Blame view
mm/zsmalloc.c
61.1 KB
61989a80f
|
1 2 3 4 |
/* * zsmalloc memory allocator * * Copyright (C) 2011 Nitin Gupta |
31fc00bb7
|
5 |
* Copyright (C) 2012, 2013 Minchan Kim |
61989a80f
|
6 7 8 9 10 11 12 |
* * This code is released using a dual license strategy: BSD/GPL * You can choose the license that better fits your requirements. * * Released under the terms of 3-clause BSD License * Released under the terms of GNU General Public License Version 2.0 */ |
2db51dae5
|
13 |
/* |
2db51dae5
|
14 15 16 17 |
* Following is how we use various fields and flags of underlying * struct page(s) to form a zspage. * * Usage of struct page fields: |
3783689a1
|
18 |
* page->private: points to zspage |
48b4800a1
|
19 20 21 |
* page->freelist(index): links together all component pages of a zspage * For the huge page, this is always 0, so we use this field * to store handle. |
fd8544639
|
22 |
* page->units: first object offset in a subpage of zspage |
2db51dae5
|
23 24 25 |
* * Usage of struct page flags: * PG_private: identifies the first component page |
399d8eebe
|
26 |
* PG_owner_priv_1: identifies the huge component page |
2db51dae5
|
27 28 |
* */ |
4abaac9b7
|
29 |
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt |
61989a80f
|
30 31 |
#include <linux/module.h> #include <linux/kernel.h> |
312fcae22
|
32 |
#include <linux/sched.h> |
50d34394c
|
33 |
#include <linux/magic.h> |
61989a80f
|
34 35 36 |
#include <linux/bitops.h> #include <linux/errno.h> #include <linux/highmem.h> |
61989a80f
|
37 38 |
#include <linux/string.h> #include <linux/slab.h> |
ca5999fde
|
39 |
#include <linux/pgtable.h> |
65fddcfca
|
40 |
#include <asm/tlbflush.h> |
61989a80f
|
41 42 |
#include <linux/cpumask.h> #include <linux/cpu.h> |
0cbb613fa
|
43 |
#include <linux/vmalloc.h> |
759b26b29
|
44 |
#include <linux/preempt.h> |
0959c63f1
|
45 |
#include <linux/spinlock.h> |
93144ca35
|
46 |
#include <linux/shrinker.h> |
0959c63f1
|
47 |
#include <linux/types.h> |
0f050d997
|
48 |
#include <linux/debugfs.h> |
bcf1647d0
|
49 |
#include <linux/zsmalloc.h> |
c795779df
|
50 |
#include <linux/zpool.h> |
48b4800a1
|
51 |
#include <linux/mount.h> |
8e9231f81
|
52 |
#include <linux/pseudo_fs.h> |
dd4123f32
|
53 |
#include <linux/migrate.h> |
701d67859
|
54 |
#include <linux/wait.h> |
48b4800a1
|
55 |
#include <linux/pagemap.h> |
cdc346b36
|
56 |
#include <linux/fs.h> |
48b4800a1
|
57 58 |
#define ZSPAGE_MAGIC 0x58 |
0959c63f1
|
59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 |
/* * This must be power of 2 and greater than of equal to sizeof(link_free). * These two conditions ensure that any 'struct link_free' itself doesn't * span more than 1 page which avoids complex case of mapping 2 pages simply * to restore link_free pointer values. */ #define ZS_ALIGN 8 /* * A single 'zspage' is composed of up to 2^N discontiguous 0-order (single) * pages. ZS_MAX_ZSPAGE_ORDER defines upper limit on N. */ #define ZS_MAX_ZSPAGE_ORDER 2 #define ZS_MAX_PAGES_PER_ZSPAGE (_AC(1, UL) << ZS_MAX_ZSPAGE_ORDER) |
2e40e163a
|
74 |
#define ZS_HANDLE_SIZE (sizeof(unsigned long)) |
0959c63f1
|
75 76 |
/* * Object location (<PFN>, <obj_idx>) is encoded as |
b956b5ac2
|
77 |
* a single (unsigned long) handle value. |
0959c63f1
|
78 |
* |
bfd093f5e
|
79 |
* Note that object index <obj_idx> starts from 0. |
0959c63f1
|
80 81 82 |
* * This is made more complicated by various memory models and PAE. */ |
02390b87a
|
83 84 85 86 |
#ifndef MAX_POSSIBLE_PHYSMEM_BITS #ifdef MAX_PHYSMEM_BITS #define MAX_POSSIBLE_PHYSMEM_BITS MAX_PHYSMEM_BITS #else |
0959c63f1
|
87 88 89 90 |
/* * If this definition of MAX_PHYSMEM_BITS is used, OBJ_INDEX_BITS will just * be PAGE_SHIFT */ |
02390b87a
|
91 |
#define MAX_POSSIBLE_PHYSMEM_BITS BITS_PER_LONG |
0959c63f1
|
92 93 |
#endif #endif |
02390b87a
|
94 95 |
#define _PFN_BITS (MAX_POSSIBLE_PHYSMEM_BITS - PAGE_SHIFT) |
312fcae22
|
96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 |
/* * Memory for allocating for handle keeps object position by * encoding <page, obj_idx> and the encoded value has a room * in least bit(ie, look at obj_to_location). * We use the bit to synchronize between object access by * user and migration. */ #define HANDLE_PIN_BIT 0 /* * Head in allocated object should have OBJ_ALLOCATED_TAG * to identify the object was allocated or not. * It's okay to add the status bit in the least bit because * header keeps handle which is 4byte-aligned address so we * have room for two bit at least. */ #define OBJ_ALLOCATED_TAG 1 #define OBJ_TAG_BITS 1 #define OBJ_INDEX_BITS (BITS_PER_LONG - _PFN_BITS - OBJ_TAG_BITS) |
0959c63f1
|
116 |
#define OBJ_INDEX_MASK ((_AC(1, UL) << OBJ_INDEX_BITS) - 1) |
cf8e0fedf
|
117 118 119 120 |
#define FULLNESS_BITS 2 #define CLASS_BITS 8 #define ISOLATED_BITS 3 #define MAGIC_VAL_BITS 8 |
0959c63f1
|
121 122 123 124 |
#define MAX(a, b) ((a) >= (b) ? (a) : (b)) /* ZS_MIN_ALLOC_SIZE must be multiple of ZS_ALIGN */ #define ZS_MIN_ALLOC_SIZE \ MAX(32, (ZS_MAX_PAGES_PER_ZSPAGE << PAGE_SHIFT >> OBJ_INDEX_BITS)) |
2e40e163a
|
125 |
/* each chunk includes extra space to keep handle */ |
7b60a6852
|
126 |
#define ZS_MAX_ALLOC_SIZE PAGE_SIZE |
0959c63f1
|
127 128 |
/* |
7eb52512a
|
129 |
* On systems with 4K page size, this gives 255 size classes! There is a |
0959c63f1
|
130 131 132 133 134 135 136 137 138 139 140 |
* trader-off here: * - Large number of size classes is potentially wasteful as free page are * spread across these classes * - Small number of size classes causes large internal fragmentation * - Probably its better to use specific size classes (empirically * determined). NOTE: all those class sizes must be set as multiple of * ZS_ALIGN to make sure link_free itself never has to span 2 pages. * * ZS_MIN_ALLOC_SIZE and ZS_SIZE_CLASS_DELTA must be multiple of ZS_ALIGN * (reason above) */ |
3783689a1
|
141 |
#define ZS_SIZE_CLASS_DELTA (PAGE_SIZE >> CLASS_BITS) |
cf8e0fedf
|
142 143 |
#define ZS_SIZE_CLASSES (DIV_ROUND_UP(ZS_MAX_ALLOC_SIZE - ZS_MIN_ALLOC_SIZE, \ ZS_SIZE_CLASS_DELTA) + 1) |
0959c63f1
|
144 |
|
0959c63f1
|
145 |
enum fullness_group { |
0959c63f1
|
146 |
ZS_EMPTY, |
48b4800a1
|
147 148 149 150 |
ZS_ALMOST_EMPTY, ZS_ALMOST_FULL, ZS_FULL, NR_ZS_FULLNESS, |
0959c63f1
|
151 |
}; |
0f050d997
|
152 |
enum zs_stat_type { |
48b4800a1
|
153 154 155 156 |
CLASS_EMPTY, CLASS_ALMOST_EMPTY, CLASS_ALMOST_FULL, CLASS_FULL, |
0f050d997
|
157 158 |
OBJ_ALLOCATED, OBJ_USED, |
48b4800a1
|
159 |
NR_ZS_STAT_TYPE, |
0f050d997
|
160 |
}; |
0f050d997
|
161 162 163 |
struct zs_size_stat { unsigned long objs[NR_ZS_STAT_TYPE]; }; |
572445941
|
164 165 |
#ifdef CONFIG_ZSMALLOC_STAT static struct dentry *zs_stat_root; |
0f050d997
|
166 |
#endif |
48b4800a1
|
167 168 169 |
#ifdef CONFIG_COMPACTION static struct vfsmount *zsmalloc_mnt; #endif |
0959c63f1
|
170 171 172 173 174 |
/* * We assign a page to ZS_ALMOST_EMPTY fullness group when: * n <= N / f, where * n = number of allocated objects * N = total number of objects zspage can store |
6dd9737e3
|
175 |
* f = fullness_threshold_frac |
0959c63f1
|
176 177 178 179 180 181 182 183 184 |
* * Similarly, we assign zspage to: * ZS_ALMOST_FULL when n > N / f * ZS_EMPTY when n == 0 * ZS_FULL when n == N * * (see: fix_fullness_group()) */ static const int fullness_threshold_frac = 4; |
010b495e2
|
185 |
static size_t huge_class_size; |
0959c63f1
|
186 187 |
struct size_class { |
572445941
|
188 |
spinlock_t lock; |
48b4800a1
|
189 |
struct list_head fullness_list[NR_ZS_FULLNESS]; |
0959c63f1
|
190 191 192 193 194 |
/* * Size of objects stored in this class. Must be multiple * of ZS_ALIGN. */ int size; |
1fc6e27d7
|
195 |
int objs_per_zspage; |
7dfa46122
|
196 197 |
/* Number of PAGE_SIZE sized pages to combine to form a 'zspage' */ int pages_per_zspage; |
48b4800a1
|
198 199 200 |
unsigned int index; struct zs_size_stat stats; |
0959c63f1
|
201 |
}; |
48b4800a1
|
202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 |
/* huge object: pages_per_zspage == 1 && maxobj_per_zspage == 1 */ static void SetPageHugeObject(struct page *page) { SetPageOwnerPriv1(page); } static void ClearPageHugeObject(struct page *page) { ClearPageOwnerPriv1(page); } static int PageHugeObject(struct page *page) { return PageOwnerPriv1(page); } |
0959c63f1
|
217 218 |
/* * Placed within free objects to form a singly linked list. |
3783689a1
|
219 |
* For every zspage, zspage->freeobj gives head of this list. |
0959c63f1
|
220 221 222 223 |
* * This must be power of 2 and less than or equal to ZS_ALIGN */ struct link_free { |
2e40e163a
|
224 225 |
union { /* |
bfd093f5e
|
226 |
* Free object index; |
2e40e163a
|
227 228 |
* It's valid for non-allocated object */ |
bfd093f5e
|
229 |
unsigned long next; |
2e40e163a
|
230 231 232 233 234 |
/* * Handle of allocated object. */ unsigned long handle; }; |
0959c63f1
|
235 236 237 |
}; struct zs_pool { |
6f3526d6d
|
238 |
const char *name; |
0f050d997
|
239 |
|
cf8e0fedf
|
240 |
struct size_class *size_class[ZS_SIZE_CLASSES]; |
2e40e163a
|
241 |
struct kmem_cache *handle_cachep; |
3783689a1
|
242 |
struct kmem_cache *zspage_cachep; |
0959c63f1
|
243 |
|
13de8933c
|
244 |
atomic_long_t pages_allocated; |
0f050d997
|
245 |
|
7d3f39382
|
246 |
struct zs_pool_stats stats; |
ab9d306d9
|
247 248 249 |
/* Compact classes */ struct shrinker shrinker; |
93144ca35
|
250 |
|
0f050d997
|
251 252 253 |
#ifdef CONFIG_ZSMALLOC_STAT struct dentry *stat_dentry; #endif |
48b4800a1
|
254 255 256 |
#ifdef CONFIG_COMPACTION struct inode *inode; struct work_struct free_work; |
701d67859
|
257 258 259 260 |
/* A wait queue for when migration races with async_free_zspage() */ struct wait_queue_head migration_wait; atomic_long_t isolated_pages; bool destroying; |
48b4800a1
|
261 |
#endif |
0959c63f1
|
262 |
}; |
61989a80f
|
263 |
|
3783689a1
|
264 265 266 |
struct zspage { struct { unsigned int fullness:FULLNESS_BITS; |
85d492f28
|
267 |
unsigned int class:CLASS_BITS + 1; |
48b4800a1
|
268 269 |
unsigned int isolated:ISOLATED_BITS; unsigned int magic:MAGIC_VAL_BITS; |
3783689a1
|
270 271 |
}; unsigned int inuse; |
bfd093f5e
|
272 |
unsigned int freeobj; |
3783689a1
|
273 274 |
struct page *first_page; struct list_head list; /* fullness list */ |
48b4800a1
|
275 276 277 |
#ifdef CONFIG_COMPACTION rwlock_t lock; #endif |
3783689a1
|
278 |
}; |
61989a80f
|
279 |
|
f553646a6
|
280 |
struct mapping_area { |
f553646a6
|
281 |
char *vm_buf; /* copy buffer for objects that span pages */ |
f553646a6
|
282 283 284 |
char *vm_addr; /* address of kmap_atomic()'ed pages */ enum zs_mapmode vm_mm; /* mapping mode */ }; |
48b4800a1
|
285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 |
#ifdef CONFIG_COMPACTION static int zs_register_migration(struct zs_pool *pool); static void zs_unregister_migration(struct zs_pool *pool); static void migrate_lock_init(struct zspage *zspage); static void migrate_read_lock(struct zspage *zspage); static void migrate_read_unlock(struct zspage *zspage); static void kick_deferred_free(struct zs_pool *pool); static void init_deferred_free(struct zs_pool *pool); static void SetZsPageMovable(struct zs_pool *pool, struct zspage *zspage); #else static int zsmalloc_mount(void) { return 0; } static void zsmalloc_unmount(void) {} static int zs_register_migration(struct zs_pool *pool) { return 0; } static void zs_unregister_migration(struct zs_pool *pool) {} static void migrate_lock_init(struct zspage *zspage) {} static void migrate_read_lock(struct zspage *zspage) {} static void migrate_read_unlock(struct zspage *zspage) {} static void kick_deferred_free(struct zs_pool *pool) {} static void init_deferred_free(struct zs_pool *pool) {} static void SetZsPageMovable(struct zs_pool *pool, struct zspage *zspage) {} #endif |
3783689a1
|
306 |
static int create_cache(struct zs_pool *pool) |
2e40e163a
|
307 308 309 |
{ pool->handle_cachep = kmem_cache_create("zs_handle", ZS_HANDLE_SIZE, 0, 0, NULL); |
3783689a1
|
310 311 312 313 314 315 316 317 318 319 320 321 |
if (!pool->handle_cachep) return 1; pool->zspage_cachep = kmem_cache_create("zspage", sizeof(struct zspage), 0, 0, NULL); if (!pool->zspage_cachep) { kmem_cache_destroy(pool->handle_cachep); pool->handle_cachep = NULL; return 1; } return 0; |
2e40e163a
|
322 |
} |
3783689a1
|
323 |
static void destroy_cache(struct zs_pool *pool) |
2e40e163a
|
324 |
{ |
cd10add00
|
325 |
kmem_cache_destroy(pool->handle_cachep); |
3783689a1
|
326 |
kmem_cache_destroy(pool->zspage_cachep); |
2e40e163a
|
327 |
} |
3783689a1
|
328 |
static unsigned long cache_alloc_handle(struct zs_pool *pool, gfp_t gfp) |
2e40e163a
|
329 330 |
{ return (unsigned long)kmem_cache_alloc(pool->handle_cachep, |
48b4800a1
|
331 |
gfp & ~(__GFP_HIGHMEM|__GFP_MOVABLE)); |
2e40e163a
|
332 |
} |
3783689a1
|
333 |
static void cache_free_handle(struct zs_pool *pool, unsigned long handle) |
2e40e163a
|
334 335 336 |
{ kmem_cache_free(pool->handle_cachep, (void *)handle); } |
3783689a1
|
337 338 |
static struct zspage *cache_alloc_zspage(struct zs_pool *pool, gfp_t flags) { |
48b4800a1
|
339 340 |
return kmem_cache_alloc(pool->zspage_cachep, flags & ~(__GFP_HIGHMEM|__GFP_MOVABLE)); |
399d8eebe
|
341 |
} |
3783689a1
|
342 343 344 345 346 |
static void cache_free_zspage(struct zs_pool *pool, struct zspage *zspage) { kmem_cache_free(pool->zspage_cachep, zspage); } |
2e40e163a
|
347 348 |
static void record_obj(unsigned long handle, unsigned long obj) { |
c102f07ca
|
349 350 351 352 353 354 |
/* * lsb of @obj represents handle lock while other bits * represent object value the handle is pointing so * updating shouldn't do store tearing. */ WRITE_ONCE(*(unsigned long *)handle, obj); |
2e40e163a
|
355 |
} |
c795779df
|
356 357 358 |
/* zpool driver */ #ifdef CONFIG_ZPOOL |
6f3526d6d
|
359 |
static void *zs_zpool_create(const char *name, gfp_t gfp, |
786727799
|
360 |
const struct zpool_ops *zpool_ops, |
479305fd7
|
361 |
struct zpool *zpool) |
c795779df
|
362 |
{ |
d0d8da2dc
|
363 364 365 366 367 368 |
/* * Ignore global gfp flags: zs_malloc() may be invoked from * different contexts and its caller must provide a valid * gfp mask. */ return zs_create_pool(name); |
c795779df
|
369 370 371 372 373 374 375 376 377 378 |
} static void zs_zpool_destroy(void *pool) { zs_destroy_pool(pool); } static int zs_zpool_malloc(void *pool, size_t size, gfp_t gfp, unsigned long *handle) { |
d0d8da2dc
|
379 |
*handle = zs_malloc(pool, size, gfp); |
c795779df
|
380 381 382 383 384 385 |
return *handle ? 0 : -1; } static void zs_zpool_free(void *pool, unsigned long handle) { zs_free(pool, handle); } |
c795779df
|
386 387 388 389 390 391 392 393 394 395 396 397 |
static void *zs_zpool_map(void *pool, unsigned long handle, enum zpool_mapmode mm) { enum zs_mapmode zs_mm; switch (mm) { case ZPOOL_MM_RO: zs_mm = ZS_MM_RO; break; case ZPOOL_MM_WO: zs_mm = ZS_MM_WO; break; |
e4a9bc589
|
398 |
case ZPOOL_MM_RW: |
c795779df
|
399 400 401 402 403 404 405 406 407 408 409 410 411 412 |
default: zs_mm = ZS_MM_RW; break; } return zs_map_object(pool, handle, zs_mm); } static void zs_zpool_unmap(void *pool, unsigned long handle) { zs_unmap_object(pool, handle); } static u64 zs_zpool_total_size(void *pool) { |
722cdc172
|
413 |
return zs_get_total_pages(pool) << PAGE_SHIFT; |
c795779df
|
414 415 416 |
} static struct zpool_driver zs_zpool_driver = { |
c165f25d2
|
417 418 419 420 421 422 423 424 425 426 |
.type = "zsmalloc", .owner = THIS_MODULE, .create = zs_zpool_create, .destroy = zs_zpool_destroy, .malloc_support_movable = true, .malloc = zs_zpool_malloc, .free = zs_zpool_free, .map = zs_zpool_map, .unmap = zs_zpool_unmap, .total_size = zs_zpool_total_size, |
c795779df
|
427 |
}; |
137f8cff5
|
428 |
MODULE_ALIAS("zpool-zsmalloc"); |
c795779df
|
429 |
#endif /* CONFIG_ZPOOL */ |
61989a80f
|
430 431 |
/* per-cpu VM mapping areas for zspage accesses that cross page boundaries */ static DEFINE_PER_CPU(struct mapping_area, zs_map_area); |
48b4800a1
|
432 433 434 435 |
static bool is_zspage_isolated(struct zspage *zspage) { return zspage->isolated; } |
3457f4147
|
436 |
static __maybe_unused int is_first_page(struct page *page) |
61989a80f
|
437 |
{ |
a27545bf0
|
438 |
return PagePrivate(page); |
61989a80f
|
439 |
} |
48b4800a1
|
440 |
/* Protected by class->lock */ |
3783689a1
|
441 |
static inline int get_zspage_inuse(struct zspage *zspage) |
4f42047bb
|
442 |
{ |
3783689a1
|
443 |
return zspage->inuse; |
4f42047bb
|
444 |
} |
4f42047bb
|
445 |
|
3783689a1
|
446 |
static inline void mod_zspage_inuse(struct zspage *zspage, int val) |
4f42047bb
|
447 |
{ |
3783689a1
|
448 |
zspage->inuse += val; |
4f42047bb
|
449 |
} |
48b4800a1
|
450 |
static inline struct page *get_first_page(struct zspage *zspage) |
4f42047bb
|
451 |
{ |
48b4800a1
|
452 |
struct page *first_page = zspage->first_page; |
3783689a1
|
453 |
|
48b4800a1
|
454 455 |
VM_BUG_ON_PAGE(!is_first_page(first_page), first_page); return first_page; |
4f42047bb
|
456 |
} |
48b4800a1
|
457 |
static inline int get_first_obj_offset(struct page *page) |
4f42047bb
|
458 |
{ |
48b4800a1
|
459 460 |
return page->units; } |
3783689a1
|
461 |
|
48b4800a1
|
462 463 464 |
static inline void set_first_obj_offset(struct page *page, int offset) { page->units = offset; |
4f42047bb
|
465 |
} |
bfd093f5e
|
466 |
static inline unsigned int get_freeobj(struct zspage *zspage) |
4f42047bb
|
467 |
{ |
bfd093f5e
|
468 |
return zspage->freeobj; |
4f42047bb
|
469 |
} |
bfd093f5e
|
470 |
static inline void set_freeobj(struct zspage *zspage, unsigned int obj) |
4f42047bb
|
471 |
{ |
bfd093f5e
|
472 |
zspage->freeobj = obj; |
4f42047bb
|
473 |
} |
3783689a1
|
474 |
static void get_zspage_mapping(struct zspage *zspage, |
a42094676
|
475 |
unsigned int *class_idx, |
61989a80f
|
476 477 |
enum fullness_group *fullness) { |
48b4800a1
|
478 |
BUG_ON(zspage->magic != ZSPAGE_MAGIC); |
3783689a1
|
479 480 |
*fullness = zspage->fullness; *class_idx = zspage->class; |
61989a80f
|
481 |
} |
3783689a1
|
482 |
static void set_zspage_mapping(struct zspage *zspage, |
a42094676
|
483 |
unsigned int class_idx, |
61989a80f
|
484 485 |
enum fullness_group fullness) { |
3783689a1
|
486 487 |
zspage->class = class_idx; zspage->fullness = fullness; |
61989a80f
|
488 |
} |
c3e3e88ad
|
489 490 491 492 493 494 495 |
/* * zsmalloc divides the pool into various size classes where each * class maintains a list of zspages where each zspage is divided * into equal sized chunks. Each allocation falls into one of these * classes depending on its size. This function returns index of the * size class which has chunk size big enough to hold the give size. */ |
61989a80f
|
496 497 498 499 500 501 502 |
static int get_size_class_index(int size) { int idx = 0; if (likely(size > ZS_MIN_ALLOC_SIZE)) idx = DIV_ROUND_UP(size - ZS_MIN_ALLOC_SIZE, ZS_SIZE_CLASS_DELTA); |
cf8e0fedf
|
503 |
return min_t(int, ZS_SIZE_CLASSES - 1, idx); |
61989a80f
|
504 |
} |
3eb95feac
|
505 |
/* type can be of enum type zs_stat_type or fullness_group */ |
248ca1b05
|
506 |
static inline void zs_stat_inc(struct size_class *class, |
3eb95feac
|
507 |
int type, unsigned long cnt) |
248ca1b05
|
508 |
{ |
48b4800a1
|
509 |
class->stats.objs[type] += cnt; |
248ca1b05
|
510 |
} |
3eb95feac
|
511 |
/* type can be of enum type zs_stat_type or fullness_group */ |
248ca1b05
|
512 |
static inline void zs_stat_dec(struct size_class *class, |
3eb95feac
|
513 |
int type, unsigned long cnt) |
248ca1b05
|
514 |
{ |
48b4800a1
|
515 |
class->stats.objs[type] -= cnt; |
248ca1b05
|
516 |
} |
3eb95feac
|
517 |
/* type can be of enum type zs_stat_type or fullness_group */ |
248ca1b05
|
518 |
static inline unsigned long zs_stat_get(struct size_class *class, |
3eb95feac
|
519 |
int type) |
248ca1b05
|
520 |
{ |
48b4800a1
|
521 |
return class->stats.objs[type]; |
248ca1b05
|
522 |
} |
572445941
|
523 |
#ifdef CONFIG_ZSMALLOC_STAT |
4abaac9b7
|
524 |
static void __init zs_stat_init(void) |
248ca1b05
|
525 |
{ |
4abaac9b7
|
526 527 528 529 530 |
if (!debugfs_initialized()) { pr_warn("debugfs not available, stat dir not created "); return; } |
248ca1b05
|
531 532 |
zs_stat_root = debugfs_create_dir("zsmalloc", NULL); |
248ca1b05
|
533 534 535 536 537 538 |
} static void __exit zs_stat_exit(void) { debugfs_remove_recursive(zs_stat_root); } |
1120ed548
|
539 |
static unsigned long zs_can_compact(struct size_class *class); |
248ca1b05
|
540 541 542 543 544 545 546 |
static int zs_stats_size_show(struct seq_file *s, void *v) { int i; struct zs_pool *pool = s->private; struct size_class *class; int objs_per_zspage; unsigned long class_almost_full, class_almost_empty; |
1120ed548
|
547 |
unsigned long obj_allocated, obj_used, pages_used, freeable; |
248ca1b05
|
548 549 |
unsigned long total_class_almost_full = 0, total_class_almost_empty = 0; unsigned long total_objs = 0, total_used_objs = 0, total_pages = 0; |
1120ed548
|
550 |
unsigned long total_freeable = 0; |
248ca1b05
|
551 |
|
1120ed548
|
552 553 |
seq_printf(s, " %5s %5s %11s %12s %13s %10s %10s %16s %8s ", |
248ca1b05
|
554 555 |
"class", "size", "almost_full", "almost_empty", "obj_allocated", "obj_used", "pages_used", |
1120ed548
|
556 |
"pages_per_zspage", "freeable"); |
248ca1b05
|
557 |
|
cf8e0fedf
|
558 |
for (i = 0; i < ZS_SIZE_CLASSES; i++) { |
248ca1b05
|
559 560 561 562 563 564 565 566 567 568 |
class = pool->size_class[i]; if (class->index != i) continue; spin_lock(&class->lock); class_almost_full = zs_stat_get(class, CLASS_ALMOST_FULL); class_almost_empty = zs_stat_get(class, CLASS_ALMOST_EMPTY); obj_allocated = zs_stat_get(class, OBJ_ALLOCATED); obj_used = zs_stat_get(class, OBJ_USED); |
1120ed548
|
569 |
freeable = zs_can_compact(class); |
248ca1b05
|
570 |
spin_unlock(&class->lock); |
b4fd07a08
|
571 |
objs_per_zspage = class->objs_per_zspage; |
248ca1b05
|
572 573 |
pages_used = obj_allocated / objs_per_zspage * class->pages_per_zspage; |
1120ed548
|
574 575 576 |
seq_printf(s, " %5u %5u %11lu %12lu %13lu" " %10lu %10lu %16d %8lu ", |
248ca1b05
|
577 578 |
i, class->size, class_almost_full, class_almost_empty, obj_allocated, obj_used, pages_used, |
1120ed548
|
579 |
class->pages_per_zspage, freeable); |
248ca1b05
|
580 581 582 583 584 585 |
total_class_almost_full += class_almost_full; total_class_almost_empty += class_almost_empty; total_objs += obj_allocated; total_used_objs += obj_used; total_pages += pages_used; |
1120ed548
|
586 |
total_freeable += freeable; |
248ca1b05
|
587 588 589 590 |
} seq_puts(s, " "); |
1120ed548
|
591 592 |
seq_printf(s, " %5s %5s %11lu %12lu %13lu %10lu %10lu %16s %8lu ", |
248ca1b05
|
593 594 |
"Total", "", total_class_almost_full, total_class_almost_empty, total_objs, |
1120ed548
|
595 |
total_used_objs, total_pages, "", total_freeable); |
248ca1b05
|
596 597 598 |
return 0; } |
5ad350936
|
599 |
DEFINE_SHOW_ATTRIBUTE(zs_stats_size); |
248ca1b05
|
600 |
|
d34f61572
|
601 |
static void zs_pool_stat_create(struct zs_pool *pool, const char *name) |
248ca1b05
|
602 |
{ |
4abaac9b7
|
603 604 605 |
if (!zs_stat_root) { pr_warn("no root stat dir, not creating <%s> stat dir ", name); |
d34f61572
|
606 |
return; |
4abaac9b7
|
607 |
} |
248ca1b05
|
608 |
|
4268509a3
|
609 610 611 612 |
pool->stat_dentry = debugfs_create_dir(name, zs_stat_root); debugfs_create_file("classes", S_IFREG | 0444, pool->stat_dentry, pool, &zs_stats_size_fops); |
248ca1b05
|
613 614 615 616 617 618 619 620 |
} static void zs_pool_stat_destroy(struct zs_pool *pool) { debugfs_remove_recursive(pool->stat_dentry); } #else /* CONFIG_ZSMALLOC_STAT */ |
4abaac9b7
|
621 |
static void __init zs_stat_init(void) |
248ca1b05
|
622 |
{ |
248ca1b05
|
623 624 625 626 627 |
} static void __exit zs_stat_exit(void) { } |
d34f61572
|
628 |
static inline void zs_pool_stat_create(struct zs_pool *pool, const char *name) |
248ca1b05
|
629 |
{ |
248ca1b05
|
630 631 632 633 634 |
} static inline void zs_pool_stat_destroy(struct zs_pool *pool) { } |
248ca1b05
|
635 |
#endif |
48b4800a1
|
636 |
|
c3e3e88ad
|
637 638 639 640 641 642 643 |
/* * For each size class, zspages are divided into different groups * depending on how "full" they are. This was done so that we could * easily find empty or nearly empty zspages when we try to shrink * the pool (not yet implemented). This function returns fullness * status of the given page. */ |
1fc6e27d7
|
644 |
static enum fullness_group get_fullness_group(struct size_class *class, |
3783689a1
|
645 |
struct zspage *zspage) |
61989a80f
|
646 |
{ |
1fc6e27d7
|
647 |
int inuse, objs_per_zspage; |
61989a80f
|
648 |
enum fullness_group fg; |
830e4bc5b
|
649 |
|
3783689a1
|
650 |
inuse = get_zspage_inuse(zspage); |
1fc6e27d7
|
651 |
objs_per_zspage = class->objs_per_zspage; |
61989a80f
|
652 653 654 |
if (inuse == 0) fg = ZS_EMPTY; |
1fc6e27d7
|
655 |
else if (inuse == objs_per_zspage) |
61989a80f
|
656 |
fg = ZS_FULL; |
1fc6e27d7
|
657 |
else if (inuse <= 3 * objs_per_zspage / fullness_threshold_frac) |
61989a80f
|
658 659 660 661 662 663 |
fg = ZS_ALMOST_EMPTY; else fg = ZS_ALMOST_FULL; return fg; } |
c3e3e88ad
|
664 665 666 667 668 669 |
/* * Each size class maintains various freelists and zspages are assigned * to one of these freelists based on the number of live objects they * have. This functions inserts the given zspage into the freelist * identified by <class, fullness_group>. */ |
251cbb951
|
670 |
static void insert_zspage(struct size_class *class, |
3783689a1
|
671 672 |
struct zspage *zspage, enum fullness_group fullness) |
61989a80f
|
673 |
{ |
3783689a1
|
674 |
struct zspage *head; |
61989a80f
|
675 |
|
48b4800a1
|
676 |
zs_stat_inc(class, fullness, 1); |
3783689a1
|
677 678 |
head = list_first_entry_or_null(&class->fullness_list[fullness], struct zspage, list); |
58f171174
|
679 |
/* |
3783689a1
|
680 681 |
* We want to see more ZS_FULL pages and less almost empty/full. * Put pages with higher ->inuse first. |
58f171174
|
682 |
*/ |
3783689a1
|
683 684 685 686 687 688 689 |
if (head) { if (get_zspage_inuse(zspage) < get_zspage_inuse(head)) { list_add(&zspage->list, &head->list); return; } } list_add(&zspage->list, &class->fullness_list[fullness]); |
61989a80f
|
690 |
} |
c3e3e88ad
|
691 692 693 694 |
/* * This function removes the given zspage from the freelist identified * by <class, fullness_group>. */ |
251cbb951
|
695 |
static void remove_zspage(struct size_class *class, |
3783689a1
|
696 697 |
struct zspage *zspage, enum fullness_group fullness) |
61989a80f
|
698 |
{ |
3783689a1
|
699 |
VM_BUG_ON(list_empty(&class->fullness_list[fullness])); |
48b4800a1
|
700 |
VM_BUG_ON(is_zspage_isolated(zspage)); |
61989a80f
|
701 |
|
3783689a1
|
702 |
list_del_init(&zspage->list); |
48b4800a1
|
703 |
zs_stat_dec(class, fullness, 1); |
61989a80f
|
704 |
} |
c3e3e88ad
|
705 706 707 708 709 710 711 712 713 |
/* * Each size class maintains zspages in different fullness groups depending * on the number of live objects they contain. When allocating or freeing * objects, the fullness status of the page can change, say, from ALMOST_FULL * to ALMOST_EMPTY when freeing an object. This function checks if such * a status change has occurred for the given page and accordingly moves the * page from the freelist of the old fullness group to that of the new * fullness group. */ |
c78062612
|
714 |
static enum fullness_group fix_fullness_group(struct size_class *class, |
3783689a1
|
715 |
struct zspage *zspage) |
61989a80f
|
716 717 |
{ int class_idx; |
61989a80f
|
718 |
enum fullness_group currfg, newfg; |
3783689a1
|
719 720 |
get_zspage_mapping(zspage, &class_idx, &currfg); newfg = get_fullness_group(class, zspage); |
61989a80f
|
721 722 |
if (newfg == currfg) goto out; |
48b4800a1
|
723 724 725 726 |
if (!is_zspage_isolated(zspage)) { remove_zspage(class, zspage, currfg); insert_zspage(class, zspage, newfg); } |
3783689a1
|
727 |
set_zspage_mapping(zspage, class_idx, newfg); |
61989a80f
|
728 729 730 731 732 733 734 735 736 737 |
out: return newfg; } /* * We have to decide on how many pages to link together * to form a zspage for each size class. This is important * to reduce wastage due to unusable space left at end of * each zspage which is given as: |
888fa374e
|
738 739 |
* wastage = Zp % class_size * usage = Zp - wastage |
61989a80f
|
740 741 742 743 744 745 |
* where Zp = zspage size = k * PAGE_SIZE where k = 1, 2, ... * * For example, for size class of 3/8 * PAGE_SIZE, we should * link together 3 PAGE_SIZE sized pages to form a zspage * since then we can perfectly fit in 8 such objects. */ |
2e3b61547
|
746 |
static int get_pages_per_zspage(int class_size) |
61989a80f
|
747 748 749 750 |
{ int i, max_usedpc = 0; /* zspage order which gives maximum used size per KB */ int max_usedpc_order = 1; |
84d4faaba
|
751 |
for (i = 1; i <= ZS_MAX_PAGES_PER_ZSPAGE; i++) { |
61989a80f
|
752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 |
int zspage_size; int waste, usedpc; zspage_size = i * PAGE_SIZE; waste = zspage_size % class_size; usedpc = (zspage_size - waste) * 100 / zspage_size; if (usedpc > max_usedpc) { max_usedpc = usedpc; max_usedpc_order = i; } } return max_usedpc_order; } |
3783689a1
|
767 |
static struct zspage *get_zspage(struct page *page) |
61989a80f
|
768 |
{ |
48b4800a1
|
769 770 771 772 |
struct zspage *zspage = (struct zspage *)page->private; BUG_ON(zspage->magic != ZSPAGE_MAGIC); return zspage; |
61989a80f
|
773 774 775 776 |
} static struct page *get_next_page(struct page *page) { |
48b4800a1
|
777 778 779 780 |
if (unlikely(PageHugeObject(page))) return NULL; return page->freelist; |
61989a80f
|
781 |
} |
bfd093f5e
|
782 783 |
/** * obj_to_location - get (<page>, <obj_idx>) from encoded object value |
e8b098fc5
|
784 |
* @obj: the encoded object value |
bfd093f5e
|
785 786 |
* @page: page object resides in zspage * @obj_idx: object index |
67296874e
|
787 |
*/ |
bfd093f5e
|
788 789 |
static void obj_to_location(unsigned long obj, struct page **page, unsigned int *obj_idx) |
61989a80f
|
790 |
{ |
bfd093f5e
|
791 792 793 794 |
obj >>= OBJ_TAG_BITS; *page = pfn_to_page(obj >> OBJ_INDEX_BITS); *obj_idx = (obj & OBJ_INDEX_MASK); } |
61989a80f
|
795 |
|
bfd093f5e
|
796 797 798 799 800 801 802 803 |
/** * location_to_obj - get obj value encoded from (<page>, <obj_idx>) * @page: page object resides in zspage * @obj_idx: object index */ static unsigned long location_to_obj(struct page *page, unsigned int obj_idx) { unsigned long obj; |
61989a80f
|
804 |
|
312fcae22
|
805 |
obj = page_to_pfn(page) << OBJ_INDEX_BITS; |
bfd093f5e
|
806 |
obj |= obj_idx & OBJ_INDEX_MASK; |
312fcae22
|
807 |
obj <<= OBJ_TAG_BITS; |
61989a80f
|
808 |
|
bfd093f5e
|
809 |
return obj; |
61989a80f
|
810 |
} |
2e40e163a
|
811 812 813 814 |
static unsigned long handle_to_obj(unsigned long handle) { return *(unsigned long *)handle; } |
48b4800a1
|
815 |
static unsigned long obj_to_head(struct page *page, void *obj) |
312fcae22
|
816 |
{ |
48b4800a1
|
817 |
if (unlikely(PageHugeObject(page))) { |
830e4bc5b
|
818 |
VM_BUG_ON_PAGE(!is_first_page(page), page); |
3783689a1
|
819 |
return page->index; |
7b60a6852
|
820 821 |
} else return *(unsigned long *)obj; |
312fcae22
|
822 |
} |
48b4800a1
|
823 824 825 826 |
static inline int testpin_tag(unsigned long handle) { return bit_spin_is_locked(HANDLE_PIN_BIT, (unsigned long *)handle); } |
312fcae22
|
827 828 |
static inline int trypin_tag(unsigned long handle) { |
1b8320b62
|
829 |
return bit_spin_trylock(HANDLE_PIN_BIT, (unsigned long *)handle); |
312fcae22
|
830 |
} |
70c7ec95b
|
831 |
static void pin_tag(unsigned long handle) __acquires(bitlock) |
312fcae22
|
832 |
{ |
1b8320b62
|
833 |
bit_spin_lock(HANDLE_PIN_BIT, (unsigned long *)handle); |
312fcae22
|
834 |
} |
bc22b18b1
|
835 |
static void unpin_tag(unsigned long handle) __releases(bitlock) |
312fcae22
|
836 |
{ |
1b8320b62
|
837 |
bit_spin_unlock(HANDLE_PIN_BIT, (unsigned long *)handle); |
312fcae22
|
838 |
} |
f4477e90b
|
839 840 |
static void reset_page(struct page *page) { |
48b4800a1
|
841 |
__ClearPageMovable(page); |
18fd06bf7
|
842 |
ClearPagePrivate(page); |
f4477e90b
|
843 |
set_page_private(page, 0); |
48b4800a1
|
844 845 846 847 |
page_mapcount_reset(page); ClearPageHugeObject(page); page->freelist = NULL; } |
4d0a5402f
|
848 |
static int trylock_zspage(struct zspage *zspage) |
48b4800a1
|
849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 |
{ struct page *cursor, *fail; for (cursor = get_first_page(zspage); cursor != NULL; cursor = get_next_page(cursor)) { if (!trylock_page(cursor)) { fail = cursor; goto unlock; } } return 1; unlock: for (cursor = get_first_page(zspage); cursor != fail; cursor = get_next_page(cursor)) unlock_page(cursor); return 0; |
f4477e90b
|
867 |
} |
48b4800a1
|
868 869 |
static void __free_zspage(struct zs_pool *pool, struct size_class *class, struct zspage *zspage) |
61989a80f
|
870 |
{ |
3783689a1
|
871 |
struct page *page, *next; |
48b4800a1
|
872 873 874 875 876 877 |
enum fullness_group fg; unsigned int class_idx; get_zspage_mapping(zspage, &class_idx, &fg); assert_spin_locked(&class->lock); |
61989a80f
|
878 |
|
3783689a1
|
879 |
VM_BUG_ON(get_zspage_inuse(zspage)); |
48b4800a1
|
880 |
VM_BUG_ON(fg != ZS_EMPTY); |
61989a80f
|
881 |
|
48b4800a1
|
882 |
next = page = get_first_page(zspage); |
3783689a1
|
883 |
do { |
48b4800a1
|
884 885 |
VM_BUG_ON_PAGE(!PageLocked(page), page); next = get_next_page(page); |
3783689a1
|
886 |
reset_page(page); |
48b4800a1
|
887 |
unlock_page(page); |
91537fee0
|
888 |
dec_zone_page_state(page, NR_ZSPAGES); |
3783689a1
|
889 890 891 |
put_page(page); page = next; } while (page != NULL); |
61989a80f
|
892 |
|
3783689a1
|
893 |
cache_free_zspage(pool, zspage); |
48b4800a1
|
894 |
|
b4fd07a08
|
895 |
zs_stat_dec(class, OBJ_ALLOCATED, class->objs_per_zspage); |
48b4800a1
|
896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 |
atomic_long_sub(class->pages_per_zspage, &pool->pages_allocated); } static void free_zspage(struct zs_pool *pool, struct size_class *class, struct zspage *zspage) { VM_BUG_ON(get_zspage_inuse(zspage)); VM_BUG_ON(list_empty(&zspage->list)); if (!trylock_zspage(zspage)) { kick_deferred_free(pool); return; } remove_zspage(class, zspage, ZS_EMPTY); __free_zspage(pool, class, zspage); |
61989a80f
|
913 914 915 |
} /* Initialize a newly allocated zspage */ |
3783689a1
|
916 |
static void init_zspage(struct size_class *class, struct zspage *zspage) |
61989a80f
|
917 |
{ |
bfd093f5e
|
918 |
unsigned int freeobj = 1; |
61989a80f
|
919 |
unsigned long off = 0; |
48b4800a1
|
920 |
struct page *page = get_first_page(zspage); |
830e4bc5b
|
921 |
|
61989a80f
|
922 923 924 |
while (page) { struct page *next_page; struct link_free *link; |
af4ee5e97
|
925 |
void *vaddr; |
61989a80f
|
926 |
|
3783689a1
|
927 |
set_first_obj_offset(page, off); |
61989a80f
|
928 |
|
af4ee5e97
|
929 930 |
vaddr = kmap_atomic(page); link = (struct link_free *)vaddr + off / sizeof(*link); |
5538c5623
|
931 932 |
while ((off += class->size) < PAGE_SIZE) { |
3b1d9ca65
|
933 |
link->next = freeobj++ << OBJ_TAG_BITS; |
5538c5623
|
934 |
link += class->size / sizeof(*link); |
61989a80f
|
935 936 937 938 939 940 941 942 |
} /* * We now come to the last (full or partial) object on this * page, which must point to the first object on the next * page (if present) */ next_page = get_next_page(page); |
bfd093f5e
|
943 |
if (next_page) { |
3b1d9ca65
|
944 |
link->next = freeobj++ << OBJ_TAG_BITS; |
bfd093f5e
|
945 946 |
} else { /* |
3b1d9ca65
|
947 |
* Reset OBJ_TAG_BITS bit to last link to tell |
bfd093f5e
|
948 949 |
* whether it's allocated object or not. */ |
01a6ad9ac
|
950 |
link->next = -1UL << OBJ_TAG_BITS; |
bfd093f5e
|
951 |
} |
af4ee5e97
|
952 |
kunmap_atomic(vaddr); |
61989a80f
|
953 |
page = next_page; |
5538c5623
|
954 |
off %= PAGE_SIZE; |
61989a80f
|
955 |
} |
bdb0af7ca
|
956 |
|
bfd093f5e
|
957 |
set_freeobj(zspage, 0); |
61989a80f
|
958 |
} |
48b4800a1
|
959 960 |
static void create_page_chain(struct size_class *class, struct zspage *zspage, struct page *pages[]) |
61989a80f
|
961 |
{ |
bdb0af7ca
|
962 963 964 |
int i; struct page *page; struct page *prev_page = NULL; |
48b4800a1
|
965 |
int nr_pages = class->pages_per_zspage; |
61989a80f
|
966 967 968 |
/* * Allocate individual pages and link them together as: |
48b4800a1
|
969 |
* 1. all pages are linked together using page->freelist |
3783689a1
|
970 |
* 2. each sub-page point to zspage using page->private |
61989a80f
|
971 |
* |
3783689a1
|
972 |
* we set PG_private to identify the first page (i.e. no other sub-page |
22c5cef16
|
973 |
* has this flag set). |
61989a80f
|
974 |
*/ |
bdb0af7ca
|
975 976 |
for (i = 0; i < nr_pages; i++) { page = pages[i]; |
3783689a1
|
977 |
set_page_private(page, (unsigned long)zspage); |
48b4800a1
|
978 |
page->freelist = NULL; |
bdb0af7ca
|
979 |
if (i == 0) { |
3783689a1
|
980 |
zspage->first_page = page; |
a27545bf0
|
981 |
SetPagePrivate(page); |
48b4800a1
|
982 983 984 |
if (unlikely(class->objs_per_zspage == 1 && class->pages_per_zspage == 1)) SetPageHugeObject(page); |
3783689a1
|
985 |
} else { |
48b4800a1
|
986 |
prev_page->freelist = page; |
61989a80f
|
987 |
} |
61989a80f
|
988 989 |
prev_page = page; } |
bdb0af7ca
|
990 |
} |
61989a80f
|
991 |
|
bdb0af7ca
|
992 993 994 |
/* * Allocate a zspage for the given size class */ |
3783689a1
|
995 996 997 |
static struct zspage *alloc_zspage(struct zs_pool *pool, struct size_class *class, gfp_t gfp) |
bdb0af7ca
|
998 999 |
{ int i; |
bdb0af7ca
|
1000 |
struct page *pages[ZS_MAX_PAGES_PER_ZSPAGE]; |
3783689a1
|
1001 1002 1003 1004 1005 1006 |
struct zspage *zspage = cache_alloc_zspage(pool, gfp); if (!zspage) return NULL; memset(zspage, 0, sizeof(struct zspage)); |
48b4800a1
|
1007 1008 |
zspage->magic = ZSPAGE_MAGIC; migrate_lock_init(zspage); |
61989a80f
|
1009 |
|
bdb0af7ca
|
1010 1011 |
for (i = 0; i < class->pages_per_zspage; i++) { struct page *page; |
61989a80f
|
1012 |
|
3783689a1
|
1013 |
page = alloc_page(gfp); |
bdb0af7ca
|
1014 |
if (!page) { |
91537fee0
|
1015 1016 |
while (--i >= 0) { dec_zone_page_state(pages[i], NR_ZSPAGES); |
bdb0af7ca
|
1017 |
__free_page(pages[i]); |
91537fee0
|
1018 |
} |
3783689a1
|
1019 |
cache_free_zspage(pool, zspage); |
bdb0af7ca
|
1020 1021 |
return NULL; } |
91537fee0
|
1022 1023 |
inc_zone_page_state(page, NR_ZSPAGES); |
bdb0af7ca
|
1024 |
pages[i] = page; |
61989a80f
|
1025 |
} |
48b4800a1
|
1026 |
create_page_chain(class, zspage, pages); |
3783689a1
|
1027 |
init_zspage(class, zspage); |
bdb0af7ca
|
1028 |
|
3783689a1
|
1029 |
return zspage; |
61989a80f
|
1030 |
} |
3783689a1
|
1031 |
static struct zspage *find_get_zspage(struct size_class *class) |
61989a80f
|
1032 1033 |
{ int i; |
3783689a1
|
1034 |
struct zspage *zspage; |
61989a80f
|
1035 |
|
48b4800a1
|
1036 |
for (i = ZS_ALMOST_FULL; i >= ZS_EMPTY; i--) { |
3783689a1
|
1037 1038 1039 |
zspage = list_first_entry_or_null(&class->fullness_list[i], struct zspage, list); if (zspage) |
61989a80f
|
1040 1041 |
break; } |
3783689a1
|
1042 |
return zspage; |
61989a80f
|
1043 |
} |
f553646a6
|
1044 1045 1046 1047 1048 1049 1050 1051 |
static inline int __zs_cpu_up(struct mapping_area *area) { /* * Make sure we don't leak memory if a cpu UP notification * and zs_init() race and both call zs_cpu_up() on the same cpu */ if (area->vm_buf) return 0; |
40f9fb8cf
|
1052 |
area->vm_buf = kmalloc(ZS_MAX_ALLOC_SIZE, GFP_KERNEL); |
f553646a6
|
1053 1054 1055 1056 1057 1058 1059 |
if (!area->vm_buf) return -ENOMEM; return 0; } static inline void __zs_cpu_down(struct mapping_area *area) { |
40f9fb8cf
|
1060 |
kfree(area->vm_buf); |
f553646a6
|
1061 1062 1063 1064 1065 |
area->vm_buf = NULL; } static void *__zs_map_object(struct mapping_area *area, struct page *pages[2], int off, int size) |
5f601902c
|
1066 |
{ |
5f601902c
|
1067 1068 |
int sizes[2]; void *addr; |
f553646a6
|
1069 |
char *buf = area->vm_buf; |
5f601902c
|
1070 |
|
f553646a6
|
1071 1072 1073 1074 1075 1076 |
/* disable page faults to match kmap_atomic() return conditions */ pagefault_disable(); /* no read fastpath */ if (area->vm_mm == ZS_MM_WO) goto out; |
5f601902c
|
1077 1078 1079 |
sizes[0] = PAGE_SIZE - off; sizes[1] = size - sizes[0]; |
5f601902c
|
1080 1081 1082 1083 1084 1085 1086 |
/* copy object to per-cpu buffer */ addr = kmap_atomic(pages[0]); memcpy(buf, addr + off, sizes[0]); kunmap_atomic(addr); addr = kmap_atomic(pages[1]); memcpy(buf + sizes[0], addr, sizes[1]); kunmap_atomic(addr); |
f553646a6
|
1087 1088 |
out: return area->vm_buf; |
5f601902c
|
1089 |
} |
f553646a6
|
1090 1091 |
static void __zs_unmap_object(struct mapping_area *area, struct page *pages[2], int off, int size) |
5f601902c
|
1092 |
{ |
5f601902c
|
1093 1094 |
int sizes[2]; void *addr; |
2e40e163a
|
1095 |
char *buf; |
5f601902c
|
1096 |
|
f553646a6
|
1097 1098 1099 |
/* no write fastpath */ if (area->vm_mm == ZS_MM_RO) goto out; |
5f601902c
|
1100 |
|
7b60a6852
|
1101 |
buf = area->vm_buf; |
a82cbf071
|
1102 1103 1104 |
buf = buf + ZS_HANDLE_SIZE; size -= ZS_HANDLE_SIZE; off += ZS_HANDLE_SIZE; |
2e40e163a
|
1105 |
|
5f601902c
|
1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 |
sizes[0] = PAGE_SIZE - off; sizes[1] = size - sizes[0]; /* copy per-cpu buffer to object */ addr = kmap_atomic(pages[0]); memcpy(addr + off, buf, sizes[0]); kunmap_atomic(addr); addr = kmap_atomic(pages[1]); memcpy(addr, buf + sizes[0], sizes[1]); kunmap_atomic(addr); |
f553646a6
|
1116 1117 1118 1119 |
out: /* enable page faults to match kunmap_atomic() return conditions */ pagefault_enable(); |
5f601902c
|
1120 |
} |
61989a80f
|
1121 |
|
215c89d05
|
1122 |
static int zs_cpu_prepare(unsigned int cpu) |
61989a80f
|
1123 |
{ |
61989a80f
|
1124 |
struct mapping_area *area; |
215c89d05
|
1125 1126 |
area = &per_cpu(zs_map_area, cpu); return __zs_cpu_up(area); |
61989a80f
|
1127 |
} |
215c89d05
|
1128 |
static int zs_cpu_dead(unsigned int cpu) |
61989a80f
|
1129 |
{ |
215c89d05
|
1130 |
struct mapping_area *area; |
40f9fb8cf
|
1131 |
|
215c89d05
|
1132 1133 1134 |
area = &per_cpu(zs_map_area, cpu); __zs_cpu_down(area); return 0; |
b1b00a5b8
|
1135 |
} |
64d90465f
|
1136 1137 |
static bool can_merge(struct size_class *prev, int pages_per_zspage, int objs_per_zspage) |
9eec4cd53
|
1138 |
{ |
64d90465f
|
1139 1140 1141 |
if (prev->pages_per_zspage == pages_per_zspage && prev->objs_per_zspage == objs_per_zspage) return true; |
9eec4cd53
|
1142 |
|
64d90465f
|
1143 |
return false; |
9eec4cd53
|
1144 |
} |
3783689a1
|
1145 |
static bool zspage_full(struct size_class *class, struct zspage *zspage) |
312fcae22
|
1146 |
{ |
3783689a1
|
1147 |
return get_zspage_inuse(zspage) == class->objs_per_zspage; |
312fcae22
|
1148 |
} |
66cdef663
|
1149 1150 1151 1152 1153 |
unsigned long zs_get_total_pages(struct zs_pool *pool) { return atomic_long_read(&pool->pages_allocated); } EXPORT_SYMBOL_GPL(zs_get_total_pages); |
4bbc0bc06
|
1154 |
/** |
66cdef663
|
1155 1156 1157 |
* zs_map_object - get address of allocated object from handle. * @pool: pool from which the object was allocated * @handle: handle returned from zs_malloc |
e8b098fc5
|
1158 |
* @mm: maping mode to use |
4bbc0bc06
|
1159 |
* |
66cdef663
|
1160 1161 1162 |
* Before using an object allocated from zs_malloc, it must be mapped using * this function. When done with the object, it must be unmapped using * zs_unmap_object. |
4bbc0bc06
|
1163 |
* |
66cdef663
|
1164 1165 1166 1167 |
* Only one object can be mapped per cpu at a time. There is no protection * against nested mappings. * * This function returns with preemption and page faults disabled. |
4bbc0bc06
|
1168 |
*/ |
66cdef663
|
1169 1170 |
void *zs_map_object(struct zs_pool *pool, unsigned long handle, enum zs_mapmode mm) |
61989a80f
|
1171 |
{ |
3783689a1
|
1172 |
struct zspage *zspage; |
66cdef663
|
1173 |
struct page *page; |
bfd093f5e
|
1174 1175 |
unsigned long obj, off; unsigned int obj_idx; |
61989a80f
|
1176 |
|
66cdef663
|
1177 1178 1179 1180 1181 |
unsigned int class_idx; enum fullness_group fg; struct size_class *class; struct mapping_area *area; struct page *pages[2]; |
2e40e163a
|
1182 |
void *ret; |
61989a80f
|
1183 |
|
9eec4cd53
|
1184 |
/* |
66cdef663
|
1185 1186 1187 |
* Because we use per-cpu mapping areas shared among the * pools/users, we can't allow mapping in interrupt context * because it can corrupt another users mappings. |
9eec4cd53
|
1188 |
*/ |
1aedcafbf
|
1189 |
BUG_ON(in_interrupt()); |
61989a80f
|
1190 |
|
312fcae22
|
1191 1192 |
/* From now on, migration cannot move the object */ pin_tag(handle); |
2e40e163a
|
1193 1194 |
obj = handle_to_obj(handle); obj_to_location(obj, &page, &obj_idx); |
3783689a1
|
1195 |
zspage = get_zspage(page); |
48b4800a1
|
1196 1197 1198 |
/* migration cannot move any subpage in this zspage */ migrate_read_lock(zspage); |
3783689a1
|
1199 |
get_zspage_mapping(zspage, &class_idx, &fg); |
66cdef663
|
1200 |
class = pool->size_class[class_idx]; |
bfd093f5e
|
1201 |
off = (class->size * obj_idx) & ~PAGE_MASK; |
df8b5bb99
|
1202 |
|
66cdef663
|
1203 1204 1205 1206 1207 |
area = &get_cpu_var(zs_map_area); area->vm_mm = mm; if (off + class->size <= PAGE_SIZE) { /* this object is contained entirely within a page */ area->vm_addr = kmap_atomic(page); |
2e40e163a
|
1208 1209 |
ret = area->vm_addr + off; goto out; |
61989a80f
|
1210 |
} |
66cdef663
|
1211 1212 1213 1214 |
/* this object spans two pages */ pages[0] = page; pages[1] = get_next_page(page); BUG_ON(!pages[1]); |
9eec4cd53
|
1215 |
|
2e40e163a
|
1216 1217 |
ret = __zs_map_object(area, pages, off, class->size); out: |
48b4800a1
|
1218 |
if (likely(!PageHugeObject(page))) |
7b60a6852
|
1219 1220 1221 |
ret += ZS_HANDLE_SIZE; return ret; |
61989a80f
|
1222 |
} |
66cdef663
|
1223 |
EXPORT_SYMBOL_GPL(zs_map_object); |
61989a80f
|
1224 |
|
66cdef663
|
1225 |
void zs_unmap_object(struct zs_pool *pool, unsigned long handle) |
61989a80f
|
1226 |
{ |
3783689a1
|
1227 |
struct zspage *zspage; |
66cdef663
|
1228 |
struct page *page; |
bfd093f5e
|
1229 1230 |
unsigned long obj, off; unsigned int obj_idx; |
61989a80f
|
1231 |
|
66cdef663
|
1232 1233 1234 1235 |
unsigned int class_idx; enum fullness_group fg; struct size_class *class; struct mapping_area *area; |
9eec4cd53
|
1236 |
|
2e40e163a
|
1237 1238 |
obj = handle_to_obj(handle); obj_to_location(obj, &page, &obj_idx); |
3783689a1
|
1239 1240 |
zspage = get_zspage(page); get_zspage_mapping(zspage, &class_idx, &fg); |
66cdef663
|
1241 |
class = pool->size_class[class_idx]; |
bfd093f5e
|
1242 |
off = (class->size * obj_idx) & ~PAGE_MASK; |
61989a80f
|
1243 |
|
66cdef663
|
1244 1245 1246 1247 1248 |
area = this_cpu_ptr(&zs_map_area); if (off + class->size <= PAGE_SIZE) kunmap_atomic(area->vm_addr); else { struct page *pages[2]; |
40f9fb8cf
|
1249 |
|
66cdef663
|
1250 1251 1252 1253 1254 1255 1256 |
pages[0] = page; pages[1] = get_next_page(page); BUG_ON(!pages[1]); __zs_unmap_object(area, pages, off, class->size); } put_cpu_var(zs_map_area); |
48b4800a1
|
1257 1258 |
migrate_read_unlock(zspage); |
312fcae22
|
1259 |
unpin_tag(handle); |
61989a80f
|
1260 |
} |
66cdef663
|
1261 |
EXPORT_SYMBOL_GPL(zs_unmap_object); |
61989a80f
|
1262 |
|
010b495e2
|
1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 |
/** * zs_huge_class_size() - Returns the size (in bytes) of the first huge * zsmalloc &size_class. * @pool: zsmalloc pool to use * * The function returns the size of the first huge class - any object of equal * or bigger size will be stored in zspage consisting of a single physical * page. * * Context: Any context. * * Return: the size (in bytes) of the first huge zsmalloc &size_class. */ size_t zs_huge_class_size(struct zs_pool *pool) { return huge_class_size; } EXPORT_SYMBOL_GPL(zs_huge_class_size); |
251cbb951
|
1281 |
static unsigned long obj_malloc(struct size_class *class, |
3783689a1
|
1282 |
struct zspage *zspage, unsigned long handle) |
c78062612
|
1283 |
{ |
bfd093f5e
|
1284 |
int i, nr_page, offset; |
c78062612
|
1285 1286 1287 1288 |
unsigned long obj; struct link_free *link; struct page *m_page; |
bfd093f5e
|
1289 |
unsigned long m_offset; |
c78062612
|
1290 |
void *vaddr; |
312fcae22
|
1291 |
handle |= OBJ_ALLOCATED_TAG; |
3783689a1
|
1292 |
obj = get_freeobj(zspage); |
bfd093f5e
|
1293 1294 1295 1296 1297 1298 1299 1300 |
offset = obj * class->size; nr_page = offset >> PAGE_SHIFT; m_offset = offset & ~PAGE_MASK; m_page = get_first_page(zspage); for (i = 0; i < nr_page; i++) m_page = get_next_page(m_page); |
c78062612
|
1301 1302 1303 |
vaddr = kmap_atomic(m_page); link = (struct link_free *)vaddr + m_offset / sizeof(*link); |
3b1d9ca65
|
1304 |
set_freeobj(zspage, link->next >> OBJ_TAG_BITS); |
48b4800a1
|
1305 |
if (likely(!PageHugeObject(m_page))) |
7b60a6852
|
1306 1307 1308 |
/* record handle in the header of allocated chunk */ link->handle = handle; else |
3783689a1
|
1309 1310 |
/* record handle to page->index */ zspage->first_page->index = handle; |
c78062612
|
1311 |
kunmap_atomic(vaddr); |
3783689a1
|
1312 |
mod_zspage_inuse(zspage, 1); |
c78062612
|
1313 |
zs_stat_inc(class, OBJ_USED, 1); |
bfd093f5e
|
1314 |
obj = location_to_obj(m_page, obj); |
c78062612
|
1315 1316 |
return obj; } |
61989a80f
|
1317 1318 1319 1320 |
/** * zs_malloc - Allocate block of given size from pool. * @pool: pool to allocate from * @size: size of block to allocate |
fd8544639
|
1321 |
* @gfp: gfp flags when allocating object |
61989a80f
|
1322 |
* |
00a61d861
|
1323 |
* On success, handle to the allocated object is returned, |
c23443483
|
1324 |
* otherwise 0. |
61989a80f
|
1325 1326 |
* Allocation requests with size > ZS_MAX_ALLOC_SIZE will fail. */ |
d0d8da2dc
|
1327 |
unsigned long zs_malloc(struct zs_pool *pool, size_t size, gfp_t gfp) |
61989a80f
|
1328 |
{ |
2e40e163a
|
1329 |
unsigned long handle, obj; |
61989a80f
|
1330 |
struct size_class *class; |
48b4800a1
|
1331 |
enum fullness_group newfg; |
3783689a1
|
1332 |
struct zspage *zspage; |
61989a80f
|
1333 |
|
7b60a6852
|
1334 |
if (unlikely(!size || size > ZS_MAX_ALLOC_SIZE)) |
2e40e163a
|
1335 |
return 0; |
3783689a1
|
1336 |
handle = cache_alloc_handle(pool, gfp); |
2e40e163a
|
1337 |
if (!handle) |
c23443483
|
1338 |
return 0; |
61989a80f
|
1339 |
|
2e40e163a
|
1340 1341 |
/* extra space in chunk to keep the handle */ size += ZS_HANDLE_SIZE; |
9eec4cd53
|
1342 |
class = pool->size_class[get_size_class_index(size)]; |
61989a80f
|
1343 1344 |
spin_lock(&class->lock); |
3783689a1
|
1345 |
zspage = find_get_zspage(class); |
48b4800a1
|
1346 1347 1348 1349 1350 |
if (likely(zspage)) { obj = obj_malloc(class, zspage, handle); /* Now move the zspage to another fullness group, if required */ fix_fullness_group(class, zspage); record_obj(handle, obj); |
61989a80f
|
1351 |
spin_unlock(&class->lock); |
61989a80f
|
1352 |
|
48b4800a1
|
1353 1354 |
return handle; } |
0f050d997
|
1355 |
|
48b4800a1
|
1356 1357 1358 1359 1360 1361 |
spin_unlock(&class->lock); zspage = alloc_zspage(pool, class, gfp); if (!zspage) { cache_free_handle(pool, handle); return 0; |
61989a80f
|
1362 |
} |
48b4800a1
|
1363 |
spin_lock(&class->lock); |
3783689a1
|
1364 |
obj = obj_malloc(class, zspage, handle); |
48b4800a1
|
1365 1366 1367 |
newfg = get_fullness_group(class, zspage); insert_zspage(class, zspage, newfg); set_zspage_mapping(zspage, class->index, newfg); |
2e40e163a
|
1368 |
record_obj(handle, obj); |
48b4800a1
|
1369 1370 |
atomic_long_add(class->pages_per_zspage, &pool->pages_allocated); |
b4fd07a08
|
1371 |
zs_stat_inc(class, OBJ_ALLOCATED, class->objs_per_zspage); |
48b4800a1
|
1372 1373 1374 |
/* We completely set up zspage so mark them as movable */ SetZsPageMovable(pool, zspage); |
61989a80f
|
1375 |
spin_unlock(&class->lock); |
2e40e163a
|
1376 |
return handle; |
61989a80f
|
1377 1378 |
} EXPORT_SYMBOL_GPL(zs_malloc); |
1ee471658
|
1379 |
static void obj_free(struct size_class *class, unsigned long obj) |
61989a80f
|
1380 1381 |
{ struct link_free *link; |
3783689a1
|
1382 1383 |
struct zspage *zspage; struct page *f_page; |
bfd093f5e
|
1384 1385 |
unsigned long f_offset; unsigned int f_objidx; |
af4ee5e97
|
1386 |
void *vaddr; |
61989a80f
|
1387 |
|
312fcae22
|
1388 |
obj &= ~OBJ_ALLOCATED_TAG; |
2e40e163a
|
1389 |
obj_to_location(obj, &f_page, &f_objidx); |
bfd093f5e
|
1390 |
f_offset = (class->size * f_objidx) & ~PAGE_MASK; |
3783689a1
|
1391 |
zspage = get_zspage(f_page); |
61989a80f
|
1392 |
|
c78062612
|
1393 |
vaddr = kmap_atomic(f_page); |
61989a80f
|
1394 1395 |
/* Insert this object in containing zspage's freelist */ |
af4ee5e97
|
1396 |
link = (struct link_free *)(vaddr + f_offset); |
3b1d9ca65
|
1397 |
link->next = get_freeobj(zspage) << OBJ_TAG_BITS; |
af4ee5e97
|
1398 |
kunmap_atomic(vaddr); |
bfd093f5e
|
1399 |
set_freeobj(zspage, f_objidx); |
3783689a1
|
1400 |
mod_zspage_inuse(zspage, -1); |
0f050d997
|
1401 |
zs_stat_dec(class, OBJ_USED, 1); |
c78062612
|
1402 1403 1404 1405 |
} void zs_free(struct zs_pool *pool, unsigned long handle) { |
3783689a1
|
1406 1407 |
struct zspage *zspage; struct page *f_page; |
bfd093f5e
|
1408 1409 |
unsigned long obj; unsigned int f_objidx; |
c78062612
|
1410 1411 1412 |
int class_idx; struct size_class *class; enum fullness_group fullness; |
48b4800a1
|
1413 |
bool isolated; |
c78062612
|
1414 1415 1416 |
if (unlikely(!handle)) return; |
312fcae22
|
1417 |
pin_tag(handle); |
c78062612
|
1418 |
obj = handle_to_obj(handle); |
c78062612
|
1419 |
obj_to_location(obj, &f_page, &f_objidx); |
3783689a1
|
1420 |
zspage = get_zspage(f_page); |
c78062612
|
1421 |
|
48b4800a1
|
1422 |
migrate_read_lock(zspage); |
3783689a1
|
1423 |
get_zspage_mapping(zspage, &class_idx, &fullness); |
c78062612
|
1424 1425 1426 |
class = pool->size_class[class_idx]; spin_lock(&class->lock); |
1ee471658
|
1427 |
obj_free(class, obj); |
3783689a1
|
1428 |
fullness = fix_fullness_group(class, zspage); |
48b4800a1
|
1429 1430 1431 |
if (fullness != ZS_EMPTY) { migrate_read_unlock(zspage); goto out; |
312fcae22
|
1432 |
} |
48b4800a1
|
1433 1434 1435 1436 1437 1438 1439 |
isolated = is_zspage_isolated(zspage); migrate_read_unlock(zspage); /* If zspage is isolated, zs_page_putback will free the zspage */ if (likely(!isolated)) free_zspage(pool, class, zspage); out: |
61989a80f
|
1440 |
spin_unlock(&class->lock); |
312fcae22
|
1441 |
unpin_tag(handle); |
3783689a1
|
1442 |
cache_free_handle(pool, handle); |
312fcae22
|
1443 1444 |
} EXPORT_SYMBOL_GPL(zs_free); |
251cbb951
|
1445 1446 |
static void zs_object_copy(struct size_class *class, unsigned long dst, unsigned long src) |
312fcae22
|
1447 1448 |
{ struct page *s_page, *d_page; |
bfd093f5e
|
1449 |
unsigned int s_objidx, d_objidx; |
312fcae22
|
1450 1451 1452 1453 1454 1455 1456 1457 1458 |
unsigned long s_off, d_off; void *s_addr, *d_addr; int s_size, d_size, size; int written = 0; s_size = d_size = class->size; obj_to_location(src, &s_page, &s_objidx); obj_to_location(dst, &d_page, &d_objidx); |
bfd093f5e
|
1459 1460 |
s_off = (class->size * s_objidx) & ~PAGE_MASK; d_off = (class->size * d_objidx) & ~PAGE_MASK; |
312fcae22
|
1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 |
if (s_off + class->size > PAGE_SIZE) s_size = PAGE_SIZE - s_off; if (d_off + class->size > PAGE_SIZE) d_size = PAGE_SIZE - d_off; s_addr = kmap_atomic(s_page); d_addr = kmap_atomic(d_page); while (1) { size = min(s_size, d_size); memcpy(d_addr + d_off, s_addr + s_off, size); written += size; if (written == class->size) break; |
495819ead
|
1478 1479 1480 1481 1482 1483 |
s_off += size; s_size -= size; d_off += size; d_size -= size; if (s_off >= PAGE_SIZE) { |
312fcae22
|
1484 1485 1486 |
kunmap_atomic(d_addr); kunmap_atomic(s_addr); s_page = get_next_page(s_page); |
312fcae22
|
1487 1488 1489 1490 |
s_addr = kmap_atomic(s_page); d_addr = kmap_atomic(d_page); s_size = class->size - written; s_off = 0; |
312fcae22
|
1491 |
} |
495819ead
|
1492 |
if (d_off >= PAGE_SIZE) { |
312fcae22
|
1493 1494 |
kunmap_atomic(d_addr); d_page = get_next_page(d_page); |
312fcae22
|
1495 1496 1497 |
d_addr = kmap_atomic(d_page); d_size = class->size - written; d_off = 0; |
312fcae22
|
1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 |
} } kunmap_atomic(d_addr); kunmap_atomic(s_addr); } /* * Find alloced object in zspage from index object and * return handle. */ |
251cbb951
|
1509 |
static unsigned long find_alloced_obj(struct size_class *class, |
cf675acb7
|
1510 |
struct page *page, int *obj_idx) |
312fcae22
|
1511 1512 1513 |
{ unsigned long head; int offset = 0; |
cf675acb7
|
1514 |
int index = *obj_idx; |
312fcae22
|
1515 1516 |
unsigned long handle = 0; void *addr = kmap_atomic(page); |
3783689a1
|
1517 |
offset = get_first_obj_offset(page); |
312fcae22
|
1518 1519 1520 |
offset += class->size * index; while (offset < PAGE_SIZE) { |
48b4800a1
|
1521 |
head = obj_to_head(page, addr + offset); |
312fcae22
|
1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 |
if (head & OBJ_ALLOCATED_TAG) { handle = head & ~OBJ_ALLOCATED_TAG; if (trypin_tag(handle)) break; handle = 0; } offset += class->size; index++; } kunmap_atomic(addr); |
cf675acb7
|
1534 1535 |
*obj_idx = index; |
312fcae22
|
1536 1537 1538 1539 |
return handle; } struct zs_compact_control { |
3783689a1
|
1540 |
/* Source spage for migration which could be a subpage of zspage */ |
312fcae22
|
1541 1542 1543 1544 1545 1546 |
struct page *s_page; /* Destination page for migration which should be a first page * of zspage. */ struct page *d_page; /* Starting object index within @s_page which used for live object * in the subpage. */ |
41b88e14c
|
1547 |
int obj_idx; |
312fcae22
|
1548 1549 1550 1551 1552 1553 1554 1555 1556 |
}; static int migrate_zspage(struct zs_pool *pool, struct size_class *class, struct zs_compact_control *cc) { unsigned long used_obj, free_obj; unsigned long handle; struct page *s_page = cc->s_page; struct page *d_page = cc->d_page; |
41b88e14c
|
1557 |
int obj_idx = cc->obj_idx; |
312fcae22
|
1558 1559 1560 |
int ret = 0; while (1) { |
cf675acb7
|
1561 |
handle = find_alloced_obj(class, s_page, &obj_idx); |
312fcae22
|
1562 1563 1564 1565 |
if (!handle) { s_page = get_next_page(s_page); if (!s_page) break; |
41b88e14c
|
1566 |
obj_idx = 0; |
312fcae22
|
1567 1568 1569 1570 |
continue; } /* Stop if there is no more space */ |
3783689a1
|
1571 |
if (zspage_full(class, get_zspage(d_page))) { |
312fcae22
|
1572 1573 1574 1575 1576 1577 |
unpin_tag(handle); ret = -ENOMEM; break; } used_obj = handle_to_obj(handle); |
3783689a1
|
1578 |
free_obj = obj_malloc(class, get_zspage(d_page), handle); |
251cbb951
|
1579 |
zs_object_copy(class, free_obj, used_obj); |
41b88e14c
|
1580 |
obj_idx++; |
c102f07ca
|
1581 1582 1583 1584 1585 1586 1587 |
/* * record_obj updates handle's value to free_obj and it will * invalidate lock bit(ie, HANDLE_PIN_BIT) of handle, which * breaks synchronization using pin_tag(e,g, zs_free) so * let's keep the lock bit. */ free_obj |= BIT(HANDLE_PIN_BIT); |
312fcae22
|
1588 1589 |
record_obj(handle, free_obj); unpin_tag(handle); |
1ee471658
|
1590 |
obj_free(class, used_obj); |
312fcae22
|
1591 1592 1593 1594 |
} /* Remember last position in this iteration */ cc->s_page = s_page; |
41b88e14c
|
1595 |
cc->obj_idx = obj_idx; |
312fcae22
|
1596 1597 1598 |
return ret; } |
3783689a1
|
1599 |
static struct zspage *isolate_zspage(struct size_class *class, bool source) |
312fcae22
|
1600 1601 |
{ int i; |
3783689a1
|
1602 1603 |
struct zspage *zspage; enum fullness_group fg[2] = {ZS_ALMOST_EMPTY, ZS_ALMOST_FULL}; |
312fcae22
|
1604 |
|
3783689a1
|
1605 1606 1607 1608 1609 1610 1611 1612 1613 |
if (!source) { fg[0] = ZS_ALMOST_FULL; fg[1] = ZS_ALMOST_EMPTY; } for (i = 0; i < 2; i++) { zspage = list_first_entry_or_null(&class->fullness_list[fg[i]], struct zspage, list); if (zspage) { |
48b4800a1
|
1614 |
VM_BUG_ON(is_zspage_isolated(zspage)); |
3783689a1
|
1615 1616 |
remove_zspage(class, zspage, fg[i]); return zspage; |
312fcae22
|
1617 1618 |
} } |
3783689a1
|
1619 |
return zspage; |
312fcae22
|
1620 |
} |
860c707dc
|
1621 |
/* |
3783689a1
|
1622 |
* putback_zspage - add @zspage into right class's fullness list |
860c707dc
|
1623 |
* @class: destination class |
3783689a1
|
1624 |
* @zspage: target page |
860c707dc
|
1625 |
* |
3783689a1
|
1626 |
* Return @zspage's fullness_group |
860c707dc
|
1627 |
*/ |
4aa409cab
|
1628 |
static enum fullness_group putback_zspage(struct size_class *class, |
3783689a1
|
1629 |
struct zspage *zspage) |
312fcae22
|
1630 |
{ |
312fcae22
|
1631 |
enum fullness_group fullness; |
48b4800a1
|
1632 |
VM_BUG_ON(is_zspage_isolated(zspage)); |
3783689a1
|
1633 1634 1635 |
fullness = get_fullness_group(class, zspage); insert_zspage(class, zspage, fullness); set_zspage_mapping(zspage, class->index, fullness); |
839373e64
|
1636 |
|
860c707dc
|
1637 |
return fullness; |
61989a80f
|
1638 |
} |
312fcae22
|
1639 |
|
48b4800a1
|
1640 |
#ifdef CONFIG_COMPACTION |
4d0a5402f
|
1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 |
/* * To prevent zspage destroy during migration, zspage freeing should * hold locks of all pages in the zspage. */ static void lock_zspage(struct zspage *zspage) { struct page *page = get_first_page(zspage); do { lock_page(page); } while ((page = get_next_page(page)) != NULL); } |
8e9231f81
|
1653 |
static int zs_init_fs_context(struct fs_context *fc) |
48b4800a1
|
1654 |
{ |
8e9231f81
|
1655 |
return init_pseudo(fc, ZSMALLOC_MAGIC) ? 0 : -ENOMEM; |
48b4800a1
|
1656 1657 1658 1659 |
} static struct file_system_type zsmalloc_fs = { .name = "zsmalloc", |
8e9231f81
|
1660 |
.init_fs_context = zs_init_fs_context, |
48b4800a1
|
1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 |
.kill_sb = kill_anon_super, }; static int zsmalloc_mount(void) { int ret = 0; zsmalloc_mnt = kern_mount(&zsmalloc_fs); if (IS_ERR(zsmalloc_mnt)) ret = PTR_ERR(zsmalloc_mnt); return ret; } static void zsmalloc_unmount(void) { kern_unmount(zsmalloc_mnt); } static void migrate_lock_init(struct zspage *zspage) { rwlock_init(&zspage->lock); } |
cfc451cfd
|
1684 |
static void migrate_read_lock(struct zspage *zspage) __acquires(&zspage->lock) |
48b4800a1
|
1685 1686 1687 |
{ read_lock(&zspage->lock); } |
8a374ccce
|
1688 |
static void migrate_read_unlock(struct zspage *zspage) __releases(&zspage->lock) |
48b4800a1
|
1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 |
{ read_unlock(&zspage->lock); } static void migrate_write_lock(struct zspage *zspage) { write_lock(&zspage->lock); } static void migrate_write_unlock(struct zspage *zspage) { write_unlock(&zspage->lock); } /* Number of isolated subpage for *page migration* in this zspage */ static void inc_zspage_isolation(struct zspage *zspage) { zspage->isolated++; } static void dec_zspage_isolation(struct zspage *zspage) { zspage->isolated--; } |
1a87aa035
|
1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 |
static void putback_zspage_deferred(struct zs_pool *pool, struct size_class *class, struct zspage *zspage) { enum fullness_group fg; fg = putback_zspage(class, zspage); if (fg == ZS_EMPTY) schedule_work(&pool->free_work); } |
701d67859
|
1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 |
static inline void zs_pool_dec_isolated(struct zs_pool *pool) { VM_BUG_ON(atomic_long_read(&pool->isolated_pages) <= 0); atomic_long_dec(&pool->isolated_pages); /* * There's no possibility of racing, since wait_for_isolated_drain() * checks the isolated count under &class->lock after enqueuing * on migration_wait. */ if (atomic_long_read(&pool->isolated_pages) == 0 && pool->destroying) wake_up_all(&pool->migration_wait); } |
48b4800a1
|
1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 |
static void replace_sub_page(struct size_class *class, struct zspage *zspage, struct page *newpage, struct page *oldpage) { struct page *page; struct page *pages[ZS_MAX_PAGES_PER_ZSPAGE] = {NULL, }; int idx = 0; page = get_first_page(zspage); do { if (page == oldpage) pages[idx] = newpage; else pages[idx] = page; idx++; } while ((page = get_next_page(page)) != NULL); create_page_chain(class, zspage, pages); set_first_obj_offset(newpage, get_first_obj_offset(oldpage)); if (unlikely(PageHugeObject(oldpage))) newpage->index = oldpage->index; __SetPageMovable(newpage, page_mapping(oldpage)); } |
4d0a5402f
|
1758 |
static bool zs_page_isolate(struct page *page, isolate_mode_t mode) |
48b4800a1
|
1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 |
{ struct zs_pool *pool; struct size_class *class; int class_idx; enum fullness_group fullness; struct zspage *zspage; struct address_space *mapping; /* * Page is locked so zspage couldn't be destroyed. For detail, look at * lock_zspage in free_zspage. */ VM_BUG_ON_PAGE(!PageMovable(page), page); VM_BUG_ON_PAGE(PageIsolated(page), page); zspage = get_zspage(page); /* * Without class lock, fullness could be stale while class_idx is okay * because class_idx is constant unless page is freed so we should get * fullness again under class lock. */ get_zspage_mapping(zspage, &class_idx, &fullness); mapping = page_mapping(page); pool = mapping->private_data; class = pool->size_class[class_idx]; spin_lock(&class->lock); if (get_zspage_inuse(zspage) == 0) { spin_unlock(&class->lock); return false; } /* zspage is isolated for object migration */ if (list_empty(&zspage->list) && !is_zspage_isolated(zspage)) { spin_unlock(&class->lock); return false; } /* * If this is first time isolation for the zspage, isolate zspage from * size_class to prevent further object allocation from the zspage. */ if (!list_empty(&zspage->list) && !is_zspage_isolated(zspage)) { get_zspage_mapping(zspage, &class_idx, &fullness); |
701d67859
|
1804 |
atomic_long_inc(&pool->isolated_pages); |
48b4800a1
|
1805 1806 1807 1808 1809 1810 1811 1812 |
remove_zspage(class, zspage, fullness); } inc_zspage_isolation(zspage); spin_unlock(&class->lock); return true; } |
4d0a5402f
|
1813 |
static int zs_page_migrate(struct address_space *mapping, struct page *newpage, |
48b4800a1
|
1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 |
struct page *page, enum migrate_mode mode) { struct zs_pool *pool; struct size_class *class; int class_idx; enum fullness_group fullness; struct zspage *zspage; struct page *dummy; void *s_addr, *d_addr, *addr; int offset, pos; unsigned long handle, head; unsigned long old_obj, new_obj; unsigned int obj_idx; int ret = -EAGAIN; |
2916ecc0f
|
1828 1829 1830 1831 1832 1833 1834 |
/* * We cannot support the _NO_COPY case here, because copy needs to * happen under the zs lock, which does not work with * MIGRATE_SYNC_NO_COPY workflow. */ if (mode == MIGRATE_SYNC_NO_COPY) return -EINVAL; |
48b4800a1
|
1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 |
VM_BUG_ON_PAGE(!PageMovable(page), page); VM_BUG_ON_PAGE(!PageIsolated(page), page); zspage = get_zspage(page); /* Concurrent compactor cannot migrate any subpage in zspage */ migrate_write_lock(zspage); get_zspage_mapping(zspage, &class_idx, &fullness); pool = mapping->private_data; class = pool->size_class[class_idx]; offset = get_first_obj_offset(page); spin_lock(&class->lock); if (!get_zspage_inuse(zspage)) { |
77ff46579
|
1849 1850 1851 1852 1853 |
/* * Set "offset" to end of the page so that every loops * skips unnecessary object scanning. */ offset = PAGE_SIZE; |
48b4800a1
|
1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 |
} pos = offset; s_addr = kmap_atomic(page); while (pos < PAGE_SIZE) { head = obj_to_head(page, s_addr + pos); if (head & OBJ_ALLOCATED_TAG) { handle = head & ~OBJ_ALLOCATED_TAG; if (!trypin_tag(handle)) goto unpin_objects; } pos += class->size; } /* * Here, any user cannot access all objects in the zspage so let's move. */ d_addr = kmap_atomic(newpage); memcpy(d_addr, s_addr, PAGE_SIZE); kunmap_atomic(d_addr); for (addr = s_addr + offset; addr < s_addr + pos; addr += class->size) { head = obj_to_head(page, addr); if (head & OBJ_ALLOCATED_TAG) { handle = head & ~OBJ_ALLOCATED_TAG; if (!testpin_tag(handle)) BUG(); old_obj = handle_to_obj(handle); obj_to_location(old_obj, &dummy, &obj_idx); new_obj = (unsigned long)location_to_obj(newpage, obj_idx); new_obj |= BIT(HANDLE_PIN_BIT); record_obj(handle, new_obj); } } replace_sub_page(class, zspage, newpage, page); get_page(newpage); dec_zspage_isolation(zspage); /* * Page migration is done so let's putback isolated zspage to * the list if @page is final isolated subpage in the zspage. */ |
701d67859
|
1901 1902 1903 1904 1905 1906 1907 |
if (!is_zspage_isolated(zspage)) { /* * We cannot race with zs_destroy_pool() here because we wait * for isolation to hit zero before we start destroying. * Also, we ensure that everyone can see pool->destroying before * we start waiting. */ |
1a87aa035
|
1908 |
putback_zspage_deferred(pool, class, zspage); |
701d67859
|
1909 1910 |
zs_pool_dec_isolated(pool); } |
48b4800a1
|
1911 |
|
ac8f05da5
|
1912 1913 1914 1915 |
if (page_zone(newpage) != page_zone(page)) { dec_zone_page_state(page, NR_ZSPAGES); inc_zone_page_state(newpage, NR_ZSPAGES); } |
48b4800a1
|
1916 1917 1918 |
reset_page(page); put_page(page); page = newpage; |
dd4123f32
|
1919 |
ret = MIGRATEPAGE_SUCCESS; |
48b4800a1
|
1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 |
unpin_objects: for (addr = s_addr + offset; addr < s_addr + pos; addr += class->size) { head = obj_to_head(page, addr); if (head & OBJ_ALLOCATED_TAG) { handle = head & ~OBJ_ALLOCATED_TAG; if (!testpin_tag(handle)) BUG(); unpin_tag(handle); } } kunmap_atomic(s_addr); |
48b4800a1
|
1932 1933 1934 1935 1936 |
spin_unlock(&class->lock); migrate_write_unlock(zspage); return ret; } |
4d0a5402f
|
1937 |
static void zs_page_putback(struct page *page) |
48b4800a1
|
1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 |
{ struct zs_pool *pool; struct size_class *class; int class_idx; enum fullness_group fg; struct address_space *mapping; struct zspage *zspage; VM_BUG_ON_PAGE(!PageMovable(page), page); VM_BUG_ON_PAGE(!PageIsolated(page), page); zspage = get_zspage(page); get_zspage_mapping(zspage, &class_idx, &fg); mapping = page_mapping(page); pool = mapping->private_data; class = pool->size_class[class_idx]; spin_lock(&class->lock); dec_zspage_isolation(zspage); if (!is_zspage_isolated(zspage)) { |
48b4800a1
|
1958 1959 1960 1961 |
/* * Due to page_lock, we cannot free zspage immediately * so let's defer. */ |
1a87aa035
|
1962 |
putback_zspage_deferred(pool, class, zspage); |
701d67859
|
1963 |
zs_pool_dec_isolated(pool); |
48b4800a1
|
1964 1965 1966 |
} spin_unlock(&class->lock); } |
4d0a5402f
|
1967 |
static const struct address_space_operations zsmalloc_aops = { |
48b4800a1
|
1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 |
.isolate_page = zs_page_isolate, .migratepage = zs_page_migrate, .putback_page = zs_page_putback, }; static int zs_register_migration(struct zs_pool *pool) { pool->inode = alloc_anon_inode(zsmalloc_mnt->mnt_sb); if (IS_ERR(pool->inode)) { pool->inode = NULL; return 1; } pool->inode->i_mapping->private_data = pool; pool->inode->i_mapping->a_ops = &zsmalloc_aops; return 0; } |
701d67859
|
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 |
static bool pool_isolated_are_drained(struct zs_pool *pool) { return atomic_long_read(&pool->isolated_pages) == 0; } /* Function for resolving migration */ static void wait_for_isolated_drain(struct zs_pool *pool) { /* * We're in the process of destroying the pool, so there are no * active allocations. zs_page_isolate() fails for completely free * zspages, so we need only wait for the zs_pool's isolated * count to hit zero. */ wait_event(pool->migration_wait, pool_isolated_are_drained(pool)); } |
48b4800a1
|
2003 2004 |
static void zs_unregister_migration(struct zs_pool *pool) { |
701d67859
|
2005 2006 2007 2008 2009 2010 2011 2012 2013 |
pool->destroying = true; /* * We need a memory barrier here to ensure global visibility of * pool->destroying. Thus pool->isolated pages will either be 0 in which * case we don't care, or it will be > 0 and pool->destroying will * ensure that we wake up once isolation hits 0. */ smp_mb(); wait_for_isolated_drain(pool); /* This can block */ |
48b4800a1
|
2014 |
flush_work(&pool->free_work); |
c3491eca3
|
2015 |
iput(pool->inode); |
48b4800a1
|
2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 |
} /* * Caller should hold page_lock of all pages in the zspage * In here, we cannot use zspage meta data. */ static void async_free_zspage(struct work_struct *work) { int i; struct size_class *class; unsigned int class_idx; enum fullness_group fullness; struct zspage *zspage, *tmp; LIST_HEAD(free_pages); struct zs_pool *pool = container_of(work, struct zs_pool, free_work); |
cf8e0fedf
|
2032 |
for (i = 0; i < ZS_SIZE_CLASSES; i++) { |
48b4800a1
|
2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 |
class = pool->size_class[i]; if (class->index != i) continue; spin_lock(&class->lock); list_splice_init(&class->fullness_list[ZS_EMPTY], &free_pages); spin_unlock(&class->lock); } list_for_each_entry_safe(zspage, tmp, &free_pages, list) { list_del(&zspage->list); lock_zspage(zspage); get_zspage_mapping(zspage, &class_idx, &fullness); VM_BUG_ON(fullness != ZS_EMPTY); class = pool->size_class[class_idx]; spin_lock(&class->lock); __free_zspage(pool, pool->size_class[class_idx], zspage); spin_unlock(&class->lock); } }; static void kick_deferred_free(struct zs_pool *pool) { schedule_work(&pool->free_work); } static void init_deferred_free(struct zs_pool *pool) { INIT_WORK(&pool->free_work, async_free_zspage); } static void SetZsPageMovable(struct zs_pool *pool, struct zspage *zspage) { struct page *page = get_first_page(zspage); do { WARN_ON(!trylock_page(page)); __SetPageMovable(page, pool->inode->i_mapping); unlock_page(page); } while ((page = get_next_page(page)) != NULL); } #endif |
04f05909e
|
2077 2078 2079 2080 |
/* * * Based on the number of unused allocated objects calculate * and return the number of pages that we can free. |
04f05909e
|
2081 2082 2083 2084 |
*/ static unsigned long zs_can_compact(struct size_class *class) { unsigned long obj_wasted; |
44f43e99f
|
2085 2086 |
unsigned long obj_allocated = zs_stat_get(class, OBJ_ALLOCATED); unsigned long obj_used = zs_stat_get(class, OBJ_USED); |
04f05909e
|
2087 |
|
44f43e99f
|
2088 2089 |
if (obj_allocated <= obj_used) return 0; |
04f05909e
|
2090 |
|
44f43e99f
|
2091 |
obj_wasted = obj_allocated - obj_used; |
b4fd07a08
|
2092 |
obj_wasted /= class->objs_per_zspage; |
04f05909e
|
2093 |
|
6cbf16b3b
|
2094 |
return obj_wasted * class->pages_per_zspage; |
04f05909e
|
2095 |
} |
7d3f39382
|
2096 |
static void __zs_compact(struct zs_pool *pool, struct size_class *class) |
312fcae22
|
2097 |
{ |
312fcae22
|
2098 |
struct zs_compact_control cc; |
3783689a1
|
2099 2100 |
struct zspage *src_zspage; struct zspage *dst_zspage = NULL; |
312fcae22
|
2101 |
|
312fcae22
|
2102 |
spin_lock(&class->lock); |
3783689a1
|
2103 |
while ((src_zspage = isolate_zspage(class, true))) { |
312fcae22
|
2104 |
|
04f05909e
|
2105 2106 |
if (!zs_can_compact(class)) break; |
41b88e14c
|
2107 |
cc.obj_idx = 0; |
48b4800a1
|
2108 |
cc.s_page = get_first_page(src_zspage); |
312fcae22
|
2109 |
|
3783689a1
|
2110 |
while ((dst_zspage = isolate_zspage(class, false))) { |
48b4800a1
|
2111 |
cc.d_page = get_first_page(dst_zspage); |
312fcae22
|
2112 |
/* |
0dc63d488
|
2113 2114 |
* If there is no more space in dst_page, resched * and see if anyone had allocated another zspage. |
312fcae22
|
2115 2116 2117 |
*/ if (!migrate_zspage(pool, class, &cc)) break; |
4aa409cab
|
2118 |
putback_zspage(class, dst_zspage); |
312fcae22
|
2119 2120 2121 |
} /* Stop if we couldn't find slot */ |
3783689a1
|
2122 |
if (dst_zspage == NULL) |
312fcae22
|
2123 |
break; |
4aa409cab
|
2124 2125 |
putback_zspage(class, dst_zspage); if (putback_zspage(class, src_zspage) == ZS_EMPTY) { |
48b4800a1
|
2126 |
free_zspage(pool, class, src_zspage); |
6cbf16b3b
|
2127 |
pool->stats.pages_compacted += class->pages_per_zspage; |
4aa409cab
|
2128 |
} |
312fcae22
|
2129 |
spin_unlock(&class->lock); |
312fcae22
|
2130 2131 2132 |
cond_resched(); spin_lock(&class->lock); } |
3783689a1
|
2133 |
if (src_zspage) |
4aa409cab
|
2134 |
putback_zspage(class, src_zspage); |
312fcae22
|
2135 |
|
7d3f39382
|
2136 |
spin_unlock(&class->lock); |
312fcae22
|
2137 2138 2139 2140 2141 |
} unsigned long zs_compact(struct zs_pool *pool) { int i; |
312fcae22
|
2142 |
struct size_class *class; |
cf8e0fedf
|
2143 |
for (i = ZS_SIZE_CLASSES - 1; i >= 0; i--) { |
312fcae22
|
2144 2145 2146 2147 2148 |
class = pool->size_class[i]; if (!class) continue; if (class->index != i) continue; |
7d3f39382
|
2149 |
__zs_compact(pool, class); |
312fcae22
|
2150 |
} |
860c707dc
|
2151 |
return pool->stats.pages_compacted; |
312fcae22
|
2152 2153 |
} EXPORT_SYMBOL_GPL(zs_compact); |
61989a80f
|
2154 |
|
7d3f39382
|
2155 2156 2157 2158 2159 |
void zs_pool_stats(struct zs_pool *pool, struct zs_pool_stats *stats) { memcpy(stats, &pool->stats, sizeof(struct zs_pool_stats)); } EXPORT_SYMBOL_GPL(zs_pool_stats); |
ab9d306d9
|
2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 |
static unsigned long zs_shrinker_scan(struct shrinker *shrinker, struct shrink_control *sc) { unsigned long pages_freed; struct zs_pool *pool = container_of(shrinker, struct zs_pool, shrinker); pages_freed = pool->stats.pages_compacted; /* * Compact classes and calculate compaction delta. * Can run concurrently with a manually triggered * (by user) compaction. */ pages_freed = zs_compact(pool) - pages_freed; return pages_freed ? pages_freed : SHRINK_STOP; } static unsigned long zs_shrinker_count(struct shrinker *shrinker, struct shrink_control *sc) { int i; struct size_class *class; unsigned long pages_to_free = 0; struct zs_pool *pool = container_of(shrinker, struct zs_pool, shrinker); |
cf8e0fedf
|
2186 |
for (i = ZS_SIZE_CLASSES - 1; i >= 0; i--) { |
ab9d306d9
|
2187 2188 2189 2190 2191 |
class = pool->size_class[i]; if (!class) continue; if (class->index != i) continue; |
ab9d306d9
|
2192 |
pages_to_free += zs_can_compact(class); |
ab9d306d9
|
2193 2194 2195 2196 2197 2198 2199 |
} return pages_to_free; } static void zs_unregister_shrinker(struct zs_pool *pool) { |
93144ca35
|
2200 |
unregister_shrinker(&pool->shrinker); |
ab9d306d9
|
2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 |
} static int zs_register_shrinker(struct zs_pool *pool) { pool->shrinker.scan_objects = zs_shrinker_scan; pool->shrinker.count_objects = zs_shrinker_count; pool->shrinker.batch = 0; pool->shrinker.seeks = DEFAULT_SEEKS; return register_shrinker(&pool->shrinker); } |
00a61d861
|
2212 |
/** |
66cdef663
|
2213 |
* zs_create_pool - Creates an allocation pool to work from. |
fd8544639
|
2214 |
* @name: pool name to be created |
166cfda75
|
2215 |
* |
66cdef663
|
2216 2217 |
* This function must be called before anything when using * the zsmalloc allocator. |
166cfda75
|
2218 |
* |
66cdef663
|
2219 2220 |
* On success, a pointer to the newly created pool is returned, * otherwise NULL. |
396b7fd6f
|
2221 |
*/ |
d0d8da2dc
|
2222 |
struct zs_pool *zs_create_pool(const char *name) |
61989a80f
|
2223 |
{ |
66cdef663
|
2224 2225 2226 |
int i; struct zs_pool *pool; struct size_class *prev_class = NULL; |
61989a80f
|
2227 |
|
66cdef663
|
2228 2229 2230 |
pool = kzalloc(sizeof(*pool), GFP_KERNEL); if (!pool) return NULL; |
61989a80f
|
2231 |
|
48b4800a1
|
2232 |
init_deferred_free(pool); |
61989a80f
|
2233 |
|
2e40e163a
|
2234 2235 2236 |
pool->name = kstrdup(name, GFP_KERNEL); if (!pool->name) goto err; |
441e254cd
|
2237 |
#ifdef CONFIG_COMPACTION |
701d67859
|
2238 |
init_waitqueue_head(&pool->migration_wait); |
441e254cd
|
2239 |
#endif |
701d67859
|
2240 |
|
3783689a1
|
2241 |
if (create_cache(pool)) |
2e40e163a
|
2242 |
goto err; |
c60369f01
|
2243 |
/* |
399d8eebe
|
2244 |
* Iterate reversely, because, size of size_class that we want to use |
66cdef663
|
2245 |
* for merging should be larger or equal to current size. |
c60369f01
|
2246 |
*/ |
cf8e0fedf
|
2247 |
for (i = ZS_SIZE_CLASSES - 1; i >= 0; i--) { |
66cdef663
|
2248 2249 |
int size; int pages_per_zspage; |
64d90465f
|
2250 |
int objs_per_zspage; |
66cdef663
|
2251 |
struct size_class *class; |
3783689a1
|
2252 |
int fullness = 0; |
c60369f01
|
2253 |
|
66cdef663
|
2254 2255 2256 2257 |
size = ZS_MIN_ALLOC_SIZE + i * ZS_SIZE_CLASS_DELTA; if (size > ZS_MAX_ALLOC_SIZE) size = ZS_MAX_ALLOC_SIZE; pages_per_zspage = get_pages_per_zspage(size); |
64d90465f
|
2258 |
objs_per_zspage = pages_per_zspage * PAGE_SIZE / size; |
61989a80f
|
2259 |
|
66cdef663
|
2260 |
/* |
010b495e2
|
2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 |
* We iterate from biggest down to smallest classes, * so huge_class_size holds the size of the first huge * class. Any object bigger than or equal to that will * endup in the huge class. */ if (pages_per_zspage != 1 && objs_per_zspage != 1 && !huge_class_size) { huge_class_size = size; /* * The object uses ZS_HANDLE_SIZE bytes to store the * handle. We need to subtract it, because zs_malloc() * unconditionally adds handle size before it performs * size class search - so object may be smaller than * huge class size, yet it still can end up in the huge * class because it grows by ZS_HANDLE_SIZE extra bytes * right before class lookup. */ huge_class_size -= (ZS_HANDLE_SIZE - 1); } /* |
66cdef663
|
2282 2283 2284 2285 2286 2287 2288 2289 2290 |
* size_class is used for normal zsmalloc operation such * as alloc/free for that size. Although it is natural that we * have one size_class for each size, there is a chance that we * can get more memory utilization if we use one size_class for * many different sizes whose size_class have same * characteristics. So, we makes size_class point to * previous size_class if possible. */ if (prev_class) { |
64d90465f
|
2291 |
if (can_merge(prev_class, pages_per_zspage, objs_per_zspage)) { |
66cdef663
|
2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 |
pool->size_class[i] = prev_class; continue; } } class = kzalloc(sizeof(struct size_class), GFP_KERNEL); if (!class) goto err; class->size = size; class->index = i; class->pages_per_zspage = pages_per_zspage; |
64d90465f
|
2304 |
class->objs_per_zspage = objs_per_zspage; |
66cdef663
|
2305 2306 |
spin_lock_init(&class->lock); pool->size_class[i] = class; |
48b4800a1
|
2307 2308 |
for (fullness = ZS_EMPTY; fullness < NR_ZS_FULLNESS; fullness++) |
3783689a1
|
2309 |
INIT_LIST_HEAD(&class->fullness_list[fullness]); |
66cdef663
|
2310 2311 |
prev_class = class; |
61989a80f
|
2312 |
} |
d34f61572
|
2313 2314 |
/* debug only, don't abort if it fails */ zs_pool_stat_create(pool, name); |
0f050d997
|
2315 |
|
48b4800a1
|
2316 2317 |
if (zs_register_migration(pool)) goto err; |
ab9d306d9
|
2318 |
/* |
93144ca35
|
2319 2320 2321 2322 |
* Not critical since shrinker is only used to trigger internal * defragmentation of the pool which is pretty optional thing. If * registration fails we still can use the pool normally and user can * trigger compaction manually. Thus, ignore return code. |
ab9d306d9
|
2323 |
*/ |
93144ca35
|
2324 |
zs_register_shrinker(pool); |
66cdef663
|
2325 2326 2327 2328 2329 |
return pool; err: zs_destroy_pool(pool); return NULL; |
61989a80f
|
2330 |
} |
66cdef663
|
2331 |
EXPORT_SYMBOL_GPL(zs_create_pool); |
61989a80f
|
2332 |
|
66cdef663
|
2333 |
void zs_destroy_pool(struct zs_pool *pool) |
61989a80f
|
2334 |
{ |
66cdef663
|
2335 |
int i; |
61989a80f
|
2336 |
|
ab9d306d9
|
2337 |
zs_unregister_shrinker(pool); |
48b4800a1
|
2338 |
zs_unregister_migration(pool); |
0f050d997
|
2339 |
zs_pool_stat_destroy(pool); |
cf8e0fedf
|
2340 |
for (i = 0; i < ZS_SIZE_CLASSES; i++) { |
66cdef663
|
2341 2342 |
int fg; struct size_class *class = pool->size_class[i]; |
61989a80f
|
2343 |
|
66cdef663
|
2344 2345 |
if (!class) continue; |
61989a80f
|
2346 |
|
66cdef663
|
2347 2348 |
if (class->index != i) continue; |
61989a80f
|
2349 |
|
48b4800a1
|
2350 |
for (fg = ZS_EMPTY; fg < NR_ZS_FULLNESS; fg++) { |
3783689a1
|
2351 |
if (!list_empty(&class->fullness_list[fg])) { |
66cdef663
|
2352 2353 2354 2355 2356 2357 2358 |
pr_info("Freeing non-empty class with size %db, fullness group %d ", class->size, fg); } } kfree(class); } |
f553646a6
|
2359 |
|
3783689a1
|
2360 |
destroy_cache(pool); |
0f050d997
|
2361 |
kfree(pool->name); |
66cdef663
|
2362 2363 2364 |
kfree(pool); } EXPORT_SYMBOL_GPL(zs_destroy_pool); |
b74185108
|
2365 |
|
66cdef663
|
2366 2367 |
static int __init zs_init(void) { |
48b4800a1
|
2368 2369 2370 2371 2372 |
int ret; ret = zsmalloc_mount(); if (ret) goto out; |
215c89d05
|
2373 2374 |
ret = cpuhp_setup_state(CPUHP_MM_ZS_PREPARE, "mm/zsmalloc:prepare", zs_cpu_prepare, zs_cpu_dead); |
0f050d997
|
2375 |
if (ret) |
215c89d05
|
2376 |
goto hp_setup_fail; |
66cdef663
|
2377 |
|
66cdef663
|
2378 2379 2380 |
#ifdef CONFIG_ZPOOL zpool_register_driver(&zs_zpool_driver); #endif |
0f050d997
|
2381 |
|
4abaac9b7
|
2382 |
zs_stat_init(); |
66cdef663
|
2383 |
return 0; |
0f050d997
|
2384 |
|
215c89d05
|
2385 |
hp_setup_fail: |
48b4800a1
|
2386 2387 |
zsmalloc_unmount(); out: |
0f050d997
|
2388 |
return ret; |
61989a80f
|
2389 |
} |
61989a80f
|
2390 |
|
66cdef663
|
2391 |
static void __exit zs_exit(void) |
61989a80f
|
2392 |
{ |
66cdef663
|
2393 2394 2395 |
#ifdef CONFIG_ZPOOL zpool_unregister_driver(&zs_zpool_driver); #endif |
48b4800a1
|
2396 |
zsmalloc_unmount(); |
215c89d05
|
2397 |
cpuhp_remove_state(CPUHP_MM_ZS_PREPARE); |
0f050d997
|
2398 2399 |
zs_stat_exit(); |
61989a80f
|
2400 |
} |
069f101fa
|
2401 2402 2403 2404 2405 2406 |
module_init(zs_init); module_exit(zs_exit); MODULE_LICENSE("Dual BSD/GPL"); MODULE_AUTHOR("Nitin Gupta <ngupta@vflare.org>"); |