Training courses

Kernel and Embedded Linux

Bootlin training courses

Embedded Linux, kernel,
Yocto Project, Buildroot, real-time,
graphics, boot time, debugging...

Bootlin logo

Elixir Cross Referencer

   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
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
/*
 * Copyright (c) Red Hat Inc.

 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sub license,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice (including the
 * next paragraph) shall be included in all copies or substantial portions
 * of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
 * DEALINGS IN THE SOFTWARE.
 *
 * Authors: Dave Airlie <airlied@redhat.com>
 *          Jerome Glisse <jglisse@redhat.com>
 *          Pauli Nieminen <suokkos@gmail.com>
 */

/* simple list based uncached page pool
 * - Pool collects resently freed pages for reuse
 * - Use page->lru to keep a free list
 * - doesn't track currently in use pages
 */

#define pr_fmt(fmt) "[TTM] " fmt

#include <linux/list.h>
#include <linux/spinlock.h>
#include <linux/highmem.h>
#include <linux/mm_types.h>
#include <linux/module.h>
#include <linux/mm.h>
#include <linux/seq_file.h> /* for seq_printf */
#include <linux/slab.h>
#include <linux/dma-mapping.h>

#include <linux/atomic.h>

#include <drm/ttm/ttm_bo_driver.h>
#include <drm/ttm/ttm_page_alloc.h>
#include <drm/ttm/ttm_set_memory.h>

#define NUM_PAGES_TO_ALLOC		(PAGE_SIZE/sizeof(struct page *))
#define SMALL_ALLOCATION		16
#define FREE_ALL_PAGES			(~0U)
/* times are in msecs */
#define PAGE_FREE_INTERVAL		1000

/**
 * struct ttm_page_pool - Pool to reuse recently allocated uc/wc pages.
 *
 * @lock: Protects the shared pool from concurrnet access. Must be used with
 * irqsave/irqrestore variants because pool allocator maybe called from
 * delayed work.
 * @fill_lock: Prevent concurrent calls to fill.
 * @list: Pool of free uc/wc pages for fast reuse.
 * @gfp_flags: Flags to pass for alloc_page.
 * @npages: Number of pages in pool.
 */
struct ttm_page_pool {
	spinlock_t		lock;
	bool			fill_lock;
	struct list_head	list;
	gfp_t			gfp_flags;
	unsigned		npages;
	char			*name;
	unsigned long		nfrees;
	unsigned long		nrefills;
	unsigned int		order;
};

/**
 * Limits for the pool. They are handled without locks because only place where
 * they may change is in sysfs store. They won't have immediate effect anyway
 * so forcing serialization to access them is pointless.
 */

struct ttm_pool_opts {
	unsigned	alloc_size;
	unsigned	max_size;
	unsigned	small;
};

#define NUM_POOLS 6

/**
 * struct ttm_pool_manager - Holds memory pools for fst allocation
 *
 * Manager is read only object for pool code so it doesn't need locking.
 *
 * @free_interval: minimum number of jiffies between freeing pages from pool.
 * @page_alloc_inited: reference counting for pool allocation.
 * @work: Work that is used to shrink the pool. Work is only run when there is
 * some pages to free.
 * @small_allocation: Limit in number of pages what is small allocation.
 *
 * @pools: All pool objects in use.
 **/
struct ttm_pool_manager {
	struct kobject		kobj;
	struct shrinker		mm_shrink;
	struct ttm_pool_opts	options;

	union {
		struct ttm_page_pool	pools[NUM_POOLS];
		struct {
			struct ttm_page_pool	wc_pool;
			struct ttm_page_pool	uc_pool;
			struct ttm_page_pool	wc_pool_dma32;
			struct ttm_page_pool	uc_pool_dma32;
			struct ttm_page_pool	wc_pool_huge;
			struct ttm_page_pool	uc_pool_huge;
		} ;
	};
};

static struct attribute ttm_page_pool_max = {
	.name = "pool_max_size",
	.mode = S_IRUGO | S_IWUSR
};
static struct attribute ttm_page_pool_small = {
	.name = "pool_small_allocation",
	.mode = S_IRUGO | S_IWUSR
};
static struct attribute ttm_page_pool_alloc_size = {
	.name = "pool_allocation_size",
	.mode = S_IRUGO | S_IWUSR
};

static struct attribute *ttm_pool_attrs[] = {
	&ttm_page_pool_max,
	&ttm_page_pool_small,
	&ttm_page_pool_alloc_size,
	NULL
};

static void ttm_pool_kobj_release(struct kobject *kobj)
{
	struct ttm_pool_manager *m =
		container_of(kobj, struct ttm_pool_manager, kobj);
	kfree(m);
}

static ssize_t ttm_pool_store(struct kobject *kobj,
		struct attribute *attr, const char *buffer, size_t size)
{
	struct ttm_pool_manager *m =
		container_of(kobj, struct ttm_pool_manager, kobj);
	int chars;
	unsigned val;
	chars = sscanf(buffer, "%u", &val);
	if (chars == 0)
		return size;

	/* Convert kb to number of pages */
	val = val / (PAGE_SIZE >> 10);

	if (attr == &ttm_page_pool_max)
		m->options.max_size = val;
	else if (attr == &ttm_page_pool_small)
		m->options.small = val;
	else if (attr == &ttm_page_pool_alloc_size) {
		if (val > NUM_PAGES_TO_ALLOC*8) {
			pr_err("Setting allocation size to %lu is not allowed. Recommended size is %lu\n",
			       NUM_PAGES_TO_ALLOC*(PAGE_SIZE >> 7),
			       NUM_PAGES_TO_ALLOC*(PAGE_SIZE >> 10));
			return size;
		} else if (val > NUM_PAGES_TO_ALLOC) {
			pr_warn("Setting allocation size to larger than %lu is not recommended\n",
				NUM_PAGES_TO_ALLOC*(PAGE_SIZE >> 10));
		}
		m->options.alloc_size = val;
	}

	return size;
}

static ssize_t ttm_pool_show(struct kobject *kobj,
		struct attribute *attr, char *buffer)
{
	struct ttm_pool_manager *m =
		container_of(kobj, struct ttm_pool_manager, kobj);
	unsigned val = 0;

	if (attr == &ttm_page_pool_max)
		val = m->options.max_size;
	else if (attr == &ttm_page_pool_small)
		val = m->options.small;
	else if (attr == &ttm_page_pool_alloc_size)
		val = m->options.alloc_size;

	val = val * (PAGE_SIZE >> 10);

	return snprintf(buffer, PAGE_SIZE, "%u\n", val);
}

static const struct sysfs_ops ttm_pool_sysfs_ops = {
	.show = &ttm_pool_show,
	.store = &ttm_pool_store,
};

static struct kobj_type ttm_pool_kobj_type = {
	.release = &ttm_pool_kobj_release,
	.sysfs_ops = &ttm_pool_sysfs_ops,
	.default_attrs = ttm_pool_attrs,
};

static struct ttm_pool_manager *_manager;

/**
 * Select the right pool or requested caching state and ttm flags. */
static struct ttm_page_pool *ttm_get_pool(int flags, bool huge,
					  enum ttm_caching_state cstate)
{
	int pool_index;

	if (cstate == tt_cached)
		return NULL;

	if (cstate == tt_wc)
		pool_index = 0x0;
	else
		pool_index = 0x1;

	if (flags & TTM_PAGE_FLAG_DMA32) {
		if (huge)
			return NULL;
		pool_index |= 0x2;

	} else if (huge) {
		pool_index |= 0x4;
	}

	return &_manager->pools[pool_index];
}

/* set memory back to wb and free the pages. */
static void ttm_pages_put(struct page *pages[], unsigned npages,
		unsigned int order)
{
	unsigned int i, pages_nr = (1 << order);

	if (order == 0) {
		if (ttm_set_pages_array_wb(pages, npages))
			pr_err("Failed to set %d pages to wb!\n", npages);
	}

	for (i = 0; i < npages; ++i) {
		if (order > 0) {
			if (ttm_set_pages_wb(pages[i], pages_nr))
				pr_err("Failed to set %d pages to wb!\n", pages_nr);
		}
		__free_pages(pages[i], order);
	}
}

static void ttm_pool_update_free_locked(struct ttm_page_pool *pool,
		unsigned freed_pages)
{
	pool->npages -= freed_pages;
	pool->nfrees += freed_pages;
}

/**
 * Free pages from pool.
 *
 * To prevent hogging the ttm_swap process we only free NUM_PAGES_TO_ALLOC
 * number of pages in one go.
 *
 * @pool: to free the pages from
 * @free_all: If set to true will free all pages in pool
 * @use_static: Safe to use static buffer
 **/
static int ttm_page_pool_free(struct ttm_page_pool *pool, unsigned nr_free,
			      bool use_static)
{
	static struct page *static_buf[NUM_PAGES_TO_ALLOC];
	unsigned long irq_flags;
	struct page *p;
	struct page **pages_to_free;
	unsigned freed_pages = 0,
		 npages_to_free = nr_free;

	if (NUM_PAGES_TO_ALLOC < nr_free)
		npages_to_free = NUM_PAGES_TO_ALLOC;

	if (use_static)
		pages_to_free = static_buf;
	else
		pages_to_free = kmalloc_array(npages_to_free,
					      sizeof(struct page *),
					      GFP_KERNEL);
	if (!pages_to_free) {
		pr_debug("Failed to allocate memory for pool free operation\n");
		return 0;
	}

restart:
	spin_lock_irqsave(&pool->lock, irq_flags);

	list_for_each_entry_reverse(p, &pool->list, lru) {
		if (freed_pages >= npages_to_free)
			break;

		pages_to_free[freed_pages++] = p;
		/* We can only remove NUM_PAGES_TO_ALLOC at a time. */
		if (freed_pages >= NUM_PAGES_TO_ALLOC) {
			/* remove range of pages from the pool */
			__list_del(p->lru.prev, &pool->list);

			ttm_pool_update_free_locked(pool, freed_pages);
			/**
			 * Because changing page caching is costly
			 * we unlock the pool to prevent stalling.
			 */
			spin_unlock_irqrestore(&pool->lock, irq_flags);

			ttm_pages_put(pages_to_free, freed_pages, pool->order);
			if (likely(nr_free != FREE_ALL_PAGES))
				nr_free -= freed_pages;

			if (NUM_PAGES_TO_ALLOC >= nr_free)
				npages_to_free = nr_free;
			else
				npages_to_free = NUM_PAGES_TO_ALLOC;

			freed_pages = 0;

			/* free all so restart the processing */
			if (nr_free)
				goto restart;

			/* Not allowed to fall through or break because
			 * following context is inside spinlock while we are
			 * outside here.
			 */
			goto out;

		}
	}

	/* remove range of pages from the pool */
	if (freed_pages) {
		__list_del(&p->lru, &pool->list);

		ttm_pool_update_free_locked(pool, freed_pages);
		nr_free -= freed_pages;
	}

	spin_unlock_irqrestore(&pool->lock, irq_flags);

	if (freed_pages)
		ttm_pages_put(pages_to_free, freed_pages, pool->order);
out:
	if (pages_to_free != static_buf)
		kfree(pages_to_free);
	return nr_free;
}

/**
 * Callback for mm to request pool to reduce number of page held.
 *
 * XXX: (dchinner) Deadlock warning!
 *
 * This code is crying out for a shrinker per pool....
 */
static unsigned long
ttm_pool_shrink_scan(struct shrinker *shrink, struct shrink_control *sc)
{
	static DEFINE_MUTEX(lock);
	static unsigned start_pool;
	unsigned i;
	unsigned pool_offset;
	struct ttm_page_pool *pool;
	int shrink_pages = sc->nr_to_scan;
	unsigned long freed = 0;
	unsigned int nr_free_pool;

	if (!mutex_trylock(&lock))
		return SHRINK_STOP;
	pool_offset = ++start_pool % NUM_POOLS;
	/* select start pool in round robin fashion */
	for (i = 0; i < NUM_POOLS; ++i) {
		unsigned nr_free = shrink_pages;
		unsigned page_nr;

		if (shrink_pages == 0)
			break;

		pool = &_manager->pools[(i + pool_offset)%NUM_POOLS];
		page_nr = (1 << pool->order);
		/* OK to use static buffer since global mutex is held. */
		nr_free_pool = roundup(nr_free, page_nr) >> pool->order;
		shrink_pages = ttm_page_pool_free(pool, nr_free_pool, true);
		freed += (nr_free_pool - shrink_pages) << pool->order;
		if (freed >= sc->nr_to_scan)
			break;
		shrink_pages <<= pool->order;
	}
	mutex_unlock(&lock);
	return freed;
}


static unsigned long
ttm_pool_shrink_count(struct shrinker *shrink, struct shrink_control *sc)
{
	unsigned i;
	unsigned long count = 0;
	struct ttm_page_pool *pool;

	for (i = 0; i < NUM_POOLS; ++i) {
		pool = &_manager->pools[i];
		count += (pool->npages << pool->order);
	}

	return count;
}

static int ttm_pool_mm_shrink_init(struct ttm_pool_manager *manager)
{
	manager->mm_shrink.count_objects = ttm_pool_shrink_count;
	manager->mm_shrink.scan_objects = ttm_pool_shrink_scan;
	manager->mm_shrink.seeks = 1;
	return register_shrinker(&manager->mm_shrink);
}

static void ttm_pool_mm_shrink_fini(struct ttm_pool_manager *manager)
{
	unregister_shrinker(&manager->mm_shrink);
}

static int ttm_set_pages_caching(struct page **pages,
		enum ttm_caching_state cstate, unsigned cpages)
{
	int r = 0;
	/* Set page caching */
	switch (cstate) {
	case tt_uncached:
		r = ttm_set_pages_array_uc(pages, cpages);
		if (r)
			pr_err("Failed to set %d pages to uc!\n", cpages);
		break;
	case tt_wc:
		r = ttm_set_pages_array_wc(pages, cpages);
		if (r)
			pr_err("Failed to set %d pages to wc!\n", cpages);
		break;
	default:
		break;
	}
	return r;
}

/**
 * Free pages the pages that failed to change the caching state. If there is
 * any pages that have changed their caching state already put them to the
 * pool.
 */
static void ttm_handle_caching_state_failure(struct list_head *pages,
		int ttm_flags, enum ttm_caching_state cstate,
		struct page **failed_pages, unsigned cpages)
{
	unsigned i;
	/* Failed pages have to be freed */
	for (i = 0; i < cpages; ++i) {
		list_del(&failed_pages[i]->lru);
		__free_page(failed_pages[i]);
	}
}

/**
 * Allocate new pages with correct caching.
 *
 * This function is reentrant if caller updates count depending on number of
 * pages returned in pages array.
 */
static int ttm_alloc_new_pages(struct list_head *pages, gfp_t gfp_flags,
			       int ttm_flags, enum ttm_caching_state cstate,
			       unsigned count, unsigned order)
{
	struct page **caching_array;
	struct page *p;
	int r = 0;
	unsigned i, j, cpages;
	unsigned npages = 1 << order;
	unsigned max_cpages = min(count << order, (unsigned)NUM_PAGES_TO_ALLOC);

	/* allocate array for page caching change */
	caching_array = kmalloc_array(max_cpages, sizeof(struct page *),
				      GFP_KERNEL);

	if (!caching_array) {
		pr_debug("Unable to allocate table for new pages\n");
		return -ENOMEM;
	}

	for (i = 0, cpages = 0; i < count; ++i) {
		p = alloc_pages(gfp_flags, order);

		if (!p) {
			pr_debug("Unable to get page %u\n", i);

			/* store already allocated pages in the pool after
			 * setting the caching state */
			if (cpages) {
				r = ttm_set_pages_caching(caching_array,
							  cstate, cpages);
				if (r)
					ttm_handle_caching_state_failure(pages,
						ttm_flags, cstate,
						caching_array, cpages);
			}
			r = -ENOMEM;
			goto out;
		}

		list_add(&p->lru, pages);

#ifdef CONFIG_HIGHMEM
		/* gfp flags of highmem page should never be dma32 so we
		 * we should be fine in such case
		 */
		if (PageHighMem(p))
			continue;

#endif
		for (j = 0; j < npages; ++j) {
			caching_array[cpages++] = p++;
			if (cpages == max_cpages) {

				r = ttm_set_pages_caching(caching_array,
						cstate, cpages);
				if (r) {
					ttm_handle_caching_state_failure(pages,
						ttm_flags, cstate,
						caching_array, cpages);
					goto out;
				}
				cpages = 0;
			}
		}
	}

	if (cpages) {
		r = ttm_set_pages_caching(caching_array, cstate, cpages);
		if (r)
			ttm_handle_caching_state_failure(pages,
					ttm_flags, cstate,
					caching_array, cpages);
	}
out:
	kfree(caching_array);

	return r;
}

/**
 * Fill the given pool if there aren't enough pages and the requested number of
 * pages is small.
 */
static void ttm_page_pool_fill_locked(struct ttm_page_pool *pool, int ttm_flags,
				      enum ttm_caching_state cstate,
				      unsigned count, unsigned long *irq_flags)
{
	struct page *p;
	int r;
	unsigned cpages = 0;
	/**
	 * Only allow one pool fill operation at a time.
	 * If pool doesn't have enough pages for the allocation new pages are
	 * allocated from outside of pool.
	 */
	if (pool->fill_lock)
		return;

	pool->fill_lock = true;

	/* If allocation request is small and there are not enough
	 * pages in a pool we fill the pool up first. */
	if (count < _manager->options.small
		&& count > pool->npages) {
		struct list_head new_pages;
		unsigned alloc_size = _manager->options.alloc_size;

		/**
		 * Can't change page caching if in irqsave context. We have to
		 * drop the pool->lock.
		 */
		spin_unlock_irqrestore(&pool->lock, *irq_flags);

		INIT_LIST_HEAD(&new_pages);
		r = ttm_alloc_new_pages(&new_pages, pool->gfp_flags, ttm_flags,
					cstate, alloc_size, 0);
		spin_lock_irqsave(&pool->lock, *irq_flags);

		if (!r) {
			list_splice(&new_pages, &pool->list);
			++pool->nrefills;
			pool->npages += alloc_size;
		} else {
			pr_debug("Failed to fill pool (%p)\n", pool);
			/* If we have any pages left put them to the pool. */
			list_for_each_entry(p, &new_pages, lru) {
				++cpages;
			}
			list_splice(&new_pages, &pool->list);
			pool->npages += cpages;
		}

	}
	pool->fill_lock = false;
}

/**
 * Allocate pages from the pool and put them on the return list.
 *
 * @return zero for success or negative error code.
 */
static int ttm_page_pool_get_pages(struct ttm_page_pool *pool,
				   struct list_head *pages,
				   int ttm_flags,
				   enum ttm_caching_state cstate,
				   unsigned count, unsigned order)
{
	unsigned long irq_flags;
	struct list_head *p;
	unsigned i;
	int r = 0;

	spin_lock_irqsave(&pool->lock, irq_flags);
	if (!order)
		ttm_page_pool_fill_locked(pool, ttm_flags, cstate, count,
					  &irq_flags);

	if (count >= pool->npages) {
		/* take all pages from the pool */
		list_splice_init(&pool->list, pages);
		count -= pool->npages;
		pool->npages = 0;
		goto out;
	}
	/* find the last pages to include for requested number of pages. Split
	 * pool to begin and halve it to reduce search space. */
	if (count <= pool->npages/2) {
		i = 0;
		list_for_each(p, &pool->list) {
			if (++i == count)
				break;
		}
	} else {
		i = pool->npages + 1;
		list_for_each_prev(p, &pool->list) {
			if (--i == count)
				break;
		}
	}
	/* Cut 'count' number of pages from the pool */
	list_cut_position(pages, &pool->list, p);
	pool->npages -= count;
	count = 0;
out:
	spin_unlock_irqrestore(&pool->lock, irq_flags);

	/* clear the pages coming from the pool if requested */
	if (ttm_flags & TTM_PAGE_FLAG_ZERO_ALLOC) {
		struct page *page;

		list_for_each_entry(page, pages, lru) {
			if (PageHighMem(page))
				clear_highpage(page);
			else
				clear_page(page_address(page));
		}
	}

	/* If pool didn't have enough pages allocate new one. */
	if (count) {
		gfp_t gfp_flags = pool->gfp_flags;

		/* set zero flag for page allocation if required */
		if (ttm_flags & TTM_PAGE_FLAG_ZERO_ALLOC)
			gfp_flags |= __GFP_ZERO;

		if (ttm_flags & TTM_PAGE_FLAG_NO_RETRY)
			gfp_flags |= __GFP_RETRY_MAYFAIL;

		/* ttm_alloc_new_pages doesn't reference pool so we can run
		 * multiple requests in parallel.
		 **/
		r = ttm_alloc_new_pages(pages, gfp_flags, ttm_flags, cstate,
					count, order);
	}

	return r;
}

/* Put all pages in pages list to correct pool to wait for reuse */
static void ttm_put_pages(struct page **pages, unsigned npages, int flags,
			  enum ttm_caching_state cstate)
{
	struct ttm_page_pool *pool = ttm_get_pool(flags, false, cstate);
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	struct ttm_page_pool *huge = ttm_get_pool(flags, true, cstate);
#endif
	unsigned long irq_flags;
	unsigned i;

	if (pool == NULL) {
		/* No pool for this memory type so free the pages */
		i = 0;
		while (i < npages) {
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
			struct page *p = pages[i];
#endif
			unsigned order = 0, j;

			if (!pages[i]) {
				++i;
				continue;
			}

#ifdef CONFIG_TRANSPARENT_HUGEPAGE
			if (!(flags & TTM_PAGE_FLAG_DMA32) &&
			    (npages - i) >= HPAGE_PMD_NR) {
				for (j = 1; j < HPAGE_PMD_NR; ++j)
					if (++p != pages[i + j])
					    break;

				if (j == HPAGE_PMD_NR)
					order = HPAGE_PMD_ORDER;
			}
#endif

			if (page_count(pages[i]) != 1)
				pr_err("Erroneous page count. Leaking pages.\n");
			__free_pages(pages[i], order);

			j = 1 << order;
			while (j) {
				pages[i++] = NULL;
				--j;
			}
		}
		return;
	}

	i = 0;
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	if (huge) {
		unsigned max_size, n2free;

		spin_lock_irqsave(&huge->lock, irq_flags);
		while ((npages - i) >= HPAGE_PMD_NR) {
			struct page *p = pages[i];
			unsigned j;

			if (!p)
				break;

			for (j = 1; j < HPAGE_PMD_NR; ++j)
				if (++p != pages[i + j])
				    break;

			if (j != HPAGE_PMD_NR)
				break;

			list_add_tail(&pages[i]->lru, &huge->list);

			for (j = 0; j < HPAGE_PMD_NR; ++j)
				pages[i++] = NULL;
			huge->npages++;
		}

		/* Check that we don't go over the pool limit */
		max_size = _manager->options.max_size;
		max_size /= HPAGE_PMD_NR;
		if (huge->npages > max_size)
			n2free = huge->npages - max_size;
		else
			n2free = 0;
		spin_unlock_irqrestore(&huge->lock, irq_flags);
		if (n2free)
			ttm_page_pool_free(huge, n2free, false);
	}
#endif

	spin_lock_irqsave(&pool->lock, irq_flags);
	while (i < npages) {
		if (pages[i]) {
			if (page_count(pages[i]) != 1)
				pr_err("Erroneous page count. Leaking pages.\n");
			list_add_tail(&pages[i]->lru, &pool->list);
			pages[i] = NULL;
			pool->npages++;
		}
		++i;
	}
	/* Check that we don't go over the pool limit */
	npages = 0;
	if (pool->npages > _manager->options.max_size) {
		npages = pool->npages - _manager->options.max_size;
		/* free at least NUM_PAGES_TO_ALLOC number of pages
		 * to reduce calls to set_memory_wb */
		if (npages < NUM_PAGES_TO_ALLOC)
			npages = NUM_PAGES_TO_ALLOC;
	}
	spin_unlock_irqrestore(&pool->lock, irq_flags);
	if (npages)
		ttm_page_pool_free(pool, npages, false);
}

/*
 * On success pages list will hold count number of correctly
 * cached pages.
 */
static int ttm_get_pages(struct page **pages, unsigned npages, int flags,
			 enum ttm_caching_state cstate)
{
	struct ttm_page_pool *pool = ttm_get_pool(flags, false, cstate);
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	struct ttm_page_pool *huge = ttm_get_pool(flags, true, cstate);
#endif
	struct list_head plist;
	struct page *p = NULL;
	unsigned count, first;
	int r;

	/* No pool for cached pages */
	if (pool == NULL) {
		gfp_t gfp_flags = GFP_USER;
		unsigned i;
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
		unsigned j;
#endif

		/* set zero flag for page allocation if required */
		if (flags & TTM_PAGE_FLAG_ZERO_ALLOC)
			gfp_flags |= __GFP_ZERO;

		if (flags & TTM_PAGE_FLAG_NO_RETRY)
			gfp_flags |= __GFP_RETRY_MAYFAIL;

		if (flags & TTM_PAGE_FLAG_DMA32)
			gfp_flags |= GFP_DMA32;
		else
			gfp_flags |= GFP_HIGHUSER;

		i = 0;
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
		if (!(gfp_flags & GFP_DMA32)) {
			while (npages >= HPAGE_PMD_NR) {
				gfp_t huge_flags = gfp_flags;

				huge_flags |= GFP_TRANSHUGE_LIGHT | __GFP_NORETRY |
					__GFP_KSWAPD_RECLAIM;
				huge_flags &= ~__GFP_MOVABLE;
				huge_flags &= ~__GFP_COMP;
				p = alloc_pages(huge_flags, HPAGE_PMD_ORDER);
				if (!p)
					break;

				for (j = 0; j < HPAGE_PMD_NR; ++j)
					pages[i++] = p++;

				npages -= HPAGE_PMD_NR;
			}
		}
#endif

		first = i;
		while (npages) {
			p = alloc_page(gfp_flags);
			if (!p) {
				pr_debug("Unable to allocate page\n");
				return -ENOMEM;
			}

			/* Swap the pages if we detect consecutive order */
			if (i > first && pages[i - 1] == p - 1)
				swap(p, pages[i - 1]);

			pages[i++] = p;
			--npages;
		}
		return 0;
	}

	count = 0;

#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	if (huge && npages >= HPAGE_PMD_NR) {
		INIT_LIST_HEAD(&plist);
		ttm_page_pool_get_pages(huge, &plist, flags, cstate,
					npages / HPAGE_PMD_NR,
					HPAGE_PMD_ORDER);

		list_for_each_entry(p, &plist, lru) {
			unsigned j;

			for (j = 0; j < HPAGE_PMD_NR; ++j)
				pages[count++] = &p[j];
		}
	}
#endif

	INIT_LIST_HEAD(&plist);
	r = ttm_page_pool_get_pages(pool, &plist, flags, cstate,
				    npages - count, 0);

	first = count;
	list_for_each_entry(p, &plist, lru) {
		struct page *tmp = p;

		/* Swap the pages if we detect consecutive order */
		if (count > first && pages[count - 1] == tmp - 1)
			swap(tmp, pages[count - 1]);
		pages[count++] = tmp;
	}

	if (r) {
		/* If there is any pages in the list put them back to
		 * the pool.
		 */
		pr_debug("Failed to allocate extra pages for large request\n");
		ttm_put_pages(pages, count, flags, cstate);
		return r;
	}

	return 0;
}

static void ttm_page_pool_init_locked(struct ttm_page_pool *pool, gfp_t flags,
		char *name, unsigned int order)
{
	spin_lock_init(&pool->lock);
	pool->fill_lock = false;
	INIT_LIST_HEAD(&pool->list);
	pool->npages = pool->nfrees = 0;
	pool->gfp_flags = flags;
	pool->name = name;
	pool->order = order;
}

int ttm_page_alloc_init(struct ttm_mem_global *glob, unsigned max_pages)
{
	int ret;
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	unsigned order = HPAGE_PMD_ORDER;
#else
	unsigned order = 0;
#endif

	WARN_ON(_manager);

	pr_info("Initializing pool allocator\n");

	_manager = kzalloc(sizeof(*_manager), GFP_KERNEL);
	if (!_manager)
		return -ENOMEM;

	ttm_page_pool_init_locked(&_manager->wc_pool, GFP_HIGHUSER, "wc", 0);

	ttm_page_pool_init_locked(&_manager->uc_pool, GFP_HIGHUSER, "uc", 0);

	ttm_page_pool_init_locked(&_manager->wc_pool_dma32,
				  GFP_USER | GFP_DMA32, "wc dma", 0);

	ttm_page_pool_init_locked(&_manager->uc_pool_dma32,
				  GFP_USER | GFP_DMA32, "uc dma", 0);

	ttm_page_pool_init_locked(&_manager->wc_pool_huge,
				  (GFP_TRANSHUGE_LIGHT | __GFP_NORETRY |
				   __GFP_KSWAPD_RECLAIM) &
				  ~(__GFP_MOVABLE | __GFP_COMP),
				  "wc huge", order);

	ttm_page_pool_init_locked(&_manager->uc_pool_huge,
				  (GFP_TRANSHUGE_LIGHT | __GFP_NORETRY |
				   __GFP_KSWAPD_RECLAIM) &
				  ~(__GFP_MOVABLE | __GFP_COMP)
				  , "uc huge", order);

	_manager->options.max_size = max_pages;
	_manager->options.small = SMALL_ALLOCATION;
	_manager->options.alloc_size = NUM_PAGES_TO_ALLOC;

	ret = kobject_init_and_add(&_manager->kobj, &ttm_pool_kobj_type,
				   &glob->kobj, "pool");
	if (unlikely(ret != 0))
		goto error;

	ret = ttm_pool_mm_shrink_init(_manager);
	if (unlikely(ret != 0))
		goto error;
	return 0;

error:
	kobject_put(&_manager->kobj);
	_manager = NULL;
	return ret;
}

void ttm_page_alloc_fini(void)
{
	int i;

	pr_info("Finalizing pool allocator\n");
	ttm_pool_mm_shrink_fini(_manager);

	/* OK to use static buffer since global mutex is no longer used. */
	for (i = 0; i < NUM_POOLS; ++i)
		ttm_page_pool_free(&_manager->pools[i], FREE_ALL_PAGES, true);

	kobject_put(&_manager->kobj);
	_manager = NULL;
}

static void
ttm_pool_unpopulate_helper(struct ttm_tt *ttm, unsigned mem_count_update)
{
	struct ttm_mem_global *mem_glob = ttm->bdev->glob->mem_glob;
	unsigned i;

	if (mem_count_update == 0)
		goto put_pages;

	for (i = 0; i < mem_count_update; ++i) {
		if (!ttm->pages[i])
			continue;

		ttm_mem_global_free_page(mem_glob, ttm->pages[i], PAGE_SIZE);
	}

put_pages:
	ttm_put_pages(ttm->pages, ttm->num_pages, ttm->page_flags,
		      ttm->caching_state);
	ttm->state = tt_unpopulated;
}

int ttm_pool_populate(struct ttm_tt *ttm, struct ttm_operation_ctx *ctx)
{
	struct ttm_mem_global *mem_glob = ttm->bdev->glob->mem_glob;
	unsigned i;
	int ret;

	if (ttm->state != tt_unpopulated)
		return 0;

	if (ttm_check_under_lowerlimit(mem_glob, ttm->num_pages, ctx))
		return -ENOMEM;

	ret = ttm_get_pages(ttm->pages, ttm->num_pages, ttm->page_flags,
			    ttm->caching_state);
	if (unlikely(ret != 0)) {
		ttm_pool_unpopulate_helper(ttm, 0);
		return ret;
	}

	for (i = 0; i < ttm->num_pages; ++i) {
		ret = ttm_mem_global_alloc_page(mem_glob, ttm->pages[i],
						PAGE_SIZE, ctx);
		if (unlikely(ret != 0)) {
			ttm_pool_unpopulate_helper(ttm, i);
			return -ENOMEM;
		}
	}

	if (unlikely(ttm->page_flags & TTM_PAGE_FLAG_SWAPPED)) {
		ret = ttm_tt_swapin(ttm);
		if (unlikely(ret != 0)) {
			ttm_pool_unpopulate(ttm);
			return ret;
		}
	}

	ttm->state = tt_unbound;
	return 0;
}
EXPORT_SYMBOL(ttm_pool_populate);

void ttm_pool_unpopulate(struct ttm_tt *ttm)
{
	ttm_pool_unpopulate_helper(ttm, ttm->num_pages);
}
EXPORT_SYMBOL(ttm_pool_unpopulate);

int ttm_populate_and_map_pages(struct device *dev, struct ttm_dma_tt *tt,
					struct ttm_operation_ctx *ctx)
{
	unsigned i, j;
	int r;

	r = ttm_pool_populate(&tt->ttm, ctx);
	if (r)
		return r;

	for (i = 0; i < tt->ttm.num_pages; ++i) {
		struct page *p = tt->ttm.pages[i];
		size_t num_pages = 1;

		for (j = i + 1; j < tt->ttm.num_pages; ++j) {
			if (++p != tt->ttm.pages[j])
				break;

			++num_pages;
		}

		tt->dma_address[i] = dma_map_page(dev, tt->ttm.pages[i],
						  0, num_pages * PAGE_SIZE,
						  DMA_BIDIRECTIONAL);
		if (dma_mapping_error(dev, tt->dma_address[i])) {
			while (i--) {
				dma_unmap_page(dev, tt->dma_address[i],
					       PAGE_SIZE, DMA_BIDIRECTIONAL);
				tt->dma_address[i] = 0;
			}
			ttm_pool_unpopulate(&tt->ttm);
			return -EFAULT;
		}

		for (j = 1; j < num_pages; ++j) {
			tt->dma_address[i + 1] = tt->dma_address[i] + PAGE_SIZE;
			++i;
		}
	}
	return 0;
}
EXPORT_SYMBOL(ttm_populate_and_map_pages);

void ttm_unmap_and_unpopulate_pages(struct device *dev, struct ttm_dma_tt *tt)
{
	unsigned i, j;

	for (i = 0; i < tt->ttm.num_pages;) {
		struct page *p = tt->ttm.pages[i];
		size_t num_pages = 1;

		if (!tt->dma_address[i] || !tt->ttm.pages[i]) {
			++i;
			continue;
		}

		for (j = i + 1; j < tt->ttm.num_pages; ++j) {
			if (++p != tt->ttm.pages[j])
				break;

			++num_pages;
		}

		dma_unmap_page(dev, tt->dma_address[i], num_pages * PAGE_SIZE,
			       DMA_BIDIRECTIONAL);

		i += num_pages;
	}
	ttm_pool_unpopulate(&tt->ttm);
}
EXPORT_SYMBOL(ttm_unmap_and_unpopulate_pages);

int ttm_page_alloc_debugfs(struct seq_file *m, void *data)
{
	struct ttm_page_pool *p;
	unsigned i;
	char *h[] = {"pool", "refills", "pages freed", "size"};
	if (!_manager) {
		seq_printf(m, "No pool allocator running.\n");
		return 0;
	}
	seq_printf(m, "%7s %12s %13s %8s\n",
			h[0], h[1], h[2], h[3]);
	for (i = 0; i < NUM_POOLS; ++i) {
		p = &_manager->pools[i];

		seq_printf(m, "%7s %12ld %13ld %8d\n",
				p->name, p->nrefills,
				p->nfrees, p->npages);
	}
	return 0;
}
EXPORT_SYMBOL(ttm_page_alloc_debugfs);