projects / deliverable / linux.git / blob
? search:


c72035e048aa9842d25af105ae0d6cab749a49ed
[deliverable/linux.git] / drivers / staging / lustre / lustre / osc / osc_cache.c
1 /*
2 * GPL HEADER START
3 *
4 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
5 *
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License version 2 only,
8 * as published by the Free Software Foundation.
9 *
10 * This program is distributed in the hope that it will be useful, but
11 * WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * General Public License version 2 for more details (a copy is included
14 * in the LICENSE file that accompanied this code).
15 *
16 * You should have received a copy of the GNU General Public License
17 * version 2 along with this program; If not, see
18 * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
19 *
20 * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
21 * CA 95054 USA or visit www.sun.com if you need additional information or
22 * have any questions.
23 *
24 * GPL HEADER END
25 */
26 /*
27 * Copyright (c) 2008, 2010, Oracle and/or its affiliates. All rights reserved.
28 * Use is subject to license terms.
29 *
30 * Copyright (c) 2012, Intel Corporation.
31 *
32 */
33 /*
34 * This file is part of Lustre, http://www.lustre.org/
35 * Lustre is a trademark of Sun Microsystems, Inc.
36 *
37 * osc cache management.
38 *
39 * Author: Jinshan Xiong <jinshan.xiong@whamcloud.com>
40 */
41
42 #define DEBUG_SUBSYSTEM S_OSC
43
44 #include "osc_cl_internal.h"
45 #include "osc_internal.h"
46
47 static int extent_debug; /* set it to be true for more debug */
48
49 static void osc_update_pending(struct osc_object *obj, int cmd, int delta);
50 static int osc_extent_wait(const struct lu_env *env, struct osc_extent *ext,
51 int state);
52 static void osc_ap_completion(const struct lu_env *env, struct client_obd *cli,
53 struct osc_async_page *oap, int sent, int rc);
54 static int osc_make_ready(const struct lu_env *env, struct osc_async_page *oap,
55 int cmd);
56 static int osc_refresh_count(const struct lu_env *env,
57 struct osc_async_page *oap, int cmd);
58 static int osc_io_unplug_async(const struct lu_env *env,
59 struct client_obd *cli, struct osc_object *osc);
60 static void osc_free_grant(struct client_obd *cli, unsigned int nr_pages,
61 unsigned int lost_grant);
62
63 static void osc_extent_tree_dump0(int level, struct osc_object *obj,
64 const char *func, int line);
65 #define osc_extent_tree_dump(lvl, obj) \
66 osc_extent_tree_dump0(lvl, obj, __func__, __LINE__)
67
68 /** \addtogroup osc
69 * @{
70 */
71
72 /* ------------------ osc extent ------------------ */
73 static inline char *ext_flags(struct osc_extent *ext, char *flags)
74 {
75 char *buf = flags;
76 *buf++ = ext->oe_rw ? 'r' : 'w';
77 if (ext->oe_intree)
78 *buf++ = 'i';
79 if (ext->oe_srvlock)
80 *buf++ = 's';
81 if (ext->oe_hp)
82 *buf++ = 'h';
83 if (ext->oe_urgent)
84 *buf++ = 'u';
85 if (ext->oe_memalloc)
86 *buf++ = 'm';
87 if (ext->oe_trunc_pending)
88 *buf++ = 't';
89 if (ext->oe_fsync_wait)
90 *buf++ = 'Y';
91 *buf = 0;
92 return flags;
93 }
94
95 static inline char list_empty_marker(struct list_head *list)
96 {
97 return list_empty(list) ? '-' : '+';
98 }
99
100 #define EXTSTR "[%lu -> %lu/%lu]"
101 #define EXTPARA(ext) (ext)->oe_start, (ext)->oe_end, (ext)->oe_max_end
102 static const char *oes_strings[] = {
103 "inv", "active", "cache", "locking", "lockdone", "rpc", "trunc", NULL };
104
105 #define OSC_EXTENT_DUMP(lvl, extent, fmt, ...) do { \
106 struct osc_extent *__ext = (extent); \
107 char __buf[16]; \
108 \
109 CDEBUG(lvl, \
110 "extent %p@{" EXTSTR ", " \
111 "[%d|%d|%c|%s|%s|%p], [%d|%d|%c|%c|%p|%u|%p]} " fmt, \
112 /* ----- extent part 0 ----- */ \
113 __ext, EXTPARA(__ext), \
114 /* ----- part 1 ----- */ \
115 atomic_read(&__ext->oe_refc), \
116 atomic_read(&__ext->oe_users), \
117 list_empty_marker(&__ext->oe_link), \
118 oes_strings[__ext->oe_state], ext_flags(__ext, __buf), \
119 __ext->oe_obj, \
120 /* ----- part 2 ----- */ \
121 __ext->oe_grants, __ext->oe_nr_pages, \
122 list_empty_marker(&__ext->oe_pages), \
123 waitqueue_active(&__ext->oe_waitq) ? '+' : '-', \
124 __ext->oe_osclock, __ext->oe_mppr, __ext->oe_owner, \
125 /* ----- part 4 ----- */ \
126 ## __VA_ARGS__); \
127 } while (0)
128
129 #undef EASSERTF
130 #define EASSERTF(expr, ext, fmt, args...) do { \
131 if (!(expr)) { \
132 OSC_EXTENT_DUMP(D_ERROR, (ext), fmt, ##args); \
133 osc_extent_tree_dump(D_ERROR, (ext)->oe_obj); \
134 LASSERT(expr); \
135 } \
136 } while (0)
137
138 #undef EASSERT
139 #define EASSERT(expr, ext) EASSERTF(expr, ext, "\n")
140
141 static inline struct osc_extent *rb_extent(struct rb_node *n)
142 {
143 if (n == NULL)
144 return NULL;
145
146 return container_of(n, struct osc_extent, oe_node);
147 }
148
149 static inline struct osc_extent *next_extent(struct osc_extent *ext)
150 {
151 if (ext == NULL)
152 return NULL;
153
154 LASSERT(ext->oe_intree);
155 return rb_extent(rb_next(&ext->oe_node));
156 }
157
158 static inline struct osc_extent *prev_extent(struct osc_extent *ext)
159 {
160 if (ext == NULL)
161 return NULL;
162
163 LASSERT(ext->oe_intree);
164 return rb_extent(rb_prev(&ext->oe_node));
165 }
166
167 static inline struct osc_extent *first_extent(struct osc_object *obj)
168 {
169 return rb_extent(rb_first(&obj->oo_root));
170 }
171
172 /* object must be locked by caller. */
173 static int osc_extent_sanity_check0(struct osc_extent *ext,
174 const char *func, const int line)
175 {
176 struct osc_object *obj = ext->oe_obj;
177 struct osc_async_page *oap;
178 int page_count;
179 int rc = 0;
180
181 if (!osc_object_is_locked(obj)) {
182 rc = 9;
183 goto out;
184 }
185
186 if (ext->oe_state >= OES_STATE_MAX) {
187 rc = 10;
188 goto out;
189 }
190
191 if (atomic_read(&ext->oe_refc) <= 0) {
192 rc = 20;
193 goto out;
194 }
195
196 if (atomic_read(&ext->oe_refc) < atomic_read(&ext->oe_users)) {
197 rc = 30;
198 goto out;
199 }
200
201 switch (ext->oe_state) {
202 case OES_INV:
203 if (ext->oe_nr_pages > 0 || !list_empty(&ext->oe_pages))
204 rc = 35;
205 else
206 rc = 0;
207 goto out;
208 case OES_ACTIVE:
209 if (atomic_read(&ext->oe_users) == 0) {
210 rc = 40;
211 goto out;
212 }
213 if (ext->oe_hp) {
214 rc = 50;
215 goto out;
216 }
217 if (ext->oe_fsync_wait && !ext->oe_urgent) {
218 rc = 55;
219 goto out;
220 }
221 break;
222 case OES_CACHE:
223 if (ext->oe_grants == 0) {
224 rc = 60;
225 goto out;
226 }
227 if (ext->oe_fsync_wait && !ext->oe_urgent && !ext->oe_hp) {
228 rc = 65;
229 goto out;
230 }
231 default:
232 if (atomic_read(&ext->oe_users) > 0) {
233 rc = 70;
234 goto out;
235 }
236 }
237
238 if (ext->oe_max_end < ext->oe_end || ext->oe_end < ext->oe_start) {
239 rc = 80;
240 goto out;
241 }
242
243 if (ext->oe_osclock == NULL && ext->oe_grants > 0) {
244 rc = 90;
245 goto out;
246 }
247
248 if (ext->oe_osclock) {
249 struct cl_lock_descr *descr;
250 descr = &ext->oe_osclock->cll_descr;
251 if (!(descr->cld_start <= ext->oe_start &&
252 descr->cld_end >= ext->oe_max_end)) {
253 rc = 100;
254 goto out;
255 }
256 }
257
258 if (ext->oe_nr_pages > ext->oe_mppr) {
259 rc = 105;
260 goto out;
261 }
262
263 /* Do not verify page list if extent is in RPC. This is because an
264 * in-RPC extent is supposed to be exclusively accessible w/o lock. */
265 if (ext->oe_state > OES_CACHE) {
266 rc = 0;
267 goto out;
268 }
269
270 if (!extent_debug) {
271 rc = 0;
272 goto out;
273 }
274
275 page_count = 0;
276 list_for_each_entry(oap, &ext->oe_pages, oap_pending_item) {
277 pgoff_t index = oap2cl_page(oap)->cp_index;
278 ++page_count;
279 if (index > ext->oe_end || index < ext->oe_start) {
280 rc = 110;
281 goto out;
282 }
283 }
284 if (page_count != ext->oe_nr_pages) {
285 rc = 120;
286 goto out;
287 }
288
289 out:
290 if (rc != 0)
291 OSC_EXTENT_DUMP(D_ERROR, ext,
292 "%s:%d sanity check %p failed with rc = %d\n",
293 func, line, ext, rc);
294 return rc;
295 }
296
297 #define sanity_check_nolock(ext) \
298 osc_extent_sanity_check0(ext, __func__, __LINE__)
299
300 #define sanity_check(ext) ({ \
301 int __res; \
302 osc_object_lock((ext)->oe_obj); \
303 __res = sanity_check_nolock(ext); \
304 osc_object_unlock((ext)->oe_obj); \
305 __res; \
306 })
307
308
309 /**
310 * sanity check - to make sure there is no overlapped extent in the tree.
311 */
312 static int osc_extent_is_overlapped(struct osc_object *obj,
313 struct osc_extent *ext)
314 {
315 struct osc_extent *tmp;
316
317 LASSERT(osc_object_is_locked(obj));
318
319 if (!extent_debug)
320 return 0;
321
322 for (tmp = first_extent(obj); tmp != NULL; tmp = next_extent(tmp)) {
323 if (tmp == ext)
324 continue;
325 if (tmp->oe_end >= ext->oe_start &&
326 tmp->oe_start <= ext->oe_end)
327 return 1;
328 }
329 return 0;
330 }
331
332 static void osc_extent_state_set(struct osc_extent *ext, int state)
333 {
334 LASSERT(osc_object_is_locked(ext->oe_obj));
335 LASSERT(state >= OES_INV && state < OES_STATE_MAX);
336
337 /* Never try to sanity check a state changing extent :-) */
338 /* LASSERT(sanity_check_nolock(ext) == 0); */
339
340 /* TODO: validate the state machine */
341 ext->oe_state = state;
342 wake_up_all(&ext->oe_waitq);
343 }
344
345 static struct osc_extent *osc_extent_alloc(struct osc_object *obj)
346 {
347 struct osc_extent *ext;
348
349 OBD_SLAB_ALLOC_PTR_GFP(ext, osc_extent_kmem, GFP_IOFS);
350 if (ext == NULL)
351 return NULL;
352
353 RB_CLEAR_NODE(&ext->oe_node);
354 ext->oe_obj = obj;
355 atomic_set(&ext->oe_refc, 1);
356 atomic_set(&ext->oe_users, 0);
357 INIT_LIST_HEAD(&ext->oe_link);
358 ext->oe_state = OES_INV;
359 INIT_LIST_HEAD(&ext->oe_pages);
360 init_waitqueue_head(&ext->oe_waitq);
361 ext->oe_osclock = NULL;
362
363 return ext;
364 }
365
366 static void osc_extent_free(struct osc_extent *ext)
367 {
368 OBD_SLAB_FREE_PTR(ext, osc_extent_kmem);
369 }
370
371 static struct osc_extent *osc_extent_get(struct osc_extent *ext)
372 {
373 LASSERT(atomic_read(&ext->oe_refc) >= 0);
374 atomic_inc(&ext->oe_refc);
375 return ext;
376 }
377
378 static void osc_extent_put(const struct lu_env *env, struct osc_extent *ext)
379 {
380 LASSERT(atomic_read(&ext->oe_refc) > 0);
381 if (atomic_dec_and_test(&ext->oe_refc)) {
382 LASSERT(list_empty(&ext->oe_link));
383 LASSERT(atomic_read(&ext->oe_users) == 0);
384 LASSERT(ext->oe_state == OES_INV);
385 LASSERT(!ext->oe_intree);
386
387 if (ext->oe_osclock) {
388 cl_lock_put(env, ext->oe_osclock);
389 ext->oe_osclock = NULL;
390 }
391 osc_extent_free(ext);
392 }
393 }
394
395 /**
396 * osc_extent_put_trust() is a special version of osc_extent_put() when
397 * it's known that the caller is not the last user. This is to address the
398 * problem of lacking of lu_env ;-).
399 */
400 static void osc_extent_put_trust(struct osc_extent *ext)
401 {
402 LASSERT(atomic_read(&ext->oe_refc) > 1);
403 LASSERT(osc_object_is_locked(ext->oe_obj));
404 atomic_dec(&ext->oe_refc);
405 }
406
407 /**
408 * Return the extent which includes pgoff @index, or return the greatest
409 * previous extent in the tree.
410 */
411 static struct osc_extent *osc_extent_search(struct osc_object *obj,
412 pgoff_t index)
413 {
414 struct rb_node *n = obj->oo_root.rb_node;
415 struct osc_extent *tmp, *p = NULL;
416
417 LASSERT(osc_object_is_locked(obj));
418 while (n != NULL) {
419 tmp = rb_extent(n);
420 if (index < tmp->oe_start) {
421 n = n->rb_left;
422 } else if (index > tmp->oe_end) {
423 p = rb_extent(n);
424 n = n->rb_right;
425 } else {
426 return tmp;
427 }
428 }
429 return p;
430 }
431
432 /*
433 * Return the extent covering @index, otherwise return NULL.
434 * caller must have held object lock.
435 */
436 static struct osc_extent *osc_extent_lookup(struct osc_object *obj,
437 pgoff_t index)
438 {
439 struct osc_extent *ext;
440
441 ext = osc_extent_search(obj, index);
442 if (ext != NULL && ext->oe_start <= index && index <= ext->oe_end)
443 return osc_extent_get(ext);
444 return NULL;
445 }
446
447 /* caller must have held object lock. */
448 static void osc_extent_insert(struct osc_object *obj, struct osc_extent *ext)
449 {
450 struct rb_node **n = &obj->oo_root.rb_node;
451 struct rb_node *parent = NULL;
452 struct osc_extent *tmp;
453
454 LASSERT(ext->oe_intree == 0);
455 LASSERT(ext->oe_obj == obj);
456 LASSERT(osc_object_is_locked(obj));
457 while (*n != NULL) {
458 tmp = rb_extent(*n);
459 parent = *n;
460
461 if (ext->oe_end < tmp->oe_start)
462 n = &(*n)->rb_left;
463 else if (ext->oe_start > tmp->oe_end)
464 n = &(*n)->rb_right;
465 else
466 EASSERTF(0, tmp, EXTSTR, EXTPARA(ext));
467 }
468 rb_link_node(&ext->oe_node, parent, n);
469 rb_insert_color(&ext->oe_node, &obj->oo_root);
470 osc_extent_get(ext);
471 ext->oe_intree = 1;
472 }
473
474 /* caller must have held object lock. */
475 static void osc_extent_erase(struct osc_extent *ext)
476 {
477 struct osc_object *obj = ext->oe_obj;
478 LASSERT(osc_object_is_locked(obj));
479 if (ext->oe_intree) {
480 rb_erase(&ext->oe_node, &obj->oo_root);
481 ext->oe_intree = 0;
482 /* rbtree held a refcount */
483 osc_extent_put_trust(ext);
484 }
485 }
486
487 static struct osc_extent *osc_extent_hold(struct osc_extent *ext)
488 {
489 struct osc_object *obj = ext->oe_obj;
490
491 LASSERT(osc_object_is_locked(obj));
492 LASSERT(ext->oe_state == OES_ACTIVE || ext->oe_state == OES_CACHE);
493 if (ext->oe_state == OES_CACHE) {
494 osc_extent_state_set(ext, OES_ACTIVE);
495 osc_update_pending(obj, OBD_BRW_WRITE, -ext->oe_nr_pages);
496 }
497 atomic_inc(&ext->oe_users);
498 list_del_init(&ext->oe_link);
499 return osc_extent_get(ext);
500 }
501
502 static void __osc_extent_remove(struct osc_extent *ext)
503 {
504 LASSERT(osc_object_is_locked(ext->oe_obj));
505 LASSERT(list_empty(&ext->oe_pages));
506 osc_extent_erase(ext);
507 list_del_init(&ext->oe_link);
508 osc_extent_state_set(ext, OES_INV);
509 OSC_EXTENT_DUMP(D_CACHE, ext, "destroyed.\n");
510 }
511
512 static void osc_extent_remove(struct osc_extent *ext)
513 {
514 struct osc_object *obj = ext->oe_obj;
515
516 osc_object_lock(obj);
517 __osc_extent_remove(ext);
518 osc_object_unlock(obj);
519 }
520
521 /**
522 * This function is used to merge extents to get better performance. It checks
523 * if @cur and @victim are contiguous at chunk level.
524 */
525 static int osc_extent_merge(const struct lu_env *env, struct osc_extent *cur,
526 struct osc_extent *victim)
527 {
528 struct osc_object *obj = cur->oe_obj;
529 pgoff_t chunk_start;
530 pgoff_t chunk_end;
531 int ppc_bits;
532
533 LASSERT(cur->oe_state == OES_CACHE);
534 LASSERT(osc_object_is_locked(obj));
535 if (victim == NULL)
536 return -EINVAL;
537
538 if (victim->oe_state != OES_CACHE || victim->oe_fsync_wait)
539 return -EBUSY;
540
541 if (cur->oe_max_end != victim->oe_max_end)
542 return -ERANGE;
543
544 LASSERT(cur->oe_osclock == victim->oe_osclock);
545 ppc_bits = osc_cli(obj)->cl_chunkbits - PAGE_CACHE_SHIFT;
546 chunk_start = cur->oe_start >> ppc_bits;
547 chunk_end = cur->oe_end >> ppc_bits;
548 if (chunk_start != (victim->oe_end >> ppc_bits) + 1 &&
549 chunk_end + 1 != victim->oe_start >> ppc_bits)
550 return -ERANGE;
551
552 OSC_EXTENT_DUMP(D_CACHE, victim, "will be merged by %p.\n", cur);
553
554 cur->oe_start = min(cur->oe_start, victim->oe_start);
555 cur->oe_end = max(cur->oe_end, victim->oe_end);
556 cur->oe_grants += victim->oe_grants;
557 cur->oe_nr_pages += victim->oe_nr_pages;
558 /* only the following bits are needed to merge */
559 cur->oe_urgent |= victim->oe_urgent;
560 cur->oe_memalloc |= victim->oe_memalloc;
561 list_splice_init(&victim->oe_pages, &cur->oe_pages);
562 list_del_init(&victim->oe_link);
563 victim->oe_nr_pages = 0;
564
565 osc_extent_get(victim);
566 __osc_extent_remove(victim);
567 osc_extent_put(env, victim);
568
569 OSC_EXTENT_DUMP(D_CACHE, cur, "after merging %p.\n", victim);
570 return 0;
571 }
572
573 /**
574 * Drop user count of osc_extent, and unplug IO asynchronously.
575 */
576 void osc_extent_release(const struct lu_env *env, struct osc_extent *ext)
577 {
578 struct osc_object *obj = ext->oe_obj;
579
580 LASSERT(atomic_read(&ext->oe_users) > 0);
581 LASSERT(sanity_check(ext) == 0);
582 LASSERT(ext->oe_grants > 0);
583
584 if (atomic_dec_and_lock(&ext->oe_users, &obj->oo_lock)) {
585 LASSERT(ext->oe_state == OES_ACTIVE);
586 if (ext->oe_trunc_pending) {
587 /* a truncate process is waiting for this extent.
588 * This may happen due to a race, check
589 * osc_cache_truncate_start(). */
590 osc_extent_state_set(ext, OES_TRUNC);
591 ext->oe_trunc_pending = 0;
592 } else {
593 osc_extent_state_set(ext, OES_CACHE);
594 osc_update_pending(obj, OBD_BRW_WRITE,
595 ext->oe_nr_pages);
596
597 /* try to merge the previous and next extent. */
598 osc_extent_merge(env, ext, prev_extent(ext));
599 osc_extent_merge(env, ext, next_extent(ext));
600
601 if (ext->oe_urgent)
602 list_move_tail(&ext->oe_link,
603 &obj->oo_urgent_exts);
604 }
605 osc_object_unlock(obj);
606
607 osc_io_unplug_async(env, osc_cli(obj), obj);
608 }
609 osc_extent_put(env, ext);
610 }
611
612 static inline int overlapped(struct osc_extent *ex1, struct osc_extent *ex2)
613 {
614 return !(ex1->oe_end < ex2->oe_start || ex2->oe_end < ex1->oe_start);
615 }
616
617 /**
618 * Find or create an extent which includes @index, core function to manage
619 * extent tree.
620 */
621 struct osc_extent *osc_extent_find(const struct lu_env *env,
622 struct osc_object *obj, pgoff_t index,
623 int *grants)
624
625 {
626 struct client_obd *cli = osc_cli(obj);
627 struct cl_lock *lock;
628 struct osc_extent *cur;
629 struct osc_extent *ext;
630 struct osc_extent *conflict = NULL;
631 struct osc_extent *found = NULL;
632 pgoff_t chunk;
633 pgoff_t max_end;
634 int max_pages; /* max_pages_per_rpc */
635 int chunksize;
636 int ppc_bits; /* pages per chunk bits */
637 int chunk_mask;
638 int rc;
639
640 cur = osc_extent_alloc(obj);
641 if (cur == NULL)
642 return ERR_PTR(-ENOMEM);
643
644 lock = cl_lock_at_pgoff(env, osc2cl(obj), index, NULL, 1, 0);
645 LASSERT(lock != NULL);
646 LASSERT(lock->cll_descr.cld_mode >= CLM_WRITE);
647
648 LASSERT(cli->cl_chunkbits >= PAGE_CACHE_SHIFT);
649 ppc_bits = cli->cl_chunkbits - PAGE_CACHE_SHIFT;
650 chunk_mask = ~((1 << ppc_bits) - 1);
651 chunksize = 1 << cli->cl_chunkbits;
652 chunk = index >> ppc_bits;
653
654 /* align end to rpc edge, rpc size may not be a power 2 integer. */
655 max_pages = cli->cl_max_pages_per_rpc;
656 LASSERT((max_pages & ~chunk_mask) == 0);
657 max_end = index - (index % max_pages) + max_pages - 1;
658 max_end = min_t(pgoff_t, max_end, lock->cll_descr.cld_end);
659
660 /* initialize new extent by parameters so far */
661 cur->oe_max_end = max_end;
662 cur->oe_start = index & chunk_mask;
663 cur->oe_end = ((index + ~chunk_mask + 1) & chunk_mask) - 1;
664 if (cur->oe_start < lock->cll_descr.cld_start)
665 cur->oe_start = lock->cll_descr.cld_start;
666 if (cur->oe_end > max_end)
667 cur->oe_end = max_end;
668 cur->oe_osclock = lock;
669 cur->oe_grants = 0;
670 cur->oe_mppr = max_pages;
671
672 /* grants has been allocated by caller */
673 LASSERTF(*grants >= chunksize + cli->cl_extent_tax,
674 "%u/%u/%u.\n", *grants, chunksize, cli->cl_extent_tax);
675 LASSERTF((max_end - cur->oe_start) < max_pages, EXTSTR, EXTPARA(cur));
676
677 restart:
678 osc_object_lock(obj);
679 ext = osc_extent_search(obj, cur->oe_start);
680 if (ext == NULL)
681 ext = first_extent(obj);
682 while (ext != NULL) {
683 loff_t ext_chk_start = ext->oe_start >> ppc_bits;
684 loff_t ext_chk_end = ext->oe_end >> ppc_bits;
685
686 LASSERT(sanity_check_nolock(ext) == 0);
687 if (chunk > ext_chk_end + 1)
688 break;
689
690 /* if covering by different locks, no chance to match */
691 if (lock != ext->oe_osclock) {
692 EASSERTF(!overlapped(ext, cur), ext,
693 EXTSTR, EXTPARA(cur));
694
695 ext = next_extent(ext);
696 continue;
697 }
698
699 /* discontiguous chunks? */
700 if (chunk + 1 < ext_chk_start) {
701 ext = next_extent(ext);
702 continue;
703 }
704
705 /* ok, from now on, ext and cur have these attrs:
706 * 1. covered by the same lock
707 * 2. contiguous at chunk level or overlapping. */
708
709 if (overlapped(ext, cur)) {
710 /* cur is the minimum unit, so overlapping means
711 * full contain. */
712 EASSERTF((ext->oe_start <= cur->oe_start &&
713 ext->oe_end >= cur->oe_end),
714 ext, EXTSTR, EXTPARA(cur));
715
716 if (ext->oe_state > OES_CACHE || ext->oe_fsync_wait) {
717 /* for simplicity, we wait for this extent to
718 * finish before going forward. */
719 conflict = osc_extent_get(ext);
720 break;
721 }
722
723 found = osc_extent_hold(ext);
724 break;
725 }
726
727 /* non-overlapped extent */
728 if (ext->oe_state != OES_CACHE || ext->oe_fsync_wait) {
729 /* we can't do anything for a non OES_CACHE extent, or
730 * if there is someone waiting for this extent to be
731 * flushed, try next one. */
732 ext = next_extent(ext);
733 continue;
734 }
735
736 /* check if they belong to the same rpc slot before trying to
737 * merge. the extents are not overlapped and contiguous at
738 * chunk level to get here. */
739 if (ext->oe_max_end != max_end) {
740 /* if they don't belong to the same RPC slot or
741 * max_pages_per_rpc has ever changed, do not merge. */
742 ext = next_extent(ext);
743 continue;
744 }
745
746 /* it's required that an extent must be contiguous at chunk
747 * level so that we know the whole extent is covered by grant
748 * (the pages in the extent are NOT required to be contiguous).
749 * Otherwise, it will be too much difficult to know which
750 * chunks have grants allocated. */
751
752 /* try to do front merge - extend ext's start */
753 if (chunk + 1 == ext_chk_start) {
754 /* ext must be chunk size aligned */
755 EASSERT((ext->oe_start & ~chunk_mask) == 0, ext);
756
757 /* pull ext's start back to cover cur */
758 ext->oe_start = cur->oe_start;
759 ext->oe_grants += chunksize;
760 *grants -= chunksize;
761
762 found = osc_extent_hold(ext);
763 } else if (chunk == ext_chk_end + 1) {
764 /* rear merge */
765 ext->oe_end = cur->oe_end;
766 ext->oe_grants += chunksize;
767 *grants -= chunksize;
768
769 /* try to merge with the next one because we just fill
770 * in a gap */
771 if (osc_extent_merge(env, ext, next_extent(ext)) == 0)
772 /* we can save extent tax from next extent */
773 *grants += cli->cl_extent_tax;
774
775 found = osc_extent_hold(ext);
776 }
777 if (found != NULL)
778 break;
779
780 ext = next_extent(ext);
781 }
782
783 osc_extent_tree_dump(D_CACHE, obj);
784 if (found != NULL) {
785 LASSERT(conflict == NULL);
786 if (!IS_ERR(found)) {
787 LASSERT(found->oe_osclock == cur->oe_osclock);
788 OSC_EXTENT_DUMP(D_CACHE, found,
789 "found caching ext for %lu.\n", index);
790 }
791 } else if (conflict == NULL) {
792 /* create a new extent */
793 EASSERT(osc_extent_is_overlapped(obj, cur) == 0, cur);
794 cur->oe_grants = chunksize + cli->cl_extent_tax;
795 *grants -= cur->oe_grants;
796 LASSERT(*grants >= 0);
797
798 cur->oe_state = OES_CACHE;
799 found = osc_extent_hold(cur);
800 osc_extent_insert(obj, cur);
801 OSC_EXTENT_DUMP(D_CACHE, cur, "add into tree %lu/%lu.\n",
802 index, lock->cll_descr.cld_end);
803 }
804 osc_object_unlock(obj);
805
806 if (conflict != NULL) {
807 LASSERT(found == NULL);
808
809 /* waiting for IO to finish. Please notice that it's impossible
810 * to be an OES_TRUNC extent. */
811 rc = osc_extent_wait(env, conflict, OES_INV);
812 osc_extent_put(env, conflict);
813 conflict = NULL;
814 if (rc < 0) {
815 found = ERR_PTR(rc);
816 goto out;
817 }
818
819 goto restart;
820 }
821
822 out:
823 osc_extent_put(env, cur);
824 LASSERT(*grants >= 0);
825 return found;
826 }
827
828 /**
829 * Called when IO is finished to an extent.
830 */
831 int osc_extent_finish(const struct lu_env *env, struct osc_extent *ext,
832 int sent, int rc)
833 {
834 struct client_obd *cli = osc_cli(ext->oe_obj);
835 struct osc_async_page *oap;
836 struct osc_async_page *tmp;
837 int nr_pages = ext->oe_nr_pages;
838 int lost_grant = 0;
839 int blocksize = cli->cl_import->imp_obd->obd_osfs.os_bsize ? : 4096;
840 __u64 last_off = 0;
841 int last_count = -1;
842
843 OSC_EXTENT_DUMP(D_CACHE, ext, "extent finished.\n");
844
845 ext->oe_rc = rc ?: ext->oe_nr_pages;
846 EASSERT(ergo(rc == 0, ext->oe_state == OES_RPC), ext);
847 list_for_each_entry_safe(oap, tmp, &ext->oe_pages,
848 oap_pending_item) {
849 list_del_init(&oap->oap_rpc_item);
850 list_del_init(&oap->oap_pending_item);
851 if (last_off <= oap->oap_obj_off) {
852 last_off = oap->oap_obj_off;
853 last_count = oap->oap_count;
854 }
855
856 --ext->oe_nr_pages;
857 osc_ap_completion(env, cli, oap, sent, rc);
858 }
859 EASSERT(ext->oe_nr_pages == 0, ext);
860
861 if (!sent) {
862 lost_grant = ext->oe_grants;
863 } else if (blocksize < PAGE_CACHE_SIZE &&
864 last_count != PAGE_CACHE_SIZE) {
865 /* For short writes we shouldn't count parts of pages that
866 * span a whole chunk on the OST side, or our accounting goes
867 * wrong. Should match the code in filter_grant_check. */
868 int offset = oap->oap_page_off & ~CFS_PAGE_MASK;
869 int count = oap->oap_count + (offset & (blocksize - 1));
870 int end = (offset + oap->oap_count) & (blocksize - 1);
871 if (end)
872 count += blocksize - end;
873
874 lost_grant = PAGE_CACHE_SIZE - count;
875 }
876 if (ext->oe_grants > 0)
877 osc_free_grant(cli, nr_pages, lost_grant);
878
879 osc_extent_remove(ext);
880 /* put the refcount for RPC */
881 osc_extent_put(env, ext);
882 return 0;
883 }
884
885 static int extent_wait_cb(struct osc_extent *ext, int state)
886 {
887 int ret;
888
889 osc_object_lock(ext->oe_obj);
890 ret = ext->oe_state == state;
891 osc_object_unlock(ext->oe_obj);
892
893 return ret;
894 }
895
896 /**
897 * Wait for the extent's state to become @state.
898 */
899 static int osc_extent_wait(const struct lu_env *env, struct osc_extent *ext,
900 int state)
901 {
902 struct osc_object *obj = ext->oe_obj;
903 struct l_wait_info lwi = LWI_TIMEOUT_INTR(cfs_time_seconds(600), NULL,
904 LWI_ON_SIGNAL_NOOP, NULL);
905 int rc = 0;
906
907 osc_object_lock(obj);
908 LASSERT(sanity_check_nolock(ext) == 0);
909 /* `Kick' this extent only if the caller is waiting for it to be
910 * written out. */
911 if (state == OES_INV && !ext->oe_urgent && !ext->oe_hp &&
912 !ext->oe_trunc_pending) {
913 if (ext->oe_state == OES_ACTIVE) {
914 ext->oe_urgent = 1;
915 } else if (ext->oe_state == OES_CACHE) {
916 ext->oe_urgent = 1;
917 osc_extent_hold(ext);
918 rc = 1;
919 }
920 }
921 osc_object_unlock(obj);
922 if (rc == 1)
923 osc_extent_release(env, ext);
924
925 /* wait for the extent until its state becomes @state */
926 rc = l_wait_event(ext->oe_waitq, extent_wait_cb(ext, state), &lwi);
927 if (rc == -ETIMEDOUT) {
928 OSC_EXTENT_DUMP(D_ERROR, ext,
929 "%s: wait ext to %d timedout, recovery in progress?\n",
930 osc_export(obj)->exp_obd->obd_name, state);
931
932 lwi = LWI_INTR(LWI_ON_SIGNAL_NOOP, NULL);
933 rc = l_wait_event(ext->oe_waitq, extent_wait_cb(ext, state),
934 &lwi);
935 }
936 if (rc == 0 && ext->oe_rc < 0)
937 rc = ext->oe_rc;
938 return rc;
939 }
940
941 /**
942 * Discard pages with index greater than @size. If @ext is overlapped with
943 * @size, then partial truncate happens.
944 */
945 static int osc_extent_truncate(struct osc_extent *ext, pgoff_t trunc_index,
946 bool partial)
947 {
948 struct cl_env_nest nest;
949 struct lu_env *env;
950 struct cl_io *io;
951 struct osc_object *obj = ext->oe_obj;
952 struct client_obd *cli = osc_cli(obj);
953 struct osc_async_page *oap;
954 struct osc_async_page *tmp;
955 int pages_in_chunk = 0;
956 int ppc_bits = cli->cl_chunkbits - PAGE_CACHE_SHIFT;
957 __u64 trunc_chunk = trunc_index >> ppc_bits;
958 int grants = 0;
959 int nr_pages = 0;
960 int rc = 0;
961
962 LASSERT(sanity_check(ext) == 0);
963 EASSERT(ext->oe_state == OES_TRUNC, ext);
964 EASSERT(!ext->oe_urgent, ext);
965
966 /* Request new lu_env.
967 * We can't use that env from osc_cache_truncate_start() because
968 * it's from lov_io_sub and not fully initialized. */
969 env = cl_env_nested_get(&nest);
970 io = &osc_env_info(env)->oti_io;
971 io->ci_obj = cl_object_top(osc2cl(obj));
972 rc = cl_io_init(env, io, CIT_MISC, io->ci_obj);
973 if (rc < 0)
974 goto out;
975
976 /* discard all pages with index greater then trunc_index */
977 list_for_each_entry_safe(oap, tmp, &ext->oe_pages,
978 oap_pending_item) {
979 struct cl_page *sub = oap2cl_page(oap);
980 struct cl_page *page = cl_page_top(sub);
981
982 LASSERT(list_empty(&oap->oap_rpc_item));
983
984 /* only discard the pages with their index greater than
985 * trunc_index, and ... */
986 if (sub->cp_index < trunc_index ||
987 (sub->cp_index == trunc_index && partial)) {
988 /* accounting how many pages remaining in the chunk
989 * so that we can calculate grants correctly. */
990 if (sub->cp_index >> ppc_bits == trunc_chunk)
991 ++pages_in_chunk;
992 continue;
993 }
994
995 list_del_init(&oap->oap_pending_item);
996
997 cl_page_get(page);
998 lu_ref_add(&page->cp_reference, "truncate", current);
999
1000 if (cl_page_own(env, io, page) == 0) {
1001 cl_page_unmap(env, io, page);
1002 cl_page_discard(env, io, page);
1003 cl_page_disown(env, io, page);
1004 } else {
1005 LASSERT(page->cp_state == CPS_FREEING);
1006 LASSERT(0);
1007 }
1008
1009 lu_ref_del(&page->cp_reference, "truncate", current);
1010 cl_page_put(env, page);
1011
1012 --ext->oe_nr_pages;
1013 ++nr_pages;
1014 }
1015 EASSERTF(ergo(ext->oe_start >= trunc_index + !!partial,
1016 ext->oe_nr_pages == 0),
1017 ext, "trunc_index %lu, partial %d\n", trunc_index, partial);
1018
1019 osc_object_lock(obj);
1020 if (ext->oe_nr_pages == 0) {
1021 LASSERT(pages_in_chunk == 0);
1022 grants = ext->oe_grants;
1023 ext->oe_grants = 0;
1024 } else { /* calculate how many grants we can free */
1025 int chunks = (ext->oe_end >> ppc_bits) - trunc_chunk;
1026 pgoff_t last_index;
1027
1028
1029 /* if there is no pages in this chunk, we can also free grants
1030 * for the last chunk */
1031 if (pages_in_chunk == 0) {
1032 /* if this is the 1st chunk and no pages in this chunk,
1033 * ext->oe_nr_pages must be zero, so we should be in
1034 * the other if-clause. */
1035 LASSERT(trunc_chunk > 0);
1036 --trunc_chunk;
1037 ++chunks;
1038 }
1039
1040 /* this is what we can free from this extent */
1041 grants = chunks << cli->cl_chunkbits;
1042 ext->oe_grants -= grants;
1043 last_index = ((trunc_chunk + 1) << ppc_bits) - 1;
1044 ext->oe_end = min(last_index, ext->oe_max_end);
1045 LASSERT(ext->oe_end >= ext->oe_start);
1046 LASSERT(ext->oe_grants > 0);
1047 }
1048 osc_object_unlock(obj);
1049
1050 if (grants > 0 || nr_pages > 0)
1051 osc_free_grant(cli, nr_pages, grants);
1052
1053 out:
1054 cl_io_fini(env, io);
1055 cl_env_nested_put(&nest, env);
1056 return rc;
1057 }
1058
1059 /**
1060 * This function is used to make the extent prepared for transfer.
1061 * A race with flushing page - ll_writepage() has to be handled cautiously.
1062 */
1063 static int osc_extent_make_ready(const struct lu_env *env,
1064 struct osc_extent *ext)
1065 {
1066 struct osc_async_page *oap;
1067 struct osc_async_page *last = NULL;
1068 struct osc_object *obj = ext->oe_obj;
1069 int page_count = 0;
1070 int rc;
1071
1072 /* we're going to grab page lock, so object lock must not be taken. */
1073 LASSERT(sanity_check(ext) == 0);
1074 /* in locking state, any process should not touch this extent. */
1075 EASSERT(ext->oe_state == OES_LOCKING, ext);
1076 EASSERT(ext->oe_owner != NULL, ext);
1077
1078 OSC_EXTENT_DUMP(D_CACHE, ext, "make ready\n");
1079
1080 list_for_each_entry(oap, &ext->oe_pages, oap_pending_item) {
1081 ++page_count;
1082 if (last == NULL || last->oap_obj_off < oap->oap_obj_off)
1083 last = oap;
1084
1085 /* checking ASYNC_READY is race safe */
1086 if ((oap->oap_async_flags & ASYNC_READY) != 0)
1087 continue;
1088
1089 rc = osc_make_ready(env, oap, OBD_BRW_WRITE);
1090 switch (rc) {
1091 case 0:
1092 spin_lock(&oap->oap_lock);
1093 oap->oap_async_flags |= ASYNC_READY;
1094 spin_unlock(&oap->oap_lock);
1095 break;
1096 case -EALREADY:
1097 LASSERT((oap->oap_async_flags & ASYNC_READY) != 0);
1098 break;
1099 default:
1100 LASSERTF(0, "unknown return code: %d\n", rc);
1101 }
1102 }
1103
1104 LASSERT(page_count == ext->oe_nr_pages);
1105 LASSERT(last != NULL);
1106 /* the last page is the only one we need to refresh its count by
1107 * the size of file. */
1108 if (!(last->oap_async_flags & ASYNC_COUNT_STABLE)) {
1109 last->oap_count = osc_refresh_count(env, last, OBD_BRW_WRITE);
1110 LASSERT(last->oap_count > 0);
1111 LASSERT(last->oap_page_off + last->oap_count <= PAGE_CACHE_SIZE);
1112 last->oap_async_flags |= ASYNC_COUNT_STABLE;
1113 }
1114
1115 /* for the rest of pages, we don't need to call osf_refresh_count()
1116 * because it's known they are not the last page */
1117 list_for_each_entry(oap, &ext->oe_pages, oap_pending_item) {
1118 if (!(oap->oap_async_flags & ASYNC_COUNT_STABLE)) {
1119 oap->oap_count = PAGE_CACHE_SIZE - oap->oap_page_off;
1120 oap->oap_async_flags |= ASYNC_COUNT_STABLE;
1121 }
1122 }
1123
1124 osc_object_lock(obj);
1125 osc_extent_state_set(ext, OES_RPC);
1126 osc_object_unlock(obj);
1127 /* get a refcount for RPC. */
1128 osc_extent_get(ext);
1129
1130 return 0;
1131 }
1132
1133 /**
1134 * Quick and simple version of osc_extent_find(). This function is frequently
1135 * called to expand the extent for the same IO. To expand the extent, the
1136 * page index must be in the same or next chunk of ext->oe_end.
1137 */
1138 static int osc_extent_expand(struct osc_extent *ext, pgoff_t index, int *grants)
1139 {
1140 struct osc_object *obj = ext->oe_obj;
1141 struct client_obd *cli = osc_cli(obj);
1142 struct osc_extent *next;
1143 int ppc_bits = cli->cl_chunkbits - PAGE_CACHE_SHIFT;
1144 pgoff_t chunk = index >> ppc_bits;
1145 pgoff_t end_chunk;
1146 pgoff_t end_index;
1147 int chunksize = 1 << cli->cl_chunkbits;
1148 int rc = 0;
1149
1150 LASSERT(ext->oe_max_end >= index && ext->oe_start <= index);
1151 osc_object_lock(obj);
1152 LASSERT(sanity_check_nolock(ext) == 0);
1153 end_chunk = ext->oe_end >> ppc_bits;
1154 if (chunk > end_chunk + 1) {
1155 rc = -ERANGE;
1156 goto out;
1157 }
1158
1159 if (end_chunk >= chunk) {
1160 rc = 0;
1161 goto out;
1162 }
1163
1164 LASSERT(end_chunk + 1 == chunk);
1165 /* try to expand this extent to cover @index */
1166 end_index = min(ext->oe_max_end, ((chunk + 1) << ppc_bits) - 1);
1167
1168 next = next_extent(ext);
1169 if (next != NULL && next->oe_start <= end_index) {
1170 /* complex mode - overlapped with the next extent,
1171 * this case will be handled by osc_extent_find() */
1172 rc = -EAGAIN;
1173 goto out;
1174 }
1175
1176 ext->oe_end = end_index;
1177 ext->oe_grants += chunksize;
1178 *grants -= chunksize;
1179 LASSERT(*grants >= 0);
1180 EASSERTF(osc_extent_is_overlapped(obj, ext) == 0, ext,
1181 "overlapped after expanding for %lu.\n", index);
1182
1183 out:
1184 osc_object_unlock(obj);
1185 return rc;
1186 }
1187
1188 static void osc_extent_tree_dump0(int level, struct osc_object *obj,
1189 const char *func, int line)
1190 {
1191 struct osc_extent *ext;
1192 int cnt;
1193
1194 CDEBUG(level, "Dump object %p extents at %s:%d, mppr: %u.\n",
1195 obj, func, line, osc_cli(obj)->cl_max_pages_per_rpc);
1196
1197 /* osc_object_lock(obj); */
1198 cnt = 1;
1199 for (ext = first_extent(obj); ext != NULL; ext = next_extent(ext))
1200 OSC_EXTENT_DUMP(level, ext, "in tree %d.\n", cnt++);
1201
1202 cnt = 1;
1203 list_for_each_entry(ext, &obj->oo_hp_exts, oe_link)
1204 OSC_EXTENT_DUMP(level, ext, "hp %d.\n", cnt++);
1205
1206 cnt = 1;
1207 list_for_each_entry(ext, &obj->oo_urgent_exts, oe_link)
1208 OSC_EXTENT_DUMP(level, ext, "urgent %d.\n", cnt++);
1209
1210 cnt = 1;
1211 list_for_each_entry(ext, &obj->oo_reading_exts, oe_link)
1212 OSC_EXTENT_DUMP(level, ext, "reading %d.\n", cnt++);
1213 /* osc_object_unlock(obj); */
1214 }
1215
1216 /* ------------------ osc extent end ------------------ */
1217
1218 static inline int osc_is_ready(struct osc_object *osc)
1219 {
1220 return !list_empty(&osc->oo_ready_item) ||
1221 !list_empty(&osc->oo_hp_ready_item);
1222 }
1223
1224 #define OSC_IO_DEBUG(OSC, STR, args...) \
1225 CDEBUG(D_CACHE, "obj %p ready %d|%c|%c wr %d|%c|%c rd %d|%c " STR, \
1226 (OSC), osc_is_ready(OSC), \
1227 list_empty_marker(&(OSC)->oo_hp_ready_item), \
1228 list_empty_marker(&(OSC)->oo_ready_item), \
1229 atomic_read(&(OSC)->oo_nr_writes), \
1230 list_empty_marker(&(OSC)->oo_hp_exts), \
1231 list_empty_marker(&(OSC)->oo_urgent_exts), \
1232 atomic_read(&(OSC)->oo_nr_reads), \
1233 list_empty_marker(&(OSC)->oo_reading_exts), \
1234 ##args)
1235
1236 static int osc_make_ready(const struct lu_env *env, struct osc_async_page *oap,
1237 int cmd)
1238 {
1239 struct osc_page *opg = oap2osc_page(oap);
1240 struct cl_page *page = cl_page_top(oap2cl_page(oap));
1241 int result;
1242
1243 LASSERT(cmd == OBD_BRW_WRITE); /* no cached reads */
1244
1245 result = cl_page_make_ready(env, page, CRT_WRITE);
1246 if (result == 0)
1247 opg->ops_submit_time = cfs_time_current();
1248 return result;
1249 }
1250
1251 static int osc_refresh_count(const struct lu_env *env,
1252 struct osc_async_page *oap, int cmd)
1253 {
1254 struct osc_page *opg = oap2osc_page(oap);
1255 struct cl_page *page = oap2cl_page(oap);
1256 struct cl_object *obj;
1257 struct cl_attr *attr = &osc_env_info(env)->oti_attr;
1258
1259 int result;
1260 loff_t kms;
1261
1262 /* readpage queues with _COUNT_STABLE, shouldn't get here. */
1263 LASSERT(!(cmd & OBD_BRW_READ));
1264 LASSERT(opg != NULL);
1265 obj = opg->ops_cl.cpl_obj;
1266
1267 cl_object_attr_lock(obj);
1268 result = cl_object_attr_get(env, obj, attr);
1269 cl_object_attr_unlock(obj);
1270 if (result < 0)
1271 return result;
1272 kms = attr->cat_kms;
1273 if (cl_offset(obj, page->cp_index) >= kms)
1274 /* catch race with truncate */
1275 return 0;
1276 else if (cl_offset(obj, page->cp_index + 1) > kms)
1277 /* catch sub-page write at end of file */
1278 return kms % PAGE_CACHE_SIZE;
1279 else
1280 return PAGE_CACHE_SIZE;
1281 }
1282
1283 static int osc_completion(const struct lu_env *env, struct osc_async_page *oap,
1284 int cmd, int rc)
1285 {
1286 struct osc_page *opg = oap2osc_page(oap);
1287 struct cl_page *page = cl_page_top(oap2cl_page(oap));
1288 struct osc_object *obj = cl2osc(opg->ops_cl.cpl_obj);
1289 enum cl_req_type crt;
1290 int srvlock;
1291
1292 cmd &= ~OBD_BRW_NOQUOTA;
1293 LASSERT(equi(page->cp_state == CPS_PAGEIN, cmd == OBD_BRW_READ));
1294 LASSERT(equi(page->cp_state == CPS_PAGEOUT, cmd == OBD_BRW_WRITE));
1295 LASSERT(opg->ops_transfer_pinned);
1296
1297 /*
1298 * page->cp_req can be NULL if io submission failed before
1299 * cl_req was allocated.
1300 */
1301 if (page->cp_req != NULL)
1302 cl_req_page_done(env, page);
1303 LASSERT(page->cp_req == NULL);
1304
1305 crt = cmd == OBD_BRW_READ ? CRT_READ : CRT_WRITE;
1306 /* Clear opg->ops_transfer_pinned before VM lock is released. */
1307 opg->ops_transfer_pinned = 0;
1308
1309 spin_lock(&obj->oo_seatbelt);
1310 LASSERT(opg->ops_submitter != NULL);
1311 LASSERT(!list_empty(&opg->ops_inflight));
1312 list_del_init(&opg->ops_inflight);
1313 opg->ops_submitter = NULL;
1314 spin_unlock(&obj->oo_seatbelt);
1315
1316 opg->ops_submit_time = 0;
1317 srvlock = oap->oap_brw_flags & OBD_BRW_SRVLOCK;
1318
1319 /* statistic */
1320 if (rc == 0 && srvlock) {
1321 struct lu_device *ld = opg->ops_cl.cpl_obj->co_lu.lo_dev;
1322 struct osc_stats *stats = &lu2osc_dev(ld)->od_stats;
1323 int bytes = oap->oap_count;
1324
1325 if (crt == CRT_READ)
1326 stats->os_lockless_reads += bytes;
1327 else
1328 stats->os_lockless_writes += bytes;
1329 }
1330
1331 /*
1332 * This has to be the last operation with the page, as locks are
1333 * released in cl_page_completion() and nothing except for the
1334 * reference counter protects page from concurrent reclaim.
1335 */
1336 lu_ref_del(&page->cp_reference, "transfer", page);
1337
1338 cl_page_completion(env, page, crt, rc);
1339
1340 return 0;
1341 }
1342
1343 #define OSC_DUMP_GRANT(cli, fmt, args...) do { \
1344 struct client_obd *__tmp = (cli); \
1345 CDEBUG(D_CACHE, "%s: { dirty: %ld/%ld dirty_pages: %d/%d " \
1346 "dropped: %ld avail: %ld, reserved: %ld, flight: %d } " fmt, \
1347 __tmp->cl_import->imp_obd->obd_name, \
1348 __tmp->cl_dirty, __tmp->cl_dirty_max, \
1349 atomic_read(&obd_dirty_pages), obd_max_dirty_pages, \
1350 __tmp->cl_lost_grant, __tmp->cl_avail_grant, \
1351 __tmp->cl_reserved_grant, __tmp->cl_w_in_flight, ##args); \
1352 } while (0)
1353
1354 /* caller must hold loi_list_lock */
1355 static void osc_consume_write_grant(struct client_obd *cli,
1356 struct brw_page *pga)
1357 {
1358 assert_spin_locked(&cli->cl_loi_list_lock.lock);
1359 LASSERT(!(pga->flag & OBD_BRW_FROM_GRANT));
1360 atomic_inc(&obd_dirty_pages);
1361 cli->cl_dirty += PAGE_CACHE_SIZE;
1362 pga->flag |= OBD_BRW_FROM_GRANT;
1363 CDEBUG(D_CACHE, "using %lu grant credits for brw %p page %p\n",
1364 PAGE_CACHE_SIZE, pga, pga->pg);
1365 osc_update_next_shrink(cli);
1366 }
1367
1368 /* the companion to osc_consume_write_grant, called when a brw has completed.
1369 * must be called with the loi lock held. */
1370 static void osc_release_write_grant(struct client_obd *cli,
1371 struct brw_page *pga)
1372 {
1373 assert_spin_locked(&cli->cl_loi_list_lock.lock);
1374 if (!(pga->flag & OBD_BRW_FROM_GRANT)) {
1375 return;
1376 }
1377
1378 pga->flag &= ~OBD_BRW_FROM_GRANT;
1379 atomic_dec(&obd_dirty_pages);
1380 cli->cl_dirty -= PAGE_CACHE_SIZE;
1381 if (pga->flag & OBD_BRW_NOCACHE) {
1382 pga->flag &= ~OBD_BRW_NOCACHE;
1383 atomic_dec(&obd_dirty_transit_pages);
1384 cli->cl_dirty_transit -= PAGE_CACHE_SIZE;
1385 }
1386 }
1387
1388 /**
1389 * To avoid sleeping with object lock held, it's good for us allocate enough
1390 * grants before entering into critical section.
1391 *
1392 * client_obd_list_lock held by caller
1393 */
1394 static int osc_reserve_grant(struct client_obd *cli, unsigned int bytes)
1395 {
1396 int rc = -EDQUOT;
1397
1398 if (cli->cl_avail_grant >= bytes) {
1399 cli->cl_avail_grant -= bytes;
1400 cli->cl_reserved_grant += bytes;
1401 rc = 0;
1402 }
1403 return rc;
1404 }
1405
1406 static void __osc_unreserve_grant(struct client_obd *cli,
1407 unsigned int reserved, unsigned int unused)
1408 {
1409 /* it's quite normal for us to get more grant than reserved.
1410 * Thinking about a case that two extents merged by adding a new
1411 * chunk, we can save one extent tax. If extent tax is greater than
1412 * one chunk, we can save more grant by adding a new chunk */
1413 cli->cl_reserved_grant -= reserved;
1414 if (unused > reserved) {
1415 cli->cl_avail_grant += reserved;
1416 cli->cl_lost_grant += unused - reserved;
1417 } else {
1418 cli->cl_avail_grant += unused;
1419 }
1420 }
1421
1422 void osc_unreserve_grant(struct client_obd *cli,
1423 unsigned int reserved, unsigned int unused)
1424 {
1425 client_obd_list_lock(&cli->cl_loi_list_lock);
1426 __osc_unreserve_grant(cli, reserved, unused);
1427 if (unused > 0)
1428 osc_wake_cache_waiters(cli);
1429 client_obd_list_unlock(&cli->cl_loi_list_lock);
1430 }
1431
1432 /**
1433 * Free grant after IO is finished or canceled.
1434 *
1435 * @lost_grant is used to remember how many grants we have allocated but not
1436 * used, we should return these grants to OST. There're two cases where grants
1437 * can be lost:
1438 * 1. truncate;
1439 * 2. blocksize at OST is less than PAGE_CACHE_SIZE and a partial page was
1440 * written. In this case OST may use less chunks to serve this partial
1441 * write. OSTs don't actually know the page size on the client side. so
1442 * clients have to calculate lost grant by the blocksize on the OST.
1443 * See filter_grant_check() for details.
1444 */
1445 static void osc_free_grant(struct client_obd *cli, unsigned int nr_pages,
1446 unsigned int lost_grant)
1447 {
1448 int grant = (1 << cli->cl_chunkbits) + cli->cl_extent_tax;
1449
1450 client_obd_list_lock(&cli->cl_loi_list_lock);
1451 atomic_sub(nr_pages, &obd_dirty_pages);
1452 cli->cl_dirty -= nr_pages << PAGE_CACHE_SHIFT;
1453 cli->cl_lost_grant += lost_grant;
1454 if (cli->cl_avail_grant < grant && cli->cl_lost_grant >= grant) {
1455 /* borrow some grant from truncate to avoid the case that
1456 * truncate uses up all avail grant */
1457 cli->cl_lost_grant -= grant;
1458 cli->cl_avail_grant += grant;
1459 }
1460 osc_wake_cache_waiters(cli);
1461 client_obd_list_unlock(&cli->cl_loi_list_lock);
1462 CDEBUG(D_CACHE, "lost %u grant: %lu avail: %lu dirty: %lu\n",
1463 lost_grant, cli->cl_lost_grant,
1464 cli->cl_avail_grant, cli->cl_dirty);
1465 }
1466
1467 /**
1468 * The companion to osc_enter_cache(), called when @oap is no longer part of
1469 * the dirty accounting due to error.
1470 */
1471 static void osc_exit_cache(struct client_obd *cli, struct osc_async_page *oap)
1472 {
1473 client_obd_list_lock(&cli->cl_loi_list_lock);
1474 osc_release_write_grant(cli, &oap->oap_brw_page);
1475 client_obd_list_unlock(&cli->cl_loi_list_lock);
1476 }
1477
1478 /**
1479 * Non-blocking version of osc_enter_cache() that consumes grant only when it
1480 * is available.
1481 */
1482 static int osc_enter_cache_try(struct client_obd *cli,
1483 struct osc_async_page *oap,
1484 int bytes, int transient)
1485 {
1486 int rc;
1487
1488 OSC_DUMP_GRANT(cli, "need:%d.\n", bytes);
1489
1490 rc = osc_reserve_grant(cli, bytes);
1491 if (rc < 0)
1492 return 0;
1493
1494 if (cli->cl_dirty + PAGE_CACHE_SIZE <= cli->cl_dirty_max &&
1495 atomic_read(&obd_dirty_pages) + 1 <= obd_max_dirty_pages) {
1496 osc_consume_write_grant(cli, &oap->oap_brw_page);
1497 if (transient) {
1498 cli->cl_dirty_transit += PAGE_CACHE_SIZE;
1499 atomic_inc(&obd_dirty_transit_pages);
1500 oap->oap_brw_flags |= OBD_BRW_NOCACHE;
1501 }
1502 rc = 1;
1503 } else {
1504 __osc_unreserve_grant(cli, bytes, bytes);
1505 rc = 0;
1506 }
1507 return rc;
1508 }
1509
1510 static int ocw_granted(struct client_obd *cli, struct osc_cache_waiter *ocw)
1511 {
1512 int rc;
1513 client_obd_list_lock(&cli->cl_loi_list_lock);
1514 rc = list_empty(&ocw->ocw_entry);
1515 client_obd_list_unlock(&cli->cl_loi_list_lock);
1516 return rc;
1517 }
1518
1519 /**
1520 * The main entry to reserve dirty page accounting. Usually the grant reserved
1521 * in this function will be freed in bulk in osc_free_grant() unless it fails
1522 * to add osc cache, in that case, it will be freed in osc_exit_cache().
1523 *
1524 * The process will be put into sleep if it's already run out of grant.
1525 */
1526 static int osc_enter_cache(const struct lu_env *env, struct client_obd *cli,
1527 struct osc_async_page *oap, int bytes)
1528 {
1529 struct osc_object *osc = oap->oap_obj;
1530 struct lov_oinfo *loi = osc->oo_oinfo;
1531 struct osc_cache_waiter ocw;
1532 struct l_wait_info lwi = LWI_INTR(LWI_ON_SIGNAL_NOOP, NULL);
1533 int rc = -EDQUOT;
1534
1535 OSC_DUMP_GRANT(cli, "need:%d.\n", bytes);
1536
1537 client_obd_list_lock(&cli->cl_loi_list_lock);
1538
1539 /* force the caller to try sync io. this can jump the list
1540 * of queued writes and create a discontiguous rpc stream */
1541 if (OBD_FAIL_CHECK(OBD_FAIL_OSC_NO_GRANT) ||
1542 cli->cl_dirty_max < PAGE_CACHE_SIZE ||
1543 cli->cl_ar.ar_force_sync || loi->loi_ar.ar_force_sync) {
1544 rc = -EDQUOT;
1545 goto out;
1546 }
1547
1548 /* Hopefully normal case - cache space and write credits available */
1549 if (osc_enter_cache_try(cli, oap, bytes, 0)) {
1550 rc = 0;
1551 goto out;
1552 }
1553
1554 /* We can get here for two reasons: too many dirty pages in cache, or
1555 * run out of grants. In both cases we should write dirty pages out.
1556 * Adding a cache waiter will trigger urgent write-out no matter what
1557 * RPC size will be.
1558 * The exiting condition is no avail grants and no dirty pages caching,
1559 * that really means there is no space on the OST. */
1560 init_waitqueue_head(&ocw.ocw_waitq);
1561 ocw.ocw_oap = oap;
1562 ocw.ocw_grant = bytes;
1563 while (cli->cl_dirty > 0 || cli->cl_w_in_flight > 0) {
1564 list_add_tail(&ocw.ocw_entry, &cli->cl_cache_waiters);
1565 ocw.ocw_rc = 0;
1566 client_obd_list_unlock(&cli->cl_loi_list_lock);
1567
1568 osc_io_unplug_async(env, cli, NULL);
1569
1570 CDEBUG(D_CACHE, "%s: sleeping for cache space @ %p for %p\n",
1571 cli->cl_import->imp_obd->obd_name, &ocw, oap);
1572
1573 rc = l_wait_event(ocw.ocw_waitq, ocw_granted(cli, &ocw), &lwi);
1574
1575 client_obd_list_lock(&cli->cl_loi_list_lock);
1576
1577 /* l_wait_event is interrupted by signal */
1578 if (rc < 0) {
1579 list_del_init(&ocw.ocw_entry);
1580 goto out;
1581 }
1582
1583 LASSERT(list_empty(&ocw.ocw_entry));
1584 rc = ocw.ocw_rc;
1585
1586 if (rc != -EDQUOT)
1587 goto out;
1588 if (osc_enter_cache_try(cli, oap, bytes, 0)) {
1589 rc = 0;
1590 goto out;
1591 }
1592 }
1593 out:
1594 client_obd_list_unlock(&cli->cl_loi_list_lock);
1595 OSC_DUMP_GRANT(cli, "returned %d.\n", rc);
1596 return rc;
1597 }
1598
1599 /* caller must hold loi_list_lock */
1600 void osc_wake_cache_waiters(struct client_obd *cli)
1601 {
1602 struct list_head *l, *tmp;
1603 struct osc_cache_waiter *ocw;
1604
1605 list_for_each_safe(l, tmp, &cli->cl_cache_waiters) {
1606 ocw = list_entry(l, struct osc_cache_waiter, ocw_entry);
1607 list_del_init(&ocw->ocw_entry);
1608
1609 ocw->ocw_rc = -EDQUOT;
1610 /* we can't dirty more */
1611 if ((cli->cl_dirty + PAGE_CACHE_SIZE > cli->cl_dirty_max) ||
1612 (atomic_read(&obd_dirty_pages) + 1 >
1613 obd_max_dirty_pages)) {
1614 CDEBUG(D_CACHE, "no dirty room: dirty: %ld osc max %ld, sys max %d\n",
1615 cli->cl_dirty,
1616 cli->cl_dirty_max, obd_max_dirty_pages);
1617 goto wakeup;
1618 }
1619
1620 ocw->ocw_rc = 0;
1621 if (!osc_enter_cache_try(cli, ocw->ocw_oap, ocw->ocw_grant, 0))
1622 ocw->ocw_rc = -EDQUOT;
1623
1624 wakeup:
1625 CDEBUG(D_CACHE, "wake up %p for oap %p, avail grant %ld, %d\n",
1626 ocw, ocw->ocw_oap, cli->cl_avail_grant, ocw->ocw_rc);
1627
1628 wake_up(&ocw->ocw_waitq);
1629 }
1630 }
1631
1632 static int osc_max_rpc_in_flight(struct client_obd *cli, struct osc_object *osc)
1633 {
1634 int hprpc = !!list_empty(&osc->oo_hp_exts);
1635 return rpcs_in_flight(cli) >= cli->cl_max_rpcs_in_flight + hprpc;
1636 }
1637
1638 /* This maintains the lists of pending pages to read/write for a given object
1639 * (lop). This is used by osc_check_rpcs->osc_next_obj() and osc_list_maint()
1640 * to quickly find objects that are ready to send an RPC. */
1641 static int osc_makes_rpc(struct client_obd *cli, struct osc_object *osc,
1642 int cmd)
1643 {
1644 int invalid_import = 0;
1645
1646 /* if we have an invalid import we want to drain the queued pages
1647 * by forcing them through rpcs that immediately fail and complete
1648 * the pages. recovery relies on this to empty the queued pages
1649 * before canceling the locks and evicting down the llite pages */
1650 if ((cli->cl_import == NULL || cli->cl_import->imp_invalid))
1651 invalid_import = 1;
1652
1653 if (cmd & OBD_BRW_WRITE) {
1654 if (atomic_read(&osc->oo_nr_writes) == 0)
1655 return 0;
1656 if (invalid_import) {
1657 CDEBUG(D_CACHE, "invalid import forcing RPC\n");
1658 return 1;
1659 }
1660 if (!list_empty(&osc->oo_hp_exts)) {
1661 CDEBUG(D_CACHE, "high prio request forcing RPC\n");
1662 return 1;
1663 }
1664 if (!list_empty(&osc->oo_urgent_exts)) {
1665 CDEBUG(D_CACHE, "urgent request forcing RPC\n");
1666 return 1;
1667 }
1668 /* trigger a write rpc stream as long as there are dirtiers
1669 * waiting for space. as they're waiting, they're not going to
1670 * create more pages to coalesce with what's waiting.. */
1671 if (!list_empty(&cli->cl_cache_waiters)) {
1672 CDEBUG(D_CACHE, "cache waiters forcing RPC\n");
1673 return 1;
1674 }
1675 if (atomic_read(&osc->oo_nr_writes) >=
1676 cli->cl_max_pages_per_rpc)
1677 return 1;
1678 } else {
1679 if (atomic_read(&osc->oo_nr_reads) == 0)
1680 return 0;
1681 if (invalid_import) {
1682 CDEBUG(D_CACHE, "invalid import forcing RPC\n");
1683 return 1;
1684 }
1685 /* all read are urgent. */
1686 if (!list_empty(&osc->oo_reading_exts))
1687 return 1;
1688 }
1689
1690 return 0;
1691 }
1692
1693 static void osc_update_pending(struct osc_object *obj, int cmd, int delta)
1694 {
1695 struct client_obd *cli = osc_cli(obj);
1696 if (cmd & OBD_BRW_WRITE) {
1697 atomic_add(delta, &obj->oo_nr_writes);
1698 atomic_add(delta, &cli->cl_pending_w_pages);
1699 LASSERT(atomic_read(&obj->oo_nr_writes) >= 0);
1700 } else {
1701 atomic_add(delta, &obj->oo_nr_reads);
1702 atomic_add(delta, &cli->cl_pending_r_pages);
1703 LASSERT(atomic_read(&obj->oo_nr_reads) >= 0);
1704 }
1705 OSC_IO_DEBUG(obj, "update pending cmd %d delta %d.\n", cmd, delta);
1706 }
1707
1708 static int osc_makes_hprpc(struct osc_object *obj)
1709 {
1710 return !list_empty(&obj->oo_hp_exts);
1711 }
1712
1713 static void on_list(struct list_head *item, struct list_head *list, int should_be_on)
1714 {
1715 if (list_empty(item) && should_be_on)
1716 list_add_tail(item, list);
1717 else if (!list_empty(item) && !should_be_on)
1718 list_del_init(item);
1719 }
1720
1721 /* maintain the osc's cli list membership invariants so that osc_send_oap_rpc
1722 * can find pages to build into rpcs quickly */
1723 static int __osc_list_maint(struct client_obd *cli, struct osc_object *osc)
1724 {
1725 if (osc_makes_hprpc(osc)) {
1726 /* HP rpc */
1727 on_list(&osc->oo_ready_item, &cli->cl_loi_ready_list, 0);
1728 on_list(&osc->oo_hp_ready_item, &cli->cl_loi_hp_ready_list, 1);
1729 } else {
1730 on_list(&osc->oo_hp_ready_item, &cli->cl_loi_hp_ready_list, 0);
1731 on_list(&osc->oo_ready_item, &cli->cl_loi_ready_list,
1732 osc_makes_rpc(cli, osc, OBD_BRW_WRITE) ||
1733 osc_makes_rpc(cli, osc, OBD_BRW_READ));
1734 }
1735
1736 on_list(&osc->oo_write_item, &cli->cl_loi_write_list,
1737 atomic_read(&osc->oo_nr_writes) > 0);
1738
1739 on_list(&osc->oo_read_item, &cli->cl_loi_read_list,
1740 atomic_read(&osc->oo_nr_reads) > 0);
1741
1742 return osc_is_ready(osc);
1743 }
1744
1745 static int osc_list_maint(struct client_obd *cli, struct osc_object *osc)
1746 {
1747 int is_ready;
1748
1749 client_obd_list_lock(&cli->cl_loi_list_lock);
1750 is_ready = __osc_list_maint(cli, osc);
1751 client_obd_list_unlock(&cli->cl_loi_list_lock);
1752
1753 return is_ready;
1754 }
1755
1756 /* this is trying to propagate async writeback errors back up to the
1757 * application. As an async write fails we record the error code for later if
1758 * the app does an fsync. As long as errors persist we force future rpcs to be
1759 * sync so that the app can get a sync error and break the cycle of queueing
1760 * pages for which writeback will fail. */
1761 static void osc_process_ar(struct osc_async_rc *ar, __u64 xid,
1762 int rc)
1763 {
1764 if (rc) {
1765 if (!ar->ar_rc)
1766 ar->ar_rc = rc;
1767
1768 ar->ar_force_sync = 1;
1769 ar->ar_min_xid = ptlrpc_sample_next_xid();
1770 return;
1771
1772 }
1773
1774 if (ar->ar_force_sync && (xid >= ar->ar_min_xid))
1775 ar->ar_force_sync = 0;
1776 }
1777
1778
1779 /* this must be called holding the loi list lock to give coverage to exit_cache,
1780 * async_flag maintenance, and oap_request */
1781 static void osc_ap_completion(const struct lu_env *env, struct client_obd *cli,
1782 struct osc_async_page *oap, int sent, int rc)
1783 {
1784 struct osc_object *osc = oap->oap_obj;
1785 struct lov_oinfo *loi = osc->oo_oinfo;
1786 __u64 xid = 0;
1787
1788 if (oap->oap_request != NULL) {
1789 xid = ptlrpc_req_xid(oap->oap_request);
1790 ptlrpc_req_finished(oap->oap_request);
1791 oap->oap_request = NULL;
1792 }
1793
1794 /* As the transfer for this page is being done, clear the flags */
1795 spin_lock(&oap->oap_lock);
1796 oap->oap_async_flags = 0;
1797 spin_unlock(&oap->oap_lock);
1798 oap->oap_interrupted = 0;
1799
1800 if (oap->oap_cmd & OBD_BRW_WRITE && xid > 0) {
1801 client_obd_list_lock(&cli->cl_loi_list_lock);
1802 osc_process_ar(&cli->cl_ar, xid, rc);
1803 osc_process_ar(&loi->loi_ar, xid, rc);
1804 client_obd_list_unlock(&cli->cl_loi_list_lock);
1805 }
1806
1807 rc = osc_completion(env, oap, oap->oap_cmd, rc);
1808 if (rc)
1809 CERROR("completion on oap %p obj %p returns %d.\n",
1810 oap, osc, rc);
1811 }
1812
1813 /**
1814 * Try to add extent to one RPC. We need to think about the following things:
1815 * - # of pages must not be over max_pages_per_rpc
1816 * - extent must be compatible with previous ones
1817 */
1818 static int try_to_add_extent_for_io(struct client_obd *cli,
1819 struct osc_extent *ext, struct list_head *rpclist,
1820 int *pc, unsigned int *max_pages)
1821 {
1822 struct osc_extent *tmp;
1823 struct osc_async_page *oap = list_first_entry(&ext->oe_pages,
1824 struct osc_async_page,
1825 oap_pending_item);
1826
1827 EASSERT((ext->oe_state == OES_CACHE || ext->oe_state == OES_LOCK_DONE),
1828 ext);
1829
1830 *max_pages = max(ext->oe_mppr, *max_pages);
1831 if (*pc + ext->oe_nr_pages > *max_pages)
1832 return 0;
1833
1834 list_for_each_entry(tmp, rpclist, oe_link) {
1835 struct osc_async_page *oap2;
1836
1837 oap2 = list_first_entry(&tmp->oe_pages, struct osc_async_page,
1838 oap_pending_item);
1839 EASSERT(tmp->oe_owner == current, tmp);
1840 if (oap2cl_page(oap)->cp_type != oap2cl_page(oap2)->cp_type) {
1841 CDEBUG(D_CACHE, "Do not permit different type of IO"
1842 " for a same RPC\n");
1843 return 0;
1844 }
1845
1846 if (tmp->oe_srvlock != ext->oe_srvlock ||
1847 !tmp->oe_grants != !ext->oe_grants)
1848 return 0;
1849
1850 /* remove break for strict check */
1851 break;
1852 }
1853
1854 *pc += ext->oe_nr_pages;
1855 list_move_tail(&ext->oe_link, rpclist);
1856 ext->oe_owner = current;
1857 return 1;
1858 }
1859
1860 /**
1861 * In order to prevent multiple ptlrpcd from breaking contiguous extents,
1862 * get_write_extent() takes all appropriate extents in atomic.
1863 *
1864 * The following policy is used to collect extents for IO:
1865 * 1. Add as many HP extents as possible;
1866 * 2. Add the first urgent extent in urgent extent list and take it out of
1867 * urgent list;
1868 * 3. Add subsequent extents of this urgent extent;
1869 * 4. If urgent list is not empty, goto 2;
1870 * 5. Traverse the extent tree from the 1st extent;
1871 * 6. Above steps exit if there is no space in this RPC.
1872 */
1873 static int get_write_extents(struct osc_object *obj, struct list_head *rpclist)
1874 {
1875 struct client_obd *cli = osc_cli(obj);
1876 struct osc_extent *ext;
1877 int page_count = 0;
1878 unsigned int max_pages = cli->cl_max_pages_per_rpc;
1879
1880 LASSERT(osc_object_is_locked(obj));
1881 while (!list_empty(&obj->oo_hp_exts)) {
1882 ext = list_entry(obj->oo_hp_exts.next, struct osc_extent,
1883 oe_link);
1884 LASSERT(ext->oe_state == OES_CACHE);
1885 if (!try_to_add_extent_for_io(cli, ext, rpclist, &page_count,
1886 &max_pages))
1887 return page_count;
1888 EASSERT(ext->oe_nr_pages <= max_pages, ext);
1889 }
1890 if (page_count == max_pages)
1891 return page_count;
1892
1893 while (!list_empty(&obj->oo_urgent_exts)) {
1894 ext = list_entry(obj->oo_urgent_exts.next,
1895 struct osc_extent, oe_link);
1896 if (!try_to_add_extent_for_io(cli, ext, rpclist, &page_count,
1897 &max_pages))
1898 return page_count;
1899
1900 if (!ext->oe_intree)
1901 continue;
1902
1903 while ((ext = next_extent(ext)) != NULL) {
1904 if ((ext->oe_state != OES_CACHE) ||
1905 (!list_empty(&ext->oe_link) &&
1906 ext->oe_owner != NULL))
1907 continue;
1908
1909 if (!try_to_add_extent_for_io(cli, ext, rpclist,
1910 &page_count, &max_pages))
1911 return page_count;
1912 }
1913 }
1914 if (page_count == max_pages)
1915 return page_count;
1916
1917 ext = first_extent(obj);
1918 while (ext != NULL) {
1919 if ((ext->oe_state != OES_CACHE) ||
1920 /* this extent may be already in current rpclist */
1921 (!list_empty(&ext->oe_link) && ext->oe_owner != NULL)) {
1922 ext = next_extent(ext);
1923 continue;
1924 }
1925
1926 if (!try_to_add_extent_for_io(cli, ext, rpclist, &page_count,
1927 &max_pages))
1928 return page_count;
1929
1930 ext = next_extent(ext);
1931 }
1932 return page_count;
1933 }
1934
1935 static int
1936 osc_send_write_rpc(const struct lu_env *env, struct client_obd *cli,
1937 struct osc_object *osc, pdl_policy_t pol)
1938 {
1939 LIST_HEAD(rpclist);
1940 struct osc_extent *ext;
1941 struct osc_extent *tmp;
1942 struct osc_extent *first = NULL;
1943 u32 page_count = 0;
1944 int srvlock = 0;
1945 int rc = 0;
1946
1947 LASSERT(osc_object_is_locked(osc));
1948
1949 page_count = get_write_extents(osc, &rpclist);
1950 LASSERT(equi(page_count == 0, list_empty(&rpclist)));
1951
1952 if (list_empty(&rpclist))
1953 return 0;
1954
1955 osc_update_pending(osc, OBD_BRW_WRITE, -page_count);
1956
1957 list_for_each_entry(ext, &rpclist, oe_link) {
1958 LASSERT(ext->oe_state == OES_CACHE ||
1959 ext->oe_state == OES_LOCK_DONE);
1960 if (ext->oe_state == OES_CACHE)
1961 osc_extent_state_set(ext, OES_LOCKING);
1962 else
1963 osc_extent_state_set(ext, OES_RPC);
1964 }
1965
1966 /* we're going to grab page lock, so release object lock because
1967 * lock order is page lock -> object lock. */
1968 osc_object_unlock(osc);
1969
1970 list_for_each_entry_safe(ext, tmp, &rpclist, oe_link) {
1971 if (ext->oe_state == OES_LOCKING) {
1972 rc = osc_extent_make_ready(env, ext);
1973 if (unlikely(rc < 0)) {
1974 list_del_init(&ext->oe_link);
1975 osc_extent_finish(env, ext, 0, rc);
1976 continue;
1977 }
1978 }
1979 if (first == NULL) {
1980 first = ext;
1981 srvlock = ext->oe_srvlock;
1982 } else {
1983 LASSERT(srvlock == ext->oe_srvlock);
1984 }
1985 }
1986
1987 if (!list_empty(&rpclist)) {
1988 LASSERT(page_count > 0);
1989 rc = osc_build_rpc(env, cli, &rpclist, OBD_BRW_WRITE, pol);
1990 LASSERT(list_empty(&rpclist));
1991 }
1992
1993 osc_object_lock(osc);
1994 return rc;
1995 }
1996
1997 /**
1998 * prepare pages for ASYNC io and put pages in send queue.
1999 *
2000 * \param cmd OBD_BRW_* macroses
2001 * \param lop pending pages
2002 *
2003 * \return zero if no page added to send queue.
2004 * \return 1 if pages successfully added to send queue.
2005 * \return negative on errors.
2006 */
2007 static int
2008 osc_send_read_rpc(const struct lu_env *env, struct client_obd *cli,
2009 struct osc_object *osc, pdl_policy_t pol)
2010 {
2011 struct osc_extent *ext;
2012 struct osc_extent *next;
2013 LIST_HEAD(rpclist);
2014 int page_count = 0;
2015 unsigned int max_pages = cli->cl_max_pages_per_rpc;
2016 int rc = 0;
2017
2018 LASSERT(osc_object_is_locked(osc));
2019 list_for_each_entry_safe(ext, next,
2020 &osc->oo_reading_exts, oe_link) {
2021 EASSERT(ext->oe_state == OES_LOCK_DONE, ext);
2022 if (!try_to_add_extent_for_io(cli, ext, &rpclist, &page_count,
2023 &max_pages))
2024 break;
2025 osc_extent_state_set(ext, OES_RPC);
2026 EASSERT(ext->oe_nr_pages <= max_pages, ext);
2027 }
2028 LASSERT(page_count <= max_pages);
2029
2030 osc_update_pending(osc, OBD_BRW_READ, -page_count);
2031
2032 if (!list_empty(&rpclist)) {
2033 osc_object_unlock(osc);
2034
2035 LASSERT(page_count > 0);
2036 rc = osc_build_rpc(env, cli, &rpclist, OBD_BRW_READ, pol);
2037 LASSERT(list_empty(&rpclist));
2038
2039 osc_object_lock(osc);
2040 }
2041 return rc;
2042 }
2043
2044 #define list_to_obj(list, item) ({ \
2045 struct list_head *__tmp = (list)->next; \
2046 list_del_init(__tmp); \
2047 list_entry(__tmp, struct osc_object, oo_##item); \
2048 })
2049
2050 /* This is called by osc_check_rpcs() to find which objects have pages that
2051 * we could be sending. These lists are maintained by osc_makes_rpc(). */
2052 static struct osc_object *osc_next_obj(struct client_obd *cli)
2053 {
2054 /* First return objects that have blocked locks so that they
2055 * will be flushed quickly and other clients can get the lock,
2056 * then objects which have pages ready to be stuffed into RPCs */
2057 if (!list_empty(&cli->cl_loi_hp_ready_list))
2058 return list_to_obj(&cli->cl_loi_hp_ready_list, hp_ready_item);
2059 if (!list_empty(&cli->cl_loi_ready_list))
2060 return list_to_obj(&cli->cl_loi_ready_list, ready_item);
2061
2062 /* then if we have cache waiters, return all objects with queued
2063 * writes. This is especially important when many small files
2064 * have filled up the cache and not been fired into rpcs because
2065 * they don't pass the nr_pending/object threshold */
2066 if (!list_empty(&cli->cl_cache_waiters) &&
2067 !list_empty(&cli->cl_loi_write_list))
2068 return list_to_obj(&cli->cl_loi_write_list, write_item);
2069
2070 /* then return all queued objects when we have an invalid import
2071 * so that they get flushed */
2072 if (cli->cl_import == NULL || cli->cl_import->imp_invalid) {
2073 if (!list_empty(&cli->cl_loi_write_list))
2074 return list_to_obj(&cli->cl_loi_write_list, write_item);
2075 if (!list_empty(&cli->cl_loi_read_list))
2076 return list_to_obj(&cli->cl_loi_read_list, read_item);
2077 }
2078 return NULL;
2079 }
2080
2081 /* called with the loi list lock held */
2082 static void osc_check_rpcs(const struct lu_env *env, struct client_obd *cli,
2083 pdl_policy_t pol)
2084 {
2085 struct osc_object *osc;
2086 int rc = 0;
2087
2088 while ((osc = osc_next_obj(cli)) != NULL) {
2089 struct cl_object *obj = osc2cl(osc);
2090 struct lu_ref_link link;
2091
2092 OSC_IO_DEBUG(osc, "%lu in flight\n", rpcs_in_flight(cli));
2093
2094 if (osc_max_rpc_in_flight(cli, osc)) {
2095 __osc_list_maint(cli, osc);
2096 break;
2097 }
2098
2099 cl_object_get(obj);
2100 client_obd_list_unlock(&cli->cl_loi_list_lock);
2101 lu_object_ref_add_at(&obj->co_lu, &link, "check",
2102 current);
2103
2104 /* attempt some read/write balancing by alternating between
2105 * reads and writes in an object. The makes_rpc checks here
2106 * would be redundant if we were getting read/write work items
2107 * instead of objects. we don't want send_oap_rpc to drain a
2108 * partial read pending queue when we're given this object to
2109 * do io on writes while there are cache waiters */
2110 osc_object_lock(osc);
2111 if (osc_makes_rpc(cli, osc, OBD_BRW_WRITE)) {
2112 rc = osc_send_write_rpc(env, cli, osc, pol);
2113 if (rc < 0) {
2114 CERROR("Write request failed with %d\n", rc);
2115
2116 /* osc_send_write_rpc failed, mostly because of
2117 * memory pressure.
2118 *
2119 * It can't break here, because if:
2120 * - a page was submitted by osc_io_submit, so
2121 * page locked;
2122 * - no request in flight
2123 * - no subsequent request
2124 * The system will be in live-lock state,
2125 * because there is no chance to call
2126 * osc_io_unplug() and osc_check_rpcs() any
2127 * more. pdflush can't help in this case,
2128 * because it might be blocked at grabbing
2129 * the page lock as we mentioned.
2130 *
2131 * Anyway, continue to drain pages. */
2132 /* break; */
2133 }
2134 }
2135 if (osc_makes_rpc(cli, osc, OBD_BRW_READ)) {
2136 rc = osc_send_read_rpc(env, cli, osc, pol);
2137 if (rc < 0)
2138 CERROR("Read request failed with %d\n", rc);
2139 }
2140 osc_object_unlock(osc);
2141
2142 osc_list_maint(cli, osc);
2143 lu_object_ref_del_at(&obj->co_lu, &link, "check",
2144 current);
2145 cl_object_put(env, obj);
2146
2147 client_obd_list_lock(&cli->cl_loi_list_lock);
2148 }
2149 }
2150
2151 static int osc_io_unplug0(const struct lu_env *env, struct client_obd *cli,
2152 struct osc_object *osc, pdl_policy_t pol, int async)
2153 {
2154 int rc = 0;
2155
2156 if (osc != NULL && osc_list_maint(cli, osc) == 0)
2157 return 0;
2158
2159 if (!async) {
2160 /* disable osc_lru_shrink() temporarily to avoid
2161 * potential stack overrun problem. LU-2859 */
2162 atomic_inc(&cli->cl_lru_shrinkers);
2163 client_obd_list_lock(&cli->cl_loi_list_lock);
2164 osc_check_rpcs(env, cli, pol);
2165 client_obd_list_unlock(&cli->cl_loi_list_lock);
2166 atomic_dec(&cli->cl_lru_shrinkers);
2167 } else {
2168 CDEBUG(D_CACHE, "Queue writeback work for client %p.\n", cli);
2169 LASSERT(cli->cl_writeback_work != NULL);
2170 rc = ptlrpcd_queue_work(cli->cl_writeback_work);
2171 }
2172 return rc;
2173 }
2174
2175 static int osc_io_unplug_async(const struct lu_env *env,
2176 struct client_obd *cli, struct osc_object *osc)
2177 {
2178 /* XXX: policy is no use actually. */
2179 return osc_io_unplug0(env, cli, osc, PDL_POLICY_ROUND, 1);
2180 }
2181
2182 void osc_io_unplug(const struct lu_env *env, struct client_obd *cli,
2183 struct osc_object *osc, pdl_policy_t pol)
2184 {
2185 (void)osc_io_unplug0(env, cli, osc, pol, 0);
2186 }
2187
2188 int osc_prep_async_page(struct osc_object *osc, struct osc_page *ops,
2189 struct page *page, loff_t offset)
2190 {
2191 struct obd_export *exp = osc_export(osc);
2192 struct osc_async_page *oap = &ops->ops_oap;
2193
2194 if (!page)
2195 return cfs_size_round(sizeof(*oap));
2196
2197 oap->oap_magic = OAP_MAGIC;
2198 oap->oap_cli = &exp->exp_obd->u.cli;
2199 oap->oap_obj = osc;
2200
2201 oap->oap_page = page;
2202 oap->oap_obj_off = offset;
2203 LASSERT(!(offset & ~CFS_PAGE_MASK));
2204
2205 if (!client_is_remote(exp) && capable(CFS_CAP_SYS_RESOURCE))
2206 oap->oap_brw_flags = OBD_BRW_NOQUOTA;
2207
2208 INIT_LIST_HEAD(&oap->oap_pending_item);
2209 INIT_LIST_HEAD(&oap->oap_rpc_item);
2210
2211 spin_lock_init(&oap->oap_lock);
2212 CDEBUG(D_INFO, "oap %p page %p obj off %llu\n",
2213 oap, page, oap->oap_obj_off);
2214 return 0;
2215 }
2216
2217 int osc_queue_async_io(const struct lu_env *env, struct cl_io *io,
2218 struct osc_page *ops)
2219 {
2220 struct osc_io *oio = osc_env_io(env);
2221 struct osc_extent *ext = NULL;
2222 struct osc_async_page *oap = &ops->ops_oap;
2223 struct client_obd *cli = oap->oap_cli;
2224 struct osc_object *osc = oap->oap_obj;
2225 pgoff_t index;
2226 int grants = 0;
2227 int brw_flags = OBD_BRW_ASYNC;
2228 int cmd = OBD_BRW_WRITE;
2229 int need_release = 0;
2230 int rc = 0;
2231
2232 if (oap->oap_magic != OAP_MAGIC)
2233 return -EINVAL;
2234
2235 if (cli->cl_import == NULL || cli->cl_import->imp_invalid)
2236 return -EIO;
2237
2238 if (!list_empty(&oap->oap_pending_item) ||
2239 !list_empty(&oap->oap_rpc_item))
2240 return -EBUSY;
2241
2242 /* Set the OBD_BRW_SRVLOCK before the page is queued. */
2243 brw_flags |= ops->ops_srvlock ? OBD_BRW_SRVLOCK : 0;
2244 if (!client_is_remote(osc_export(osc)) &&
2245 capable(CFS_CAP_SYS_RESOURCE)) {
2246 brw_flags |= OBD_BRW_NOQUOTA;
2247 cmd |= OBD_BRW_NOQUOTA;
2248 }
2249
2250 /* check if the file's owner/group is over quota */
2251 if (!(cmd & OBD_BRW_NOQUOTA)) {
2252 struct cl_object *obj;
2253 struct cl_attr *attr;
2254 unsigned int qid[MAXQUOTAS];
2255
2256 obj = cl_object_top(&osc->oo_cl);
2257 attr = &osc_env_info(env)->oti_attr;
2258
2259 cl_object_attr_lock(obj);
2260 rc = cl_object_attr_get(env, obj, attr);
2261 cl_object_attr_unlock(obj);
2262
2263 qid[USRQUOTA] = attr->cat_uid;
2264 qid[GRPQUOTA] = attr->cat_gid;
2265 if (rc == 0 && osc_quota_chkdq(cli, qid) == NO_QUOTA)
2266 rc = -EDQUOT;
2267 if (rc)
2268 return rc;
2269 }
2270
2271 oap->oap_cmd = cmd;
2272 oap->oap_page_off = ops->ops_from;
2273 oap->oap_count = ops->ops_to - ops->ops_from;
2274 oap->oap_async_flags = 0;
2275 oap->oap_brw_flags = brw_flags;
2276
2277 OSC_IO_DEBUG(osc, "oap %p page %p added for cmd %d\n",
2278 oap, oap->oap_page, oap->oap_cmd & OBD_BRW_RWMASK);
2279
2280 index = oap2cl_page(oap)->cp_index;
2281
2282 /* Add this page into extent by the following steps:
2283 * 1. if there exists an active extent for this IO, mostly this page
2284 * can be added to the active extent and sometimes we need to
2285 * expand extent to accommodate this page;
2286 * 2. otherwise, a new extent will be allocated. */
2287
2288 ext = oio->oi_active;
2289 if (ext != NULL && ext->oe_start <= index && ext->oe_max_end >= index) {
2290 /* one chunk plus extent overhead must be enough to write this
2291 * page */
2292 grants = (1 << cli->cl_chunkbits) + cli->cl_extent_tax;
2293 if (ext->oe_end >= index)
2294 grants = 0;
2295
2296 /* it doesn't need any grant to dirty this page */
2297 client_obd_list_lock(&cli->cl_loi_list_lock);
2298 rc = osc_enter_cache_try(cli, oap, grants, 0);
2299 client_obd_list_unlock(&cli->cl_loi_list_lock);
2300 if (rc == 0) { /* try failed */
2301 grants = 0;
2302 need_release = 1;
2303 } else if (ext->oe_end < index) {
2304 int tmp = grants;
2305 /* try to expand this extent */
2306 rc = osc_extent_expand(ext, index, &tmp);
2307 if (rc < 0) {
2308 need_release = 1;
2309 /* don't free reserved grant */
2310 } else {
2311 OSC_EXTENT_DUMP(D_CACHE, ext,
2312 "expanded for %lu.\n", index);
2313 osc_unreserve_grant(cli, grants, tmp);
2314 grants = 0;
2315 }
2316 }
2317 rc = 0;
2318 } else if (ext != NULL) {
2319 /* index is located outside of active extent */
2320 need_release = 1;
2321 }
2322 if (need_release) {
2323 osc_extent_release(env, ext);
2324 oio->oi_active = NULL;
2325 ext = NULL;
2326 }
2327
2328 if (ext == NULL) {
2329 int tmp = (1 << cli->cl_chunkbits) + cli->cl_extent_tax;
2330
2331 /* try to find new extent to cover this page */
2332 LASSERT(oio->oi_active == NULL);
2333 /* we may have allocated grant for this page if we failed
2334 * to expand the previous active extent. */
2335 LASSERT(ergo(grants > 0, grants >= tmp));
2336
2337 rc = 0;
2338 if (grants == 0) {
2339 /* we haven't allocated grant for this page. */
2340 rc = osc_enter_cache(env, cli, oap, tmp);
2341 if (rc == 0)
2342 grants = tmp;
2343 }
2344
2345 tmp = grants;
2346 if (rc == 0) {
2347 ext = osc_extent_find(env, osc, index, &tmp);
2348 if (IS_ERR(ext)) {
2349 LASSERT(tmp == grants);
2350 osc_exit_cache(cli, oap);
2351 rc = PTR_ERR(ext);
2352 ext = NULL;
2353 } else {
2354 oio->oi_active = ext;
2355 }
2356 }
2357 if (grants > 0)
2358 osc_unreserve_grant(cli, grants, tmp);
2359 }
2360
2361 LASSERT(ergo(rc == 0, ext != NULL));
2362 if (ext != NULL) {
2363 EASSERTF(ext->oe_end >= index && ext->oe_start <= index,
2364 ext, "index = %lu.\n", index);
2365 LASSERT((oap->oap_brw_flags & OBD_BRW_FROM_GRANT) != 0);
2366
2367 osc_object_lock(osc);
2368 if (ext->oe_nr_pages == 0)
2369 ext->oe_srvlock = ops->ops_srvlock;
2370 else
2371 LASSERT(ext->oe_srvlock == ops->ops_srvlock);
2372 ++ext->oe_nr_pages;
2373 list_add_tail(&oap->oap_pending_item, &ext->oe_pages);
2374 osc_object_unlock(osc);
2375 }
2376 return rc;
2377 }
2378
2379 int osc_teardown_async_page(const struct lu_env *env,
2380 struct osc_object *obj, struct osc_page *ops)
2381 {
2382 struct osc_async_page *oap = &ops->ops_oap;
2383 struct osc_extent *ext = NULL;
2384 int rc = 0;
2385
2386 LASSERT(oap->oap_magic == OAP_MAGIC);
2387
2388 CDEBUG(D_INFO, "teardown oap %p page %p at index %lu.\n",
2389 oap, ops, oap2cl_page(oap)->cp_index);
2390
2391 osc_object_lock(obj);
2392 if (!list_empty(&oap->oap_rpc_item)) {
2393 CDEBUG(D_CACHE, "oap %p is not in cache.\n", oap);
2394 rc = -EBUSY;
2395 } else if (!list_empty(&oap->oap_pending_item)) {
2396 ext = osc_extent_lookup(obj, oap2cl_page(oap)->cp_index);
2397 /* only truncated pages are allowed to be taken out.
2398 * See osc_extent_truncate() and osc_cache_truncate_start()
2399 * for details. */
2400 if (ext != NULL && ext->oe_state != OES_TRUNC) {
2401 OSC_EXTENT_DUMP(D_ERROR, ext, "trunc at %lu.\n",
2402 oap2cl_page(oap)->cp_index);
2403 rc = -EBUSY;
2404 }
2405 }
2406 osc_object_unlock(obj);
2407 if (ext != NULL)
2408 osc_extent_put(env, ext);
2409 return rc;
2410 }
2411
2412 /**
2413 * This is called when a page is picked up by kernel to write out.
2414 *
2415 * We should find out the corresponding extent and add the whole extent
2416 * into urgent list. The extent may be being truncated or used, handle it
2417 * carefully.
2418 */
2419 int osc_flush_async_page(const struct lu_env *env, struct cl_io *io,
2420 struct osc_page *ops)
2421 {
2422 struct osc_extent *ext = NULL;
2423 struct osc_object *obj = cl2osc(ops->ops_cl.cpl_obj);
2424 struct cl_page *cp = ops->ops_cl.cpl_page;
2425 pgoff_t index = cp->cp_index;
2426 struct osc_async_page *oap = &ops->ops_oap;
2427 bool unplug = false;
2428 int rc = 0;
2429
2430 osc_object_lock(obj);
2431 ext = osc_extent_lookup(obj, index);
2432 if (ext == NULL) {
2433 osc_extent_tree_dump(D_ERROR, obj);
2434 LASSERTF(0, "page index %lu is NOT covered.\n", index);
2435 }
2436
2437 switch (ext->oe_state) {
2438 case OES_RPC:
2439 case OES_LOCK_DONE:
2440 CL_PAGE_DEBUG(D_ERROR, env, cl_page_top(cp),
2441 "flush an in-rpc page?\n");
2442 LASSERT(0);
2443 break;
2444 case OES_LOCKING:
2445 /* If we know this extent is being written out, we should abort
2446 * so that the writer can make this page ready. Otherwise, there
2447 * exists a deadlock problem because other process can wait for
2448 * page writeback bit holding page lock; and meanwhile in
2449 * vvp_page_make_ready(), we need to grab page lock before
2450 * really sending the RPC. */
2451 case OES_TRUNC:
2452 /* race with truncate, page will be redirtied */
2453 case OES_ACTIVE:
2454 /* The extent is active so we need to abort and let the caller
2455 * re-dirty the page. If we continued on here, and we were the
2456 * one making the extent active, we could deadlock waiting for
2457 * the page writeback to clear but it won't because the extent
2458 * is active and won't be written out. */
2459 rc = -EAGAIN;
2460 goto out;
2461 default:
2462 break;
2463 }
2464
2465 rc = cl_page_prep(env, io, cl_page_top(cp), CRT_WRITE);
2466 if (rc)
2467 goto out;
2468
2469 spin_lock(&oap->oap_lock);
2470 oap->oap_async_flags |= ASYNC_READY|ASYNC_URGENT;
2471 spin_unlock(&oap->oap_lock);
2472
2473 if (memory_pressure_get())
2474 ext->oe_memalloc = 1;
2475
2476 ext->oe_urgent = 1;
2477 if (ext->oe_state == OES_CACHE) {
2478 OSC_EXTENT_DUMP(D_CACHE, ext,
2479 "flush page %p make it urgent.\n", oap);
2480 if (list_empty(&ext->oe_link))
2481 list_add_tail(&ext->oe_link, &obj->oo_urgent_exts);
2482 unplug = true;
2483 }
2484 rc = 0;
2485
2486 out:
2487 osc_object_unlock(obj);
2488 osc_extent_put(env, ext);
2489 if (unplug)
2490 osc_io_unplug_async(env, osc_cli(obj), obj);
2491 return rc;
2492 }
2493
2494 /**
2495 * this is called when a sync waiter receives an interruption. Its job is to
2496 * get the caller woken as soon as possible. If its page hasn't been put in an
2497 * rpc yet it can dequeue immediately. Otherwise it has to mark the rpc as
2498 * desiring interruption which will forcefully complete the rpc once the rpc
2499 * has timed out.
2500 */
2501 int osc_cancel_async_page(const struct lu_env *env, struct osc_page *ops)
2502 {
2503 struct osc_async_page *oap = &ops->ops_oap;
2504 struct osc_object *obj = oap->oap_obj;
2505 struct client_obd *cli = osc_cli(obj);
2506 struct osc_extent *ext;
2507 struct osc_extent *found = NULL;
2508 struct list_head *plist;
2509 pgoff_t index = oap2cl_page(oap)->cp_index;
2510 int rc = -EBUSY;
2511 int cmd;
2512
2513 LASSERT(!oap->oap_interrupted);
2514 oap->oap_interrupted = 1;
2515
2516 /* Find out the caching extent */
2517 osc_object_lock(obj);
2518 if (oap->oap_cmd & OBD_BRW_WRITE) {
2519 plist = &obj->oo_urgent_exts;
2520 cmd = OBD_BRW_WRITE;
2521 } else {
2522 plist = &obj->oo_reading_exts;
2523 cmd = OBD_BRW_READ;
2524 }
2525 list_for_each_entry(ext, plist, oe_link) {
2526 if (ext->oe_start <= index && ext->oe_end >= index) {
2527 LASSERT(ext->oe_state == OES_LOCK_DONE);
2528 /* For OES_LOCK_DONE state extent, it has already held
2529 * a refcount for RPC. */
2530 found = osc_extent_get(ext);
2531 break;
2532 }
2533 }
2534 if (found != NULL) {
2535 list_del_init(&found->oe_link);
2536 osc_update_pending(obj, cmd, -found->oe_nr_pages);
2537 osc_object_unlock(obj);
2538
2539 osc_extent_finish(env, found, 0, -EINTR);
2540 osc_extent_put(env, found);
2541 rc = 0;
2542 } else {
2543 osc_object_unlock(obj);
2544 /* ok, it's been put in an rpc. only one oap gets a request
2545 * reference */
2546 if (oap->oap_request != NULL) {
2547 ptlrpc_mark_interrupted(oap->oap_request);
2548 ptlrpcd_wake(oap->oap_request);
2549 ptlrpc_req_finished(oap->oap_request);
2550 oap->oap_request = NULL;
2551 }
2552 }
2553
2554 osc_list_maint(cli, obj);
2555 return rc;
2556 }
2557
2558 int osc_queue_sync_pages(const struct lu_env *env, struct osc_object *obj,
2559 struct list_head *list, int cmd, int brw_flags)
2560 {
2561 struct client_obd *cli = osc_cli(obj);
2562 struct osc_extent *ext;
2563 struct osc_async_page *oap, *tmp;
2564 int page_count = 0;
2565 int mppr = cli->cl_max_pages_per_rpc;
2566 pgoff_t start = CL_PAGE_EOF;
2567 pgoff_t end = 0;
2568
2569 list_for_each_entry(oap, list, oap_pending_item) {
2570 struct cl_page *cp = oap2cl_page(oap);
2571 if (cp->cp_index > end)
2572 end = cp->cp_index;
2573 if (cp->cp_index < start)
2574 start = cp->cp_index;
2575 ++page_count;
2576 mppr <<= (page_count > mppr);
2577 }
2578
2579 ext = osc_extent_alloc(obj);
2580 if (ext == NULL) {
2581 list_for_each_entry_safe(oap, tmp, list, oap_pending_item) {
2582 list_del_init(&oap->oap_pending_item);
2583 osc_ap_completion(env, cli, oap, 0, -ENOMEM);
2584 }
2585 return -ENOMEM;
2586 }
2587
2588 ext->oe_rw = !!(cmd & OBD_BRW_READ);
2589 ext->oe_urgent = 1;
2590 ext->oe_start = start;
2591 ext->oe_end = ext->oe_max_end = end;
2592 ext->oe_obj = obj;
2593 ext->oe_srvlock = !!(brw_flags & OBD_BRW_SRVLOCK);
2594 ext->oe_nr_pages = page_count;
2595 ext->oe_mppr = mppr;
2596 list_splice_init(list, &ext->oe_pages);
2597
2598 osc_object_lock(obj);
2599 /* Reuse the initial refcount for RPC, don't drop it */
2600 osc_extent_state_set(ext, OES_LOCK_DONE);
2601 if (cmd & OBD_BRW_WRITE) {
2602 list_add_tail(&ext->oe_link, &obj->oo_urgent_exts);
2603 osc_update_pending(obj, OBD_BRW_WRITE, page_count);
2604 } else {
2605 list_add_tail(&ext->oe_link, &obj->oo_reading_exts);
2606 osc_update_pending(obj, OBD_BRW_READ, page_count);
2607 }
2608 osc_object_unlock(obj);
2609
2610 osc_io_unplug_async(env, cli, obj);
2611 return 0;
2612 }
2613
2614 /**
2615 * Called by osc_io_setattr_start() to freeze and destroy covering extents.
2616 */
2617 int osc_cache_truncate_start(const struct lu_env *env, struct osc_io *oio,
2618 struct osc_object *obj, __u64 size)
2619 {
2620 struct client_obd *cli = osc_cli(obj);
2621 struct osc_extent *ext;
2622 struct osc_extent *waiting = NULL;
2623 pgoff_t index;
2624 LIST_HEAD(list);
2625 int result = 0;
2626 bool partial;
2627
2628 /* pages with index greater or equal to index will be truncated. */
2629 index = cl_index(osc2cl(obj), size);
2630 partial = size > cl_offset(osc2cl(obj), index);
2631
2632 again:
2633 osc_object_lock(obj);
2634 ext = osc_extent_search(obj, index);
2635 if (ext == NULL)
2636 ext = first_extent(obj);
2637 else if (ext->oe_end < index)
2638 ext = next_extent(ext);
2639 while (ext != NULL) {
2640 EASSERT(ext->oe_state != OES_TRUNC, ext);
2641
2642 if (ext->oe_state > OES_CACHE || ext->oe_urgent) {
2643 /* if ext is in urgent state, it means there must exist
2644 * a page already having been flushed by write_page().
2645 * We have to wait for this extent because we can't
2646 * truncate that page. */
2647 LASSERT(!ext->oe_hp);
2648 OSC_EXTENT_DUMP(D_CACHE, ext,
2649 "waiting for busy extent\n");
2650 waiting = osc_extent_get(ext);
2651 break;
2652 }
2653
2654 OSC_EXTENT_DUMP(D_CACHE, ext, "try to trunc:%llu.\n", size);
2655
2656 osc_extent_get(ext);
2657 if (ext->oe_state == OES_ACTIVE) {
2658 /* though we grab inode mutex for write path, but we
2659 * release it before releasing extent(in osc_io_end()),
2660 * so there is a race window that an extent is still
2661 * in OES_ACTIVE when truncate starts. */
2662 LASSERT(!ext->oe_trunc_pending);
2663 ext->oe_trunc_pending = 1;
2664 } else {
2665 EASSERT(ext->oe_state == OES_CACHE, ext);
2666 osc_extent_state_set(ext, OES_TRUNC);
2667 osc_update_pending(obj, OBD_BRW_WRITE,
2668 -ext->oe_nr_pages);
2669 }
2670 EASSERT(list_empty(&ext->oe_link), ext);
2671 list_add_tail(&ext->oe_link, &list);
2672
2673 ext = next_extent(ext);
2674 }
2675 osc_object_unlock(obj);
2676
2677 osc_list_maint(cli, obj);
2678
2679 while (!list_empty(&list)) {
2680 int rc;
2681
2682 ext = list_entry(list.next, struct osc_extent, oe_link);
2683 list_del_init(&ext->oe_link);
2684
2685 /* extent may be in OES_ACTIVE state because inode mutex
2686 * is released before osc_io_end() in file write case */
2687 if (ext->oe_state != OES_TRUNC)
2688 osc_extent_wait(env, ext, OES_TRUNC);
2689
2690 rc = osc_extent_truncate(ext, index, partial);
2691 if (rc < 0) {
2692 if (result == 0)
2693 result = rc;
2694
2695 OSC_EXTENT_DUMP(D_ERROR, ext,
2696 "truncate error %d\n", rc);
2697 } else if (ext->oe_nr_pages == 0) {
2698 osc_extent_remove(ext);
2699 } else {
2700 /* this must be an overlapped extent which means only
2701 * part of pages in this extent have been truncated.
2702 */
2703 EASSERTF(ext->oe_start <= index, ext,
2704 "trunc index = %lu/%d.\n", index, partial);
2705 /* fix index to skip this partially truncated extent */
2706 index = ext->oe_end + 1;
2707 partial = false;
2708
2709 /* we need to hold this extent in OES_TRUNC state so
2710 * that no writeback will happen. This is to avoid
2711 * BUG 17397. */
2712 LASSERT(oio->oi_trunc == NULL);
2713 oio->oi_trunc = osc_extent_get(ext);
2714 OSC_EXTENT_DUMP(D_CACHE, ext,
2715 "trunc at %llu\n", size);
2716 }
2717 osc_extent_put(env, ext);
2718 }
2719 if (waiting != NULL) {
2720 int rc;
2721
2722 /* ignore the result of osc_extent_wait the write initiator
2723 * should take care of it. */
2724 rc = osc_extent_wait(env, waiting, OES_INV);
2725 if (rc < 0)
2726 OSC_EXTENT_DUMP(D_CACHE, waiting, "error: %d.\n", rc);
2727
2728 osc_extent_put(env, waiting);
2729 waiting = NULL;
2730 goto again;
2731 }
2732 return result;
2733 }
2734
2735 /**
2736 * Called after osc_io_setattr_end to add oio->oi_trunc back to cache.
2737 */
2738 void osc_cache_truncate_end(const struct lu_env *env, struct osc_io *oio,
2739 struct osc_object *obj)
2740 {
2741 struct osc_extent *ext = oio->oi_trunc;
2742
2743 oio->oi_trunc = NULL;
2744 if (ext != NULL) {
2745 bool unplug = false;
2746
2747 EASSERT(ext->oe_nr_pages > 0, ext);
2748 EASSERT(ext->oe_state == OES_TRUNC, ext);
2749 EASSERT(!ext->oe_urgent, ext);
2750
2751 OSC_EXTENT_DUMP(D_CACHE, ext, "trunc -> cache.\n");
2752 osc_object_lock(obj);
2753 osc_extent_state_set(ext, OES_CACHE);
2754 if (ext->oe_fsync_wait && !ext->oe_urgent) {
2755 ext->oe_urgent = 1;
2756 list_move_tail(&ext->oe_link, &obj->oo_urgent_exts);
2757 unplug = true;
2758 }
2759 osc_update_pending(obj, OBD_BRW_WRITE, ext->oe_nr_pages);
2760 osc_object_unlock(obj);
2761 osc_extent_put(env, ext);
2762
2763 if (unplug)
2764 osc_io_unplug_async(env, osc_cli(obj), obj);
2765 }
2766 }
2767
2768 /**
2769 * Wait for extents in a specific range to be written out.
2770 * The caller must have called osc_cache_writeback_range() to issue IO
2771 * otherwise it will take a long time for this function to finish.
2772 *
2773 * Caller must hold inode_mutex , or cancel exclusive dlm lock so that
2774 * nobody else can dirty this range of file while we're waiting for
2775 * extents to be written.
2776 */
2777 int osc_cache_wait_range(const struct lu_env *env, struct osc_object *obj,
2778 pgoff_t start, pgoff_t end)
2779 {
2780 struct osc_extent *ext;
2781 pgoff_t index = start;
2782 int result = 0;
2783
2784 again:
2785 osc_object_lock(obj);
2786 ext = osc_extent_search(obj, index);
2787 if (ext == NULL)
2788 ext = first_extent(obj);
2789 else if (ext->oe_end < index)
2790 ext = next_extent(ext);
2791 while (ext != NULL) {
2792 int rc;
2793
2794 if (ext->oe_start > end)
2795 break;
2796
2797 if (!ext->oe_fsync_wait) {
2798 ext = next_extent(ext);
2799 continue;
2800 }
2801
2802 EASSERT(ergo(ext->oe_state == OES_CACHE,
2803 ext->oe_hp || ext->oe_urgent), ext);
2804 EASSERT(ergo(ext->oe_state == OES_ACTIVE,
2805 !ext->oe_hp && ext->oe_urgent), ext);
2806
2807 index = ext->oe_end + 1;
2808 osc_extent_get(ext);
2809 osc_object_unlock(obj);
2810
2811 rc = osc_extent_wait(env, ext, OES_INV);
2812 if (result == 0)
2813 result = rc;
2814 osc_extent_put(env, ext);
2815 goto again;
2816 }
2817 osc_object_unlock(obj);
2818
2819 OSC_IO_DEBUG(obj, "sync file range.\n");
2820 return result;
2821 }
2822
2823 /**
2824 * Called to write out a range of osc object.
2825 *
2826 * @hp : should be set this is caused by lock cancel;
2827 * @discard: is set if dirty pages should be dropped - file will be deleted or
2828 * truncated, this implies there is no partially discarding extents.
2829 *
2830 * Return how many pages will be issued, or error code if error occurred.
2831 */
2832 int osc_cache_writeback_range(const struct lu_env *env, struct osc_object *obj,
2833 pgoff_t start, pgoff_t end, int hp, int discard)
2834 {
2835 struct osc_extent *ext;
2836 LIST_HEAD(discard_list);
2837 bool unplug = false;
2838 int result = 0;
2839
2840 osc_object_lock(obj);
2841 ext = osc_extent_search(obj, start);
2842 if (ext == NULL)
2843 ext = first_extent(obj);
2844 else if (ext->oe_end < start)
2845 ext = next_extent(ext);
2846 while (ext != NULL) {
2847 if (ext->oe_start > end)
2848 break;
2849
2850 ext->oe_fsync_wait = 1;
2851 switch (ext->oe_state) {
2852 case OES_CACHE:
2853 result += ext->oe_nr_pages;
2854 if (!discard) {
2855 struct list_head *list = NULL;
2856 if (hp) {
2857 EASSERT(!ext->oe_hp, ext);
2858 ext->oe_hp = 1;
2859 list = &obj->oo_hp_exts;
2860 } else if (!ext->oe_urgent) {
2861 ext->oe_urgent = 1;
2862 list = &obj->oo_urgent_exts;
2863 }
2864 if (list != NULL)
2865 list_move_tail(&ext->oe_link, list);
2866 unplug = true;
2867 } else {
2868 /* the only discarder is lock cancelling, so
2869 * [start, end] must contain this extent */
2870 EASSERT(ext->oe_start >= start &&
2871 ext->oe_max_end <= end, ext);
2872 osc_extent_state_set(ext, OES_LOCKING);
2873 ext->oe_owner = current;
2874 list_move_tail(&ext->oe_link,
2875 &discard_list);
2876 osc_update_pending(obj, OBD_BRW_WRITE,
2877 -ext->oe_nr_pages);
2878 }
2879 break;
2880 case OES_ACTIVE:
2881 /* It's pretty bad to wait for ACTIVE extents, because
2882 * we don't know how long we will wait for it to be
2883 * flushed since it may be blocked at awaiting more
2884 * grants. We do this for the correctness of fsync. */
2885 LASSERT(hp == 0 && discard == 0);
2886 ext->oe_urgent = 1;
2887 break;
2888 case OES_TRUNC:
2889 /* this extent is being truncated, can't do anything
2890 * for it now. it will be set to urgent after truncate
2891 * is finished in osc_cache_truncate_end(). */
2892 default:
2893 break;
2894 }
2895 ext = next_extent(ext);
2896 }
2897 osc_object_unlock(obj);
2898
2899 LASSERT(ergo(!discard, list_empty(&discard_list)));
2900 if (!list_empty(&discard_list)) {
2901 struct osc_extent *tmp;
2902 int rc;
2903
2904 osc_list_maint(osc_cli(obj), obj);
2905 list_for_each_entry_safe(ext, tmp, &discard_list, oe_link) {
2906 list_del_init(&ext->oe_link);
2907 EASSERT(ext->oe_state == OES_LOCKING, ext);
2908
2909 /* Discard caching pages. We don't actually write this
2910 * extent out but we complete it as if we did. */
2911 rc = osc_extent_make_ready(env, ext);
2912 if (unlikely(rc < 0)) {
2913 OSC_EXTENT_DUMP(D_ERROR, ext,
2914 "make_ready returned %d\n", rc);
2915 if (result >= 0)
2916 result = rc;
2917 }
2918
2919 /* finish the extent as if the pages were sent */
2920 osc_extent_finish(env, ext, 0, 0);
2921 }
2922 }
2923
2924 if (unplug)
2925 osc_io_unplug(env, osc_cli(obj), obj, PDL_POLICY_ROUND);
2926
2927 if (hp || discard) {
2928 int rc;
2929 rc = osc_cache_wait_range(env, obj, start, end);
2930 if (result >= 0 && rc < 0)
2931 result = rc;
2932 }
2933
2934 OSC_IO_DEBUG(obj, "cache page out.\n");
2935 return result;
2936 }
2937
2938 /** @} osc */
Linux kernel - Deliverables
This page took 0.095392 seconds and 4 git commands to generate.