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 #include "sunrpc.h"
36
37 #define RPCDBG_FACILITY RPCDBG_SVCDSP
38
39 static void svc_unregister(const struct svc_serv *serv, struct net *net);
40
41 #define SVC_POOL_DEFAULT SVC_POOL_GLOBAL
42
43 /*
44 * Mode for mapping cpus to pools.
45 */
46 enum {
47 SVC_POOL_AUTO = -1, /* choose one of the others */
48 SVC_POOL_GLOBAL, /* no mapping, just a single global pool
49 * (legacy & UP mode) */
50 SVC_POOL_PERCPU, /* one pool per cpu */
51 SVC_POOL_PERNODE /* one pool per numa node */
52 };
53
54 /*
55 * Structure for mapping cpus to pools and vice versa.
56 * Setup once during sunrpc initialisation.
57 */
58
59 struct svc_pool_map {
60 int count; /* How many svc_servs use us */
61 int mode; /* Note: int not enum to avoid
62 * warnings about "enumeration value
63 * not handled in switch" */
64 unsigned int npools;
65 unsigned int *pool_to; /* maps pool id to cpu or node */
66 unsigned int *to_pool; /* maps cpu or node to pool id */
67 };
68
69 static struct svc_pool_map svc_pool_map = {
70 .mode = SVC_POOL_DEFAULT
71 };
72
73 static DEFINE_MUTEX(svc_pool_map_mutex);/* protects svc_pool_map.count only */
74
75 static int
__param_set_pool_mode(const char * val,struct svc_pool_map * m)76 __param_set_pool_mode(const char *val, struct svc_pool_map *m)
77 {
78 int err, mode;
79
80 mutex_lock(&svc_pool_map_mutex);
81
82 err = 0;
83 if (!strncmp(val, "auto", 4))
84 mode = SVC_POOL_AUTO;
85 else if (!strncmp(val, "global", 6))
86 mode = SVC_POOL_GLOBAL;
87 else if (!strncmp(val, "percpu", 6))
88 mode = SVC_POOL_PERCPU;
89 else if (!strncmp(val, "pernode", 7))
90 mode = SVC_POOL_PERNODE;
91 else
92 err = -EINVAL;
93
94 if (err)
95 goto out;
96
97 if (m->count == 0)
98 m->mode = mode;
99 else if (mode != m->mode)
100 err = -EBUSY;
101 out:
102 mutex_unlock(&svc_pool_map_mutex);
103 return err;
104 }
105
106 static int
param_set_pool_mode(const char * val,const struct kernel_param * kp)107 param_set_pool_mode(const char *val, const struct kernel_param *kp)
108 {
109 struct svc_pool_map *m = kp->arg;
110
111 return __param_set_pool_mode(val, m);
112 }
113
sunrpc_set_pool_mode(const char * val)114 int sunrpc_set_pool_mode(const char *val)
115 {
116 return __param_set_pool_mode(val, &svc_pool_map);
117 }
118 EXPORT_SYMBOL(sunrpc_set_pool_mode);
119
120 /**
121 * sunrpc_get_pool_mode - get the current pool_mode for the host
122 * @buf: where to write the current pool_mode
123 * @size: size of @buf
124 *
125 * Grab the current pool_mode from the svc_pool_map and write
126 * the resulting string to @buf. Returns the number of characters
127 * written to @buf (a'la snprintf()).
128 */
129 int
sunrpc_get_pool_mode(char * buf,size_t size)130 sunrpc_get_pool_mode(char *buf, size_t size)
131 {
132 struct svc_pool_map *m = &svc_pool_map;
133
134 switch (m->mode)
135 {
136 case SVC_POOL_AUTO:
137 return snprintf(buf, size, "auto");
138 case SVC_POOL_GLOBAL:
139 return snprintf(buf, size, "global");
140 case SVC_POOL_PERCPU:
141 return snprintf(buf, size, "percpu");
142 case SVC_POOL_PERNODE:
143 return snprintf(buf, size, "pernode");
144 default:
145 return snprintf(buf, size, "%d", m->mode);
146 }
147 }
148 EXPORT_SYMBOL(sunrpc_get_pool_mode);
149
150 static int
param_get_pool_mode(char * buf,const struct kernel_param * kp)151 param_get_pool_mode(char *buf, const struct kernel_param *kp)
152 {
153 char str[16];
154 int len;
155
156 len = sunrpc_get_pool_mode(str, ARRAY_SIZE(str));
157
158 /* Ensure we have room for newline and NUL */
159 len = min_t(int, len, ARRAY_SIZE(str) - 2);
160
161 /* tack on the newline */
162 str[len] = '\n';
163 str[len + 1] = '\0';
164
165 return sysfs_emit(buf, "%s", str);
166 }
167
168 module_param_call(pool_mode, param_set_pool_mode, param_get_pool_mode,
169 &svc_pool_map, 0644);
170
171 /*
172 * Detect best pool mapping mode heuristically,
173 * according to the machine's topology.
174 */
175 static int
svc_pool_map_choose_mode(void)176 svc_pool_map_choose_mode(void)
177 {
178 unsigned int node;
179
180 if (nr_online_nodes > 1) {
181 /*
182 * Actually have multiple NUMA nodes,
183 * so split pools on NUMA node boundaries
184 */
185 return SVC_POOL_PERNODE;
186 }
187
188 node = first_online_node;
189 if (nr_cpus_node(node) > 2) {
190 /*
191 * Non-trivial SMP, or CONFIG_NUMA on
192 * non-NUMA hardware, e.g. with a generic
193 * x86_64 kernel on Xeons. In this case we
194 * want to divide the pools on cpu boundaries.
195 */
196 return SVC_POOL_PERCPU;
197 }
198
199 /* default: one global pool */
200 return SVC_POOL_GLOBAL;
201 }
202
203 /*
204 * Allocate the to_pool[] and pool_to[] arrays.
205 * Returns 0 on success or an errno.
206 */
207 static int
svc_pool_map_alloc_arrays(struct svc_pool_map * m,unsigned int maxpools)208 svc_pool_map_alloc_arrays(struct svc_pool_map *m, unsigned int maxpools)
209 {
210 m->to_pool = kcalloc(maxpools, sizeof(unsigned int), GFP_KERNEL);
211 if (!m->to_pool)
212 goto fail;
213 m->pool_to = kcalloc(maxpools, sizeof(unsigned int), GFP_KERNEL);
214 if (!m->pool_to)
215 goto fail_free;
216
217 return 0;
218
219 fail_free:
220 kfree(m->to_pool);
221 m->to_pool = NULL;
222 fail:
223 return -ENOMEM;
224 }
225
226 /*
227 * Initialise the pool map for SVC_POOL_PERCPU mode.
228 * Returns number of pools or <0 on error.
229 */
230 static int
svc_pool_map_init_percpu(struct svc_pool_map * m)231 svc_pool_map_init_percpu(struct svc_pool_map *m)
232 {
233 unsigned int maxpools = nr_cpu_ids;
234 unsigned int pidx = 0;
235 unsigned int cpu;
236 int err;
237
238 err = svc_pool_map_alloc_arrays(m, maxpools);
239 if (err)
240 return err;
241
242 for_each_online_cpu(cpu) {
243 BUG_ON(pidx >= maxpools);
244 m->to_pool[cpu] = pidx;
245 m->pool_to[pidx] = cpu;
246 pidx++;
247 }
248 /* cpus brought online later all get mapped to pool0, sorry */
249
250 return pidx;
251 };
252
253
254 /*
255 * Initialise the pool map for SVC_POOL_PERNODE mode.
256 * Returns number of pools or <0 on error.
257 */
258 static int
svc_pool_map_init_pernode(struct svc_pool_map * m)259 svc_pool_map_init_pernode(struct svc_pool_map *m)
260 {
261 unsigned int maxpools = nr_node_ids;
262 unsigned int pidx = 0;
263 unsigned int node;
264 int err;
265
266 err = svc_pool_map_alloc_arrays(m, maxpools);
267 if (err)
268 return err;
269
270 for_each_node_with_cpus(node) {
271 /* some architectures (e.g. SN2) have cpuless nodes */
272 BUG_ON(pidx > maxpools);
273 m->to_pool[node] = pidx;
274 m->pool_to[pidx] = node;
275 pidx++;
276 }
277 /* nodes brought online later all get mapped to pool0, sorry */
278
279 return pidx;
280 }
281
282
283 /*
284 * Add a reference to the global map of cpus to pools (and
285 * vice versa) if pools are in use.
286 * Initialise the map if we're the first user.
287 * Returns the number of pools. If this is '1', no reference
288 * was taken.
289 */
290 static unsigned int
svc_pool_map_get(void)291 svc_pool_map_get(void)
292 {
293 struct svc_pool_map *m = &svc_pool_map;
294 int npools = -1;
295
296 mutex_lock(&svc_pool_map_mutex);
297 if (m->count++) {
298 mutex_unlock(&svc_pool_map_mutex);
299 return m->npools;
300 }
301
302 if (m->mode == SVC_POOL_AUTO)
303 m->mode = svc_pool_map_choose_mode();
304
305 switch (m->mode) {
306 case SVC_POOL_PERCPU:
307 npools = svc_pool_map_init_percpu(m);
308 break;
309 case SVC_POOL_PERNODE:
310 npools = svc_pool_map_init_pernode(m);
311 break;
312 }
313
314 if (npools <= 0) {
315 /* default, or memory allocation failure */
316 npools = 1;
317 m->mode = SVC_POOL_GLOBAL;
318 }
319 m->npools = npools;
320 mutex_unlock(&svc_pool_map_mutex);
321 return npools;
322 }
323
324 /*
325 * Drop a reference to the global map of cpus to pools.
326 * When the last reference is dropped, the map data is
327 * freed; this allows the sysadmin to change the pool.
328 */
329 static void
svc_pool_map_put(void)330 svc_pool_map_put(void)
331 {
332 struct svc_pool_map *m = &svc_pool_map;
333
334 mutex_lock(&svc_pool_map_mutex);
335 if (!--m->count) {
336 kfree(m->to_pool);
337 m->to_pool = NULL;
338 kfree(m->pool_to);
339 m->pool_to = NULL;
340 m->npools = 0;
341 }
342 mutex_unlock(&svc_pool_map_mutex);
343 }
344
svc_pool_map_get_node(unsigned int pidx)345 static int svc_pool_map_get_node(unsigned int pidx)
346 {
347 const struct svc_pool_map *m = &svc_pool_map;
348
349 if (m->count) {
350 if (m->mode == SVC_POOL_PERCPU)
351 return cpu_to_node(m->pool_to[pidx]);
352 if (m->mode == SVC_POOL_PERNODE)
353 return m->pool_to[pidx];
354 }
355 return NUMA_NO_NODE;
356 }
357 /*
358 * Set the given thread's cpus_allowed mask so that it
359 * will only run on cpus in the given pool.
360 */
361 static inline void
svc_pool_map_set_cpumask(struct task_struct * task,unsigned int pidx)362 svc_pool_map_set_cpumask(struct task_struct *task, unsigned int pidx)
363 {
364 struct svc_pool_map *m = &svc_pool_map;
365 unsigned int node = m->pool_to[pidx];
366
367 /*
368 * The caller checks for sv_nrpools > 1, which
369 * implies that we've been initialized.
370 */
371 WARN_ON_ONCE(m->count == 0);
372 if (m->count == 0)
373 return;
374
375 switch (m->mode) {
376 case SVC_POOL_PERCPU:
377 {
378 set_cpus_allowed_ptr(task, cpumask_of(node));
379 break;
380 }
381 case SVC_POOL_PERNODE:
382 {
383 set_cpus_allowed_ptr(task, cpumask_of_node(node));
384 break;
385 }
386 }
387 }
388
389 /**
390 * svc_pool_for_cpu - Select pool to run a thread on this cpu
391 * @serv: An RPC service
392 *
393 * Use the active CPU and the svc_pool_map's mode setting to
394 * select the svc thread pool to use. Once initialized, the
395 * svc_pool_map does not change.
396 *
397 * Return value:
398 * A pointer to an svc_pool
399 */
svc_pool_for_cpu(struct svc_serv * serv)400 struct svc_pool *svc_pool_for_cpu(struct svc_serv *serv)
401 {
402 struct svc_pool_map *m = &svc_pool_map;
403 int cpu = raw_smp_processor_id();
404 unsigned int pidx = 0;
405
406 if (serv->sv_nrpools <= 1)
407 return serv->sv_pools;
408
409 switch (m->mode) {
410 case SVC_POOL_PERCPU:
411 pidx = m->to_pool[cpu];
412 break;
413 case SVC_POOL_PERNODE:
414 pidx = m->to_pool[cpu_to_node(cpu)];
415 break;
416 }
417
418 return &serv->sv_pools[pidx % serv->sv_nrpools];
419 }
420
svc_rpcb_setup(struct svc_serv * serv,struct net * net)421 static int svc_rpcb_setup(struct svc_serv *serv, struct net *net)
422 {
423 int err;
424
425 err = rpcb_create_local(net);
426 if (err)
427 return err;
428
429 /* Remove any stale portmap registrations */
430 svc_unregister(serv, net);
431 return 0;
432 }
433
svc_rpcb_cleanup(struct svc_serv * serv,struct net * net)434 void svc_rpcb_cleanup(struct svc_serv *serv, struct net *net)
435 {
436 svc_unregister(serv, net);
437 rpcb_put_local(net);
438 }
439 EXPORT_SYMBOL_GPL(svc_rpcb_cleanup);
440
svc_uses_rpcbind(struct svc_serv * serv)441 static int svc_uses_rpcbind(struct svc_serv *serv)
442 {
443 unsigned int p, i;
444
445 for (p = 0; p < serv->sv_nprogs; p++) {
446 struct svc_program *progp = &serv->sv_programs[p];
447
448 for (i = 0; i < progp->pg_nvers; i++) {
449 if (progp->pg_vers[i] == NULL)
450 continue;
451 if (!progp->pg_vers[i]->vs_hidden)
452 return 1;
453 }
454 }
455
456 return 0;
457 }
458
svc_bind(struct svc_serv * serv,struct net * net)459 int svc_bind(struct svc_serv *serv, struct net *net)
460 {
461 if (!svc_uses_rpcbind(serv))
462 return 0;
463 return svc_rpcb_setup(serv, net);
464 }
465 EXPORT_SYMBOL_GPL(svc_bind);
466
467 #if defined(CONFIG_SUNRPC_BACKCHANNEL)
468 static void
__svc_init_bc(struct svc_serv * serv)469 __svc_init_bc(struct svc_serv *serv)
470 {
471 lwq_init(&serv->sv_cb_list);
472 }
473 #else
474 static void
__svc_init_bc(struct svc_serv * serv)475 __svc_init_bc(struct svc_serv *serv)
476 {
477 }
478 #endif
479
480 /*
481 * Create an RPC service
482 */
483 static struct svc_serv *
__svc_create(struct svc_program * prog,int nprogs,struct svc_stat * stats,unsigned int bufsize,int npools,int (* threadfn)(void * data))484 __svc_create(struct svc_program *prog, int nprogs, struct svc_stat *stats,
485 unsigned int bufsize, int npools, int (*threadfn)(void *data))
486 {
487 struct svc_serv *serv;
488 unsigned int vers;
489 unsigned int xdrsize;
490 unsigned int i;
491
492 if (!(serv = kzalloc(sizeof(*serv), GFP_KERNEL)))
493 return NULL;
494 serv->sv_name = prog->pg_name;
495 serv->sv_programs = prog;
496 serv->sv_nprogs = nprogs;
497 serv->sv_stats = stats;
498 if (bufsize > RPCSVC_MAXPAYLOAD)
499 bufsize = RPCSVC_MAXPAYLOAD;
500 serv->sv_max_payload = bufsize? bufsize : 4096;
501 serv->sv_max_mesg = roundup(serv->sv_max_payload + PAGE_SIZE, PAGE_SIZE);
502 serv->sv_threadfn = threadfn;
503 xdrsize = 0;
504 for (i = 0; i < nprogs; i++) {
505 struct svc_program *progp = &prog[i];
506
507 progp->pg_lovers = progp->pg_nvers-1;
508 for (vers = 0; vers < progp->pg_nvers ; vers++)
509 if (progp->pg_vers[vers]) {
510 progp->pg_hivers = vers;
511 if (progp->pg_lovers > vers)
512 progp->pg_lovers = vers;
513 if (progp->pg_vers[vers]->vs_xdrsize > xdrsize)
514 xdrsize = progp->pg_vers[vers]->vs_xdrsize;
515 }
516 }
517 serv->sv_xdrsize = xdrsize;
518 INIT_LIST_HEAD(&serv->sv_tempsocks);
519 INIT_LIST_HEAD(&serv->sv_permsocks);
520 timer_setup(&serv->sv_temptimer, NULL, 0);
521 spin_lock_init(&serv->sv_lock);
522
523 __svc_init_bc(serv);
524
525 serv->sv_nrpools = npools;
526 serv->sv_pools =
527 kcalloc(serv->sv_nrpools, sizeof(struct svc_pool),
528 GFP_KERNEL);
529 if (!serv->sv_pools) {
530 kfree(serv);
531 return NULL;
532 }
533
534 for (i = 0; i < serv->sv_nrpools; i++) {
535 struct svc_pool *pool = &serv->sv_pools[i];
536
537 dprintk("svc: initialising pool %u for %s\n",
538 i, serv->sv_name);
539
540 pool->sp_id = i;
541 lwq_init(&pool->sp_xprts);
542 INIT_LIST_HEAD(&pool->sp_all_threads);
543 init_llist_head(&pool->sp_idle_threads);
544
545 percpu_counter_init(&pool->sp_messages_arrived, 0, GFP_KERNEL);
546 percpu_counter_init(&pool->sp_sockets_queued, 0, GFP_KERNEL);
547 percpu_counter_init(&pool->sp_threads_woken, 0, GFP_KERNEL);
548 }
549
550 return serv;
551 }
552
553 /**
554 * svc_create - Create an RPC service
555 * @prog: the RPC program the new service will handle
556 * @bufsize: maximum message size for @prog
557 * @threadfn: a function to service RPC requests for @prog
558 *
559 * Returns an instantiated struct svc_serv object or NULL.
560 */
svc_create(struct svc_program * prog,unsigned int bufsize,int (* threadfn)(void * data))561 struct svc_serv *svc_create(struct svc_program *prog, unsigned int bufsize,
562 int (*threadfn)(void *data))
563 {
564 return __svc_create(prog, 1, NULL, bufsize, 1, threadfn);
565 }
566 EXPORT_SYMBOL_GPL(svc_create);
567
568 /**
569 * svc_create_pooled - Create an RPC service with pooled threads
570 * @prog: Array of RPC programs the new service will handle
571 * @nprogs: Number of programs in the array
572 * @stats: the stats struct if desired
573 * @bufsize: maximum message size for @prog
574 * @threadfn: a function to service RPC requests for @prog
575 *
576 * Returns an instantiated struct svc_serv object or NULL.
577 */
svc_create_pooled(struct svc_program * prog,unsigned int nprogs,struct svc_stat * stats,unsigned int bufsize,int (* threadfn)(void * data))578 struct svc_serv *svc_create_pooled(struct svc_program *prog,
579 unsigned int nprogs,
580 struct svc_stat *stats,
581 unsigned int bufsize,
582 int (*threadfn)(void *data))
583 {
584 struct svc_serv *serv;
585 unsigned int npools = svc_pool_map_get();
586
587 serv = __svc_create(prog, nprogs, stats, bufsize, npools, threadfn);
588 if (!serv)
589 goto out_err;
590 serv->sv_is_pooled = true;
591 return serv;
592 out_err:
593 svc_pool_map_put();
594 return NULL;
595 }
596 EXPORT_SYMBOL_GPL(svc_create_pooled);
597
598 /*
599 * Destroy an RPC service. Should be called with appropriate locking to
600 * protect sv_permsocks and sv_tempsocks.
601 */
602 void
svc_destroy(struct svc_serv ** servp)603 svc_destroy(struct svc_serv **servp)
604 {
605 struct svc_serv *serv = *servp;
606 unsigned int i;
607
608 *servp = NULL;
609
610 dprintk("svc: svc_destroy(%s)\n", serv->sv_programs->pg_name);
611 timer_shutdown_sync(&serv->sv_temptimer);
612
613 /*
614 * Remaining transports at this point are not expected.
615 */
616 WARN_ONCE(!list_empty(&serv->sv_permsocks),
617 "SVC: permsocks remain for %s\n", serv->sv_programs->pg_name);
618 WARN_ONCE(!list_empty(&serv->sv_tempsocks),
619 "SVC: tempsocks remain for %s\n", serv->sv_programs->pg_name);
620
621 cache_clean_deferred(serv);
622
623 if (serv->sv_is_pooled)
624 svc_pool_map_put();
625
626 for (i = 0; i < serv->sv_nrpools; i++) {
627 struct svc_pool *pool = &serv->sv_pools[i];
628
629 percpu_counter_destroy(&pool->sp_messages_arrived);
630 percpu_counter_destroy(&pool->sp_sockets_queued);
631 percpu_counter_destroy(&pool->sp_threads_woken);
632 }
633 kfree(serv->sv_pools);
634 kfree(serv);
635 }
636 EXPORT_SYMBOL_GPL(svc_destroy);
637
638 static bool
svc_init_buffer(struct svc_rqst * rqstp,unsigned int size,int node)639 svc_init_buffer(struct svc_rqst *rqstp, unsigned int size, int node)
640 {
641 unsigned long pages, ret;
642
643 /* bc_xprt uses fore channel allocated buffers */
644 if (svc_is_backchannel(rqstp))
645 return true;
646
647 pages = size / PAGE_SIZE + 1; /* extra page as we hold both request and reply.
648 * We assume one is at most one page
649 */
650 WARN_ON_ONCE(pages > RPCSVC_MAXPAGES);
651 if (pages > RPCSVC_MAXPAGES)
652 pages = RPCSVC_MAXPAGES;
653
654 ret = alloc_pages_bulk_array_node(GFP_KERNEL, node, pages,
655 rqstp->rq_pages);
656 return ret == pages;
657 }
658
659 /*
660 * Release an RPC server buffer
661 */
662 static void
svc_release_buffer(struct svc_rqst * rqstp)663 svc_release_buffer(struct svc_rqst *rqstp)
664 {
665 unsigned int i;
666
667 for (i = 0; i < ARRAY_SIZE(rqstp->rq_pages); i++)
668 if (rqstp->rq_pages[i])
669 put_page(rqstp->rq_pages[i]);
670 }
671
672 static void
svc_rqst_free(struct svc_rqst * rqstp)673 svc_rqst_free(struct svc_rqst *rqstp)
674 {
675 folio_batch_release(&rqstp->rq_fbatch);
676 svc_release_buffer(rqstp);
677 if (rqstp->rq_scratch_page)
678 put_page(rqstp->rq_scratch_page);
679 kfree(rqstp->rq_resp);
680 kfree(rqstp->rq_argp);
681 kfree(rqstp->rq_auth_data);
682 kfree_rcu(rqstp, rq_rcu_head);
683 }
684
685 static struct svc_rqst *
svc_prepare_thread(struct svc_serv * serv,struct svc_pool * pool,int node)686 svc_prepare_thread(struct svc_serv *serv, struct svc_pool *pool, int node)
687 {
688 struct svc_rqst *rqstp;
689
690 rqstp = kzalloc_node(sizeof(*rqstp), GFP_KERNEL, node);
691 if (!rqstp)
692 return rqstp;
693
694 folio_batch_init(&rqstp->rq_fbatch);
695
696 rqstp->rq_server = serv;
697 rqstp->rq_pool = pool;
698
699 rqstp->rq_scratch_page = alloc_pages_node(node, GFP_KERNEL, 0);
700 if (!rqstp->rq_scratch_page)
701 goto out_enomem;
702
703 rqstp->rq_argp = kmalloc_node(serv->sv_xdrsize, GFP_KERNEL, node);
704 if (!rqstp->rq_argp)
705 goto out_enomem;
706
707 rqstp->rq_resp = kmalloc_node(serv->sv_xdrsize, GFP_KERNEL, node);
708 if (!rqstp->rq_resp)
709 goto out_enomem;
710
711 if (!svc_init_buffer(rqstp, serv->sv_max_mesg, node))
712 goto out_enomem;
713
714 rqstp->rq_err = -EAGAIN; /* No error yet */
715
716 serv->sv_nrthreads += 1;
717 pool->sp_nrthreads += 1;
718
719 /* Protected by whatever lock the service uses when calling
720 * svc_set_num_threads()
721 */
722 list_add_rcu(&rqstp->rq_all, &pool->sp_all_threads);
723
724 return rqstp;
725
726 out_enomem:
727 svc_rqst_free(rqstp);
728 return NULL;
729 }
730
731 /**
732 * svc_pool_wake_idle_thread - Awaken an idle thread in @pool
733 * @pool: service thread pool
734 *
735 * Can be called from soft IRQ or process context. Finding an idle
736 * service thread and marking it BUSY is atomic with respect to
737 * other calls to svc_pool_wake_idle_thread().
738 *
739 */
svc_pool_wake_idle_thread(struct svc_pool * pool)740 void svc_pool_wake_idle_thread(struct svc_pool *pool)
741 {
742 struct svc_rqst *rqstp;
743 struct llist_node *ln;
744
745 rcu_read_lock();
746 ln = READ_ONCE(pool->sp_idle_threads.first);
747 if (ln) {
748 rqstp = llist_entry(ln, struct svc_rqst, rq_idle);
749 WRITE_ONCE(rqstp->rq_qtime, ktime_get());
750 if (!task_is_running(rqstp->rq_task)) {
751 wake_up_process(rqstp->rq_task);
752 trace_svc_wake_up(rqstp->rq_task->pid);
753 percpu_counter_inc(&pool->sp_threads_woken);
754 }
755 rcu_read_unlock();
756 return;
757 }
758 rcu_read_unlock();
759
760 }
761 EXPORT_SYMBOL_GPL(svc_pool_wake_idle_thread);
762
763 static struct svc_pool *
svc_pool_next(struct svc_serv * serv,struct svc_pool * pool,unsigned int * state)764 svc_pool_next(struct svc_serv *serv, struct svc_pool *pool, unsigned int *state)
765 {
766 return pool ? pool : &serv->sv_pools[(*state)++ % serv->sv_nrpools];
767 }
768
769 static struct svc_pool *
svc_pool_victim(struct svc_serv * serv,struct svc_pool * target_pool,unsigned int * state)770 svc_pool_victim(struct svc_serv *serv, struct svc_pool *target_pool,
771 unsigned int *state)
772 {
773 struct svc_pool *pool;
774 unsigned int i;
775
776 pool = target_pool;
777
778 if (!pool) {
779 for (i = 0; i < serv->sv_nrpools; i++) {
780 pool = &serv->sv_pools[--(*state) % serv->sv_nrpools];
781 if (pool->sp_nrthreads)
782 break;
783 }
784 }
785
786 if (pool && pool->sp_nrthreads) {
787 set_bit(SP_VICTIM_REMAINS, &pool->sp_flags);
788 set_bit(SP_NEED_VICTIM, &pool->sp_flags);
789 return pool;
790 }
791 return NULL;
792 }
793
794 static int
svc_start_kthreads(struct svc_serv * serv,struct svc_pool * pool,int nrservs)795 svc_start_kthreads(struct svc_serv *serv, struct svc_pool *pool, int nrservs)
796 {
797 struct svc_rqst *rqstp;
798 struct task_struct *task;
799 struct svc_pool *chosen_pool;
800 unsigned int state = serv->sv_nrthreads-1;
801 int node;
802 int err;
803
804 do {
805 nrservs--;
806 chosen_pool = svc_pool_next(serv, pool, &state);
807 node = svc_pool_map_get_node(chosen_pool->sp_id);
808
809 rqstp = svc_prepare_thread(serv, chosen_pool, node);
810 if (!rqstp)
811 return -ENOMEM;
812 task = kthread_create_on_node(serv->sv_threadfn, rqstp,
813 node, "%s", serv->sv_name);
814 if (IS_ERR(task)) {
815 svc_exit_thread(rqstp);
816 return PTR_ERR(task);
817 }
818
819 rqstp->rq_task = task;
820 if (serv->sv_nrpools > 1)
821 svc_pool_map_set_cpumask(task, chosen_pool->sp_id);
822
823 svc_sock_update_bufs(serv);
824 wake_up_process(task);
825
826 wait_var_event(&rqstp->rq_err, rqstp->rq_err != -EAGAIN);
827 err = rqstp->rq_err;
828 if (err) {
829 svc_exit_thread(rqstp);
830 return err;
831 }
832 } while (nrservs > 0);
833
834 return 0;
835 }
836
837 static int
svc_stop_kthreads(struct svc_serv * serv,struct svc_pool * pool,int nrservs)838 svc_stop_kthreads(struct svc_serv *serv, struct svc_pool *pool, int nrservs)
839 {
840 unsigned int state = serv->sv_nrthreads-1;
841 struct svc_pool *victim;
842
843 do {
844 victim = svc_pool_victim(serv, pool, &state);
845 if (!victim)
846 break;
847 svc_pool_wake_idle_thread(victim);
848 wait_on_bit(&victim->sp_flags, SP_VICTIM_REMAINS,
849 TASK_IDLE);
850 nrservs++;
851 } while (nrservs < 0);
852 return 0;
853 }
854
855 /**
856 * svc_set_num_threads - adjust number of threads per RPC service
857 * @serv: RPC service to adjust
858 * @pool: Specific pool from which to choose threads, or NULL
859 * @nrservs: New number of threads for @serv (0 or less means kill all threads)
860 *
861 * Create or destroy threads to make the number of threads for @serv the
862 * given number. If @pool is non-NULL, change only threads in that pool;
863 * otherwise, round-robin between all pools for @serv. @serv's
864 * sv_nrthreads is adjusted for each thread created or destroyed.
865 *
866 * Caller must ensure mutual exclusion between this and server startup or
867 * shutdown.
868 *
869 * Returns zero on success or a negative errno if an error occurred while
870 * starting a thread.
871 */
872 int
svc_set_num_threads(struct svc_serv * serv,struct svc_pool * pool,int nrservs)873 svc_set_num_threads(struct svc_serv *serv, struct svc_pool *pool, int nrservs)
874 {
875 if (!pool)
876 nrservs -= serv->sv_nrthreads;
877 else
878 nrservs -= pool->sp_nrthreads;
879
880 if (nrservs > 0)
881 return svc_start_kthreads(serv, pool, nrservs);
882 if (nrservs < 0)
883 return svc_stop_kthreads(serv, pool, nrservs);
884 return 0;
885 }
886 EXPORT_SYMBOL_GPL(svc_set_num_threads);
887
888 /**
889 * svc_rqst_replace_page - Replace one page in rq_pages[]
890 * @rqstp: svc_rqst with pages to replace
891 * @page: replacement page
892 *
893 * When replacing a page in rq_pages, batch the release of the
894 * replaced pages to avoid hammering the page allocator.
895 *
896 * Return values:
897 * %true: page replaced
898 * %false: array bounds checking failed
899 */
svc_rqst_replace_page(struct svc_rqst * rqstp,struct page * page)900 bool svc_rqst_replace_page(struct svc_rqst *rqstp, struct page *page)
901 {
902 struct page **begin = rqstp->rq_pages;
903 struct page **end = &rqstp->rq_pages[RPCSVC_MAXPAGES];
904
905 if (unlikely(rqstp->rq_next_page < begin || rqstp->rq_next_page > end)) {
906 trace_svc_replace_page_err(rqstp);
907 return false;
908 }
909
910 if (*rqstp->rq_next_page) {
911 if (!folio_batch_add(&rqstp->rq_fbatch,
912 page_folio(*rqstp->rq_next_page)))
913 __folio_batch_release(&rqstp->rq_fbatch);
914 }
915
916 get_page(page);
917 *(rqstp->rq_next_page++) = page;
918 return true;
919 }
920 EXPORT_SYMBOL_GPL(svc_rqst_replace_page);
921
922 /**
923 * svc_rqst_release_pages - Release Reply buffer pages
924 * @rqstp: RPC transaction context
925 *
926 * Release response pages that might still be in flight after
927 * svc_send, and any spliced filesystem-owned pages.
928 */
svc_rqst_release_pages(struct svc_rqst * rqstp)929 void svc_rqst_release_pages(struct svc_rqst *rqstp)
930 {
931 int i, count = rqstp->rq_next_page - rqstp->rq_respages;
932
933 if (count) {
934 release_pages(rqstp->rq_respages, count);
935 for (i = 0; i < count; i++)
936 rqstp->rq_respages[i] = NULL;
937 }
938 }
939
940 /**
941 * svc_exit_thread - finalise the termination of a sunrpc server thread
942 * @rqstp: the svc_rqst which represents the thread.
943 *
944 * When a thread started with svc_new_thread() exits it must call
945 * svc_exit_thread() as its last act. This must be done with the
946 * service mutex held. Normally this is held by a DIFFERENT thread, the
947 * one that is calling svc_set_num_threads() and which will wait for
948 * SP_VICTIM_REMAINS to be cleared before dropping the mutex. If the
949 * thread exits for any reason other than svc_thread_should_stop()
950 * returning %true (which indicated that svc_set_num_threads() is
951 * waiting for it to exit), then it must take the service mutex itself,
952 * which can only safely be done using mutex_try_lock().
953 */
954 void
svc_exit_thread(struct svc_rqst * rqstp)955 svc_exit_thread(struct svc_rqst *rqstp)
956 {
957 struct svc_serv *serv = rqstp->rq_server;
958 struct svc_pool *pool = rqstp->rq_pool;
959
960 list_del_rcu(&rqstp->rq_all);
961
962 pool->sp_nrthreads -= 1;
963 serv->sv_nrthreads -= 1;
964 svc_sock_update_bufs(serv);
965
966 svc_rqst_free(rqstp);
967
968 clear_and_wake_up_bit(SP_VICTIM_REMAINS, &pool->sp_flags);
969 }
970 EXPORT_SYMBOL_GPL(svc_exit_thread);
971
972 /*
973 * Register an "inet" protocol family netid with the local
974 * rpcbind daemon via an rpcbind v4 SET request.
975 *
976 * No netconfig infrastructure is available in the kernel, so
977 * we map IP_ protocol numbers to netids by hand.
978 *
979 * Returns zero on success; a negative errno value is returned
980 * if any error occurs.
981 */
__svc_rpcb_register4(struct net * net,const u32 program,const u32 version,const unsigned short protocol,const unsigned short port)982 static int __svc_rpcb_register4(struct net *net, const u32 program,
983 const u32 version,
984 const unsigned short protocol,
985 const unsigned short port)
986 {
987 const struct sockaddr_in sin = {
988 .sin_family = AF_INET,
989 .sin_addr.s_addr = htonl(INADDR_ANY),
990 .sin_port = htons(port),
991 };
992 const char *netid;
993 int error;
994
995 switch (protocol) {
996 case IPPROTO_UDP:
997 netid = RPCBIND_NETID_UDP;
998 break;
999 case IPPROTO_TCP:
1000 netid = RPCBIND_NETID_TCP;
1001 break;
1002 default:
1003 return -ENOPROTOOPT;
1004 }
1005
1006 error = rpcb_v4_register(net, program, version,
1007 (const struct sockaddr *)&sin, netid);
1008
1009 /*
1010 * User space didn't support rpcbind v4, so retry this
1011 * registration request with the legacy rpcbind v2 protocol.
1012 */
1013 if (error == -EPROTONOSUPPORT)
1014 error = rpcb_register(net, program, version, protocol, port);
1015
1016 return error;
1017 }
1018
1019 #if IS_ENABLED(CONFIG_IPV6)
1020 /*
1021 * Register an "inet6" protocol family netid with the local
1022 * rpcbind daemon via an rpcbind v4 SET request.
1023 *
1024 * No netconfig infrastructure is available in the kernel, so
1025 * we map IP_ protocol numbers to netids by hand.
1026 *
1027 * Returns zero on success; a negative errno value is returned
1028 * if any error occurs.
1029 */
__svc_rpcb_register6(struct net * net,const u32 program,const u32 version,const unsigned short protocol,const unsigned short port)1030 static int __svc_rpcb_register6(struct net *net, const u32 program,
1031 const u32 version,
1032 const unsigned short protocol,
1033 const unsigned short port)
1034 {
1035 const struct sockaddr_in6 sin6 = {
1036 .sin6_family = AF_INET6,
1037 .sin6_addr = IN6ADDR_ANY_INIT,
1038 .sin6_port = htons(port),
1039 };
1040 const char *netid;
1041 int error;
1042
1043 switch (protocol) {
1044 case IPPROTO_UDP:
1045 netid = RPCBIND_NETID_UDP6;
1046 break;
1047 case IPPROTO_TCP:
1048 netid = RPCBIND_NETID_TCP6;
1049 break;
1050 default:
1051 return -ENOPROTOOPT;
1052 }
1053
1054 error = rpcb_v4_register(net, program, version,
1055 (const struct sockaddr *)&sin6, netid);
1056
1057 /*
1058 * User space didn't support rpcbind version 4, so we won't
1059 * use a PF_INET6 listener.
1060 */
1061 if (error == -EPROTONOSUPPORT)
1062 error = -EAFNOSUPPORT;
1063
1064 return error;
1065 }
1066 #endif /* IS_ENABLED(CONFIG_IPV6) */
1067
1068 /*
1069 * Register a kernel RPC service via rpcbind version 4.
1070 *
1071 * Returns zero on success; a negative errno value is returned
1072 * if any error occurs.
1073 */
__svc_register(struct net * net,const char * progname,const u32 program,const u32 version,const int family,const unsigned short protocol,const unsigned short port)1074 static int __svc_register(struct net *net, const char *progname,
1075 const u32 program, const u32 version,
1076 const int family,
1077 const unsigned short protocol,
1078 const unsigned short port)
1079 {
1080 int error = -EAFNOSUPPORT;
1081
1082 switch (family) {
1083 case PF_INET:
1084 error = __svc_rpcb_register4(net, program, version,
1085 protocol, port);
1086 break;
1087 #if IS_ENABLED(CONFIG_IPV6)
1088 case PF_INET6:
1089 error = __svc_rpcb_register6(net, program, version,
1090 protocol, port);
1091 #endif
1092 }
1093
1094 trace_svc_register(progname, version, family, protocol, port, error);
1095 return error;
1096 }
1097
1098 static
svc_rpcbind_set_version(struct net * net,const struct svc_program * progp,u32 version,int family,unsigned short proto,unsigned short port)1099 int svc_rpcbind_set_version(struct net *net,
1100 const struct svc_program *progp,
1101 u32 version, int family,
1102 unsigned short proto,
1103 unsigned short port)
1104 {
1105 return __svc_register(net, progp->pg_name, progp->pg_prog,
1106 version, family, proto, port);
1107
1108 }
1109
svc_generic_rpcbind_set(struct net * net,const struct svc_program * progp,u32 version,int family,unsigned short proto,unsigned short port)1110 int svc_generic_rpcbind_set(struct net *net,
1111 const struct svc_program *progp,
1112 u32 version, int family,
1113 unsigned short proto,
1114 unsigned short port)
1115 {
1116 const struct svc_version *vers = progp->pg_vers[version];
1117 int error;
1118
1119 if (vers == NULL)
1120 return 0;
1121
1122 if (vers->vs_hidden) {
1123 trace_svc_noregister(progp->pg_name, version, proto,
1124 port, family, 0);
1125 return 0;
1126 }
1127
1128 /*
1129 * Don't register a UDP port if we need congestion
1130 * control.
1131 */
1132 if (vers->vs_need_cong_ctrl && proto == IPPROTO_UDP)
1133 return 0;
1134
1135 error = svc_rpcbind_set_version(net, progp, version,
1136 family, proto, port);
1137
1138 return (vers->vs_rpcb_optnl) ? 0 : error;
1139 }
1140 EXPORT_SYMBOL_GPL(svc_generic_rpcbind_set);
1141
1142 /**
1143 * svc_register - register an RPC service with the local portmapper
1144 * @serv: svc_serv struct for the service to register
1145 * @net: net namespace for the service to register
1146 * @family: protocol family of service's listener socket
1147 * @proto: transport protocol number to advertise
1148 * @port: port to advertise
1149 *
1150 * Service is registered for any address in the passed-in protocol family
1151 */
svc_register(const struct svc_serv * serv,struct net * net,const int family,const unsigned short proto,const unsigned short port)1152 int svc_register(const struct svc_serv *serv, struct net *net,
1153 const int family, const unsigned short proto,
1154 const unsigned short port)
1155 {
1156 unsigned int p, i;
1157 int error = 0;
1158
1159 WARN_ON_ONCE(proto == 0 && port == 0);
1160 if (proto == 0 && port == 0)
1161 return -EINVAL;
1162
1163 for (p = 0; p < serv->sv_nprogs; p++) {
1164 struct svc_program *progp = &serv->sv_programs[p];
1165
1166 for (i = 0; i < progp->pg_nvers; i++) {
1167
1168 error = progp->pg_rpcbind_set(net, progp, i,
1169 family, proto, port);
1170 if (error < 0) {
1171 printk(KERN_WARNING "svc: failed to register "
1172 "%sv%u RPC service (errno %d).\n",
1173 progp->pg_name, i, -error);
1174 break;
1175 }
1176 }
1177 }
1178
1179 return error;
1180 }
1181
1182 /*
1183 * If user space is running rpcbind, it should take the v4 UNSET
1184 * and clear everything for this [program, version]. If user space
1185 * is running portmap, it will reject the v4 UNSET, but won't have
1186 * any "inet6" entries anyway. So a PMAP_UNSET should be sufficient
1187 * in this case to clear all existing entries for [program, version].
1188 */
__svc_unregister(struct net * net,const u32 program,const u32 version,const char * progname)1189 static void __svc_unregister(struct net *net, const u32 program, const u32 version,
1190 const char *progname)
1191 {
1192 int error;
1193
1194 error = rpcb_v4_register(net, program, version, NULL, "");
1195
1196 /*
1197 * User space didn't support rpcbind v4, so retry this
1198 * request with the legacy rpcbind v2 protocol.
1199 */
1200 if (error == -EPROTONOSUPPORT)
1201 error = rpcb_register(net, program, version, 0, 0);
1202
1203 trace_svc_unregister(progname, version, error);
1204 }
1205
1206 /*
1207 * All netids, bind addresses and ports registered for [program, version]
1208 * are removed from the local rpcbind database (if the service is not
1209 * hidden) to make way for a new instance of the service.
1210 *
1211 * The result of unregistration is reported via dprintk for those who want
1212 * verification of the result, but is otherwise not important.
1213 */
svc_unregister(const struct svc_serv * serv,struct net * net)1214 static void svc_unregister(const struct svc_serv *serv, struct net *net)
1215 {
1216 struct sighand_struct *sighand;
1217 unsigned long flags;
1218 unsigned int p, i;
1219
1220 clear_thread_flag(TIF_SIGPENDING);
1221
1222 for (p = 0; p < serv->sv_nprogs; p++) {
1223 struct svc_program *progp = &serv->sv_programs[p];
1224
1225 for (i = 0; i < progp->pg_nvers; i++) {
1226 if (progp->pg_vers[i] == NULL)
1227 continue;
1228 if (progp->pg_vers[i]->vs_hidden)
1229 continue;
1230 __svc_unregister(net, progp->pg_prog, i, progp->pg_name);
1231 }
1232 }
1233
1234 rcu_read_lock();
1235 sighand = rcu_dereference(current->sighand);
1236 spin_lock_irqsave(&sighand->siglock, flags);
1237 recalc_sigpending();
1238 spin_unlock_irqrestore(&sighand->siglock, flags);
1239 rcu_read_unlock();
1240 }
1241
1242 /*
1243 * dprintk the given error with the address of the client that caused it.
1244 */
1245 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
1246 static __printf(2, 3)
svc_printk(struct svc_rqst * rqstp,const char * fmt,...)1247 void svc_printk(struct svc_rqst *rqstp, const char *fmt, ...)
1248 {
1249 struct va_format vaf;
1250 va_list args;
1251 char buf[RPC_MAX_ADDRBUFLEN];
1252
1253 va_start(args, fmt);
1254
1255 vaf.fmt = fmt;
1256 vaf.va = &args;
1257
1258 dprintk("svc: %s: %pV", svc_print_addr(rqstp, buf, sizeof(buf)), &vaf);
1259
1260 va_end(args);
1261 }
1262 #else
svc_printk(struct svc_rqst * rqstp,const char * fmt,...)1263 static __printf(2,3) void svc_printk(struct svc_rqst *rqstp, const char *fmt, ...) {}
1264 #endif
1265
1266 __be32
svc_generic_init_request(struct svc_rqst * rqstp,const struct svc_program * progp,struct svc_process_info * ret)1267 svc_generic_init_request(struct svc_rqst *rqstp,
1268 const struct svc_program *progp,
1269 struct svc_process_info *ret)
1270 {
1271 const struct svc_version *versp = NULL; /* compiler food */
1272 const struct svc_procedure *procp = NULL;
1273
1274 if (rqstp->rq_vers >= progp->pg_nvers )
1275 goto err_bad_vers;
1276 versp = progp->pg_vers[rqstp->rq_vers];
1277 if (!versp)
1278 goto err_bad_vers;
1279
1280 /*
1281 * Some protocol versions (namely NFSv4) require some form of
1282 * congestion control. (See RFC 7530 section 3.1 paragraph 2)
1283 * In other words, UDP is not allowed. We mark those when setting
1284 * up the svc_xprt, and verify that here.
1285 *
1286 * The spec is not very clear about what error should be returned
1287 * when someone tries to access a server that is listening on UDP
1288 * for lower versions. RPC_PROG_MISMATCH seems to be the closest
1289 * fit.
1290 */
1291 if (versp->vs_need_cong_ctrl && rqstp->rq_xprt &&
1292 !test_bit(XPT_CONG_CTRL, &rqstp->rq_xprt->xpt_flags))
1293 goto err_bad_vers;
1294
1295 if (rqstp->rq_proc >= versp->vs_nproc)
1296 goto err_bad_proc;
1297 rqstp->rq_procinfo = procp = &versp->vs_proc[rqstp->rq_proc];
1298
1299 /* Initialize storage for argp and resp */
1300 memset(rqstp->rq_argp, 0, procp->pc_argzero);
1301 memset(rqstp->rq_resp, 0, procp->pc_ressize);
1302
1303 /* Bump per-procedure stats counter */
1304 this_cpu_inc(versp->vs_count[rqstp->rq_proc]);
1305
1306 ret->dispatch = versp->vs_dispatch;
1307 return rpc_success;
1308 err_bad_vers:
1309 ret->mismatch.lovers = progp->pg_lovers;
1310 ret->mismatch.hivers = progp->pg_hivers;
1311 return rpc_prog_mismatch;
1312 err_bad_proc:
1313 return rpc_proc_unavail;
1314 }
1315 EXPORT_SYMBOL_GPL(svc_generic_init_request);
1316
1317 /*
1318 * Common routine for processing the RPC request.
1319 */
1320 static int
svc_process_common(struct svc_rqst * rqstp)1321 svc_process_common(struct svc_rqst *rqstp)
1322 {
1323 struct xdr_stream *xdr = &rqstp->rq_res_stream;
1324 struct svc_program *progp = NULL;
1325 const struct svc_procedure *procp = NULL;
1326 struct svc_serv *serv = rqstp->rq_server;
1327 struct svc_process_info process;
1328 enum svc_auth_status auth_res;
1329 unsigned int aoffset;
1330 int pr, rc;
1331 __be32 *p;
1332
1333 /* Will be turned off only when NFSv4 Sessions are used */
1334 set_bit(RQ_USEDEFERRAL, &rqstp->rq_flags);
1335 clear_bit(RQ_DROPME, &rqstp->rq_flags);
1336
1337 /* Construct the first words of the reply: */
1338 svcxdr_init_encode(rqstp);
1339 xdr_stream_encode_be32(xdr, rqstp->rq_xid);
1340 xdr_stream_encode_be32(xdr, rpc_reply);
1341
1342 p = xdr_inline_decode(&rqstp->rq_arg_stream, XDR_UNIT * 4);
1343 if (unlikely(!p))
1344 goto err_short_len;
1345 if (*p++ != cpu_to_be32(RPC_VERSION))
1346 goto err_bad_rpc;
1347
1348 xdr_stream_encode_be32(xdr, rpc_msg_accepted);
1349
1350 rqstp->rq_prog = be32_to_cpup(p++);
1351 rqstp->rq_vers = be32_to_cpup(p++);
1352 rqstp->rq_proc = be32_to_cpup(p);
1353
1354 for (pr = 0; pr < serv->sv_nprogs; pr++)
1355 if (rqstp->rq_prog == serv->sv_programs[pr].pg_prog)
1356 progp = &serv->sv_programs[pr];
1357
1358 /*
1359 * Decode auth data, and add verifier to reply buffer.
1360 * We do this before anything else in order to get a decent
1361 * auth verifier.
1362 */
1363 auth_res = svc_authenticate(rqstp);
1364 /* Also give the program a chance to reject this call: */
1365 if (auth_res == SVC_OK && progp)
1366 auth_res = progp->pg_authenticate(rqstp);
1367 trace_svc_authenticate(rqstp, auth_res);
1368 switch (auth_res) {
1369 case SVC_OK:
1370 break;
1371 case SVC_GARBAGE:
1372 goto err_garbage_args;
1373 case SVC_SYSERR:
1374 goto err_system_err;
1375 case SVC_DENIED:
1376 goto err_bad_auth;
1377 case SVC_CLOSE:
1378 goto close;
1379 case SVC_DROP:
1380 goto dropit;
1381 case SVC_COMPLETE:
1382 goto sendit;
1383 default:
1384 pr_warn_once("Unexpected svc_auth_status (%d)\n", auth_res);
1385 goto err_system_err;
1386 }
1387
1388 if (progp == NULL)
1389 goto err_bad_prog;
1390
1391 switch (progp->pg_init_request(rqstp, progp, &process)) {
1392 case rpc_success:
1393 break;
1394 case rpc_prog_unavail:
1395 goto err_bad_prog;
1396 case rpc_prog_mismatch:
1397 goto err_bad_vers;
1398 case rpc_proc_unavail:
1399 goto err_bad_proc;
1400 }
1401
1402 procp = rqstp->rq_procinfo;
1403 /* Should this check go into the dispatcher? */
1404 if (!procp || !procp->pc_func)
1405 goto err_bad_proc;
1406
1407 /* Syntactic check complete */
1408 if (serv->sv_stats)
1409 serv->sv_stats->rpccnt++;
1410 trace_svc_process(rqstp, progp->pg_name);
1411
1412 aoffset = xdr_stream_pos(xdr);
1413
1414 /* un-reserve some of the out-queue now that we have a
1415 * better idea of reply size
1416 */
1417 if (procp->pc_xdrressize)
1418 svc_reserve_auth(rqstp, procp->pc_xdrressize<<2);
1419
1420 /* Call the function that processes the request. */
1421 rc = process.dispatch(rqstp);
1422 if (procp->pc_release)
1423 procp->pc_release(rqstp);
1424 xdr_finish_decode(xdr);
1425
1426 if (!rc)
1427 goto dropit;
1428 if (rqstp->rq_auth_stat != rpc_auth_ok)
1429 goto err_bad_auth;
1430
1431 if (*rqstp->rq_accept_statp != rpc_success)
1432 xdr_truncate_encode(xdr, aoffset);
1433
1434 if (procp->pc_encode == NULL)
1435 goto dropit;
1436
1437 sendit:
1438 if (svc_authorise(rqstp))
1439 goto close_xprt;
1440 return 1; /* Caller can now send it */
1441
1442 dropit:
1443 svc_authorise(rqstp); /* doesn't hurt to call this twice */
1444 dprintk("svc: svc_process dropit\n");
1445 return 0;
1446
1447 close:
1448 svc_authorise(rqstp);
1449 close_xprt:
1450 if (rqstp->rq_xprt && test_bit(XPT_TEMP, &rqstp->rq_xprt->xpt_flags))
1451 svc_xprt_close(rqstp->rq_xprt);
1452 dprintk("svc: svc_process close\n");
1453 return 0;
1454
1455 err_short_len:
1456 svc_printk(rqstp, "short len %u, dropping request\n",
1457 rqstp->rq_arg.len);
1458 goto close_xprt;
1459
1460 err_bad_rpc:
1461 if (serv->sv_stats)
1462 serv->sv_stats->rpcbadfmt++;
1463 xdr_stream_encode_u32(xdr, RPC_MSG_DENIED);
1464 xdr_stream_encode_u32(xdr, RPC_MISMATCH);
1465 /* Only RPCv2 supported */
1466 xdr_stream_encode_u32(xdr, RPC_VERSION);
1467 xdr_stream_encode_u32(xdr, RPC_VERSION);
1468 return 1; /* don't wrap */
1469
1470 err_bad_auth:
1471 dprintk("svc: authentication failed (%d)\n",
1472 be32_to_cpu(rqstp->rq_auth_stat));
1473 if (serv->sv_stats)
1474 serv->sv_stats->rpcbadauth++;
1475 /* Restore write pointer to location of reply status: */
1476 xdr_truncate_encode(xdr, XDR_UNIT * 2);
1477 xdr_stream_encode_u32(xdr, RPC_MSG_DENIED);
1478 xdr_stream_encode_u32(xdr, RPC_AUTH_ERROR);
1479 xdr_stream_encode_be32(xdr, rqstp->rq_auth_stat);
1480 goto sendit;
1481
1482 err_bad_prog:
1483 dprintk("svc: unknown program %d\n", rqstp->rq_prog);
1484 if (serv->sv_stats)
1485 serv->sv_stats->rpcbadfmt++;
1486 *rqstp->rq_accept_statp = rpc_prog_unavail;
1487 goto sendit;
1488
1489 err_bad_vers:
1490 svc_printk(rqstp, "unknown version (%d for prog %d, %s)\n",
1491 rqstp->rq_vers, rqstp->rq_prog, progp->pg_name);
1492
1493 if (serv->sv_stats)
1494 serv->sv_stats->rpcbadfmt++;
1495 *rqstp->rq_accept_statp = rpc_prog_mismatch;
1496
1497 /*
1498 * svc_authenticate() has already added the verifier and
1499 * advanced the stream just past rq_accept_statp.
1500 */
1501 xdr_stream_encode_u32(xdr, process.mismatch.lovers);
1502 xdr_stream_encode_u32(xdr, process.mismatch.hivers);
1503 goto sendit;
1504
1505 err_bad_proc:
1506 svc_printk(rqstp, "unknown procedure (%d)\n", rqstp->rq_proc);
1507
1508 if (serv->sv_stats)
1509 serv->sv_stats->rpcbadfmt++;
1510 *rqstp->rq_accept_statp = rpc_proc_unavail;
1511 goto sendit;
1512
1513 err_garbage_args:
1514 svc_printk(rqstp, "failed to decode RPC header\n");
1515
1516 if (serv->sv_stats)
1517 serv->sv_stats->rpcbadfmt++;
1518 *rqstp->rq_accept_statp = rpc_garbage_args;
1519 goto sendit;
1520
1521 err_system_err:
1522 if (serv->sv_stats)
1523 serv->sv_stats->rpcbadfmt++;
1524 *rqstp->rq_accept_statp = rpc_system_err;
1525 goto sendit;
1526 }
1527
1528 /*
1529 * Drop request
1530 */
svc_drop(struct svc_rqst * rqstp)1531 static void svc_drop(struct svc_rqst *rqstp)
1532 {
1533 trace_svc_drop(rqstp);
1534 }
1535
1536 /**
1537 * svc_process - Execute one RPC transaction
1538 * @rqstp: RPC transaction context
1539 *
1540 */
svc_process(struct svc_rqst * rqstp)1541 void svc_process(struct svc_rqst *rqstp)
1542 {
1543 struct kvec *resv = &rqstp->rq_res.head[0];
1544 __be32 *p;
1545
1546 #if IS_ENABLED(CONFIG_FAIL_SUNRPC)
1547 if (!fail_sunrpc.ignore_server_disconnect &&
1548 should_fail(&fail_sunrpc.attr, 1))
1549 svc_xprt_deferred_close(rqstp->rq_xprt);
1550 #endif
1551
1552 /*
1553 * Setup response xdr_buf.
1554 * Initially it has just one page
1555 */
1556 rqstp->rq_next_page = &rqstp->rq_respages[1];
1557 resv->iov_base = page_address(rqstp->rq_respages[0]);
1558 resv->iov_len = 0;
1559 rqstp->rq_res.pages = rqstp->rq_next_page;
1560 rqstp->rq_res.len = 0;
1561 rqstp->rq_res.page_base = 0;
1562 rqstp->rq_res.page_len = 0;
1563 rqstp->rq_res.buflen = PAGE_SIZE;
1564 rqstp->rq_res.tail[0].iov_base = NULL;
1565 rqstp->rq_res.tail[0].iov_len = 0;
1566
1567 svcxdr_init_decode(rqstp);
1568 p = xdr_inline_decode(&rqstp->rq_arg_stream, XDR_UNIT * 2);
1569 if (unlikely(!p))
1570 goto out_drop;
1571 rqstp->rq_xid = *p++;
1572 if (unlikely(*p != rpc_call))
1573 goto out_baddir;
1574
1575 if (!svc_process_common(rqstp))
1576 goto out_drop;
1577 svc_send(rqstp);
1578 return;
1579
1580 out_baddir:
1581 svc_printk(rqstp, "bad direction 0x%08x, dropping request\n",
1582 be32_to_cpu(*p));
1583 if (rqstp->rq_server->sv_stats)
1584 rqstp->rq_server->sv_stats->rpcbadfmt++;
1585 out_drop:
1586 svc_drop(rqstp);
1587 }
1588
1589 #if defined(CONFIG_SUNRPC_BACKCHANNEL)
1590 /**
1591 * svc_process_bc - process a reverse-direction RPC request
1592 * @req: RPC request to be used for client-side processing
1593 * @rqstp: server-side execution context
1594 *
1595 */
svc_process_bc(struct rpc_rqst * req,struct svc_rqst * rqstp)1596 void svc_process_bc(struct rpc_rqst *req, struct svc_rqst *rqstp)
1597 {
1598 struct rpc_timeout timeout = {
1599 .to_increment = 0,
1600 };
1601 struct rpc_task *task;
1602 int proc_error;
1603
1604 /* Build the svc_rqst used by the common processing routine */
1605 rqstp->rq_xid = req->rq_xid;
1606 rqstp->rq_prot = req->rq_xprt->prot;
1607 rqstp->rq_bc_net = req->rq_xprt->xprt_net;
1608
1609 rqstp->rq_addrlen = sizeof(req->rq_xprt->addr);
1610 memcpy(&rqstp->rq_addr, &req->rq_xprt->addr, rqstp->rq_addrlen);
1611 memcpy(&rqstp->rq_arg, &req->rq_rcv_buf, sizeof(rqstp->rq_arg));
1612 memcpy(&rqstp->rq_res, &req->rq_snd_buf, sizeof(rqstp->rq_res));
1613
1614 /* Adjust the argument buffer length */
1615 rqstp->rq_arg.len = req->rq_private_buf.len;
1616 if (rqstp->rq_arg.len <= rqstp->rq_arg.head[0].iov_len) {
1617 rqstp->rq_arg.head[0].iov_len = rqstp->rq_arg.len;
1618 rqstp->rq_arg.page_len = 0;
1619 } else if (rqstp->rq_arg.len <= rqstp->rq_arg.head[0].iov_len +
1620 rqstp->rq_arg.page_len)
1621 rqstp->rq_arg.page_len = rqstp->rq_arg.len -
1622 rqstp->rq_arg.head[0].iov_len;
1623 else
1624 rqstp->rq_arg.len = rqstp->rq_arg.head[0].iov_len +
1625 rqstp->rq_arg.page_len;
1626
1627 /* Reset the response buffer */
1628 rqstp->rq_res.head[0].iov_len = 0;
1629
1630 /*
1631 * Skip the XID and calldir fields because they've already
1632 * been processed by the caller.
1633 */
1634 svcxdr_init_decode(rqstp);
1635 if (!xdr_inline_decode(&rqstp->rq_arg_stream, XDR_UNIT * 2))
1636 return;
1637
1638 /* Parse and execute the bc call */
1639 proc_error = svc_process_common(rqstp);
1640
1641 atomic_dec(&req->rq_xprt->bc_slot_count);
1642 if (!proc_error) {
1643 /* Processing error: drop the request */
1644 xprt_free_bc_request(req);
1645 return;
1646 }
1647 /* Finally, send the reply synchronously */
1648 if (rqstp->bc_to_initval > 0) {
1649 timeout.to_initval = rqstp->bc_to_initval;
1650 timeout.to_retries = rqstp->bc_to_retries;
1651 } else {
1652 timeout.to_initval = req->rq_xprt->timeout->to_initval;
1653 timeout.to_retries = req->rq_xprt->timeout->to_retries;
1654 }
1655 timeout.to_maxval = timeout.to_initval;
1656 memcpy(&req->rq_snd_buf, &rqstp->rq_res, sizeof(req->rq_snd_buf));
1657 task = rpc_run_bc_task(req, &timeout);
1658
1659 if (IS_ERR(task))
1660 return;
1661
1662 WARN_ON_ONCE(atomic_read(&task->tk_count) != 1);
1663 rpc_put_task(task);
1664 }
1665 #endif /* CONFIG_SUNRPC_BACKCHANNEL */
1666
1667 /**
1668 * svc_max_payload - Return transport-specific limit on the RPC payload
1669 * @rqstp: RPC transaction context
1670 *
1671 * Returns the maximum number of payload bytes the current transport
1672 * allows.
1673 */
svc_max_payload(const struct svc_rqst * rqstp)1674 u32 svc_max_payload(const struct svc_rqst *rqstp)
1675 {
1676 u32 max = rqstp->rq_xprt->xpt_class->xcl_max_payload;
1677
1678 if (rqstp->rq_server->sv_max_payload < max)
1679 max = rqstp->rq_server->sv_max_payload;
1680 return max;
1681 }
1682 EXPORT_SYMBOL_GPL(svc_max_payload);
1683
1684 /**
1685 * svc_proc_name - Return RPC procedure name in string form
1686 * @rqstp: svc_rqst to operate on
1687 *
1688 * Return value:
1689 * Pointer to a NUL-terminated string
1690 */
svc_proc_name(const struct svc_rqst * rqstp)1691 const char *svc_proc_name(const struct svc_rqst *rqstp)
1692 {
1693 if (rqstp && rqstp->rq_procinfo)
1694 return rqstp->rq_procinfo->pc_name;
1695 return "unknown";
1696 }
1697
1698
1699 /**
1700 * svc_encode_result_payload - mark a range of bytes as a result payload
1701 * @rqstp: svc_rqst to operate on
1702 * @offset: payload's byte offset in rqstp->rq_res
1703 * @length: size of payload, in bytes
1704 *
1705 * Returns zero on success, or a negative errno if a permanent
1706 * error occurred.
1707 */
svc_encode_result_payload(struct svc_rqst * rqstp,unsigned int offset,unsigned int length)1708 int svc_encode_result_payload(struct svc_rqst *rqstp, unsigned int offset,
1709 unsigned int length)
1710 {
1711 return rqstp->rq_xprt->xpt_ops->xpo_result_payload(rqstp, offset,
1712 length);
1713 }
1714 EXPORT_SYMBOL_GPL(svc_encode_result_payload);
1715
1716 /**
1717 * svc_fill_write_vector - Construct data argument for VFS write call
1718 * @rqstp: svc_rqst to operate on
1719 * @payload: xdr_buf containing only the write data payload
1720 *
1721 * Fills in rqstp::rq_vec, and returns the number of elements.
1722 */
svc_fill_write_vector(struct svc_rqst * rqstp,struct xdr_buf * payload)1723 unsigned int svc_fill_write_vector(struct svc_rqst *rqstp,
1724 struct xdr_buf *payload)
1725 {
1726 struct page **pages = payload->pages;
1727 struct kvec *first = payload->head;
1728 struct kvec *vec = rqstp->rq_vec;
1729 size_t total = payload->len;
1730 unsigned int i;
1731
1732 /* Some types of transport can present the write payload
1733 * entirely in rq_arg.pages. In this case, @first is empty.
1734 */
1735 i = 0;
1736 if (first->iov_len) {
1737 vec[i].iov_base = first->iov_base;
1738 vec[i].iov_len = min_t(size_t, total, first->iov_len);
1739 total -= vec[i].iov_len;
1740 ++i;
1741 }
1742
1743 while (total) {
1744 vec[i].iov_base = page_address(*pages);
1745 vec[i].iov_len = min_t(size_t, total, PAGE_SIZE);
1746 total -= vec[i].iov_len;
1747 ++i;
1748 ++pages;
1749 }
1750
1751 WARN_ON_ONCE(i > ARRAY_SIZE(rqstp->rq_vec));
1752 return i;
1753 }
1754 EXPORT_SYMBOL_GPL(svc_fill_write_vector);
1755
1756 /**
1757 * svc_fill_symlink_pathname - Construct pathname argument for VFS symlink call
1758 * @rqstp: svc_rqst to operate on
1759 * @first: buffer containing first section of pathname
1760 * @p: buffer containing remaining section of pathname
1761 * @total: total length of the pathname argument
1762 *
1763 * The VFS symlink API demands a NUL-terminated pathname in mapped memory.
1764 * Returns pointer to a NUL-terminated string, or an ERR_PTR. Caller must free
1765 * the returned string.
1766 */
svc_fill_symlink_pathname(struct svc_rqst * rqstp,struct kvec * first,void * p,size_t total)1767 char *svc_fill_symlink_pathname(struct svc_rqst *rqstp, struct kvec *first,
1768 void *p, size_t total)
1769 {
1770 size_t len, remaining;
1771 char *result, *dst;
1772
1773 result = kmalloc(total + 1, GFP_KERNEL);
1774 if (!result)
1775 return ERR_PTR(-ESERVERFAULT);
1776
1777 dst = result;
1778 remaining = total;
1779
1780 len = min_t(size_t, total, first->iov_len);
1781 if (len) {
1782 memcpy(dst, first->iov_base, len);
1783 dst += len;
1784 remaining -= len;
1785 }
1786
1787 if (remaining) {
1788 len = min_t(size_t, remaining, PAGE_SIZE);
1789 memcpy(dst, p, len);
1790 dst += len;
1791 }
1792
1793 *dst = '\0';
1794
1795 /* Sanity check: Linux doesn't allow the pathname argument to
1796 * contain a NUL byte.
1797 */
1798 if (strlen(result) != total) {
1799 kfree(result);
1800 return ERR_PTR(-EINVAL);
1801 }
1802 return result;
1803 }
1804 EXPORT_SYMBOL_GPL(svc_fill_symlink_pathname);
1805