-
Notifications
You must be signed in to change notification settings - Fork 861
/
subr_pool.c
2289 lines (1892 loc) · 51.8 KB
/
subr_pool.c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/* $OpenBSD: subr_pool.c,v 1.230 2020/01/24 06:31:17 cheloha Exp $ */
/* $NetBSD: subr_pool.c,v 1.61 2001/09/26 07:14:56 chs Exp $ */
/*-
* Copyright (c) 1997, 1999, 2000 The NetBSD Foundation, Inc.
* All rights reserved.
*
* This code is derived from software contributed to The NetBSD Foundation
* by Paul Kranenburg; by Jason R. Thorpe of the Numerical Aerospace
* Simulation Facility, NASA Ames Research Center.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
* ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
* TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/errno.h>
#include <sys/kernel.h>
#include <sys/malloc.h>
#include <sys/pool.h>
#include <sys/proc.h>
#include <sys/syslog.h>
#include <sys/sysctl.h>
#include <sys/task.h>
#include <sys/timeout.h>
#include <sys/percpu.h>
#include <uvm/uvm_extern.h>
/*
* Pool resource management utility.
*
* Memory is allocated in pages which are split into pieces according to
* the pool item size. Each page is kept on one of three lists in the
* pool structure: `pr_emptypages', `pr_fullpages' and `pr_partpages',
* for empty, full and partially-full pages respectively. The individual
* pool items are on a linked list headed by `ph_items' in each page
* header. The memory for building the page list is either taken from
* the allocated pages themselves (for small pool items) or taken from
* an internal pool of page headers (`phpool').
*/
/* List of all pools */
SIMPLEQ_HEAD(,pool) pool_head = SIMPLEQ_HEAD_INITIALIZER(pool_head);
/*
* Every pool gets a unique serial number assigned to it. If this counter
* wraps, we're screwed, but we shouldn't create so many pools anyway.
*/
unsigned int pool_serial;
unsigned int pool_count;
/* Lock the previous variables making up the global pool state */
struct rwlock pool_lock = RWLOCK_INITIALIZER("pools");
/* Private pool for page header structures */
struct pool phpool;
struct pool_lock_ops {
void (*pl_init)(struct pool *, union pool_lock *,
const struct lock_type *);
void (*pl_enter)(union pool_lock *);
int (*pl_enter_try)(union pool_lock *);
void (*pl_leave)(union pool_lock *);
void (*pl_assert_locked)(union pool_lock *);
void (*pl_assert_unlocked)(union pool_lock *);
int (*pl_sleep)(void *, union pool_lock *, int, const char *);
};
static const struct pool_lock_ops pool_lock_ops_mtx;
static const struct pool_lock_ops pool_lock_ops_rw;
#ifdef WITNESS
#define pl_init(pp, pl) do { \
static const struct lock_type __lock_type = { .lt_name = #pl }; \
(pp)->pr_lock_ops->pl_init(pp, pl, &__lock_type); \
} while (0)
#else /* WITNESS */
#define pl_init(pp, pl) (pp)->pr_lock_ops->pl_init(pp, pl, NULL)
#endif /* WITNESS */
static inline void
pl_enter(struct pool *pp, union pool_lock *pl)
{
pp->pr_lock_ops->pl_enter(pl);
}
static inline int
pl_enter_try(struct pool *pp, union pool_lock *pl)
{
return pp->pr_lock_ops->pl_enter_try(pl);
}
static inline void
pl_leave(struct pool *pp, union pool_lock *pl)
{
pp->pr_lock_ops->pl_leave(pl);
}
static inline void
pl_assert_locked(struct pool *pp, union pool_lock *pl)
{
pp->pr_lock_ops->pl_assert_locked(pl);
}
static inline void
pl_assert_unlocked(struct pool *pp, union pool_lock *pl)
{
pp->pr_lock_ops->pl_assert_unlocked(pl);
}
static inline int
pl_sleep(struct pool *pp, void *ident, union pool_lock *lock, int priority,
const char *wmesg)
{
return pp->pr_lock_ops->pl_sleep(ident, lock, priority, wmesg);
}
struct pool_item {
u_long pi_magic;
XSIMPLEQ_ENTRY(pool_item) pi_list;
};
#define POOL_IMAGIC(ph, pi) ((u_long)(pi) ^ (ph)->ph_magic)
struct pool_page_header {
/* Page headers */
TAILQ_ENTRY(pool_page_header)
ph_entry; /* pool page list */
XSIMPLEQ_HEAD(, pool_item)
ph_items; /* free items on the page */
RBT_ENTRY(pool_page_header)
ph_node; /* off-page page headers */
unsigned int ph_nmissing; /* # of chunks in use */
caddr_t ph_page; /* this page's address */
caddr_t ph_colored; /* page's colored address */
unsigned long ph_magic;
int ph_tick;
};
#define POOL_MAGICBIT (1 << 3) /* keep away from perturbed low bits */
#define POOL_PHPOISON(ph) ISSET((ph)->ph_magic, POOL_MAGICBIT)
#ifdef MULTIPROCESSOR
struct pool_cache_item {
struct pool_cache_item *ci_next; /* next item in list */
unsigned long ci_nitems; /* number of items in list */
TAILQ_ENTRY(pool_cache_item)
ci_nextl; /* entry in list of lists */
};
/* we store whether the cached item is poisoned in the high bit of nitems */
#define POOL_CACHE_ITEM_NITEMS_MASK 0x7ffffffUL
#define POOL_CACHE_ITEM_NITEMS_POISON 0x8000000UL
#define POOL_CACHE_ITEM_NITEMS(_ci) \
((_ci)->ci_nitems & POOL_CACHE_ITEM_NITEMS_MASK)
#define POOL_CACHE_ITEM_POISONED(_ci) \
ISSET((_ci)->ci_nitems, POOL_CACHE_ITEM_NITEMS_POISON)
struct pool_cache {
struct pool_cache_item *pc_actv; /* active list of items */
unsigned long pc_nactv; /* actv head nitems cache */
struct pool_cache_item *pc_prev; /* previous list of items */
uint64_t pc_gen; /* generation number */
uint64_t pc_nget; /* # of successful requests */
uint64_t pc_nfail; /* # of unsuccessful reqs */
uint64_t pc_nput; /* # of releases */
uint64_t pc_nlget; /* # of list requests */
uint64_t pc_nlfail; /* # of fails getting a list */
uint64_t pc_nlput; /* # of list releases */
int pc_nout;
};
void *pool_cache_get(struct pool *);
void pool_cache_put(struct pool *, void *);
void pool_cache_destroy(struct pool *);
void pool_cache_gc(struct pool *);
#endif
void pool_cache_pool_info(struct pool *, struct kinfo_pool *);
int pool_cache_info(struct pool *, void *, size_t *);
int pool_cache_cpus_info(struct pool *, void *, size_t *);
#ifdef POOL_DEBUG
int pool_debug = 1;
#else
int pool_debug = 0;
#endif
#define POOL_INPGHDR(pp) ((pp)->pr_phoffset != 0)
struct pool_page_header *
pool_p_alloc(struct pool *, int, int *);
void pool_p_insert(struct pool *, struct pool_page_header *);
void pool_p_remove(struct pool *, struct pool_page_header *);
void pool_p_free(struct pool *, struct pool_page_header *);
void pool_update_curpage(struct pool *);
void *pool_do_get(struct pool *, int, int *);
void pool_do_put(struct pool *, void *);
int pool_chk_page(struct pool *, struct pool_page_header *, int);
int pool_chk(struct pool *);
void pool_get_done(struct pool *, void *, void *);
void pool_runqueue(struct pool *, int);
void *pool_allocator_alloc(struct pool *, int, int *);
void pool_allocator_free(struct pool *, void *);
/*
* The default pool allocator.
*/
void *pool_page_alloc(struct pool *, int, int *);
void pool_page_free(struct pool *, void *);
/*
* safe for interrupts; this is the default allocator
*/
struct pool_allocator pool_allocator_single = {
pool_page_alloc,
pool_page_free,
POOL_ALLOC_SIZE(PAGE_SIZE, POOL_ALLOC_ALIGNED)
};
void *pool_multi_alloc(struct pool *, int, int *);
void pool_multi_free(struct pool *, void *);
struct pool_allocator pool_allocator_multi = {
pool_multi_alloc,
pool_multi_free,
POOL_ALLOC_SIZES(PAGE_SIZE, (1UL << 31), POOL_ALLOC_ALIGNED)
};
void *pool_multi_alloc_ni(struct pool *, int, int *);
void pool_multi_free_ni(struct pool *, void *);
struct pool_allocator pool_allocator_multi_ni = {
pool_multi_alloc_ni,
pool_multi_free_ni,
POOL_ALLOC_SIZES(PAGE_SIZE, (1UL << 31), POOL_ALLOC_ALIGNED)
};
#ifdef DDB
void pool_print_pagelist(struct pool_pagelist *, int (*)(const char *, ...)
__attribute__((__format__(__kprintf__,1,2))));
void pool_print1(struct pool *, const char *, int (*)(const char *, ...)
__attribute__((__format__(__kprintf__,1,2))));
#endif
/* stale page garbage collectors */
void pool_gc_sched(void *);
struct timeout pool_gc_tick = TIMEOUT_INITIALIZER(pool_gc_sched, NULL);
void pool_gc_pages(void *);
struct task pool_gc_task = TASK_INITIALIZER(pool_gc_pages, NULL);
int pool_wait_free = 1;
int pool_wait_gc = 8;
RBT_PROTOTYPE(phtree, pool_page_header, ph_node, phtree_compare);
static inline int
phtree_compare(const struct pool_page_header *a,
const struct pool_page_header *b)
{
vaddr_t va = (vaddr_t)a->ph_page;
vaddr_t vb = (vaddr_t)b->ph_page;
/* the compares in this order are important for the NFIND to work */
if (vb < va)
return (-1);
if (vb > va)
return (1);
return (0);
}
RBT_GENERATE(phtree, pool_page_header, ph_node, phtree_compare);
/*
* Return the pool page header based on page address.
*/
static inline struct pool_page_header *
pr_find_pagehead(struct pool *pp, void *v)
{
struct pool_page_header *ph, key;
if (POOL_INPGHDR(pp)) {
caddr_t page;
page = (caddr_t)((vaddr_t)v & pp->pr_pgmask);
return ((struct pool_page_header *)(page + pp->pr_phoffset));
}
key.ph_page = v;
ph = RBT_NFIND(phtree, &pp->pr_phtree, &key);
if (ph == NULL)
panic("%s: %s: page header missing", __func__, pp->pr_wchan);
KASSERT(ph->ph_page <= (caddr_t)v);
if (ph->ph_page + pp->pr_pgsize <= (caddr_t)v)
panic("%s: %s: incorrect page", __func__, pp->pr_wchan);
return (ph);
}
/*
* Initialize the given pool resource structure.
*
* We export this routine to allow other kernel parts to declare
* static pools that must be initialized before malloc() is available.
*/
void
pool_init(struct pool *pp, size_t size, u_int align, int ipl, int flags,
const char *wchan, struct pool_allocator *palloc)
{
int off = 0, space;
unsigned int pgsize = PAGE_SIZE, items;
size_t pa_pagesz;
#ifdef DIAGNOSTIC
struct pool *iter;
#endif
if (align == 0)
align = ALIGN(1);
if (size < sizeof(struct pool_item))
size = sizeof(struct pool_item);
size = roundup(size, align);
while (size * 8 > pgsize)
pgsize <<= 1;
if (palloc == NULL) {
if (pgsize > PAGE_SIZE) {
palloc = ISSET(flags, PR_WAITOK) ?
&pool_allocator_multi_ni : &pool_allocator_multi;
} else
palloc = &pool_allocator_single;
pa_pagesz = palloc->pa_pagesz;
} else {
size_t pgsizes;
pa_pagesz = palloc->pa_pagesz;
if (pa_pagesz == 0)
pa_pagesz = POOL_ALLOC_DEFAULT;
pgsizes = pa_pagesz & ~POOL_ALLOC_ALIGNED;
/* make sure the allocator can fit at least one item */
if (size > pgsizes) {
panic("%s: pool %s item size 0x%zx > "
"allocator %p sizes 0x%zx", __func__, wchan,
size, palloc, pgsizes);
}
/* shrink pgsize until it fits into the range */
while (!ISSET(pgsizes, pgsize))
pgsize >>= 1;
}
KASSERT(ISSET(pa_pagesz, pgsize));
items = pgsize / size;
/*
* Decide whether to put the page header off page to avoid
* wasting too large a part of the page. Off-page page headers
* go into an RB tree, so we can match a returned item with
* its header based on the page address.
*/
if (ISSET(pa_pagesz, POOL_ALLOC_ALIGNED)) {
if (pgsize - (size * items) >
sizeof(struct pool_page_header)) {
off = pgsize - sizeof(struct pool_page_header);
} else if (sizeof(struct pool_page_header) * 2 >= size) {
off = pgsize - sizeof(struct pool_page_header);
items = off / size;
}
}
KASSERT(items > 0);
/*
* Initialize the pool structure.
*/
memset(pp, 0, sizeof(*pp));
if (ISSET(flags, PR_RWLOCK)) {
KASSERT(flags & PR_WAITOK);
pp->pr_lock_ops = &pool_lock_ops_rw;
} else
pp->pr_lock_ops = &pool_lock_ops_mtx;
TAILQ_INIT(&pp->pr_emptypages);
TAILQ_INIT(&pp->pr_fullpages);
TAILQ_INIT(&pp->pr_partpages);
pp->pr_curpage = NULL;
pp->pr_npages = 0;
pp->pr_minitems = 0;
pp->pr_minpages = 0;
pp->pr_maxpages = 8;
pp->pr_size = size;
pp->pr_pgsize = pgsize;
pp->pr_pgmask = ~0UL ^ (pgsize - 1);
pp->pr_phoffset = off;
pp->pr_itemsperpage = items;
pp->pr_wchan = wchan;
pp->pr_alloc = palloc;
pp->pr_nitems = 0;
pp->pr_nout = 0;
pp->pr_hardlimit = UINT_MAX;
pp->pr_hardlimit_warning = NULL;
pp->pr_hardlimit_ratecap.tv_sec = 0;
pp->pr_hardlimit_ratecap.tv_usec = 0;
pp->pr_hardlimit_warning_last.tv_sec = 0;
pp->pr_hardlimit_warning_last.tv_usec = 0;
RBT_INIT(phtree, &pp->pr_phtree);
/*
* Use the space between the chunks and the page header
* for cache coloring.
*/
space = POOL_INPGHDR(pp) ? pp->pr_phoffset : pp->pr_pgsize;
space -= pp->pr_itemsperpage * pp->pr_size;
pp->pr_align = align;
pp->pr_maxcolors = (space / align) + 1;
pp->pr_nget = 0;
pp->pr_nfail = 0;
pp->pr_nput = 0;
pp->pr_npagealloc = 0;
pp->pr_npagefree = 0;
pp->pr_hiwat = 0;
pp->pr_nidle = 0;
pp->pr_ipl = ipl;
pp->pr_flags = flags;
pl_init(pp, &pp->pr_lock);
pl_init(pp, &pp->pr_requests_lock);
TAILQ_INIT(&pp->pr_requests);
if (phpool.pr_size == 0) {
pool_init(&phpool, sizeof(struct pool_page_header), 0,
IPL_HIGH, 0, "phpool", NULL);
/* make sure phpool wont "recurse" */
KASSERT(POOL_INPGHDR(&phpool));
}
/* pglistalloc/constraint parameters */
pp->pr_crange = &kp_dirty;
/* Insert this into the list of all pools. */
rw_enter_write(&pool_lock);
#ifdef DIAGNOSTIC
SIMPLEQ_FOREACH(iter, &pool_head, pr_poollist) {
if (iter == pp)
panic("%s: pool %s already on list", __func__, wchan);
}
#endif
pp->pr_serial = ++pool_serial;
if (pool_serial == 0)
panic("%s: too much uptime", __func__);
SIMPLEQ_INSERT_HEAD(&pool_head, pp, pr_poollist);
pool_count++;
rw_exit_write(&pool_lock);
}
/*
* Decommission a pool resource.
*/
void
pool_destroy(struct pool *pp)
{
struct pool_page_header *ph;
struct pool *prev, *iter;
#ifdef MULTIPROCESSOR
if (pp->pr_cache != NULL)
pool_cache_destroy(pp);
#endif
#ifdef DIAGNOSTIC
if (pp->pr_nout != 0)
panic("%s: pool busy: still out: %u", __func__, pp->pr_nout);
#endif
/* Remove from global pool list */
rw_enter_write(&pool_lock);
pool_count--;
if (pp == SIMPLEQ_FIRST(&pool_head))
SIMPLEQ_REMOVE_HEAD(&pool_head, pr_poollist);
else {
prev = SIMPLEQ_FIRST(&pool_head);
SIMPLEQ_FOREACH(iter, &pool_head, pr_poollist) {
if (iter == pp) {
SIMPLEQ_REMOVE_AFTER(&pool_head, prev,
pr_poollist);
break;
}
prev = iter;
}
}
rw_exit_write(&pool_lock);
/* Remove all pages */
while ((ph = TAILQ_FIRST(&pp->pr_emptypages)) != NULL) {
pl_enter(pp, &pp->pr_lock);
pool_p_remove(pp, ph);
pl_leave(pp, &pp->pr_lock);
pool_p_free(pp, ph);
}
KASSERT(TAILQ_EMPTY(&pp->pr_fullpages));
KASSERT(TAILQ_EMPTY(&pp->pr_partpages));
}
void
pool_request_init(struct pool_request *pr,
void (*handler)(struct pool *, void *, void *), void *cookie)
{
pr->pr_handler = handler;
pr->pr_cookie = cookie;
pr->pr_item = NULL;
}
void
pool_request(struct pool *pp, struct pool_request *pr)
{
pl_enter(pp, &pp->pr_requests_lock);
TAILQ_INSERT_TAIL(&pp->pr_requests, pr, pr_entry);
pool_runqueue(pp, PR_NOWAIT);
pl_leave(pp, &pp->pr_requests_lock);
}
struct pool_get_memory {
union pool_lock lock;
void * volatile v;
};
/*
* Grab an item from the pool.
*/
void *
pool_get(struct pool *pp, int flags)
{
void *v = NULL;
int slowdown = 0;
KASSERT(flags & (PR_WAITOK | PR_NOWAIT));
if (pp->pr_flags & PR_RWLOCK)
KASSERT(flags & PR_WAITOK);
#ifdef MULTIPROCESSOR
if (pp->pr_cache != NULL) {
v = pool_cache_get(pp);
if (v != NULL)
goto good;
}
#endif
pl_enter(pp, &pp->pr_lock);
if (pp->pr_nout >= pp->pr_hardlimit) {
if (ISSET(flags, PR_NOWAIT|PR_LIMITFAIL))
goto fail;
} else if ((v = pool_do_get(pp, flags, &slowdown)) == NULL) {
if (ISSET(flags, PR_NOWAIT))
goto fail;
}
pl_leave(pp, &pp->pr_lock);
if ((slowdown || pool_debug == 2) && ISSET(flags, PR_WAITOK))
yield();
if (v == NULL) {
struct pool_get_memory mem = { .v = NULL };
struct pool_request pr;
#ifdef DIAGNOSTIC
if (ISSET(flags, PR_WAITOK) && curproc == &proc0)
panic("%s: cannot sleep for memory during boot",
__func__);
#endif
pl_init(pp, &mem.lock);
pool_request_init(&pr, pool_get_done, &mem);
pool_request(pp, &pr);
pl_enter(pp, &mem.lock);
while (mem.v == NULL)
pl_sleep(pp, &mem, &mem.lock, PSWP, pp->pr_wchan);
pl_leave(pp, &mem.lock);
v = mem.v;
}
#ifdef MULTIPROCESSOR
good:
#endif
if (ISSET(flags, PR_ZERO))
memset(v, 0, pp->pr_size);
return (v);
fail:
pp->pr_nfail++;
pl_leave(pp, &pp->pr_lock);
return (NULL);
}
void
pool_get_done(struct pool *pp, void *xmem, void *v)
{
struct pool_get_memory *mem = xmem;
pl_enter(pp, &mem->lock);
mem->v = v;
pl_leave(pp, &mem->lock);
wakeup_one(mem);
}
void
pool_runqueue(struct pool *pp, int flags)
{
struct pool_requests prl = TAILQ_HEAD_INITIALIZER(prl);
struct pool_request *pr;
pl_assert_unlocked(pp, &pp->pr_lock);
pl_assert_locked(pp, &pp->pr_requests_lock);
if (pp->pr_requesting++)
return;
do {
pp->pr_requesting = 1;
TAILQ_CONCAT(&prl, &pp->pr_requests, pr_entry);
if (TAILQ_EMPTY(&prl))
continue;
pl_leave(pp, &pp->pr_requests_lock);
pl_enter(pp, &pp->pr_lock);
pr = TAILQ_FIRST(&prl);
while (pr != NULL) {
int slowdown = 0;
if (pp->pr_nout >= pp->pr_hardlimit)
break;
pr->pr_item = pool_do_get(pp, flags, &slowdown);
if (pr->pr_item == NULL) /* || slowdown ? */
break;
pr = TAILQ_NEXT(pr, pr_entry);
}
pl_leave(pp, &pp->pr_lock);
while ((pr = TAILQ_FIRST(&prl)) != NULL &&
pr->pr_item != NULL) {
TAILQ_REMOVE(&prl, pr, pr_entry);
(*pr->pr_handler)(pp, pr->pr_cookie, pr->pr_item);
}
pl_enter(pp, &pp->pr_requests_lock);
} while (--pp->pr_requesting);
TAILQ_CONCAT(&pp->pr_requests, &prl, pr_entry);
}
void *
pool_do_get(struct pool *pp, int flags, int *slowdown)
{
struct pool_item *pi;
struct pool_page_header *ph;
pl_assert_locked(pp, &pp->pr_lock);
splassert(pp->pr_ipl);
/*
* Account for this item now to avoid races if we need to give up
* pr_lock to allocate a page.
*/
pp->pr_nout++;
if (pp->pr_curpage == NULL) {
pl_leave(pp, &pp->pr_lock);
ph = pool_p_alloc(pp, flags, slowdown);
pl_enter(pp, &pp->pr_lock);
if (ph == NULL) {
pp->pr_nout--;
return (NULL);
}
pool_p_insert(pp, ph);
}
ph = pp->pr_curpage;
pi = XSIMPLEQ_FIRST(&ph->ph_items);
if (__predict_false(pi == NULL))
panic("%s: %s: page empty", __func__, pp->pr_wchan);
if (__predict_false(pi->pi_magic != POOL_IMAGIC(ph, pi))) {
panic("%s: %s free list modified: "
"page %p; item addr %p; offset 0x%x=0x%lx != 0x%lx",
__func__, pp->pr_wchan, ph->ph_page, pi,
0, pi->pi_magic, POOL_IMAGIC(ph, pi));
}
XSIMPLEQ_REMOVE_HEAD(&ph->ph_items, pi_list);
#ifdef DIAGNOSTIC
if (pool_debug && POOL_PHPOISON(ph)) {
size_t pidx;
uint32_t pval;
if (poison_check(pi + 1, pp->pr_size - sizeof(*pi),
&pidx, &pval)) {
int *ip = (int *)(pi + 1);
panic("%s: %s free list modified: "
"page %p; item addr %p; offset 0x%zx=0x%x",
__func__, pp->pr_wchan, ph->ph_page, pi,
(pidx * sizeof(int)) + sizeof(*pi), ip[pidx]);
}
}
#endif /* DIAGNOSTIC */
if (ph->ph_nmissing++ == 0) {
/*
* This page was previously empty. Move it to the list of
* partially-full pages. This page is already curpage.
*/
TAILQ_REMOVE(&pp->pr_emptypages, ph, ph_entry);
TAILQ_INSERT_TAIL(&pp->pr_partpages, ph, ph_entry);
pp->pr_nidle--;
}
if (ph->ph_nmissing == pp->pr_itemsperpage) {
/*
* This page is now full. Move it to the full list
* and select a new current page.
*/
TAILQ_REMOVE(&pp->pr_partpages, ph, ph_entry);
TAILQ_INSERT_TAIL(&pp->pr_fullpages, ph, ph_entry);
pool_update_curpage(pp);
}
pp->pr_nget++;
return (pi);
}
/*
* Return resource to the pool.
*/
void
pool_put(struct pool *pp, void *v)
{
struct pool_page_header *ph, *freeph = NULL;
#ifdef DIAGNOSTIC
if (v == NULL)
panic("%s: NULL item", __func__);
#endif
#ifdef MULTIPROCESSOR
if (pp->pr_cache != NULL && TAILQ_EMPTY(&pp->pr_requests)) {
pool_cache_put(pp, v);
return;
}
#endif
pl_enter(pp, &pp->pr_lock);
pool_do_put(pp, v);
pp->pr_nout--;
pp->pr_nput++;
/* is it time to free a page? */
if (pp->pr_nidle > pp->pr_maxpages &&
(ph = TAILQ_FIRST(&pp->pr_emptypages)) != NULL &&
(ticks - ph->ph_tick) > (hz * pool_wait_free)) {
freeph = ph;
pool_p_remove(pp, freeph);
}
pl_leave(pp, &pp->pr_lock);
if (freeph != NULL)
pool_p_free(pp, freeph);
pool_wakeup(pp);
}
void
pool_wakeup(struct pool *pp)
{
if (!TAILQ_EMPTY(&pp->pr_requests)) {
pl_enter(pp, &pp->pr_requests_lock);
pool_runqueue(pp, PR_NOWAIT);
pl_leave(pp, &pp->pr_requests_lock);
}
}
void
pool_do_put(struct pool *pp, void *v)
{
struct pool_item *pi = v;
struct pool_page_header *ph;
splassert(pp->pr_ipl);
ph = pr_find_pagehead(pp, v);
#ifdef DIAGNOSTIC
if (pool_debug) {
struct pool_item *qi;
XSIMPLEQ_FOREACH(qi, &ph->ph_items, pi_list) {
if (pi == qi) {
panic("%s: %s: double pool_put: %p", __func__,
pp->pr_wchan, pi);
}
}
}
#endif /* DIAGNOSTIC */
pi->pi_magic = POOL_IMAGIC(ph, pi);
XSIMPLEQ_INSERT_HEAD(&ph->ph_items, pi, pi_list);
#ifdef DIAGNOSTIC
if (POOL_PHPOISON(ph))
poison_mem(pi + 1, pp->pr_size - sizeof(*pi));
#endif /* DIAGNOSTIC */
if (ph->ph_nmissing-- == pp->pr_itemsperpage) {
/*
* The page was previously completely full, move it to the
* partially-full list.
*/
TAILQ_REMOVE(&pp->pr_fullpages, ph, ph_entry);
TAILQ_INSERT_TAIL(&pp->pr_partpages, ph, ph_entry);
}
if (ph->ph_nmissing == 0) {
/*
* The page is now empty, so move it to the empty page list.
*/
pp->pr_nidle++;
ph->ph_tick = ticks;
TAILQ_REMOVE(&pp->pr_partpages, ph, ph_entry);
TAILQ_INSERT_TAIL(&pp->pr_emptypages, ph, ph_entry);
pool_update_curpage(pp);
}
}
/*
* Add N items to the pool.
*/
int
pool_prime(struct pool *pp, int n)
{
struct pool_pagelist pl = TAILQ_HEAD_INITIALIZER(pl);
struct pool_page_header *ph;
int newpages;
newpages = roundup(n, pp->pr_itemsperpage) / pp->pr_itemsperpage;
while (newpages-- > 0) {
int slowdown = 0;
ph = pool_p_alloc(pp, PR_NOWAIT, &slowdown);
if (ph == NULL) /* or slowdown? */
break;
TAILQ_INSERT_TAIL(&pl, ph, ph_entry);
}
pl_enter(pp, &pp->pr_lock);
while ((ph = TAILQ_FIRST(&pl)) != NULL) {
TAILQ_REMOVE(&pl, ph, ph_entry);
pool_p_insert(pp, ph);
}
pl_leave(pp, &pp->pr_lock);
return (0);
}
struct pool_page_header *
pool_p_alloc(struct pool *pp, int flags, int *slowdown)
{
struct pool_page_header *ph;
struct pool_item *pi;
caddr_t addr;
unsigned int order;
int o;
int n;
pl_assert_unlocked(pp, &pp->pr_lock);
KASSERT(pp->pr_size >= sizeof(*pi));
addr = pool_allocator_alloc(pp, flags, slowdown);
if (addr == NULL)
return (NULL);
if (POOL_INPGHDR(pp))
ph = (struct pool_page_header *)(addr + pp->pr_phoffset);
else {
ph = pool_get(&phpool, flags);
if (ph == NULL) {
pool_allocator_free(pp, addr);
return (NULL);
}
}
XSIMPLEQ_INIT(&ph->ph_items);
ph->ph_page = addr;
addr += pp->pr_align * (pp->pr_npagealloc % pp->pr_maxcolors);
ph->ph_colored = addr;
ph->ph_nmissing = 0;
arc4random_buf(&ph->ph_magic, sizeof(ph->ph_magic));
#ifdef DIAGNOSTIC
/* use a bit in ph_magic to record if we poison page items */
if (pool_debug)
SET(ph->ph_magic, POOL_MAGICBIT);
else
CLR(ph->ph_magic, POOL_MAGICBIT);
#endif /* DIAGNOSTIC */
n = pp->pr_itemsperpage;
o = 32;
while (n--) {
pi = (struct pool_item *)addr;
pi->pi_magic = POOL_IMAGIC(ph, pi);
if (o == 32) {
order = arc4random();
o = 0;
}
if (ISSET(order, 1 << o++))
XSIMPLEQ_INSERT_TAIL(&ph->ph_items, pi, pi_list);
else
XSIMPLEQ_INSERT_HEAD(&ph->ph_items, pi, pi_list);
#ifdef DIAGNOSTIC
if (POOL_PHPOISON(ph))
poison_mem(pi + 1, pp->pr_size - sizeof(*pi));
#endif /* DIAGNOSTIC */
addr += pp->pr_size;
}
return (ph);
}
void
pool_p_free(struct pool *pp, struct pool_page_header *ph)
{
struct pool_item *pi;
pl_assert_unlocked(pp, &pp->pr_lock);
KASSERT(ph->ph_nmissing == 0);
XSIMPLEQ_FOREACH(pi, &ph->ph_items, pi_list) {
if (__predict_false(pi->pi_magic != POOL_IMAGIC(ph, pi))) {
panic("%s: %s free list modified: "
"page %p; item addr %p; offset 0x%x=0x%lx",
__func__, pp->pr_wchan, ph->ph_page, pi,
0, pi->pi_magic);
}
#ifdef DIAGNOSTIC
if (POOL_PHPOISON(ph)) {
size_t pidx;
uint32_t pval;
if (poison_check(pi + 1, pp->pr_size - sizeof(*pi),
&pidx, &pval)) {
int *ip = (int *)(pi + 1);
panic("%s: %s free list modified: "
"page %p; item addr %p; offset 0x%zx=0x%x",
__func__, pp->pr_wchan, ph->ph_page, pi,
pidx * sizeof(int), ip[pidx]);
}