthreads--;
869 	if (!test_and_set_bit(RQ_VICTIM, &rqstp->rq_flags))
870 		list_del_rcu(&rqstp->rq_all);
871 	spin_unlock_bh(&pool->sp_lock);
872 
873 	spin_lock_bh(&serv->sv_lock);
874 	serv->sv_nrthreads -= 1;
875 	spin_unlock_bh(&serv->sv_lock);
876 	svc_sock_update_bufs(serv);
877 
878 	svc_rqst_free(rqstp);
879 
880 	svc_put(serv);
881 }
882 EXPORT_SYMBOL_GPL(svc_exit_thread);
883 
884 /*
885  * Register an "inet" protocol family netid with the local
886  * rpcbind daemon via an rpcbind v4 SET request.
887  *
888  * No netconfig infrastructure is available in the kernel, so
889  * we map IP_ protocol numbers to netids by hand.
890  *
891  * Returns zero on success; a negative errno value is returned
892  * if any error occurs.
893  */
894 static int __svc_rpcb_register4(struct net *net, const u32 program,
895 				const u32 version,
896 				const unsigned short protocol,
897 				const unsigned short port)
898 {
899 	const struct sockaddr_in sin = {
900 		.sin_family		= AF_INET,
901 		.sin_addr.s_addr	= htonl(INADDR_ANY),
902 		.sin_port		= htons(port),
903 	};
904 	const char *netid;
905 	int error;
906 
907 	switch (protocol) {
908 	case IPPROTO_UDP:
909 		netid = RPCBIND_NETID_UDP;
910 		break;
911 	case IPPROTO_TCP:
912 		netid = RPCBIND_NETID_TCP;
913 		break;
914 	default:
915 		return -ENOPROTOOPT;
916 	}
917 
918 	error = rpcb_v4_register(net, program, version,
919 					(const struct sockaddr *)&sin, netid);
920 
921 	/*
922 	 * User space didn't support rpcbind v4, so retry this
923 	 * registration request with the legacy rpcbind v2 protocol.
924 	 */
925 	if (error == -EPROTONOSUPPORT)
926 		error = rpcb_register(net, program, version, protocol, port);
927 
928 	return error;
929 }
930 
931 #if IS_ENABLED(CONFIG_IPV6)
932 /*
933  * Register an "inet6" protocol family netid with the local
934  * rpcbind daemon via an rpcbind v4 SET request.
935  *
936  * No netconfig infrastructure is available in the kernel, so
937  * we map IP_ protocol numbers to netids by hand.
938  *
939  * Returns zero on success; a negative errno value is returned
940  * if any error occurs.
941  */
942 static int __svc_rpcb_register6(struct net *net, const u32 program,
943 				const u32 version,
944 				const unsigned short protocol,
945 				const unsigned short port)
946 {
947 	const struct sockaddr_in6 sin6 = {
948 		.sin6_family		= AF_INET6,
949 		.sin6_addr		= IN6ADDR_ANY_INIT,
950 		.sin6_port		= htons(port),
951 	};
952 	const char *netid;
953 	int error;
954 
955 	switch (protocol) {
956 	case IPPROTO_UDP:
957 		netid = RPCBIND_NETID_UDP6;
958 		break;
959 	case IPPROTO_TCP:
960 		netid = RPCBIND_NETID_TCP6;
961 		break;
962 	default:
963 		return -ENOPROTOOPT;
964 	}
965 
966 	error = rpcb_v4_register(net, program, version,
967 					(const struct sockaddr *)&sin6, netid);
968 
969 	/*
970 	 * User space didn't support rpcbind version 4, so we won't
971 	 * use a PF_INET6 listener.
972 	 */
973 	if (error == -EPROTONOSUPPORT)
974 		error = -EAFNOSUPPORT;
975 
976 	return error;
977 }
978 #endif	/* IS_ENABLED(CONFIG_IPV6) */
979 
980 /*
981  * Register a kernel RPC service via rpcbind version 4.
982  *
983  * Returns zero on success; a negative errno value is returned
984  * if any error occurs.
985  */
986 static int __svc_register(struct net *net, const char *progname,
987 			  const u32 program, const u32 version,
988 			  const int family,
989 			  const unsigned short protocol,
990 			  const unsigned short port)
991 {
992 	int error = -EAFNOSUPPORT;
993 
994 	switch (family) {
995 	case PF_INET:
996 		error = __svc_rpcb_register4(net, program, version,
997 						protocol, port);
998 		break;
999 #if IS_ENABLED(CONFIG_IPV6)
1000 	case PF_INET6:
1001 		error = __svc_rpcb_register6(net, program, version,
1002 						protocol, port);
1003 #endif
1004 	}
1005 
1006 	trace_svc_register(progname, version, protocol, port, family, error);
1007 	return error;
1008 }
1009 
1010 int svc_rpcbind_set_version(struct net *net,
1011 			    const struct svc_program *progp,
1012 			    u32 version, int family,
1013 			    unsigned short proto,
1014 			    unsigned short port)
1015 {
1016 	return __svc_register(net, progp->pg_name, progp->pg_prog,
1017 				version, family, proto, port);
1018 
1019 }
1020 EXPORT_SYMBOL_GPL(svc_rpcbind_set_version);
1021 
1022 int svc_generic_rpcbind_set(struct net *net,
1023 			    const struct svc_program *progp,
1024 			    u32 version, int family,
1025 			    unsigned short proto,
1026 			    unsigned short port)
1027 {
1028 	const struct svc_version *vers = progp->pg_vers[version];
1029 	int error;
1030 
1031 	if (vers == NULL)
1032 		return 0;
1033 
1034 	if (vers->vs_hidden) {
1035 		trace_svc_noregister(progp->pg_name, version, proto,
1036 				     port, family, 0);
1037 		return 0;
1038 	}
1039 
1040 	/*
1041 	 * Don't register a UDP port if we need congestion
1042 	 * control.
1043 	 */
1044 	if (vers->vs_need_cong_ctrl && proto == IPPROTO_UDP)
1045 		return 0;
1046 
1047 	error = svc_rpcbind_set_version(net, progp, version,
1048 					family, proto, port);
1049 
1050 	return (vers->vs_rpcb_optnl) ? 0 : error;
1051 }
1052 EXPORT_SYMBOL_GPL(svc_generic_rpcbind_set);
1053 
1054 /**
1055  * svc_register - register an RPC service with the local portmapper
1056  * @serv: svc_serv struct for the service to register
1057  * @net: net namespace for the service to register
1058  * @family: protocol family of service's listener socket
1059  * @proto: transport protocol number to advertise
1060  * @port: port to advertise
1061  *
1062  * Service is registered for any address in the passed-in protocol family
1063  */
1064 int svc_register(const struct svc_serv *serv, struct net *net,
1065 		 const int family, const unsigned short proto,
1066 		 const unsigned short port)
1067 {
1068 	struct svc_program	*progp;
1069 	unsigned int		i;
1070 	int			error = 0;
1071 
1072 	WARN_ON_ONCE(proto == 0 && port == 0);
1073 	if (proto == 0 && port == 0)
1074 		return -EINVAL;
1075 
1076 	for (progp = serv->sv_program; progp; progp = progp->pg_next) {
1077 		for (i = 0; i < progp->pg_nvers; i++) {
1078 
1079 			error = progp->pg_rpcbind_set(net, progp, i,
1080 					family, proto, port);
1081 			if (error < 0) {
1082 				printk(KERN_WARNING "svc: failed to register "
1083 					"%sv%u RPC service (errno %d).\n",
1084 					progp->pg_name, i, -error);
1085 				break;
1086 			}
1087 		}
1088 	}
1089 
1090 	return error;
1091 }
1092 
1093 /*
1094  * If user space is running rpcbind, it should take the v4 UNSET
1095  * and clear everything for this [program, version].  If user space
1096  * is running portmap, it will reject the v4 UNSET, but won't have
1097  * any "inet6" entries anyway.  So a PMAP_UNSET should be sufficient
1098  * in this case to clear all existing entries for [program, version].
1099  */
1100 static void __svc_unregister(struct net *net, const u32 program, const u32 version,
1101 			     const char *progname)
1102 {
1103 	int error;
1104 
1105 	error = rpcb_v4_register(net, program, version, NULL, "");
1106 
1107 	/*
1108 	 * User space didn't support rpcbind v4, so retry this
1109 	 * request with the legacy rpcbind v2 protocol.
1110 	 */
1111 	if (error == -EPROTONOSUPPORT)
1112 		error = rpcb_register(net, program, version, 0, 0);
1113 
1114 	trace_svc_unregister(progname, version, error);
1115 }
1116 
1117 /*
1118  * All netids, bind addresses and ports registered for [program, version]
1119  * are removed from the local rpcbind database (if the service is not
1120  * hidden) to make way for a new instance of the service.
1121  *
1122  * The result of unregistration is reported via dprintk for those who want
1123  * verification of the result, but is otherwise not important.
1124  */
1125 static void svc_unregister(const struct svc_serv *serv, struct net *net)
1126 {
1127 	struct svc_program *progp;
1128 	unsigned long flags;
1129 	unsigned int i;
1130 
1131 	clear_thread_flag(TIF_SIGPENDING);
1132 
1133 	for (progp = serv->sv_program; progp; progp = progp->pg_next) {
1134 		for (i = 0; i < progp->pg_nvers; i++) {
1135 			if (progp->pg_vers[i] == NULL)
1136 				continue;
1137 			if (progp->pg_vers[i]->vs_hidden)
1138 				continue;
1139 			__svc_unregister(net, progp->pg_prog, i, progp->pg_name);
1140 		}
1141 	}
1142 
1143 	spin_lock_irqsave(&current->sighand->siglock, flags);
1144 	recalc_sigpending();
1145 	spin_unlock_irqrestore(&current->sighand->siglock, flags);
1146 }
1147 
1148 /*
1149  * dprintk the given error with the address of the client that caused it.
1150  */
1151 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
1152 static __printf(2, 3)
1153 void svc_printk(struct svc_rqst *rqstp, const char *fmt, ...)
1154 {
1155 	struct va_format vaf;
1156 	va_list args;
1157 	char 	buf[RPC_MAX_ADDRBUFLEN];
1158 
1159 	va_start(args, fmt);
1160 
1161 	vaf.fmt = fmt;
1162 	vaf.va = &args;
1163 
1164 	dprintk("svc: %s: %pV", svc_print_addr(rqstp, buf, sizeof(buf)), &vaf);
1165 
1166 	va_end(args);
1167 }
1168 #else
1169 static __printf(2,3) void svc_printk(struct svc_rqst *rqstp, const char *fmt, ...) {}
1170 #endif
1171 
1172 __be32
1173 svc_generic_init_request(struct svc_rqst *rqstp,
1174 		const struct svc_program *progp,
1175 		struct svc_process_info *ret)
1176 {
1177 	const struct svc_version *versp = NULL;	/* compiler food */
1178 	const struct svc_procedure *procp = NULL;
1179 
1180 	if (rqstp->rq_vers >= progp->pg_nvers )
1181 		goto err_bad_vers;
1182 	versp = progp->pg_vers[rqstp->rq_vers];
1183 	if (!versp)
1184 		goto err_bad_vers;
1185 
1186 	/*
1187 	 * Some protocol versions (namely NFSv4) require some form of
1188 	 * congestion control.  (See RFC 7530 section 3.1 paragraph 2)
1189 	 * In other words, UDP is not allowed. We mark those when setting
1190 	 * up the svc_xprt, and verify that here.
1191 	 *
1192 	 * The spec is not very clear about what error should be returned
1193 	 * when someone tries to access a server that is listening on UDP
1194 	 * for lower versions. RPC_PROG_MISMATCH seems to be the closest
1195 	 * fit.
1196 	 */
1197 	if (versp->vs_need_cong_ctrl && rqstp->rq_xprt &&
1198 	    !test_bit(XPT_CONG_CTRL, &rqstp->rq_xprt->xpt_flags))
1199 		goto err_bad_vers;
1200 
1201 	if (rqstp->rq_proc >= versp->vs_nproc)
1202 		goto err_bad_proc;
1203 	rqstp->rq_procinfo = procp = &versp->vs_proc[rqstp->rq_proc];
1204 	if (!procp)
1205 		goto err_bad_proc;
1206 
1207 	/* Initialize storage for argp and resp */
1208 	memset(rqstp->rq_argp, 0, procp->pc_argzero);
1209 	memset(rqstp->rq_resp, 0, procp->pc_ressize);
1210 
1211 	/* Bump per-procedure stats counter */
1212 	versp->vs_count[rqstp->rq_proc]++;
1213 
1214 	ret->dispatch = versp->vs_dispatch;
1215 	return rpc_success;
1216 err_bad_vers:
1217 	ret->mismatch.lovers = progp->pg_lovers;
1218 	ret->mismatch.hivers = progp->pg_hivers;
1219 	return rpc_prog_mismatch;
1220 err_bad_proc:
1221 	return rpc_proc_unavail;
1222 }
1223 EXPORT_SYMBOL_GPL(svc_generic_init_request);
1224 
1225 /*
1226  * Common routine for processing the RPC request.
1227  */
1228 static int
1229 svc_process_common(struct svc_rqst *rqstp, struct kvec *argv, struct kvec *resv)
1230 {
1231 	struct svc_program	*progp;
1232 	const struct svc_procedure *procp = NULL;
1233 	struct svc_serv		*serv = rqstp->rq_server;
1234 	struct svc_process_info process;
1235 	__be32			*statp;
1236 	u32			prog, vers;
1237 	__be32			rpc_stat;
1238 	int			auth_res, rc;
1239 	__be32			*reply_statp;
1240 
1241 	rpc_stat = rpc_success;
1242 
1243 	if (argv->iov_len < 6*4)
1244 		goto err_short_len;
1245 
1246 	/* Will be turned off by GSS integrity and privacy services */
1247 	__set_bit(RQ_SPLICE_OK, &rqstp->rq_flags);
1248 	/* Will be turned off only when NFSv4 Sessions are used */
1249 	__set_bit(RQ_USEDEFERRAL, &rqstp->rq_flags);
1250 	__clear_bit(RQ_DROPME, &rqstp->rq_flags);
1251 
1252 	svc_putu32(resv, rqstp->rq_xid);
1253 
1254 	vers = svc_getnl(argv);
1255 
1256 	/* First words of reply: */
1257 	svc_putnl(resv, 1);		/* REPLY */
1258 
1259 	if (vers != 2)		/* RPC version number */
1260 		goto err_bad_rpc;
1261 
1262 	/* Save position in case we later decide to reject: */
1263 	reply_statp = resv->iov_base + resv->iov_len;
1264 
1265 	svc_putnl(resv, 0);		/* ACCEPT */
1266 
1267 	rqstp->rq_prog = prog = svc_getnl(argv);	/* program number */
1268 	rqstp->rq_vers = svc_getnl(argv);	/* version number */
1269 	rqstp->rq_proc = svc_getnl(argv);	/* procedure number */
1270 
1271 	for (progp = serv->sv_program; progp; progp = progp->pg_next)
1272 		if (prog == progp->pg_prog)
1273 			break;
1274 
1275 	/*
1276 	 * Decode auth data, and add verifier to reply buffer.
1277 	 * We do this before anything else in order to get a decent
1278 	 * auth verifier.
1279 	 */
1280 	auth_res = svc_authenticate(rqstp);
1281 	/* Also give the program a chance to reject this call: */
1282 	if (auth_res == SVC_OK && progp)
1283 		auth_res = progp->pg_authenticate(rqstp);
1284 	if (auth_res != SVC_OK)
1285 		trace_svc_authenticate(rqstp, auth_res);
1286 	switch (auth_res) {
1287 	case SVC_OK:
1288 		break;
1289 	case SVC_GARBAGE:
1290 		goto err_garbage;
1291 	case SVC_SYSERR:
1292 		rpc_stat = rpc_system_err;
1293 		goto err_bad;
1294 	case SVC_DENIED:
1295 		goto err_bad_auth;
1296 	case SVC_CLOSE:
1297 		goto close;
1298 	case SVC_DROP:
1299 		goto dropit;
1300 	case SVC_COMPLETE:
1301 		goto sendit;
1302 	}
1303 
1304 	if (progp == NULL)
1305 		goto err_bad_prog;
1306 
1307 	rpc_stat = progp->pg_init_request(rqstp, progp, &process);
1308 	switch (rpc_stat) {
1309 	case rpc_success:
1310 		break;
1311 	case rpc_prog_unavail:
1312 		goto err_bad_prog;
1313 	case rpc_prog_mismatch:
1314 		goto err_bad_vers;
1315 	case rpc_proc_unavail:
1316 		goto err_bad_proc;
1317 	}
1318 
1319 	procp = rqstp->rq_procinfo;
1320 	/* Should this check go into the dispatcher? */
1321 	if (!procp || !procp->pc_func)
1322 		goto err_bad_proc;
1323 
1324 	/* Syntactic check complete */
1325 	serv->sv_stats->rpccnt++;
1326 	trace_svc_process(rqstp, progp->pg_name);
1327 
1328 	/* Build the reply header. */
1329 	statp = resv->iov_base +resv->iov_len;
1330 	svc_putnl(resv, RPC_SUCCESS);
1331 
1332 	/* un-reserve some of the out-queue now that we have a
1333 	 * better idea of reply size
1334 	 */
1335 	if (procp->pc_xdrressize)
1336 		svc_reserve_auth(rqstp, procp->pc_xdrressize<<2);
1337 
1338 	/* Call the function that processes the request. */
1339 	rc = process.dispatch(rqstp, statp);
1340 	if (procp->pc_release)
1341 		procp->pc_release(rqstp);
1342 	if (!rc)
1343 		goto dropit;
1344 	if (rqstp->rq_auth_stat != rpc_auth_ok)
1345 		goto err_bad_auth;
1346 
1347 	/* Check RPC status result */
1348 	if (*statp != rpc_success)
1349 		resv->iov_len = ((void*)statp)  - resv->iov_base + 4;
1350 
1351 	if (procp->pc_encode == NULL)
1352 		goto dropit;
1353 
1354  sendit:
1355 	if (svc_authorise(rqstp))
1356 		goto close_xprt;
1357 	return 1;		/* Caller can now send it */
1358 
1359  dropit:
1360 	svc_authorise(rqstp);	/* doesn't hurt to call this twice */
1361 	dprintk("svc: svc_process dropit\n");
1362 	return 0;
1363 
1364  close:
1365 	svc_authorise(rqstp);
1366 close_xprt:
1367 	if (rqstp->rq_xprt && test_bit(XPT_TEMP, &rqstp->rq_xprt->xpt_flags))
1368 		svc_xprt_close(rqstp->rq_xprt);
1369 	dprintk("svc: svc_process close\n");
1370 	return 0;
1371 
1372 err_short_len:
1373 	svc_printk(rqstp, "short len %zd, dropping request\n",
1374 			argv->iov_len);
1375 	goto close_xprt;
1376 
1377 err_bad_rpc:
1378 	serv->sv_stats->rpcbadfmt++;
1379 	svc_putnl(resv, 1);	/* REJECT */
1380 	svc_putnl(resv, 0);	/* RPC_MISMATCH */
1381 	svc_putnl(resv, 2);	/* Only RPCv2 supported */
1382 	svc_putnl(resv, 2);
1383 	goto sendit;
1384 
1385 err_bad_auth:
1386 	dprintk("svc: authentication failed (%d)\n",
1387 		be32_to_cpu(rqstp->rq_auth_stat));
1388 	serv->sv_stats->rpcbadauth++;
1389 	/* Restore write pointer to location of accept status: */
1390 	xdr_ressize_check(rqstp, reply_statp);
1391 	svc_putnl(resv, 1);	/* REJECT */
1392 	svc_putnl(resv, 1);	/* AUTH_ERROR */
1393 	svc_putu32(resv, rqstp->rq_auth_stat);	/* status */
1394 	goto sendit;
1395 
1396 err_bad_prog:
1397 	dprintk("svc: unknown program %d\n", prog);
1398 	serv->sv_stats->rpcbadfmt++;
1399 	svc_putnl(resv, RPC_PROG_UNAVAIL);
1400 	goto sendit;
1401 
1402 err_bad_vers:
1403 	svc_printk(rqstp, "unknown version (%d for prog %d, %s)\n",
1404 		       rqstp->rq_vers, rqstp->rq_prog, progp->pg_name);
1405 
1406 	serv->sv_stats->rpcbadfmt++;
1407 	svc_putnl(resv, RPC_PROG_MISMATCH);
1408 	svc_putnl(resv, process.mismatch.lovers);
1409 	svc_putnl(resv, process.mismatch.hivers);
1410 	goto sendit;
1411 
1412 err_bad_proc:
1413 	svc_printk(rqstp, "unknown procedure (%d)\n", rqstp->rq_proc);
1414 
1415 	serv->sv_stats->rpcbadfmt++;
1416 	svc_putnl(resv, RPC_PROC_UNAVAIL);
1417 	goto sendit;
1418 
1419 err_garbage:
1420 	svc_printk(rqstp, "failed to decode args\n");
1421 
1422 	rpc_stat = rpc_garbage_args;
1423 err_bad:
1424 	serv->sv_stats->rpcbadfmt++;
1425 	svc_putnl(resv, ntohl(rpc_stat));
1426 	goto sendit;
1427 }
1428 
1429 /*
1430  * Process the RPC request.
1431  */
1432 int
1433 svc_process(struct svc_rqst *rqstp)
1434 {
1435 	struct kvec		*argv = &rqstp->rq_arg.head[0];
1436 	struct kvec		*resv = &rqstp->rq_res.head[0];
1437 	__be32			dir;
1438 
1439 #if IS_ENABLED(CONFIG_FAIL_SUNRPC)
1440 	if (!fail_sunrpc.ignore_server_disconnect &&
1441 	    should_fail(&fail_sunrpc.attr, 1))
1442 		svc_xprt_deferred_close(rqstp->rq_xprt);
1443 #endif
1444 
1445 	/*
1446 	 * Setup response xdr_buf.
1447 	 * Initially it has just one page
1448 	 */
1449 	rqstp->rq_next_page = &rqstp->rq_respages[1];
1450 	resv->iov_base = page_address(rqstp->rq_respages[0]);
1451 	resv->iov_len = 0;
1452 	rqstp->rq_res.pages = rqstp->rq_next_page;
1453 	rqstp->rq_res.len = 0;
1454 	rqstp->rq_res.page_base = 0;
1455 	rqstp->rq_res.page_len = 0;
1456 	rqstp->rq_res.buflen = PAGE_SIZE;
1457 	rqstp->rq_res.tail[0].iov_base = NULL;
1458 	rqstp->rq_res.tail[0].iov_len = 0;
1459 
1460 	dir = svc_getu32(argv);
1461 	if (dir != rpc_call)
1462 		goto out_baddir;
1463 	if (!svc_process_common(rqstp, argv, resv))
1464 		goto out_drop;
1465 	return svc_send(rqstp);
1466 
1467 out_baddir:
1468 	svc_printk(rqstp, "bad direction 0x%08x, dropping request\n",
1469 		   be32_to_cpu(dir));
1470 	rqstp->rq_server->sv_stats->rpcbadfmt++;
1471 out_drop:
1472 	svc_drop(rqstp);
1473 	return 0;
1474 }
1475 EXPORT_SYMBOL_GPL(svc_process);
1476 
1477 #if defined(CONFIG_SUNRPC_BACKCHANNEL)
1478 /*
1479  * Process a backchannel RPC request that arrived over an existing
1480  * outbound connection
1481  */
1482 int
1483 bc_svc_process(struct svc_serv *serv, struct rpc_rqst *req,
1484 	       struct svc_rqst *rqstp)
1485 {
1486 	struct kvec	*argv = &rqstp->rq_arg.head[0];
1487 	struct kvec	*resv = &rqstp->rq_res.head[0];
1488 	struct rpc_task *task;
1489 	int proc_error;
1490 	int error;
1491 
1492 	dprintk("svc: %s(%p)\n", __func__, req);
1493 
1494 	/* Build the svc_rqst used by the common processing routine */
1495 	rqstp->rq_xid = req->rq_xid;
1496 	rqstp->rq_prot = req->rq_xprt->prot;
1497 	rqstp->rq_server = serv;
1498 	rqstp->rq_bc_net = req->rq_xprt->xprt_net;
1499 
1500 	rqstp->rq_addrlen = sizeof(req->rq_xprt->addr);
1501 	memcpy(&rqstp->rq_addr, &req->rq_xprt->addr, rqstp->rq_addrlen);
1502 	memcpy(&rqstp->rq_arg, &req->rq_rcv_buf, sizeof(rqstp->rq_arg));
1503 	memcpy(&rqstp->rq_res, &req->rq_snd_buf, sizeof(rqstp->rq_res));
1504 
1505 	/* Adjust the argument buffer length */
1506 	rqstp->rq_arg.len = req->rq_private_buf.len;
1507 	if (rqstp->rq_arg.len <= rqstp->rq_arg.head[0].iov_len) {
1508 		rqstp->rq_arg.head[0].iov_len = rqstp->rq_arg.len;
1509 		rqstp->rq_arg.page_len = 0;
1510 	} else if (rqstp->rq_arg.len <= rqstp->rq_arg.head[0].iov_len +
1511 			rqstp->rq_arg.page_len)
1512 		rqstp->rq_arg.page_len = rqstp->rq_arg.len -
1513 			rqstp->rq_arg.head[0].iov_len;
1514 	else
1515 		rqstp->rq_arg.len = rqstp->rq_arg.head[0].iov_len +
1516 			rqstp->rq_arg.page_len;
1517 
1518 	/* reset result send buffer "put" position */
1519 	resv->iov_len = 0;
1520 
1521 	/*
1522 	 * Skip the next two words because they've already been
1523 	 * processed in the transport
1524 	 */
1525 	svc_getu32(argv);	/* XID */
1526 	svc_getnl(argv);	/* CALLDIR */
1527 
1528 	/* Parse and execute the bc call */
1529 	proc_error = svc_process_common(rqstp, argv, resv);
1530 
1531 	atomic_dec(&req->rq_xprt->bc_slot_count);
1532 	if (!proc_error) {
1533 		/* Processing error: drop the request */
1534 		xprt_free_bc_request(req);
1535 		error = -EINVAL;
1536 		goto out;
1537 	}
1538 	/* Finally, send the reply synchronously */
1539 	memcpy(&req->rq_snd_buf, &rqstp->rq_res, sizeof(req->rq_snd_buf));
1540 	task = rpc_run_bc_task(req);
1541 	if (IS_ERR(task)) {
1542 		error = PTR_ERR(task);
1543 		goto out;
1544 	}
1545 
1546 	WARN_ON_ONCE(atomic_read(&task->tk_count) != 1);
1547 	error = task->tk_status;
1548 	rpc_put_task(task);
1549 
1550 out:
1551 	dprintk("svc: %s(), error=%d\n", __func__, error);
1552 	return error;
1553 }
1554 EXPORT_SYMBOL_GPL(bc_svc_process);
1555 #endif /* CONFIG_SUNRPC_BACKCHANNEL */
1556 
1557 /**
1558  * svc_max_payload - Return transport-specific limit on the RPC payload
1559  * @rqstp: RPC transaction context
1560  *
1561  * Returns the maximum number of payload bytes the current transport
1562  * allows.
1563  */
1564 u32 svc_max_payload(const struct svc_rqst *rqstp)
1565 {
1566 	u32 max = rqstp->rq_xprt->xpt_class->xcl_max_payload;
1567 
1568 	if (rqstp->rq_server->sv_max_payload < max)
1569 		max = rqstp->rq_server->sv_max_payload;
1570 	return max;
1571 }
1572 EXPORT_SYMBOL_GPL(svc_max_payload);
1573 
1574 /**
1575  * svc_proc_name - Return RPC procedure name in string form
1576  * @rqstp: svc_rqst to operate on
1577  *
1578  * Return value:
1579  *   Pointer to a NUL-terminated string
1580  */
1581 const char *svc_proc_name(const struct svc_rqst *rqstp)
1582 {
1583 	if (rqstp && rqstp->rq_procinfo)
1584 		return rqstp->rq_procinfo->pc_name;
1585 	return "unknown";
1586 }
1587 
1588 
1589 /**
1590  * svc_encode_result_payload - mark a range of bytes as a result payload
1591  * @rqstp: svc_rqst to operate on
1592  * @offset: payload's byte offset in rqstp->rq_res
1593  * @length: size of payload, in bytes
1594  *
1595  * Returns zero on success, or a negative errno if a permanent
1596  * error occurred.
1597  */
1598 int svc_encode_result_payload(struct svc_rqst *rqstp, unsigned int offset,
1599 			      unsigned int length)
1600 {
1601 	return rqstp->rq_xprt->xpt_ops->xpo_result_payload(rqstp, offset,
1602 							   length);
1603 }
1604 EXPORT_SYMBOL_GPL(svc_encode_result_payload);
1605 
1606 /**
1607  * svc_fill_write_vector - Construct data argument for VFS write call
1608  * @rqstp: svc_rqst to operate on
1609  * @payload: xdr_buf containing only the write data payload
1610  *
1611  * Fills in rqstp::rq_vec, and returns the number of elements.
1612  */
1613 unsigned int svc_fill_write_vector(struct svc_rqst *rqstp,
1614 				   struct xdr_buf *payload)
1615 {
1616 	struct page **pages = payload->pages;
1617 	struct kvec *first = payload->head;
1618 	struct kvec *vec = rqstp->rq_vec;
1619 	size_t total = payload->len;
1620 	unsigned int i;
1621 
1622 	/* Some types of transport can present the write payload
1623 	 * entirely in rq_arg.pages. In this case, @first is empty.
1624 	 */
1625 	i = 0;
1626 	if (first->iov_len) {
1627 		vec[i].iov_base = first->iov_base;
1628 		vec[i].iov_len = min_t(size_t, total, first->iov_len);
1629 		total -= vec[i].iov_len;
1630 		++i;
1631 	}
1632 
1633 	while (total) {
1634 		vec[i].iov_base = page_address(*pages);
1635 		vec[i].iov_len = min_t(size_t, total, PAGE_SIZE);
1636 		total -= vec[i].iov_len;
1637 		++i;
1638 		++pages;
1639 	}
1640 
1641 	WARN_ON_ONCE(i > ARRAY_SIZE(rqstp->rq_vec));
1642 	return i;
1643 }
1644 EXPORT_SYMBOL_GPL(svc_fill_write_vector);
1645 
1646 /**
1647  * svc_fill_symlink_pathname - Construct pathname argument for VFS symlink call
1648  * @rqstp: svc_rqst to operate on
1649  * @first: buffer containing first section of pathname
1650  * @p: buffer containing remaining section of pathname
1651  * @total: total length of the pathname argument
1652  *
1653  * The VFS symlink API demands a NUL-terminated pathname in mapped memory.
1654  * Returns pointer to a NUL-terminated string, or an ERR_PTR. Caller must free
1655  * the returned string.
1656  */
1657 char *svc_fill_symlink_pathname(struct svc_rqst *rqstp, struct kvec *first,
1658 				void *p, size_t total)
1659 {
1660 	size_t len, remaining;
1661 	char *result, *dst;
1662 
1663 	result = kmalloc(total + 1, GFP_KERNEL);
1664 	if (!result)
1665 		return ERR_PTR(-ESERVERFAULT);
1666 
1667 	dst = result;
1668 	remaining = total;
1669 
1670 	len = min_t(size_t, total, first->iov_len);
1671 	if (len) {
1672 		memcpy(dst, first->iov_base, len);
1673 		dst += len;
1674 		remaining -= len;
1675 	}
1676 
1677 	if (remaining) {
1678 		len = min_t(size_t, remaining, PAGE_SIZE);
1679 		memcpy(dst, p, len);
1680 		dst += len;
1681 	}
1682 
1683 	*dst = '\0';
1684 
1685 	/* Sanity check: Linux doesn't allow the pathname argument to
1686 	 * contain a NUL byte.
1687 	 */
1688 	if (strlen(result) != total) {
1689 		kfree(result);
1690 		return ERR_PTR(-EINVAL);
1691 	}
1692 	return result;
1693 }
1694 EXPORT_SYMBOL_GPL(svc_fill_symlink_pathname);
1695