Blame view
mm/ksm.c
63.5 KB
f8af4da3b
|
1 |
/* |
31dbd01f3
|
2 3 4 5 6 |
* Memory merging support. * * This code enables dynamic sharing of identical pages found in different * memory areas, even if they are not shared by fork() * |
36b2528dc
|
7 |
* Copyright (C) 2008-2009 Red Hat, Inc. |
31dbd01f3
|
8 9 10 11 |
* Authors: * Izik Eidus * Andrea Arcangeli * Chris Wright |
36b2528dc
|
12 |
* Hugh Dickins |
31dbd01f3
|
13 14 |
* * This work is licensed under the terms of the GNU GPL, version 2. |
f8af4da3b
|
15 16 17 |
*/ #include <linux/errno.h> |
31dbd01f3
|
18 19 |
#include <linux/mm.h> #include <linux/fs.h> |
f8af4da3b
|
20 |
#include <linux/mman.h> |
31dbd01f3
|
21 22 23 24 25 26 27 28 29 30 31 |
#include <linux/sched.h> #include <linux/rwsem.h> #include <linux/pagemap.h> #include <linux/rmap.h> #include <linux/spinlock.h> #include <linux/jhash.h> #include <linux/delay.h> #include <linux/kthread.h> #include <linux/wait.h> #include <linux/slab.h> #include <linux/rbtree.h> |
62b61f611
|
32 |
#include <linux/memory.h> |
31dbd01f3
|
33 |
#include <linux/mmu_notifier.h> |
2c6854fda
|
34 |
#include <linux/swap.h> |
f8af4da3b
|
35 |
#include <linux/ksm.h> |
4ca3a69bc
|
36 |
#include <linux/hashtable.h> |
878aee7d6
|
37 |
#include <linux/freezer.h> |
72788c385
|
38 |
#include <linux/oom.h> |
90bd6fd31
|
39 |
#include <linux/numa.h> |
f8af4da3b
|
40 |
|
31dbd01f3
|
41 |
#include <asm/tlbflush.h> |
73848b468
|
42 |
#include "internal.h" |
31dbd01f3
|
43 |
|
e850dcf53
|
44 45 46 47 48 49 50 |
#ifdef CONFIG_NUMA #define NUMA(x) (x) #define DO_NUMA(x) do { (x); } while (0) #else #define NUMA(x) (0) #define DO_NUMA(x) do { } while (0) #endif |
31dbd01f3
|
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 |
/* * A few notes about the KSM scanning process, * to make it easier to understand the data structures below: * * In order to reduce excessive scanning, KSM sorts the memory pages by their * contents into a data structure that holds pointers to the pages' locations. * * Since the contents of the pages may change at any moment, KSM cannot just * insert the pages into a normal sorted tree and expect it to find anything. * Therefore KSM uses two data structures - the stable and the unstable tree. * * The stable tree holds pointers to all the merged pages (ksm pages), sorted * by their contents. Because each such page is write-protected, searching on * this tree is fully assured to be working (except when pages are unmapped), * and therefore this tree is called the stable tree. * * In addition to the stable tree, KSM uses a second data structure called the * unstable tree: this tree holds pointers to pages which have been found to * be "unchanged for a period of time". The unstable tree sorts these pages * by their contents, but since they are not write-protected, KSM cannot rely * upon the unstable tree to work correctly - the unstable tree is liable to * be corrupted as its contents are modified, and so it is called unstable. * * KSM solves this problem by several techniques: * * 1) The unstable tree is flushed every time KSM completes scanning all * memory areas, and then the tree is rebuilt again from the beginning. * 2) KSM will only insert into the unstable tree, pages whose hash value * has not changed since the previous scan of all memory areas. * 3) The unstable tree is a RedBlack Tree - so its balancing is based on the * colors of the nodes and not on their contents, assuring that even when * the tree gets "corrupted" it won't get out of balance, so scanning time * remains the same (also, searching and inserting nodes in an rbtree uses * the same algorithm, so we have no overhead when we flush and rebuild). * 4) KSM never flushes the stable tree, which means that even if it were to * take 10 attempts to find a page in the unstable tree, once it is found, * it is secured in the stable tree. (When we scan a new page, we first * compare it against the stable tree, and then against the unstable tree.) |
8fdb3dbf0
|
89 90 91 |
* * If the merge_across_nodes tunable is unset, then KSM maintains multiple * stable trees and multiple unstable trees: one of each for each NUMA node. |
31dbd01f3
|
92 93 94 95 96 97 |
*/ /** * struct mm_slot - ksm information per mm that is being scanned * @link: link to the mm_slots hash list * @mm_list: link into the mm_slots list, rooted in ksm_mm_head |
6514d511d
|
98 |
* @rmap_list: head for this mm_slot's singly-linked list of rmap_items |
31dbd01f3
|
99 100 101 102 103 |
* @mm: the mm that this information is valid for */ struct mm_slot { struct hlist_node link; struct list_head mm_list; |
6514d511d
|
104 |
struct rmap_item *rmap_list; |
31dbd01f3
|
105 106 107 108 109 110 111 |
struct mm_struct *mm; }; /** * struct ksm_scan - cursor for scanning * @mm_slot: the current mm_slot we are scanning * @address: the next address inside that to be scanned |
6514d511d
|
112 |
* @rmap_list: link to the next rmap to be scanned in the rmap_list |
31dbd01f3
|
113 114 115 116 117 118 119 |
* @seqnr: count of completed full scans (needed when removing unstable node) * * There is only the one ksm_scan instance of this cursor structure. */ struct ksm_scan { struct mm_slot *mm_slot; unsigned long address; |
6514d511d
|
120 |
struct rmap_item **rmap_list; |
31dbd01f3
|
121 122 123 124 |
unsigned long seqnr; }; /** |
7b6ba2c7d
|
125 126 |
* struct stable_node - node of the stable rbtree * @node: rb node of this ksm page in the stable tree |
4146d2d67
|
127 128 |
* @head: (overlaying parent) &migrate_nodes indicates temporarily on that list * @list: linked into migrate_nodes, pending placement in the proper node tree |
7b6ba2c7d
|
129 |
* @hlist: hlist head of rmap_items using this ksm page |
4146d2d67
|
130 131 |
* @kpfn: page frame number of this ksm page (perhaps temporarily on wrong nid) * @nid: NUMA node id of stable tree in which linked (may not match kpfn) |
7b6ba2c7d
|
132 133 |
*/ struct stable_node { |
4146d2d67
|
134 135 136 137 138 139 140 |
union { struct rb_node node; /* when node of stable tree */ struct { /* when listed for migration */ struct list_head *head; struct list_head list; }; }; |
7b6ba2c7d
|
141 |
struct hlist_head hlist; |
62b61f611
|
142 |
unsigned long kpfn; |
4146d2d67
|
143 144 145 |
#ifdef CONFIG_NUMA int nid; #endif |
7b6ba2c7d
|
146 147 148 |
}; /** |
31dbd01f3
|
149 |
* struct rmap_item - reverse mapping item for virtual addresses |
6514d511d
|
150 |
* @rmap_list: next rmap_item in mm_slot's singly-linked rmap_list |
db114b83a
|
151 |
* @anon_vma: pointer to anon_vma for this mm,address, when in stable tree |
bc56620b4
|
152 |
* @nid: NUMA node id of unstable tree in which linked (may not match page) |
31dbd01f3
|
153 154 155 |
* @mm: the memory structure this rmap_item is pointing into * @address: the virtual address this rmap_item tracks (+ flags in low bits) * @oldchecksum: previous checksum of the page at that virtual address |
7b6ba2c7d
|
156 157 158 |
* @node: rb node of this rmap_item in the unstable tree * @head: pointer to stable_node heading this list in the stable tree * @hlist: link into hlist of rmap_items hanging off that stable_node |
31dbd01f3
|
159 160 |
*/ struct rmap_item { |
6514d511d
|
161 |
struct rmap_item *rmap_list; |
bc56620b4
|
162 163 164 165 166 167 |
union { struct anon_vma *anon_vma; /* when stable */ #ifdef CONFIG_NUMA int nid; /* when node of unstable tree */ #endif }; |
31dbd01f3
|
168 169 |
struct mm_struct *mm; unsigned long address; /* + low bits used for flags below */ |
7b6ba2c7d
|
170 |
unsigned int oldchecksum; /* when unstable */ |
31dbd01f3
|
171 |
union { |
7b6ba2c7d
|
172 173 174 175 176 |
struct rb_node node; /* when node of unstable tree */ struct { /* when listed from stable tree */ struct stable_node *head; struct hlist_node hlist; }; |
31dbd01f3
|
177 178 179 180 |
}; }; #define SEQNR_MASK 0x0ff /* low bits of unstable tree seqnr */ |
7b6ba2c7d
|
181 182 |
#define UNSTABLE_FLAG 0x100 /* is a node of the unstable tree */ #define STABLE_FLAG 0x200 /* is listed from the stable tree */ |
31dbd01f3
|
183 184 |
/* The stable and unstable tree heads */ |
ef53d16cd
|
185 186 187 188 |
static struct rb_root one_stable_tree[1] = { RB_ROOT }; static struct rb_root one_unstable_tree[1] = { RB_ROOT }; static struct rb_root *root_stable_tree = one_stable_tree; static struct rb_root *root_unstable_tree = one_unstable_tree; |
31dbd01f3
|
189 |
|
4146d2d67
|
190 191 |
/* Recently migrated nodes of stable tree, pending proper placement */ static LIST_HEAD(migrate_nodes); |
4ca3a69bc
|
192 193 |
#define MM_SLOTS_HASH_BITS 10 static DEFINE_HASHTABLE(mm_slots_hash, MM_SLOTS_HASH_BITS); |
31dbd01f3
|
194 195 196 197 198 199 200 201 202 |
static struct mm_slot ksm_mm_head = { .mm_list = LIST_HEAD_INIT(ksm_mm_head.mm_list), }; static struct ksm_scan ksm_scan = { .mm_slot = &ksm_mm_head, }; static struct kmem_cache *rmap_item_cache; |
7b6ba2c7d
|
203 |
static struct kmem_cache *stable_node_cache; |
31dbd01f3
|
204 205 206 |
static struct kmem_cache *mm_slot_cache; /* The number of nodes in the stable tree */ |
b40282603
|
207 |
static unsigned long ksm_pages_shared; |
31dbd01f3
|
208 |
|
e178dfde3
|
209 |
/* The number of page slots additionally sharing those nodes */ |
b40282603
|
210 |
static unsigned long ksm_pages_sharing; |
31dbd01f3
|
211 |
|
473b0ce4d
|
212 213 214 215 216 |
/* The number of nodes in the unstable tree */ static unsigned long ksm_pages_unshared; /* The number of rmap_items in use: to calculate pages_volatile */ static unsigned long ksm_rmap_items; |
31dbd01f3
|
217 |
/* Number of pages ksmd should scan in one batch */ |
2c6854fda
|
218 |
static unsigned int ksm_thread_pages_to_scan = 100; |
31dbd01f3
|
219 220 |
/* Milliseconds ksmd should sleep between batches */ |
2ffd8679c
|
221 |
static unsigned int ksm_thread_sleep_millisecs = 20; |
31dbd01f3
|
222 |
|
e850dcf53
|
223 |
#ifdef CONFIG_NUMA |
90bd6fd31
|
224 225 |
/* Zeroed when merging across nodes is not allowed */ static unsigned int ksm_merge_across_nodes = 1; |
ef53d16cd
|
226 |
static int ksm_nr_node_ids = 1; |
e850dcf53
|
227 228 |
#else #define ksm_merge_across_nodes 1U |
ef53d16cd
|
229 |
#define ksm_nr_node_ids 1 |
e850dcf53
|
230 |
#endif |
90bd6fd31
|
231 |
|
31dbd01f3
|
232 233 234 |
#define KSM_RUN_STOP 0 #define KSM_RUN_MERGE 1 #define KSM_RUN_UNMERGE 2 |
ef4d43a80
|
235 236 237 |
#define KSM_RUN_OFFLINE 4 static unsigned long ksm_run = KSM_RUN_STOP; static void wait_while_offlining(void); |
31dbd01f3
|
238 239 240 241 242 243 244 245 246 247 248 249 250 251 |
static DECLARE_WAIT_QUEUE_HEAD(ksm_thread_wait); static DEFINE_MUTEX(ksm_thread_mutex); static DEFINE_SPINLOCK(ksm_mmlist_lock); #define KSM_KMEM_CACHE(__struct, __flags) kmem_cache_create("ksm_"#__struct,\ sizeof(struct __struct), __alignof__(struct __struct),\ (__flags), NULL) static int __init ksm_slab_init(void) { rmap_item_cache = KSM_KMEM_CACHE(rmap_item, 0); if (!rmap_item_cache) goto out; |
7b6ba2c7d
|
252 253 254 |
stable_node_cache = KSM_KMEM_CACHE(stable_node, 0); if (!stable_node_cache) goto out_free1; |
31dbd01f3
|
255 256 |
mm_slot_cache = KSM_KMEM_CACHE(mm_slot, 0); if (!mm_slot_cache) |
7b6ba2c7d
|
257 |
goto out_free2; |
31dbd01f3
|
258 259 |
return 0; |
7b6ba2c7d
|
260 261 262 |
out_free2: kmem_cache_destroy(stable_node_cache); out_free1: |
31dbd01f3
|
263 264 265 266 267 268 269 270 |
kmem_cache_destroy(rmap_item_cache); out: return -ENOMEM; } static void __init ksm_slab_free(void) { kmem_cache_destroy(mm_slot_cache); |
7b6ba2c7d
|
271 |
kmem_cache_destroy(stable_node_cache); |
31dbd01f3
|
272 273 274 275 276 277 |
kmem_cache_destroy(rmap_item_cache); mm_slot_cache = NULL; } static inline struct rmap_item *alloc_rmap_item(void) { |
473b0ce4d
|
278 |
struct rmap_item *rmap_item; |
5b398e416
|
279 280 |
rmap_item = kmem_cache_zalloc(rmap_item_cache, GFP_KERNEL | __GFP_NORETRY | __GFP_NOWARN); |
473b0ce4d
|
281 282 283 |
if (rmap_item) ksm_rmap_items++; return rmap_item; |
31dbd01f3
|
284 285 286 287 |
} static inline void free_rmap_item(struct rmap_item *rmap_item) { |
473b0ce4d
|
288 |
ksm_rmap_items--; |
31dbd01f3
|
289 290 291 |
rmap_item->mm = NULL; /* debug safety */ kmem_cache_free(rmap_item_cache, rmap_item); } |
7b6ba2c7d
|
292 293 |
static inline struct stable_node *alloc_stable_node(void) { |
6213055f2
|
294 295 296 297 298 299 |
/* * The allocation can take too long with GFP_KERNEL when memory is under * pressure, which may lead to hung task warnings. Adding __GFP_HIGH * grants access to memory reserves, helping to avoid this problem. */ return kmem_cache_alloc(stable_node_cache, GFP_KERNEL | __GFP_HIGH); |
7b6ba2c7d
|
300 301 302 303 304 305 |
} static inline void free_stable_node(struct stable_node *stable_node) { kmem_cache_free(stable_node_cache, stable_node); } |
31dbd01f3
|
306 307 308 309 310 311 312 313 314 315 316 |
static inline struct mm_slot *alloc_mm_slot(void) { if (!mm_slot_cache) /* initialization failed */ return NULL; return kmem_cache_zalloc(mm_slot_cache, GFP_KERNEL); } static inline void free_mm_slot(struct mm_slot *mm_slot) { kmem_cache_free(mm_slot_cache, mm_slot); } |
31dbd01f3
|
317 318 |
static struct mm_slot *get_mm_slot(struct mm_struct *mm) { |
4ca3a69bc
|
319 |
struct mm_slot *slot; |
b67bfe0d4
|
320 |
hash_for_each_possible(mm_slots_hash, slot, link, (unsigned long)mm) |
4ca3a69bc
|
321 322 |
if (slot->mm == mm) return slot; |
31dbd01f3
|
323 |
|
31dbd01f3
|
324 325 326 327 328 329 |
return NULL; } static void insert_to_mm_slots_hash(struct mm_struct *mm, struct mm_slot *mm_slot) { |
31dbd01f3
|
330 |
mm_slot->mm = mm; |
4ca3a69bc
|
331 |
hash_add(mm_slots_hash, &mm_slot->link, (unsigned long)mm); |
31dbd01f3
|
332 |
} |
31dbd01f3
|
333 |
/* |
a913e182a
|
334 335 336 337 338 339 340 341 342 343 344 345 346 |
* ksmd, and unmerge_and_remove_all_rmap_items(), must not touch an mm's * page tables after it has passed through ksm_exit() - which, if necessary, * takes mmap_sem briefly to serialize against them. ksm_exit() does not set * a special flag: they can just back out as soon as mm_users goes to zero. * ksm_test_exit() is used throughout to make this test for exit: in some * places for correctness, in some places just to avoid unnecessary work. */ static inline bool ksm_test_exit(struct mm_struct *mm) { return atomic_read(&mm->mm_users) == 0; } /* |
31dbd01f3
|
347 348 |
* We use break_ksm to break COW on a ksm page: it's a stripped down * |
d4edcf0d5
|
349 |
* if (get_user_pages(addr, 1, 1, 1, &page, NULL) == 1) |
31dbd01f3
|
350 351 352 353 354 355 |
* put_page(page); * * but taking great care only to touch a ksm page, in a VM_MERGEABLE vma, * in case the application has unmapped and remapped mm,addr meanwhile. * Could a ksm page appear anywhere else? Actually yes, in a VM_PFNMAP * mmap of /dev/mem or /dev/kmem, where we would not want to touch it. |
1b2ee1266
|
356 357 358 359 |
* * FAULT_FLAG/FOLL_REMOTE are because we do this outside the context * of the process that owns 'vma'. We also do not want to enforce * protection keys here anyway. |
31dbd01f3
|
360 |
*/ |
d952b7913
|
361 |
static int break_ksm(struct vm_area_struct *vma, unsigned long addr) |
31dbd01f3
|
362 363 |
{ struct page *page; |
d952b7913
|
364 |
int ret = 0; |
31dbd01f3
|
365 366 367 |
do { cond_resched(); |
1b2ee1266
|
368 369 |
page = follow_page(vma, addr, FOLL_GET | FOLL_MIGRATION | FOLL_REMOTE); |
22eccdd7d
|
370 |
if (IS_ERR_OR_NULL(page)) |
31dbd01f3
|
371 372 |
break; if (PageKsm(page)) |
dcddffd41
|
373 374 |
ret = handle_mm_fault(vma, addr, FAULT_FLAG_WRITE | FAULT_FLAG_REMOTE); |
31dbd01f3
|
375 376 377 |
else ret = VM_FAULT_WRITE; put_page(page); |
33692f275
|
378 |
} while (!(ret & (VM_FAULT_WRITE | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | VM_FAULT_OOM))); |
d952b7913
|
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 |
/* * We must loop because handle_mm_fault() may back out if there's * any difficulty e.g. if pte accessed bit gets updated concurrently. * * VM_FAULT_WRITE is what we have been hoping for: it indicates that * COW has been broken, even if the vma does not permit VM_WRITE; * but note that a concurrent fault might break PageKsm for us. * * VM_FAULT_SIGBUS could occur if we race with truncation of the * backing file, which also invalidates anonymous pages: that's * okay, that truncation will have unmapped the PageKsm for us. * * VM_FAULT_OOM: at the time of writing (late July 2009), setting * aside mem_cgroup limits, VM_FAULT_OOM would only be set if the * current task has TIF_MEMDIE set, and will be OOM killed on return * to user; and ksmd, having no mm, would never be chosen for that. * * But if the mm is in a limited mem_cgroup, then the fault may fail * with VM_FAULT_OOM even if the current task is not TIF_MEMDIE; and * even ksmd can fail in this way - though it's usually breaking ksm * just to undo a merge it made a moment before, so unlikely to oom. * * That's a pity: we might therefore have more kernel pages allocated * than we're counting as nodes in the stable tree; but ksm_do_scan * will retry to break_cow on each pass, so should recover the page * in due course. The important thing is to not let VM_MERGEABLE * be cleared while any such pages might remain in the area. */ return (ret & VM_FAULT_OOM) ? -ENOMEM : 0; |
31dbd01f3
|
408 |
} |
ef6942224
|
409 410 411 412 413 414 415 416 417 418 419 420 421 |
static struct vm_area_struct *find_mergeable_vma(struct mm_struct *mm, unsigned long addr) { struct vm_area_struct *vma; if (ksm_test_exit(mm)) return NULL; vma = find_vma(mm, addr); if (!vma || vma->vm_start > addr) return NULL; if (!(vma->vm_flags & VM_MERGEABLE) || !vma->anon_vma) return NULL; return vma; } |
8dd3557a5
|
422 |
static void break_cow(struct rmap_item *rmap_item) |
31dbd01f3
|
423 |
{ |
8dd3557a5
|
424 425 |
struct mm_struct *mm = rmap_item->mm; unsigned long addr = rmap_item->address; |
31dbd01f3
|
426 |
struct vm_area_struct *vma; |
4035c07a8
|
427 428 429 430 |
/* * It is not an accident that whenever we want to break COW * to undo, we also need to drop a reference to the anon_vma. */ |
9e60109f1
|
431 |
put_anon_vma(rmap_item->anon_vma); |
4035c07a8
|
432 |
|
81464e306
|
433 |
down_read(&mm->mmap_sem); |
ef6942224
|
434 435 436 |
vma = find_mergeable_vma(mm, addr); if (vma) break_ksm(vma, addr); |
31dbd01f3
|
437 438 439 440 441 442 443 444 445 446 447 |
up_read(&mm->mmap_sem); } static struct page *get_mergeable_page(struct rmap_item *rmap_item) { struct mm_struct *mm = rmap_item->mm; unsigned long addr = rmap_item->address; struct vm_area_struct *vma; struct page *page; down_read(&mm->mmap_sem); |
ef6942224
|
448 449 |
vma = find_mergeable_vma(mm, addr); if (!vma) |
31dbd01f3
|
450 451 452 |
goto out; page = follow_page(vma, addr, FOLL_GET); |
22eccdd7d
|
453 |
if (IS_ERR_OR_NULL(page)) |
31dbd01f3
|
454 |
goto out; |
f765f5405
|
455 |
if (PageAnon(page)) { |
31dbd01f3
|
456 457 458 459 |
flush_anon_page(vma, page, addr); flush_dcache_page(page); } else { put_page(page); |
c8f95ed1a
|
460 461 |
out: page = NULL; |
31dbd01f3
|
462 463 464 465 |
} up_read(&mm->mmap_sem); return page; } |
90bd6fd31
|
466 467 468 469 470 471 472 473 |
/* * This helper is used for getting right index into array of tree roots. * When merge_across_nodes knob is set to 1, there are only two rb-trees for * stable and unstable pages from all nodes with roots in index 0. Otherwise, * every node has its own stable and unstable tree. */ static inline int get_kpfn_nid(unsigned long kpfn) { |
d8fc16a82
|
474 |
return ksm_merge_across_nodes ? 0 : NUMA(pfn_to_nid(kpfn)); |
90bd6fd31
|
475 |
} |
4035c07a8
|
476 477 478 |
static void remove_node_from_stable_tree(struct stable_node *stable_node) { struct rmap_item *rmap_item; |
4035c07a8
|
479 |
|
b67bfe0d4
|
480 |
hlist_for_each_entry(rmap_item, &stable_node->hlist, hlist) { |
4035c07a8
|
481 482 483 484 |
if (rmap_item->hlist.next) ksm_pages_sharing--; else ksm_pages_shared--; |
9e60109f1
|
485 |
put_anon_vma(rmap_item->anon_vma); |
4035c07a8
|
486 487 488 |
rmap_item->address &= PAGE_MASK; cond_resched(); } |
4146d2d67
|
489 490 491 492 |
if (stable_node->head == &migrate_nodes) list_del(&stable_node->list); else rb_erase(&stable_node->node, |
ef53d16cd
|
493 |
root_stable_tree + NUMA(stable_node->nid)); |
4035c07a8
|
494 495 496 497 498 499 500 501 502 |
free_stable_node(stable_node); } /* * get_ksm_page: checks if the page indicated by the stable node * is still its ksm page, despite having held no reference to it. * In which case we can trust the content of the page, and it * returns the gotten page; but if the page has now been zapped, * remove the stale node from the stable tree and return NULL. |
c8d6553b9
|
503 |
* But beware, the stable node's page might be being migrated. |
4035c07a8
|
504 505 506 507 508 509 510 511 512 513 |
* * You would expect the stable_node to hold a reference to the ksm page. * But if it increments the page's count, swapping out has to wait for * ksmd to come around again before it can free the page, which may take * seconds or even minutes: much too unresponsive. So instead we use a * "keyhole reference": access to the ksm page from the stable node peeps * out through its keyhole to see if that page still holds the right key, * pointing back to this stable node. This relies on freeing a PageAnon * page to reset its page->mapping to NULL, and relies on no other use of * a page to put something that might look like our key in page->mapping. |
4035c07a8
|
514 515 |
* is on its way to being freed; but it is an anomaly to bear in mind. */ |
8fdb3dbf0
|
516 |
static struct page *get_ksm_page(struct stable_node *stable_node, bool lock_it) |
4035c07a8
|
517 518 519 |
{ struct page *page; void *expected_mapping; |
c8d6553b9
|
520 |
unsigned long kpfn; |
4035c07a8
|
521 |
|
bda807d44
|
522 523 |
expected_mapping = (void *)((unsigned long)stable_node | PAGE_MAPPING_KSM); |
c8d6553b9
|
524 |
again: |
4db0c3c29
|
525 |
kpfn = READ_ONCE(stable_node->kpfn); |
c8d6553b9
|
526 527 528 529 530 531 532 533 |
page = pfn_to_page(kpfn); /* * page is computed from kpfn, so on most architectures reading * page->mapping is naturally ordered after reading node->kpfn, * but on Alpha we need to be more careful. */ smp_read_barrier_depends(); |
4db0c3c29
|
534 |
if (READ_ONCE(page->mapping) != expected_mapping) |
4035c07a8
|
535 |
goto stale; |
c8d6553b9
|
536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 |
/* * We cannot do anything with the page while its refcount is 0. * Usually 0 means free, or tail of a higher-order page: in which * case this node is no longer referenced, and should be freed; * however, it might mean that the page is under page_freeze_refs(). * The __remove_mapping() case is easy, again the node is now stale; * but if page is swapcache in migrate_page_move_mapping(), it might * still be our page, in which case it's essential to keep the node. */ while (!get_page_unless_zero(page)) { /* * Another check for page->mapping != expected_mapping would * work here too. We have chosen the !PageSwapCache test to * optimize the common case, when the page is or is about to * be freed: PageSwapCache is cleared (under spin_lock_irq) * in the freeze_refs section of __remove_mapping(); but Anon * page->mapping reset to NULL later, in free_pages_prepare(). */ if (!PageSwapCache(page)) goto stale; cpu_relax(); } |
4db0c3c29
|
559 |
if (READ_ONCE(page->mapping) != expected_mapping) { |
4035c07a8
|
560 561 562 |
put_page(page); goto stale; } |
c8d6553b9
|
563 |
|
8fdb3dbf0
|
564 |
if (lock_it) { |
8aafa6a48
|
565 |
lock_page(page); |
4db0c3c29
|
566 |
if (READ_ONCE(page->mapping) != expected_mapping) { |
8aafa6a48
|
567 568 569 570 571 |
unlock_page(page); put_page(page); goto stale; } } |
4035c07a8
|
572 |
return page; |
c8d6553b9
|
573 |
|
4035c07a8
|
574 |
stale: |
c8d6553b9
|
575 576 577 578 579 580 581 |
/* * We come here from above when page->mapping or !PageSwapCache * suggests that the node is stale; but it might be under migration. * We need smp_rmb(), matching the smp_wmb() in ksm_migrate_page(), * before checking whether node->kpfn has been changed. */ smp_rmb(); |
4db0c3c29
|
582 |
if (READ_ONCE(stable_node->kpfn) != kpfn) |
c8d6553b9
|
583 |
goto again; |
4035c07a8
|
584 585 586 |
remove_node_from_stable_tree(stable_node); return NULL; } |
31dbd01f3
|
587 |
/* |
31dbd01f3
|
588 589 590 591 592 |
* Removing rmap_item from stable or unstable tree. * This function will clean the information from the stable/unstable tree. */ static void remove_rmap_item_from_tree(struct rmap_item *rmap_item) { |
7b6ba2c7d
|
593 594 |
if (rmap_item->address & STABLE_FLAG) { struct stable_node *stable_node; |
5ad646880
|
595 |
struct page *page; |
31dbd01f3
|
596 |
|
7b6ba2c7d
|
597 |
stable_node = rmap_item->head; |
8aafa6a48
|
598 |
page = get_ksm_page(stable_node, true); |
4035c07a8
|
599 600 |
if (!page) goto out; |
5ad646880
|
601 |
|
7b6ba2c7d
|
602 |
hlist_del(&rmap_item->hlist); |
4035c07a8
|
603 604 |
unlock_page(page); put_page(page); |
08beca44d
|
605 |
|
98666f8a2
|
606 |
if (!hlist_empty(&stable_node->hlist)) |
4035c07a8
|
607 608 |
ksm_pages_sharing--; else |
7b6ba2c7d
|
609 |
ksm_pages_shared--; |
31dbd01f3
|
610 |
|
9e60109f1
|
611 |
put_anon_vma(rmap_item->anon_vma); |
93d17715a
|
612 |
rmap_item->address &= PAGE_MASK; |
31dbd01f3
|
613 |
|
7b6ba2c7d
|
614 |
} else if (rmap_item->address & UNSTABLE_FLAG) { |
31dbd01f3
|
615 616 |
unsigned char age; /* |
9ba692948
|
617 |
* Usually ksmd can and must skip the rb_erase, because |
31dbd01f3
|
618 |
* root_unstable_tree was already reset to RB_ROOT. |
9ba692948
|
619 620 621 |
* But be careful when an mm is exiting: do the rb_erase * if this rmap_item was inserted by this scan, rather * than left over from before. |
31dbd01f3
|
622 623 |
*/ age = (unsigned char)(ksm_scan.seqnr - rmap_item->address); |
cd551f975
|
624 |
BUG_ON(age > 1); |
31dbd01f3
|
625 |
if (!age) |
90bd6fd31
|
626 |
rb_erase(&rmap_item->node, |
ef53d16cd
|
627 |
root_unstable_tree + NUMA(rmap_item->nid)); |
473b0ce4d
|
628 |
ksm_pages_unshared--; |
93d17715a
|
629 |
rmap_item->address &= PAGE_MASK; |
31dbd01f3
|
630 |
} |
4035c07a8
|
631 |
out: |
31dbd01f3
|
632 633 |
cond_resched(); /* we're called from many long loops */ } |
31dbd01f3
|
634 |
static void remove_trailing_rmap_items(struct mm_slot *mm_slot, |
6514d511d
|
635 |
struct rmap_item **rmap_list) |
31dbd01f3
|
636 |
{ |
6514d511d
|
637 638 639 |
while (*rmap_list) { struct rmap_item *rmap_item = *rmap_list; *rmap_list = rmap_item->rmap_list; |
31dbd01f3
|
640 |
remove_rmap_item_from_tree(rmap_item); |
31dbd01f3
|
641 642 643 644 645 |
free_rmap_item(rmap_item); } } /* |
e850dcf53
|
646 |
* Though it's very tempting to unmerge rmap_items from stable tree rather |
31dbd01f3
|
647 648 649 650 651 |
* than check every pte of a given vma, the locking doesn't quite work for * that - an rmap_item is assigned to the stable tree after inserting ksm * page and upping mmap_sem. Nor does it fit with the way we skip dup'ing * rmap_items from parent to child at fork time (so as not to waste time * if exit comes before the next scan reaches it). |
81464e306
|
652 653 654 655 656 |
* * Similarly, although we'd like to remove rmap_items (so updating counts * and freeing memory) when unmerging an area, it's easier to leave that * to the next pass of ksmd - consider, for example, how ksmd might be * in cmp_and_merge_page on one of the rmap_items we would be removing. |
31dbd01f3
|
657 |
*/ |
d952b7913
|
658 659 |
static int unmerge_ksm_pages(struct vm_area_struct *vma, unsigned long start, unsigned long end) |
31dbd01f3
|
660 661 |
{ unsigned long addr; |
d952b7913
|
662 |
int err = 0; |
31dbd01f3
|
663 |
|
d952b7913
|
664 |
for (addr = start; addr < end && !err; addr += PAGE_SIZE) { |
9ba692948
|
665 666 |
if (ksm_test_exit(vma->vm_mm)) break; |
d952b7913
|
667 668 669 670 671 672 |
if (signal_pending(current)) err = -ERESTARTSYS; else err = break_ksm(vma, addr); } return err; |
31dbd01f3
|
673 |
} |
2ffd8679c
|
674 675 676 677 |
#ifdef CONFIG_SYSFS /* * Only called through the sysfs control interface: */ |
cbf86cfe0
|
678 679 680 681 682 683 684 685 686 687 688 689 |
static int remove_stable_node(struct stable_node *stable_node) { struct page *page; int err; page = get_ksm_page(stable_node, true); if (!page) { /* * get_ksm_page did remove_node_from_stable_tree itself. */ return 0; } |
8fdb3dbf0
|
690 691 692 693 694 |
if (WARN_ON_ONCE(page_mapped(page))) { /* * This should not happen: but if it does, just refuse to let * merge_across_nodes be switched - there is no need to panic. */ |
cbf86cfe0
|
695 |
err = -EBUSY; |
8fdb3dbf0
|
696 |
} else { |
cbf86cfe0
|
697 |
/* |
8fdb3dbf0
|
698 699 700 |
* The stable node did not yet appear stale to get_ksm_page(), * since that allows for an unmapped ksm page to be recognized * right up until it is freed; but the node is safe to remove. |
cbf86cfe0
|
701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 |
* This page might be in a pagevec waiting to be freed, * or it might be PageSwapCache (perhaps under writeback), * or it might have been removed from swapcache a moment ago. */ set_page_stable_node(page, NULL); remove_node_from_stable_tree(stable_node); err = 0; } unlock_page(page); put_page(page); return err; } static int remove_all_stable_nodes(void) { |
036404183
|
717 |
struct stable_node *stable_node, *next; |
cbf86cfe0
|
718 719 |
int nid; int err = 0; |
ef53d16cd
|
720 |
for (nid = 0; nid < ksm_nr_node_ids; nid++) { |
cbf86cfe0
|
721 722 723 724 725 726 727 728 729 730 |
while (root_stable_tree[nid].rb_node) { stable_node = rb_entry(root_stable_tree[nid].rb_node, struct stable_node, node); if (remove_stable_node(stable_node)) { err = -EBUSY; break; /* proceed to next nid */ } cond_resched(); } } |
036404183
|
731 |
list_for_each_entry_safe(stable_node, next, &migrate_nodes, list) { |
4146d2d67
|
732 733 734 735 |
if (remove_stable_node(stable_node)) err = -EBUSY; cond_resched(); } |
cbf86cfe0
|
736 737 |
return err; } |
d952b7913
|
738 |
static int unmerge_and_remove_all_rmap_items(void) |
31dbd01f3
|
739 740 741 742 |
{ struct mm_slot *mm_slot; struct mm_struct *mm; struct vm_area_struct *vma; |
d952b7913
|
743 744 745 |
int err = 0; spin_lock(&ksm_mmlist_lock); |
9ba692948
|
746 |
ksm_scan.mm_slot = list_entry(ksm_mm_head.mm_list.next, |
d952b7913
|
747 748 |
struct mm_slot, mm_list); spin_unlock(&ksm_mmlist_lock); |
31dbd01f3
|
749 |
|
9ba692948
|
750 751 |
for (mm_slot = ksm_scan.mm_slot; mm_slot != &ksm_mm_head; mm_slot = ksm_scan.mm_slot) { |
31dbd01f3
|
752 753 754 |
mm = mm_slot->mm; down_read(&mm->mmap_sem); for (vma = mm->mmap; vma; vma = vma->vm_next) { |
9ba692948
|
755 756 |
if (ksm_test_exit(mm)) break; |
31dbd01f3
|
757 758 |
if (!(vma->vm_flags & VM_MERGEABLE) || !vma->anon_vma) continue; |
d952b7913
|
759 760 |
err = unmerge_ksm_pages(vma, vma->vm_start, vma->vm_end); |
9ba692948
|
761 762 |
if (err) goto error; |
31dbd01f3
|
763 |
} |
9ba692948
|
764 |
|
6514d511d
|
765 |
remove_trailing_rmap_items(mm_slot, &mm_slot->rmap_list); |
7496fea9a
|
766 |
up_read(&mm->mmap_sem); |
d952b7913
|
767 768 |
spin_lock(&ksm_mmlist_lock); |
9ba692948
|
769 |
ksm_scan.mm_slot = list_entry(mm_slot->mm_list.next, |
d952b7913
|
770 |
struct mm_slot, mm_list); |
9ba692948
|
771 |
if (ksm_test_exit(mm)) { |
4ca3a69bc
|
772 |
hash_del(&mm_slot->link); |
9ba692948
|
773 774 775 776 777 |
list_del(&mm_slot->mm_list); spin_unlock(&ksm_mmlist_lock); free_mm_slot(mm_slot); clear_bit(MMF_VM_MERGEABLE, &mm->flags); |
9ba692948
|
778 |
mmdrop(mm); |
7496fea9a
|
779 |
} else |
9ba692948
|
780 |
spin_unlock(&ksm_mmlist_lock); |
31dbd01f3
|
781 |
} |
cbf86cfe0
|
782 783 |
/* Clean up stable nodes, but don't worry if some are still busy */ remove_all_stable_nodes(); |
d952b7913
|
784 |
ksm_scan.seqnr = 0; |
9ba692948
|
785 786 787 788 |
return 0; error: up_read(&mm->mmap_sem); |
31dbd01f3
|
789 |
spin_lock(&ksm_mmlist_lock); |
d952b7913
|
790 |
ksm_scan.mm_slot = &ksm_mm_head; |
31dbd01f3
|
791 |
spin_unlock(&ksm_mmlist_lock); |
d952b7913
|
792 |
return err; |
31dbd01f3
|
793 |
} |
2ffd8679c
|
794 |
#endif /* CONFIG_SYSFS */ |
31dbd01f3
|
795 |
|
31dbd01f3
|
796 797 798 |
static u32 calc_checksum(struct page *page) { u32 checksum; |
9b04c5fec
|
799 |
void *addr = kmap_atomic(page); |
31dbd01f3
|
800 |
checksum = jhash2(addr, PAGE_SIZE / 4, 17); |
9b04c5fec
|
801 |
kunmap_atomic(addr); |
31dbd01f3
|
802 803 804 805 806 807 808 |
return checksum; } static int memcmp_pages(struct page *page1, struct page *page2) { char *addr1, *addr2; int ret; |
9b04c5fec
|
809 810 |
addr1 = kmap_atomic(page1); addr2 = kmap_atomic(page2); |
31dbd01f3
|
811 |
ret = memcmp(addr1, addr2, PAGE_SIZE); |
9b04c5fec
|
812 813 |
kunmap_atomic(addr2); kunmap_atomic(addr1); |
31dbd01f3
|
814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 |
return ret; } static inline int pages_identical(struct page *page1, struct page *page2) { return !memcmp_pages(page1, page2); } static int write_protect_page(struct vm_area_struct *vma, struct page *page, pte_t *orig_pte) { struct mm_struct *mm = vma->vm_mm; unsigned long addr; pte_t *ptep; spinlock_t *ptl; int swapped; int err = -EFAULT; |
6bdb913f0
|
831 832 |
unsigned long mmun_start; /* For mmu_notifiers */ unsigned long mmun_end; /* For mmu_notifiers */ |
31dbd01f3
|
833 834 835 836 |
addr = page_address_in_vma(page, vma); if (addr == -EFAULT) goto out; |
29ad768cf
|
837 |
BUG_ON(PageTransCompound(page)); |
6bdb913f0
|
838 839 840 841 |
mmun_start = addr; mmun_end = addr + PAGE_SIZE; mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); |
31dbd01f3
|
842 843 |
ptep = page_check_address(page, mm, addr, &ptl, 0); if (!ptep) |
6bdb913f0
|
844 |
goto out_mn; |
31dbd01f3
|
845 |
|
4e31635c3
|
846 |
if (pte_write(*ptep) || pte_dirty(*ptep)) { |
31dbd01f3
|
847 848 849 850 851 |
pte_t entry; swapped = PageSwapCache(page); flush_cache_page(vma, addr, page_to_pfn(page)); /* |
25985edce
|
852 |
* Ok this is tricky, when get_user_pages_fast() run it doesn't |
31dbd01f3
|
853 854 855 856 857 858 859 |
* take any lock, therefore the check that we are going to make * with the pagecount against the mapcount is racey and * O_DIRECT can happen right after the check. * So we clear the pte and flush the tlb before the check * this assure us that no O_DIRECT can happen after the check * or in the middle of the check. */ |
34ee645e8
|
860 |
entry = ptep_clear_flush_notify(vma, addr, ptep); |
31dbd01f3
|
861 862 863 864 |
/* * Check that no O_DIRECT or similar I/O is in progress on the * page */ |
31e855ea7
|
865 |
if (page_mapcount(page) + 1 + swapped != page_count(page)) { |
cb5323751
|
866 |
set_pte_at(mm, addr, ptep, entry); |
31dbd01f3
|
867 868 |
goto out_unlock; } |
4e31635c3
|
869 870 871 |
if (pte_dirty(entry)) set_page_dirty(page); entry = pte_mkclean(pte_wrprotect(entry)); |
31dbd01f3
|
872 873 874 875 876 877 878 |
set_pte_at_notify(mm, addr, ptep, entry); } *orig_pte = *ptep; err = 0; out_unlock: pte_unmap_unlock(ptep, ptl); |
6bdb913f0
|
879 880 |
out_mn: mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
31dbd01f3
|
881 882 883 884 885 886 |
out: return err; } /** * replace_page - replace page in vma by new ksm page |
8dd3557a5
|
887 888 889 |
* @vma: vma that holds the pte pointing to page * @page: the page we are replacing by kpage * @kpage: the ksm page we replace page by |
31dbd01f3
|
890 891 892 893 |
* @orig_pte: the original value of the pte * * Returns 0 on success, -EFAULT on failure. */ |
8dd3557a5
|
894 895 |
static int replace_page(struct vm_area_struct *vma, struct page *page, struct page *kpage, pte_t orig_pte) |
31dbd01f3
|
896 897 |
{ struct mm_struct *mm = vma->vm_mm; |
31dbd01f3
|
898 899 900 901 |
pmd_t *pmd; pte_t *ptep; spinlock_t *ptl; unsigned long addr; |
31dbd01f3
|
902 |
int err = -EFAULT; |
6bdb913f0
|
903 904 |
unsigned long mmun_start; /* For mmu_notifiers */ unsigned long mmun_end; /* For mmu_notifiers */ |
31dbd01f3
|
905 |
|
8dd3557a5
|
906 |
addr = page_address_in_vma(page, vma); |
31dbd01f3
|
907 908 |
if (addr == -EFAULT) goto out; |
6219049ae
|
909 910 |
pmd = mm_find_pmd(mm, addr); if (!pmd) |
31dbd01f3
|
911 |
goto out; |
31dbd01f3
|
912 |
|
6bdb913f0
|
913 914 915 |
mmun_start = addr; mmun_end = addr + PAGE_SIZE; mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); |
31dbd01f3
|
916 917 918 |
ptep = pte_offset_map_lock(mm, pmd, addr, &ptl); if (!pte_same(*ptep, orig_pte)) { pte_unmap_unlock(ptep, ptl); |
6bdb913f0
|
919 |
goto out_mn; |
31dbd01f3
|
920 |
} |
8dd3557a5
|
921 |
get_page(kpage); |
d281ee614
|
922 |
page_add_anon_rmap(kpage, vma, addr, false); |
31dbd01f3
|
923 924 |
flush_cache_page(vma, addr, pte_pfn(*ptep)); |
34ee645e8
|
925 |
ptep_clear_flush_notify(vma, addr, ptep); |
8dd3557a5
|
926 |
set_pte_at_notify(mm, addr, ptep, mk_pte(kpage, vma->vm_page_prot)); |
31dbd01f3
|
927 |
|
d281ee614
|
928 |
page_remove_rmap(page, false); |
ae52a2adb
|
929 930 |
if (!page_mapped(page)) try_to_free_swap(page); |
8dd3557a5
|
931 |
put_page(page); |
31dbd01f3
|
932 933 934 |
pte_unmap_unlock(ptep, ptl); err = 0; |
6bdb913f0
|
935 936 |
out_mn: mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
31dbd01f3
|
937 938 939 940 941 942 |
out: return err; } /* * try_to_merge_one_page - take two pages and merge them into one |
8dd3557a5
|
943 944 |
* @vma: the vma that holds the pte pointing to page * @page: the PageAnon page that we want to replace with kpage |
80e148226
|
945 946 |
* @kpage: the PageKsm page that we want to map instead of page, * or NULL the first time when we want to use page as kpage. |
31dbd01f3
|
947 948 949 950 |
* * This function returns 0 if the pages were merged, -EFAULT otherwise. */ static int try_to_merge_one_page(struct vm_area_struct *vma, |
8dd3557a5
|
951 |
struct page *page, struct page *kpage) |
31dbd01f3
|
952 953 954 |
{ pte_t orig_pte = __pte(0); int err = -EFAULT; |
db114b83a
|
955 956 |
if (page == kpage) /* ksm page forked */ return 0; |
8dd3557a5
|
957 |
if (!PageAnon(page)) |
31dbd01f3
|
958 |
goto out; |
31dbd01f3
|
959 960 961 962 963 964 965 |
/* * We need the page lock to read a stable PageSwapCache in * write_protect_page(). We use trylock_page() instead of * lock_page() because we don't want to wait here - we * prefer to continue scanning and merging different pages, * then come back to this page when it is unlocked. */ |
8dd3557a5
|
966 |
if (!trylock_page(page)) |
31e855ea7
|
967 |
goto out; |
f765f5405
|
968 969 970 971 972 973 |
if (PageTransCompound(page)) { err = split_huge_page(page); if (err) goto out_unlock; } |
31dbd01f3
|
974 975 976 977 978 979 |
/* * If this anonymous page is mapped only here, its pte may need * to be write-protected. If it's mapped elsewhere, all of its * ptes are necessarily already write-protected. But in either * case, we need to lock and check page_count is not raised. */ |
80e148226
|
980 981 982 983 984 985 986 987 988 |
if (write_protect_page(vma, page, &orig_pte) == 0) { if (!kpage) { /* * While we hold page lock, upgrade page from * PageAnon+anon_vma to PageKsm+NULL stable_node: * stable_tree_insert() will update stable_node. */ set_page_stable_node(page, NULL); mark_page_accessed(page); |
337ed7eb5
|
989 990 991 992 993 994 |
/* * Page reclaim just frees a clean page with no dirty * ptes: make sure that the ksm page would be swapped. */ if (!PageDirty(page)) SetPageDirty(page); |
80e148226
|
995 996 997 998 |
err = 0; } else if (pages_identical(page, kpage)) err = replace_page(vma, page, kpage, orig_pte); } |
31dbd01f3
|
999 |
|
80e148226
|
1000 |
if ((vma->vm_flags & VM_LOCKED) && kpage && !err) { |
73848b468
|
1001 |
munlock_vma_page(page); |
5ad646880
|
1002 1003 |
if (!PageMlocked(kpage)) { unlock_page(page); |
5ad646880
|
1004 1005 1006 1007 1008 |
lock_page(kpage); mlock_vma_page(kpage); page = kpage; /* for final unlock */ } } |
73848b468
|
1009 |
|
f765f5405
|
1010 |
out_unlock: |
8dd3557a5
|
1011 |
unlock_page(page); |
31dbd01f3
|
1012 1013 1014 1015 1016 |
out: return err; } /* |
81464e306
|
1017 1018 |
* try_to_merge_with_ksm_page - like try_to_merge_two_pages, * but no new kernel page is allocated: kpage must already be a ksm page. |
8dd3557a5
|
1019 1020 |
* * This function returns 0 if the pages were merged, -EFAULT otherwise. |
81464e306
|
1021 |
*/ |
8dd3557a5
|
1022 1023 |
static int try_to_merge_with_ksm_page(struct rmap_item *rmap_item, struct page *page, struct page *kpage) |
81464e306
|
1024 |
{ |
8dd3557a5
|
1025 |
struct mm_struct *mm = rmap_item->mm; |
81464e306
|
1026 1027 |
struct vm_area_struct *vma; int err = -EFAULT; |
8dd3557a5
|
1028 |
down_read(&mm->mmap_sem); |
85c6e8dd2
|
1029 1030 |
vma = find_mergeable_vma(mm, rmap_item->address); if (!vma) |
81464e306
|
1031 |
goto out; |
8dd3557a5
|
1032 |
err = try_to_merge_one_page(vma, page, kpage); |
db114b83a
|
1033 1034 |
if (err) goto out; |
bc56620b4
|
1035 1036 |
/* Unstable nid is in union with stable anon_vma: remove first */ remove_rmap_item_from_tree(rmap_item); |
db114b83a
|
1037 |
/* Must get reference to anon_vma while still holding mmap_sem */ |
9e60109f1
|
1038 1039 |
rmap_item->anon_vma = vma->anon_vma; get_anon_vma(vma->anon_vma); |
81464e306
|
1040 |
out: |
8dd3557a5
|
1041 |
up_read(&mm->mmap_sem); |
81464e306
|
1042 1043 1044 1045 |
return err; } /* |
31dbd01f3
|
1046 1047 1048 |
* try_to_merge_two_pages - take two identical pages and prepare them * to be merged into one page. * |
8dd3557a5
|
1049 1050 |
* This function returns the kpage if we successfully merged two identical * pages into one ksm page, NULL otherwise. |
31dbd01f3
|
1051 |
* |
80e148226
|
1052 |
* Note that this function upgrades page to ksm page: if one of the pages |
31dbd01f3
|
1053 1054 |
* is already a ksm page, try_to_merge_with_ksm_page should be used. */ |
8dd3557a5
|
1055 1056 1057 1058 |
static struct page *try_to_merge_two_pages(struct rmap_item *rmap_item, struct page *page, struct rmap_item *tree_rmap_item, struct page *tree_page) |
31dbd01f3
|
1059 |
{ |
80e148226
|
1060 |
int err; |
31dbd01f3
|
1061 |
|
80e148226
|
1062 |
err = try_to_merge_with_ksm_page(rmap_item, page, NULL); |
31dbd01f3
|
1063 |
if (!err) { |
8dd3557a5
|
1064 |
err = try_to_merge_with_ksm_page(tree_rmap_item, |
80e148226
|
1065 |
tree_page, page); |
31dbd01f3
|
1066 |
/* |
81464e306
|
1067 1068 |
* If that fails, we have a ksm page with only one pte * pointing to it: so break it. |
31dbd01f3
|
1069 |
*/ |
4035c07a8
|
1070 |
if (err) |
8dd3557a5
|
1071 |
break_cow(rmap_item); |
31dbd01f3
|
1072 |
} |
80e148226
|
1073 |
return err ? NULL : page; |
31dbd01f3
|
1074 1075 1076 |
} /* |
8dd3557a5
|
1077 |
* stable_tree_search - search for page inside the stable tree |
31dbd01f3
|
1078 1079 1080 1081 |
* * This function checks if there is a page inside the stable tree * with identical content to the page that we are scanning right now. * |
7b6ba2c7d
|
1082 |
* This function returns the stable tree node of identical content if found, |
31dbd01f3
|
1083 1084 |
* NULL otherwise. */ |
62b61f611
|
1085 |
static struct page *stable_tree_search(struct page *page) |
31dbd01f3
|
1086 |
{ |
90bd6fd31
|
1087 |
int nid; |
ef53d16cd
|
1088 |
struct rb_root *root; |
4146d2d67
|
1089 1090 1091 1092 |
struct rb_node **new; struct rb_node *parent; struct stable_node *stable_node; struct stable_node *page_node; |
31dbd01f3
|
1093 |
|
4146d2d67
|
1094 1095 1096 |
page_node = page_stable_node(page); if (page_node && page_node->head != &migrate_nodes) { /* ksm page forked */ |
08beca44d
|
1097 |
get_page(page); |
62b61f611
|
1098 |
return page; |
08beca44d
|
1099 |
} |
90bd6fd31
|
1100 |
nid = get_kpfn_nid(page_to_pfn(page)); |
ef53d16cd
|
1101 |
root = root_stable_tree + nid; |
4146d2d67
|
1102 |
again: |
ef53d16cd
|
1103 |
new = &root->rb_node; |
4146d2d67
|
1104 |
parent = NULL; |
90bd6fd31
|
1105 |
|
4146d2d67
|
1106 |
while (*new) { |
4035c07a8
|
1107 |
struct page *tree_page; |
31dbd01f3
|
1108 |
int ret; |
08beca44d
|
1109 |
cond_resched(); |
4146d2d67
|
1110 |
stable_node = rb_entry(*new, struct stable_node, node); |
8aafa6a48
|
1111 |
tree_page = get_ksm_page(stable_node, false); |
f2e5ff85e
|
1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 |
if (!tree_page) { /* * If we walked over a stale stable_node, * get_ksm_page() will call rb_erase() and it * may rebalance the tree from under us. So * restart the search from scratch. Returning * NULL would be safe too, but we'd generate * false negative insertions just because some * stable_node was stale. */ goto again; } |
31dbd01f3
|
1124 |
|
4035c07a8
|
1125 |
ret = memcmp_pages(page, tree_page); |
c8d6553b9
|
1126 |
put_page(tree_page); |
31dbd01f3
|
1127 |
|
4146d2d67
|
1128 |
parent = *new; |
c8d6553b9
|
1129 |
if (ret < 0) |
4146d2d67
|
1130 |
new = &parent->rb_left; |
c8d6553b9
|
1131 |
else if (ret > 0) |
4146d2d67
|
1132 |
new = &parent->rb_right; |
c8d6553b9
|
1133 1134 1135 1136 1137 1138 1139 1140 1141 |
else { /* * Lock and unlock the stable_node's page (which * might already have been migrated) so that page * migration is sure to notice its raised count. * It would be more elegant to return stable_node * than kpage, but that involves more changes. */ tree_page = get_ksm_page(stable_node, true); |
4146d2d67
|
1142 |
if (tree_page) { |
c8d6553b9
|
1143 |
unlock_page(tree_page); |
4146d2d67
|
1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 |
if (get_kpfn_nid(stable_node->kpfn) != NUMA(stable_node->nid)) { put_page(tree_page); goto replace; } return tree_page; } /* * There is now a place for page_node, but the tree may * have been rebalanced, so re-evaluate parent and new. */ if (page_node) goto again; return NULL; |
c8d6553b9
|
1158 |
} |
31dbd01f3
|
1159 |
} |
4146d2d67
|
1160 1161 1162 1163 1164 1165 |
if (!page_node) return NULL; list_del(&page_node->list); DO_NUMA(page_node->nid = nid); rb_link_node(&page_node->node, parent, new); |
ef53d16cd
|
1166 |
rb_insert_color(&page_node->node, root); |
4146d2d67
|
1167 1168 1169 1170 1171 1172 1173 |
get_page(page); return page; replace: if (page_node) { list_del(&page_node->list); DO_NUMA(page_node->nid = nid); |
ef53d16cd
|
1174 |
rb_replace_node(&stable_node->node, &page_node->node, root); |
4146d2d67
|
1175 1176 |
get_page(page); } else { |
ef53d16cd
|
1177 |
rb_erase(&stable_node->node, root); |
4146d2d67
|
1178 1179 1180 1181 1182 |
page = NULL; } stable_node->head = &migrate_nodes; list_add(&stable_node->list, stable_node->head); return page; |
31dbd01f3
|
1183 1184 1185 |
} /* |
e850dcf53
|
1186 |
* stable_tree_insert - insert stable tree node pointing to new ksm page |
31dbd01f3
|
1187 1188 |
* into the stable tree. * |
7b6ba2c7d
|
1189 1190 |
* This function returns the stable tree node just allocated on success, * NULL otherwise. |
31dbd01f3
|
1191 |
*/ |
7b6ba2c7d
|
1192 |
static struct stable_node *stable_tree_insert(struct page *kpage) |
31dbd01f3
|
1193 |
{ |
90bd6fd31
|
1194 1195 |
int nid; unsigned long kpfn; |
ef53d16cd
|
1196 |
struct rb_root *root; |
90bd6fd31
|
1197 |
struct rb_node **new; |
f2e5ff85e
|
1198 |
struct rb_node *parent; |
7b6ba2c7d
|
1199 |
struct stable_node *stable_node; |
31dbd01f3
|
1200 |
|
90bd6fd31
|
1201 1202 |
kpfn = page_to_pfn(kpage); nid = get_kpfn_nid(kpfn); |
ef53d16cd
|
1203 |
root = root_stable_tree + nid; |
f2e5ff85e
|
1204 1205 |
again: parent = NULL; |
ef53d16cd
|
1206 |
new = &root->rb_node; |
90bd6fd31
|
1207 |
|
31dbd01f3
|
1208 |
while (*new) { |
4035c07a8
|
1209 |
struct page *tree_page; |
31dbd01f3
|
1210 |
int ret; |
08beca44d
|
1211 |
cond_resched(); |
7b6ba2c7d
|
1212 |
stable_node = rb_entry(*new, struct stable_node, node); |
8aafa6a48
|
1213 |
tree_page = get_ksm_page(stable_node, false); |
f2e5ff85e
|
1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 |
if (!tree_page) { /* * If we walked over a stale stable_node, * get_ksm_page() will call rb_erase() and it * may rebalance the tree from under us. So * restart the search from scratch. Returning * NULL would be safe too, but we'd generate * false negative insertions just because some * stable_node was stale. */ goto again; } |
31dbd01f3
|
1226 |
|
4035c07a8
|
1227 1228 |
ret = memcmp_pages(kpage, tree_page); put_page(tree_page); |
31dbd01f3
|
1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 |
parent = *new; if (ret < 0) new = &parent->rb_left; else if (ret > 0) new = &parent->rb_right; else { /* * It is not a bug that stable_tree_search() didn't * find this node: because at that time our page was * not yet write-protected, so may have changed since. */ return NULL; } } |
7b6ba2c7d
|
1244 1245 1246 |
stable_node = alloc_stable_node(); if (!stable_node) return NULL; |
31dbd01f3
|
1247 |
|
7b6ba2c7d
|
1248 |
INIT_HLIST_HEAD(&stable_node->hlist); |
90bd6fd31
|
1249 |
stable_node->kpfn = kpfn; |
08beca44d
|
1250 |
set_page_stable_node(kpage, stable_node); |
4146d2d67
|
1251 |
DO_NUMA(stable_node->nid = nid); |
e850dcf53
|
1252 |
rb_link_node(&stable_node->node, parent, new); |
ef53d16cd
|
1253 |
rb_insert_color(&stable_node->node, root); |
08beca44d
|
1254 |
|
7b6ba2c7d
|
1255 |
return stable_node; |
31dbd01f3
|
1256 1257 1258 |
} /* |
8dd3557a5
|
1259 1260 |
* unstable_tree_search_insert - search for identical page, * else insert rmap_item into the unstable tree. |
31dbd01f3
|
1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 |
* * This function searches for a page in the unstable tree identical to the * page currently being scanned; and if no identical page is found in the * tree, we insert rmap_item as a new object into the unstable tree. * * This function returns pointer to rmap_item found to be identical * to the currently scanned page, NULL otherwise. * * This function does both searching and inserting, because they share * the same walking algorithm in an rbtree. */ |
8dd3557a5
|
1272 1273 1274 1275 |
static struct rmap_item *unstable_tree_search_insert(struct rmap_item *rmap_item, struct page *page, struct page **tree_pagep) |
31dbd01f3
|
1276 |
{ |
90bd6fd31
|
1277 1278 |
struct rb_node **new; struct rb_root *root; |
31dbd01f3
|
1279 |
struct rb_node *parent = NULL; |
90bd6fd31
|
1280 1281 1282 |
int nid; nid = get_kpfn_nid(page_to_pfn(page)); |
ef53d16cd
|
1283 |
root = root_unstable_tree + nid; |
90bd6fd31
|
1284 |
new = &root->rb_node; |
31dbd01f3
|
1285 1286 1287 |
while (*new) { struct rmap_item *tree_rmap_item; |
8dd3557a5
|
1288 |
struct page *tree_page; |
31dbd01f3
|
1289 |
int ret; |
d178f27fc
|
1290 |
cond_resched(); |
31dbd01f3
|
1291 |
tree_rmap_item = rb_entry(*new, struct rmap_item, node); |
8dd3557a5
|
1292 |
tree_page = get_mergeable_page(tree_rmap_item); |
c8f95ed1a
|
1293 |
if (!tree_page) |
31dbd01f3
|
1294 1295 1296 |
return NULL; /* |
8dd3557a5
|
1297 |
* Don't substitute a ksm page for a forked page. |
31dbd01f3
|
1298 |
*/ |
8dd3557a5
|
1299 1300 |
if (page == tree_page) { put_page(tree_page); |
31dbd01f3
|
1301 1302 |
return NULL; } |
8dd3557a5
|
1303 |
ret = memcmp_pages(page, tree_page); |
31dbd01f3
|
1304 1305 1306 |
parent = *new; if (ret < 0) { |
8dd3557a5
|
1307 |
put_page(tree_page); |
31dbd01f3
|
1308 1309 |
new = &parent->rb_left; } else if (ret > 0) { |
8dd3557a5
|
1310 |
put_page(tree_page); |
31dbd01f3
|
1311 |
new = &parent->rb_right; |
b599cbdf1
|
1312 1313 1314 1315 1316 1317 1318 1319 1320 |
} else if (!ksm_merge_across_nodes && page_to_nid(tree_page) != nid) { /* * If tree_page has been migrated to another NUMA node, * it will be flushed out and put in the right unstable * tree next time: only merge with it when across_nodes. */ put_page(tree_page); return NULL; |
31dbd01f3
|
1321 |
} else { |
8dd3557a5
|
1322 |
*tree_pagep = tree_page; |
31dbd01f3
|
1323 1324 1325 |
return tree_rmap_item; } } |
7b6ba2c7d
|
1326 |
rmap_item->address |= UNSTABLE_FLAG; |
31dbd01f3
|
1327 |
rmap_item->address |= (ksm_scan.seqnr & SEQNR_MASK); |
e850dcf53
|
1328 |
DO_NUMA(rmap_item->nid = nid); |
31dbd01f3
|
1329 |
rb_link_node(&rmap_item->node, parent, new); |
90bd6fd31
|
1330 |
rb_insert_color(&rmap_item->node, root); |
31dbd01f3
|
1331 |
|
473b0ce4d
|
1332 |
ksm_pages_unshared++; |
31dbd01f3
|
1333 1334 1335 1336 1337 1338 1339 1340 1341 |
return NULL; } /* * stable_tree_append - add another rmap_item to the linked list of * rmap_items hanging off a given node of the stable tree, all sharing * the same ksm page. */ static void stable_tree_append(struct rmap_item *rmap_item, |
7b6ba2c7d
|
1342 |
struct stable_node *stable_node) |
31dbd01f3
|
1343 |
{ |
7b6ba2c7d
|
1344 |
rmap_item->head = stable_node; |
31dbd01f3
|
1345 |
rmap_item->address |= STABLE_FLAG; |
7b6ba2c7d
|
1346 |
hlist_add_head(&rmap_item->hlist, &stable_node->hlist); |
e178dfde3
|
1347 |
|
7b6ba2c7d
|
1348 1349 1350 1351 |
if (rmap_item->hlist.next) ksm_pages_sharing++; else ksm_pages_shared++; |
31dbd01f3
|
1352 1353 1354 |
} /* |
81464e306
|
1355 1356 1357 1358 |
* cmp_and_merge_page - first see if page can be merged into the stable tree; * if not, compare checksum to previous and if it's the same, see if page can * be inserted into the unstable tree, or merged with a page already there and * both transferred to the stable tree. |
31dbd01f3
|
1359 1360 1361 1362 1363 1364 |
* * @page: the page that we are searching identical page to. * @rmap_item: the reverse mapping into the virtual address of this page */ static void cmp_and_merge_page(struct page *page, struct rmap_item *rmap_item) { |
31dbd01f3
|
1365 |
struct rmap_item *tree_rmap_item; |
8dd3557a5
|
1366 |
struct page *tree_page = NULL; |
7b6ba2c7d
|
1367 |
struct stable_node *stable_node; |
8dd3557a5
|
1368 |
struct page *kpage; |
31dbd01f3
|
1369 1370 |
unsigned int checksum; int err; |
4146d2d67
|
1371 1372 1373 1374 1375 |
stable_node = page_stable_node(page); if (stable_node) { if (stable_node->head != &migrate_nodes && get_kpfn_nid(stable_node->kpfn) != NUMA(stable_node->nid)) { rb_erase(&stable_node->node, |
ef53d16cd
|
1376 |
root_stable_tree + NUMA(stable_node->nid)); |
4146d2d67
|
1377 1378 1379 1380 1381 1382 1383 |
stable_node->head = &migrate_nodes; list_add(&stable_node->list, stable_node->head); } if (stable_node->head != &migrate_nodes && rmap_item->head == stable_node) return; } |
31dbd01f3
|
1384 1385 |
/* We first start with searching the page inside the stable tree */ |
62b61f611
|
1386 |
kpage = stable_tree_search(page); |
4146d2d67
|
1387 1388 1389 1390 1391 1392 |
if (kpage == page && rmap_item->head == stable_node) { put_page(kpage); return; } remove_rmap_item_from_tree(rmap_item); |
62b61f611
|
1393 |
if (kpage) { |
08beca44d
|
1394 |
err = try_to_merge_with_ksm_page(rmap_item, page, kpage); |
31dbd01f3
|
1395 1396 1397 1398 1399 |
if (!err) { /* * The page was successfully merged: * add its rmap_item to the stable tree. */ |
5ad646880
|
1400 |
lock_page(kpage); |
62b61f611
|
1401 |
stable_tree_append(rmap_item, page_stable_node(kpage)); |
5ad646880
|
1402 |
unlock_page(kpage); |
31dbd01f3
|
1403 |
} |
8dd3557a5
|
1404 |
put_page(kpage); |
31dbd01f3
|
1405 1406 1407 1408 |
return; } /* |
4035c07a8
|
1409 1410 1411 1412 |
* If the hash value of the page has changed from the last time * we calculated it, this page is changing frequently: therefore we * don't want to insert it in the unstable tree, and we don't want * to waste our time searching for something identical to it there. |
31dbd01f3
|
1413 1414 1415 1416 1417 1418 |
*/ checksum = calc_checksum(page); if (rmap_item->oldchecksum != checksum) { rmap_item->oldchecksum = checksum; return; } |
8dd3557a5
|
1419 1420 |
tree_rmap_item = unstable_tree_search_insert(rmap_item, page, &tree_page); |
31dbd01f3
|
1421 |
if (tree_rmap_item) { |
8dd3557a5
|
1422 1423 1424 |
kpage = try_to_merge_two_pages(rmap_item, page, tree_rmap_item, tree_page); put_page(tree_page); |
8dd3557a5
|
1425 |
if (kpage) { |
bc56620b4
|
1426 1427 1428 1429 |
/* * The pages were successfully merged: insert new * node in the stable tree and add both rmap_items. */ |
5ad646880
|
1430 |
lock_page(kpage); |
7b6ba2c7d
|
1431 1432 1433 1434 1435 |
stable_node = stable_tree_insert(kpage); if (stable_node) { stable_tree_append(tree_rmap_item, stable_node); stable_tree_append(rmap_item, stable_node); } |
5ad646880
|
1436 |
unlock_page(kpage); |
7b6ba2c7d
|
1437 |
|
31dbd01f3
|
1438 1439 1440 1441 1442 1443 |
/* * If we fail to insert the page into the stable tree, * we will have 2 virtual addresses that are pointing * to a ksm page left outside the stable tree, * in which case we need to break_cow on both. */ |
7b6ba2c7d
|
1444 |
if (!stable_node) { |
8dd3557a5
|
1445 1446 |
break_cow(tree_rmap_item); break_cow(rmap_item); |
31dbd01f3
|
1447 1448 |
} } |
31dbd01f3
|
1449 1450 1451 1452 |
} } static struct rmap_item *get_next_rmap_item(struct mm_slot *mm_slot, |
6514d511d
|
1453 |
struct rmap_item **rmap_list, |
31dbd01f3
|
1454 1455 1456 |
unsigned long addr) { struct rmap_item *rmap_item; |
6514d511d
|
1457 1458 |
while (*rmap_list) { rmap_item = *rmap_list; |
93d17715a
|
1459 |
if ((rmap_item->address & PAGE_MASK) == addr) |
31dbd01f3
|
1460 |
return rmap_item; |
31dbd01f3
|
1461 1462 |
if (rmap_item->address > addr) break; |
6514d511d
|
1463 |
*rmap_list = rmap_item->rmap_list; |
31dbd01f3
|
1464 |
remove_rmap_item_from_tree(rmap_item); |
31dbd01f3
|
1465 1466 1467 1468 1469 1470 1471 1472 |
free_rmap_item(rmap_item); } rmap_item = alloc_rmap_item(); if (rmap_item) { /* It has already been zeroed */ rmap_item->mm = mm_slot->mm; rmap_item->address = addr; |
6514d511d
|
1473 1474 |
rmap_item->rmap_list = *rmap_list; *rmap_list = rmap_item; |
31dbd01f3
|
1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 |
} return rmap_item; } static struct rmap_item *scan_get_next_rmap_item(struct page **page) { struct mm_struct *mm; struct mm_slot *slot; struct vm_area_struct *vma; struct rmap_item *rmap_item; |
90bd6fd31
|
1485 |
int nid; |
31dbd01f3
|
1486 1487 1488 1489 1490 1491 |
if (list_empty(&ksm_mm_head.mm_list)) return NULL; slot = ksm_scan.mm_slot; if (slot == &ksm_mm_head) { |
2919bfd07
|
1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 |
/* * A number of pages can hang around indefinitely on per-cpu * pagevecs, raised page count preventing write_protect_page * from merging them. Though it doesn't really matter much, * it is puzzling to see some stuck in pages_volatile until * other activity jostles them out, and they also prevented * LTP's KSM test from succeeding deterministically; so drain * them here (here rather than on entry to ksm_do_scan(), * so we don't IPI too often when pages_to_scan is set low). */ lru_add_drain_all(); |
4146d2d67
|
1503 1504 1505 1506 1507 1508 1509 |
/* * Whereas stale stable_nodes on the stable_tree itself * get pruned in the regular course of stable_tree_search(), * those moved out to the migrate_nodes list can accumulate: * so prune them once before each full scan. */ if (!ksm_merge_across_nodes) { |
036404183
|
1510 |
struct stable_node *stable_node, *next; |
4146d2d67
|
1511 |
struct page *page; |
036404183
|
1512 1513 |
list_for_each_entry_safe(stable_node, next, &migrate_nodes, list) { |
4146d2d67
|
1514 1515 1516 1517 1518 1519 |
page = get_ksm_page(stable_node, false); if (page) put_page(page); cond_resched(); } } |
ef53d16cd
|
1520 |
for (nid = 0; nid < ksm_nr_node_ids; nid++) |
90bd6fd31
|
1521 |
root_unstable_tree[nid] = RB_ROOT; |
31dbd01f3
|
1522 1523 1524 1525 1526 |
spin_lock(&ksm_mmlist_lock); slot = list_entry(slot->mm_list.next, struct mm_slot, mm_list); ksm_scan.mm_slot = slot; spin_unlock(&ksm_mmlist_lock); |
2b472611a
|
1527 1528 1529 1530 1531 1532 |
/* * Although we tested list_empty() above, a racing __ksm_exit * of the last mm on the list may have removed it since then. */ if (slot == &ksm_mm_head) return NULL; |
31dbd01f3
|
1533 1534 |
next_mm: ksm_scan.address = 0; |
6514d511d
|
1535 |
ksm_scan.rmap_list = &slot->rmap_list; |
31dbd01f3
|
1536 1537 1538 1539 |
} mm = slot->mm; down_read(&mm->mmap_sem); |
9ba692948
|
1540 1541 1542 1543 1544 1545 |
if (ksm_test_exit(mm)) vma = NULL; else vma = find_vma(mm, ksm_scan.address); for (; vma; vma = vma->vm_next) { |
31dbd01f3
|
1546 1547 1548 1549 1550 1551 1552 1553 |
if (!(vma->vm_flags & VM_MERGEABLE)) continue; if (ksm_scan.address < vma->vm_start) ksm_scan.address = vma->vm_start; if (!vma->anon_vma) ksm_scan.address = vma->vm_end; while (ksm_scan.address < vma->vm_end) { |
9ba692948
|
1554 1555 |
if (ksm_test_exit(mm)) break; |
31dbd01f3
|
1556 |
*page = follow_page(vma, ksm_scan.address, FOLL_GET); |
21ae5b017
|
1557 1558 1559 1560 1561 |
if (IS_ERR_OR_NULL(*page)) { ksm_scan.address += PAGE_SIZE; cond_resched(); continue; } |
f765f5405
|
1562 |
if (PageAnon(*page)) { |
31dbd01f3
|
1563 1564 1565 |
flush_anon_page(vma, *page, ksm_scan.address); flush_dcache_page(*page); rmap_item = get_next_rmap_item(slot, |
6514d511d
|
1566 |
ksm_scan.rmap_list, ksm_scan.address); |
31dbd01f3
|
1567 |
if (rmap_item) { |
6514d511d
|
1568 1569 |
ksm_scan.rmap_list = &rmap_item->rmap_list; |
31dbd01f3
|
1570 1571 1572 1573 1574 1575 |
ksm_scan.address += PAGE_SIZE; } else put_page(*page); up_read(&mm->mmap_sem); return rmap_item; } |
21ae5b017
|
1576 |
put_page(*page); |
31dbd01f3
|
1577 1578 1579 1580 |
ksm_scan.address += PAGE_SIZE; cond_resched(); } } |
9ba692948
|
1581 1582 |
if (ksm_test_exit(mm)) { ksm_scan.address = 0; |
6514d511d
|
1583 |
ksm_scan.rmap_list = &slot->rmap_list; |
9ba692948
|
1584 |
} |
31dbd01f3
|
1585 1586 1587 1588 |
/* * Nuke all the rmap_items that are above this current rmap: * because there were no VM_MERGEABLE vmas with such addresses. */ |
6514d511d
|
1589 |
remove_trailing_rmap_items(slot, ksm_scan.rmap_list); |
31dbd01f3
|
1590 1591 |
spin_lock(&ksm_mmlist_lock); |
cd551f975
|
1592 1593 1594 1595 1596 1597 1598 |
ksm_scan.mm_slot = list_entry(slot->mm_list.next, struct mm_slot, mm_list); if (ksm_scan.address == 0) { /* * We've completed a full scan of all vmas, holding mmap_sem * throughout, and found no VM_MERGEABLE: so do the same as * __ksm_exit does to remove this mm from all our lists now. |
9ba692948
|
1599 1600 1601 1602 |
* This applies either when cleaning up after __ksm_exit * (but beware: we can reach here even before __ksm_exit), * or when all VM_MERGEABLE areas have been unmapped (and * mmap_sem then protects against race with MADV_MERGEABLE). |
cd551f975
|
1603 |
*/ |
4ca3a69bc
|
1604 |
hash_del(&slot->link); |
cd551f975
|
1605 |
list_del(&slot->mm_list); |
9ba692948
|
1606 |
spin_unlock(&ksm_mmlist_lock); |
cd551f975
|
1607 1608 |
free_mm_slot(slot); clear_bit(MMF_VM_MERGEABLE, &mm->flags); |
9ba692948
|
1609 1610 1611 |
up_read(&mm->mmap_sem); mmdrop(mm); } else { |
9ba692948
|
1612 |
up_read(&mm->mmap_sem); |
7496fea9a
|
1613 1614 1615 1616 1617 1618 1619 1620 |
/* * up_read(&mm->mmap_sem) first because after * spin_unlock(&ksm_mmlist_lock) run, the "mm" may * already have been freed under us by __ksm_exit() * because the "mm_slot" is still hashed and * ksm_scan.mm_slot doesn't point to it anymore. */ spin_unlock(&ksm_mmlist_lock); |
cd551f975
|
1621 |
} |
31dbd01f3
|
1622 1623 |
/* Repeat until we've completed scanning the whole list */ |
cd551f975
|
1624 |
slot = ksm_scan.mm_slot; |
31dbd01f3
|
1625 1626 |
if (slot != &ksm_mm_head) goto next_mm; |
31dbd01f3
|
1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 |
ksm_scan.seqnr++; return NULL; } /** * ksm_do_scan - the ksm scanner main worker function. * @scan_npages - number of pages we want to scan before we return. */ static void ksm_do_scan(unsigned int scan_npages) { struct rmap_item *rmap_item; |
22eccdd7d
|
1638 |
struct page *uninitialized_var(page); |
31dbd01f3
|
1639 |
|
878aee7d6
|
1640 |
while (scan_npages-- && likely(!freezing(current))) { |
31dbd01f3
|
1641 1642 1643 1644 |
cond_resched(); rmap_item = scan_get_next_rmap_item(&page); if (!rmap_item) return; |
4146d2d67
|
1645 |
cmp_and_merge_page(page, rmap_item); |
31dbd01f3
|
1646 1647 1648 |
put_page(page); } } |
6e1583842
|
1649 1650 1651 1652 |
static int ksmd_should_run(void) { return (ksm_run & KSM_RUN_MERGE) && !list_empty(&ksm_mm_head.mm_list); } |
31dbd01f3
|
1653 1654 |
static int ksm_scan_thread(void *nothing) { |
878aee7d6
|
1655 |
set_freezable(); |
339aa6246
|
1656 |
set_user_nice(current, 5); |
31dbd01f3
|
1657 1658 |
while (!kthread_should_stop()) { |
6e1583842
|
1659 |
mutex_lock(&ksm_thread_mutex); |
ef4d43a80
|
1660 |
wait_while_offlining(); |
6e1583842
|
1661 |
if (ksmd_should_run()) |
31dbd01f3
|
1662 |
ksm_do_scan(ksm_thread_pages_to_scan); |
6e1583842
|
1663 |
mutex_unlock(&ksm_thread_mutex); |
878aee7d6
|
1664 |
try_to_freeze(); |
6e1583842
|
1665 |
if (ksmd_should_run()) { |
31dbd01f3
|
1666 1667 1668 |
schedule_timeout_interruptible( msecs_to_jiffies(ksm_thread_sleep_millisecs)); } else { |
878aee7d6
|
1669 |
wait_event_freezable(ksm_thread_wait, |
6e1583842
|
1670 |
ksmd_should_run() || kthread_should_stop()); |
31dbd01f3
|
1671 1672 1673 1674 |
} } return 0; } |
f8af4da3b
|
1675 1676 1677 1678 |
int ksm_madvise(struct vm_area_struct *vma, unsigned long start, unsigned long end, int advice, unsigned long *vm_flags) { struct mm_struct *mm = vma->vm_mm; |
d952b7913
|
1679 |
int err; |
f8af4da3b
|
1680 1681 1682 1683 1684 1685 1686 1687 |
switch (advice) { case MADV_MERGEABLE: /* * Be somewhat over-protective for now! */ if (*vm_flags & (VM_MERGEABLE | VM_SHARED | VM_MAYSHARE | VM_PFNMAP | VM_IO | VM_DONTEXPAND | |
0661a3361
|
1688 |
VM_HUGETLB | VM_MIXEDMAP)) |
f8af4da3b
|
1689 |
return 0; /* just ignore the advice */ |
cc2383ec0
|
1690 1691 1692 1693 |
#ifdef VM_SAO if (*vm_flags & VM_SAO) return 0; #endif |
d952b7913
|
1694 1695 1696 1697 1698 |
if (!test_bit(MMF_VM_MERGEABLE, &mm->flags)) { err = __ksm_enter(mm); if (err) return err; } |
f8af4da3b
|
1699 1700 1701 1702 1703 1704 1705 |
*vm_flags |= VM_MERGEABLE; break; case MADV_UNMERGEABLE: if (!(*vm_flags & VM_MERGEABLE)) return 0; /* just ignore the advice */ |
d952b7913
|
1706 1707 1708 1709 1710 |
if (vma->anon_vma) { err = unmerge_ksm_pages(vma, start, end); if (err) return err; } |
f8af4da3b
|
1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 |
*vm_flags &= ~VM_MERGEABLE; break; } return 0; } int __ksm_enter(struct mm_struct *mm) { |
6e1583842
|
1721 1722 1723 1724 |
struct mm_slot *mm_slot; int needs_wakeup; mm_slot = alloc_mm_slot(); |
31dbd01f3
|
1725 1726 |
if (!mm_slot) return -ENOMEM; |
6e1583842
|
1727 1728 |
/* Check ksm_run too? Would need tighter locking */ needs_wakeup = list_empty(&ksm_mm_head.mm_list); |
31dbd01f3
|
1729 1730 1731 |
spin_lock(&ksm_mmlist_lock); insert_to_mm_slots_hash(mm, mm_slot); /* |
cbf86cfe0
|
1732 1733 |
* When KSM_RUN_MERGE (or KSM_RUN_STOP), * insert just behind the scanning cursor, to let the area settle |
31dbd01f3
|
1734 1735 |
* down a little; when fork is followed by immediate exec, we don't * want ksmd to waste time setting up and tearing down an rmap_list. |
cbf86cfe0
|
1736 1737 1738 1739 |
* * But when KSM_RUN_UNMERGE, it's important to insert ahead of its * scanning cursor, otherwise KSM pages in newly forked mms will be * missed: then we might as well insert at the end of the list. |
31dbd01f3
|
1740 |
*/ |
cbf86cfe0
|
1741 1742 1743 1744 |
if (ksm_run & KSM_RUN_UNMERGE) list_add_tail(&mm_slot->mm_list, &ksm_mm_head.mm_list); else list_add_tail(&mm_slot->mm_list, &ksm_scan.mm_slot->mm_list); |
31dbd01f3
|
1745 |
spin_unlock(&ksm_mmlist_lock); |
f8af4da3b
|
1746 |
set_bit(MMF_VM_MERGEABLE, &mm->flags); |
9ba692948
|
1747 |
atomic_inc(&mm->mm_count); |
6e1583842
|
1748 1749 1750 |
if (needs_wakeup) wake_up_interruptible(&ksm_thread_wait); |
f8af4da3b
|
1751 1752 |
return 0; } |
1c2fb7a4c
|
1753 |
void __ksm_exit(struct mm_struct *mm) |
f8af4da3b
|
1754 |
{ |
cd551f975
|
1755 |
struct mm_slot *mm_slot; |
9ba692948
|
1756 |
int easy_to_free = 0; |
cd551f975
|
1757 |
|
31dbd01f3
|
1758 |
/* |
9ba692948
|
1759 1760 1761 1762 1763 1764 |
* This process is exiting: if it's straightforward (as is the * case when ksmd was never running), free mm_slot immediately. * But if it's at the cursor or has rmap_items linked to it, use * mmap_sem to synchronize with any break_cows before pagetables * are freed, and leave the mm_slot on the list for ksmd to free. * Beware: ksm may already have noticed it exiting and freed the slot. |
31dbd01f3
|
1765 |
*/ |
9ba692948
|
1766 |
|
cd551f975
|
1767 1768 |
spin_lock(&ksm_mmlist_lock); mm_slot = get_mm_slot(mm); |
9ba692948
|
1769 |
if (mm_slot && ksm_scan.mm_slot != mm_slot) { |
6514d511d
|
1770 |
if (!mm_slot->rmap_list) { |
4ca3a69bc
|
1771 |
hash_del(&mm_slot->link); |
9ba692948
|
1772 1773 1774 1775 1776 1777 |
list_del(&mm_slot->mm_list); easy_to_free = 1; } else { list_move(&mm_slot->mm_list, &ksm_scan.mm_slot->mm_list); } |
cd551f975
|
1778 |
} |
cd551f975
|
1779 |
spin_unlock(&ksm_mmlist_lock); |
9ba692948
|
1780 1781 1782 1783 1784 |
if (easy_to_free) { free_mm_slot(mm_slot); clear_bit(MMF_VM_MERGEABLE, &mm->flags); mmdrop(mm); } else if (mm_slot) { |
9ba692948
|
1785 1786 |
down_write(&mm->mmap_sem); up_write(&mm->mmap_sem); |
9ba692948
|
1787 |
} |
31dbd01f3
|
1788 |
} |
cbf86cfe0
|
1789 |
struct page *ksm_might_need_to_copy(struct page *page, |
5ad646880
|
1790 1791 |
struct vm_area_struct *vma, unsigned long address) { |
cbf86cfe0
|
1792 |
struct anon_vma *anon_vma = page_anon_vma(page); |
5ad646880
|
1793 |
struct page *new_page; |
cbf86cfe0
|
1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 |
if (PageKsm(page)) { if (page_stable_node(page) && !(ksm_run & KSM_RUN_UNMERGE)) return page; /* no need to copy it */ } else if (!anon_vma) { return page; /* no need to copy it */ } else if (anon_vma->root == vma->anon_vma->root && page->index == linear_page_index(vma, address)) { return page; /* still no need to copy it */ } if (!PageUptodate(page)) return page; /* let do_swap_page report the error */ |
5ad646880
|
1806 1807 1808 1809 1810 1811 |
new_page = alloc_page_vma(GFP_HIGHUSER_MOVABLE, vma, address); if (new_page) { copy_user_highpage(new_page, page, address, vma); SetPageDirty(new_page); __SetPageUptodate(new_page); |
48c935ad8
|
1812 |
__SetPageLocked(new_page); |
5ad646880
|
1813 |
} |
5ad646880
|
1814 1815 |
return new_page; } |
051ac83ad
|
1816 |
int rmap_walk_ksm(struct page *page, struct rmap_walk_control *rwc) |
e9995ef97
|
1817 1818 |
{ struct stable_node *stable_node; |
e9995ef97
|
1819 1820 1821 |
struct rmap_item *rmap_item; int ret = SWAP_AGAIN; int search_new_forks = 0; |
309381fea
|
1822 |
VM_BUG_ON_PAGE(!PageKsm(page), page); |
9f32624be
|
1823 1824 1825 1826 1827 |
/* * Rely on the page lock to protect against concurrent modifications * to that page's node of the stable tree. */ |
309381fea
|
1828 |
VM_BUG_ON_PAGE(!PageLocked(page), page); |
e9995ef97
|
1829 1830 1831 1832 1833 |
stable_node = page_stable_node(page); if (!stable_node) return ret; again: |
b67bfe0d4
|
1834 |
hlist_for_each_entry(rmap_item, &stable_node->hlist, hlist) { |
e9995ef97
|
1835 |
struct anon_vma *anon_vma = rmap_item->anon_vma; |
5beb49305
|
1836 |
struct anon_vma_chain *vmac; |
e9995ef97
|
1837 |
struct vm_area_struct *vma; |
ad12695f1
|
1838 |
cond_resched(); |
b6b19f25f
|
1839 |
anon_vma_lock_read(anon_vma); |
bf181b9f9
|
1840 1841 |
anon_vma_interval_tree_foreach(vmac, &anon_vma->rb_root, 0, ULONG_MAX) { |
ad12695f1
|
1842 |
cond_resched(); |
5beb49305
|
1843 |
vma = vmac->vma; |
e9995ef97
|
1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 |
if (rmap_item->address < vma->vm_start || rmap_item->address >= vma->vm_end) continue; /* * Initially we examine only the vma which covers this * rmap_item; but later, if there is still work to do, * we examine covering vmas in other mms: in case they * were forked from the original since ksmd passed. */ if ((rmap_item->mm == vma->vm_mm) == search_new_forks) continue; |
0dd1c7bbc
|
1855 1856 |
if (rwc->invalid_vma && rwc->invalid_vma(vma, rwc->arg)) continue; |
051ac83ad
|
1857 1858 |
ret = rwc->rmap_one(page, vma, rmap_item->address, rwc->arg); |
e9995ef97
|
1859 |
if (ret != SWAP_AGAIN) { |
b6b19f25f
|
1860 |
anon_vma_unlock_read(anon_vma); |
e9995ef97
|
1861 1862 |
goto out; } |
0dd1c7bbc
|
1863 1864 1865 1866 |
if (rwc->done && rwc->done(page)) { anon_vma_unlock_read(anon_vma); goto out; } |
e9995ef97
|
1867 |
} |
b6b19f25f
|
1868 |
anon_vma_unlock_read(anon_vma); |
e9995ef97
|
1869 1870 1871 1872 1873 1874 |
} if (!search_new_forks++) goto again; out: return ret; } |
526295064
|
1875 |
#ifdef CONFIG_MIGRATION |
e9995ef97
|
1876 1877 1878 |
void ksm_migrate_page(struct page *newpage, struct page *oldpage) { struct stable_node *stable_node; |
309381fea
|
1879 1880 1881 |
VM_BUG_ON_PAGE(!PageLocked(oldpage), oldpage); VM_BUG_ON_PAGE(!PageLocked(newpage), newpage); VM_BUG_ON_PAGE(newpage->mapping != oldpage->mapping, newpage); |
e9995ef97
|
1882 1883 1884 |
stable_node = page_stable_node(newpage); if (stable_node) { |
309381fea
|
1885 |
VM_BUG_ON_PAGE(stable_node->kpfn != page_to_pfn(oldpage), oldpage); |
62b61f611
|
1886 |
stable_node->kpfn = page_to_pfn(newpage); |
c8d6553b9
|
1887 1888 1889 1890 1891 1892 1893 1894 |
/* * newpage->mapping was set in advance; now we need smp_wmb() * to make sure that the new stable_node->kpfn is visible * to get_ksm_page() before it can see that oldpage->mapping * has gone stale (or that PageSwapCache has been cleared). */ smp_wmb(); set_page_stable_node(oldpage, NULL); |
e9995ef97
|
1895 1896 1897 |
} } #endif /* CONFIG_MIGRATION */ |
62b61f611
|
1898 |
#ifdef CONFIG_MEMORY_HOTREMOVE |
ef4d43a80
|
1899 1900 1901 1902 1903 |
static void wait_while_offlining(void) { while (ksm_run & KSM_RUN_OFFLINE) { mutex_unlock(&ksm_thread_mutex); wait_on_bit(&ksm_run, ilog2(KSM_RUN_OFFLINE), |
743162013
|
1904 |
TASK_UNINTERRUPTIBLE); |
ef4d43a80
|
1905 1906 1907 |
mutex_lock(&ksm_thread_mutex); } } |
ee0ea59cf
|
1908 1909 |
static void ksm_check_stable_tree(unsigned long start_pfn, unsigned long end_pfn) |
62b61f611
|
1910 |
{ |
036404183
|
1911 |
struct stable_node *stable_node, *next; |
62b61f611
|
1912 |
struct rb_node *node; |
90bd6fd31
|
1913 |
int nid; |
62b61f611
|
1914 |
|
ef53d16cd
|
1915 1916 |
for (nid = 0; nid < ksm_nr_node_ids; nid++) { node = rb_first(root_stable_tree + nid); |
ee0ea59cf
|
1917 |
while (node) { |
90bd6fd31
|
1918 1919 |
stable_node = rb_entry(node, struct stable_node, node); if (stable_node->kpfn >= start_pfn && |
ee0ea59cf
|
1920 1921 1922 1923 1924 1925 |
stable_node->kpfn < end_pfn) { /* * Don't get_ksm_page, page has already gone: * which is why we keep kpfn instead of page* */ remove_node_from_stable_tree(stable_node); |
ef53d16cd
|
1926 |
node = rb_first(root_stable_tree + nid); |
ee0ea59cf
|
1927 1928 1929 |
} else node = rb_next(node); cond_resched(); |
90bd6fd31
|
1930 |
} |
ee0ea59cf
|
1931 |
} |
036404183
|
1932 |
list_for_each_entry_safe(stable_node, next, &migrate_nodes, list) { |
4146d2d67
|
1933 1934 1935 1936 1937 |
if (stable_node->kpfn >= start_pfn && stable_node->kpfn < end_pfn) remove_node_from_stable_tree(stable_node); cond_resched(); } |
62b61f611
|
1938 1939 1940 1941 1942 1943 |
} static int ksm_memory_callback(struct notifier_block *self, unsigned long action, void *arg) { struct memory_notify *mn = arg; |
62b61f611
|
1944 1945 1946 1947 |
switch (action) { case MEM_GOING_OFFLINE: /* |
ef4d43a80
|
1948 1949 1950 1951 1952 |
* Prevent ksm_do_scan(), unmerge_and_remove_all_rmap_items() * and remove_all_stable_nodes() while memory is going offline: * it is unsafe for them to touch the stable tree at this time. * But unmerge_ksm_pages(), rmap lookups and other entry points * which do not need the ksm_thread_mutex are all safe. |
62b61f611
|
1953 |
*/ |
ef4d43a80
|
1954 1955 1956 |
mutex_lock(&ksm_thread_mutex); ksm_run |= KSM_RUN_OFFLINE; mutex_unlock(&ksm_thread_mutex); |
62b61f611
|
1957 1958 1959 1960 1961 1962 |
break; case MEM_OFFLINE: /* * Most of the work is done by page migration; but there might * be a few stable_nodes left over, still pointing to struct |
ee0ea59cf
|
1963 1964 1965 |
* pages which have been offlined: prune those from the tree, * otherwise get_ksm_page() might later try to access a * non-existent struct page. |
62b61f611
|
1966 |
*/ |
ee0ea59cf
|
1967 1968 |
ksm_check_stable_tree(mn->start_pfn, mn->start_pfn + mn->nr_pages); |
62b61f611
|
1969 1970 1971 |
/* fallthrough */ case MEM_CANCEL_OFFLINE: |
ef4d43a80
|
1972 1973 |
mutex_lock(&ksm_thread_mutex); ksm_run &= ~KSM_RUN_OFFLINE; |
62b61f611
|
1974 |
mutex_unlock(&ksm_thread_mutex); |
ef4d43a80
|
1975 1976 1977 |
smp_mb(); /* wake_up_bit advises this */ wake_up_bit(&ksm_run, ilog2(KSM_RUN_OFFLINE)); |
62b61f611
|
1978 1979 1980 1981 |
break; } return NOTIFY_OK; } |
ef4d43a80
|
1982 1983 1984 1985 |
#else static void wait_while_offlining(void) { } |
62b61f611
|
1986 |
#endif /* CONFIG_MEMORY_HOTREMOVE */ |
2ffd8679c
|
1987 1988 1989 1990 |
#ifdef CONFIG_SYSFS /* * This all compiles without CONFIG_SYSFS, but is a waste of space. */ |
31dbd01f3
|
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 |
#define KSM_ATTR_RO(_name) \ static struct kobj_attribute _name##_attr = __ATTR_RO(_name) #define KSM_ATTR(_name) \ static struct kobj_attribute _name##_attr = \ __ATTR(_name, 0644, _name##_show, _name##_store) static ssize_t sleep_millisecs_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return sprintf(buf, "%u ", ksm_thread_sleep_millisecs); } static ssize_t sleep_millisecs_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { unsigned long msecs; int err; |
3dbb95f78
|
2010 |
err = kstrtoul(buf, 10, &msecs); |
31dbd01f3
|
2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 |
if (err || msecs > UINT_MAX) return -EINVAL; ksm_thread_sleep_millisecs = msecs; return count; } KSM_ATTR(sleep_millisecs); static ssize_t pages_to_scan_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return sprintf(buf, "%u ", ksm_thread_pages_to_scan); } static ssize_t pages_to_scan_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { int err; unsigned long nr_pages; |
3dbb95f78
|
2033 |
err = kstrtoul(buf, 10, &nr_pages); |
31dbd01f3
|
2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 |
if (err || nr_pages > UINT_MAX) return -EINVAL; ksm_thread_pages_to_scan = nr_pages; return count; } KSM_ATTR(pages_to_scan); static ssize_t run_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { |
ef4d43a80
|
2046 2047 |
return sprintf(buf, "%lu ", ksm_run); |
31dbd01f3
|
2048 2049 2050 2051 2052 2053 2054 |
} static ssize_t run_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { int err; unsigned long flags; |
3dbb95f78
|
2055 |
err = kstrtoul(buf, 10, &flags); |
31dbd01f3
|
2056 2057 2058 2059 2060 2061 2062 2063 |
if (err || flags > UINT_MAX) return -EINVAL; if (flags > KSM_RUN_UNMERGE) return -EINVAL; /* * KSM_RUN_MERGE sets ksmd running, and 0 stops it running. * KSM_RUN_UNMERGE stops it running and unmerges all rmap_items, |
d0f209f68
|
2064 2065 |
* breaking COW to free the pages_shared (but leaves mm_slots * on the list for when ksmd may be set running again). |
31dbd01f3
|
2066 2067 2068 |
*/ mutex_lock(&ksm_thread_mutex); |
ef4d43a80
|
2069 |
wait_while_offlining(); |
31dbd01f3
|
2070 2071 |
if (ksm_run != flags) { ksm_run = flags; |
d952b7913
|
2072 |
if (flags & KSM_RUN_UNMERGE) { |
e1e12d2f3
|
2073 |
set_current_oom_origin(); |
d952b7913
|
2074 |
err = unmerge_and_remove_all_rmap_items(); |
e1e12d2f3
|
2075 |
clear_current_oom_origin(); |
d952b7913
|
2076 2077 2078 2079 2080 |
if (err) { ksm_run = KSM_RUN_STOP; count = err; } } |
31dbd01f3
|
2081 2082 2083 2084 2085 2086 2087 2088 2089 |
} mutex_unlock(&ksm_thread_mutex); if (flags & KSM_RUN_MERGE) wake_up_interruptible(&ksm_thread_wait); return count; } KSM_ATTR(run); |
90bd6fd31
|
2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 |
#ifdef CONFIG_NUMA static ssize_t merge_across_nodes_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return sprintf(buf, "%u ", ksm_merge_across_nodes); } static ssize_t merge_across_nodes_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { int err; unsigned long knob; err = kstrtoul(buf, 10, &knob); if (err) return err; if (knob > 1) return -EINVAL; mutex_lock(&ksm_thread_mutex); |
ef4d43a80
|
2112 |
wait_while_offlining(); |
90bd6fd31
|
2113 |
if (ksm_merge_across_nodes != knob) { |
cbf86cfe0
|
2114 |
if (ksm_pages_shared || remove_all_stable_nodes()) |
90bd6fd31
|
2115 |
err = -EBUSY; |
ef53d16cd
|
2116 2117 2118 2119 2120 2121 2122 2123 2124 |
else if (root_stable_tree == one_stable_tree) { struct rb_root *buf; /* * This is the first time that we switch away from the * default of merging across nodes: must now allocate * a buffer to hold as many roots as may be needed. * Allocate stable and unstable together: * MAXSMP NODES_SHIFT 10 will use 16kB. */ |
bafe1e144
|
2125 2126 |
buf = kcalloc(nr_node_ids + nr_node_ids, sizeof(*buf), GFP_KERNEL); |
ef53d16cd
|
2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 |
/* Let us assume that RB_ROOT is NULL is zero */ if (!buf) err = -ENOMEM; else { root_stable_tree = buf; root_unstable_tree = buf + nr_node_ids; /* Stable tree is empty but not the unstable */ root_unstable_tree[0] = one_unstable_tree[0]; } } if (!err) { |
90bd6fd31
|
2138 |
ksm_merge_across_nodes = knob; |
ef53d16cd
|
2139 2140 |
ksm_nr_node_ids = knob ? 1 : nr_node_ids; } |
90bd6fd31
|
2141 2142 2143 2144 2145 2146 2147 |
} mutex_unlock(&ksm_thread_mutex); return err ? err : count; } KSM_ATTR(merge_across_nodes); #endif |
b40282603
|
2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 |
static ssize_t pages_shared_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return sprintf(buf, "%lu ", ksm_pages_shared); } KSM_ATTR_RO(pages_shared); static ssize_t pages_sharing_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { |
e178dfde3
|
2159 2160 |
return sprintf(buf, "%lu ", ksm_pages_sharing); |
b40282603
|
2161 2162 |
} KSM_ATTR_RO(pages_sharing); |
473b0ce4d
|
2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 |
static ssize_t pages_unshared_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return sprintf(buf, "%lu ", ksm_pages_unshared); } KSM_ATTR_RO(pages_unshared); static ssize_t pages_volatile_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { long ksm_pages_volatile; ksm_pages_volatile = ksm_rmap_items - ksm_pages_shared - ksm_pages_sharing - ksm_pages_unshared; /* * It was not worth any locking to calculate that statistic, * but it might therefore sometimes be negative: conceal that. */ if (ksm_pages_volatile < 0) ksm_pages_volatile = 0; return sprintf(buf, "%ld ", ksm_pages_volatile); } KSM_ATTR_RO(pages_volatile); static ssize_t full_scans_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return sprintf(buf, "%lu ", ksm_scan.seqnr); } KSM_ATTR_RO(full_scans); |
31dbd01f3
|
2196 2197 2198 2199 |
static struct attribute *ksm_attrs[] = { &sleep_millisecs_attr.attr, &pages_to_scan_attr.attr, &run_attr.attr, |
b40282603
|
2200 2201 |
&pages_shared_attr.attr, &pages_sharing_attr.attr, |
473b0ce4d
|
2202 2203 2204 |
&pages_unshared_attr.attr, &pages_volatile_attr.attr, &full_scans_attr.attr, |
90bd6fd31
|
2205 2206 2207 |
#ifdef CONFIG_NUMA &merge_across_nodes_attr.attr, #endif |
31dbd01f3
|
2208 2209 2210 2211 2212 2213 2214 |
NULL, }; static struct attribute_group ksm_attr_group = { .attrs = ksm_attrs, .name = "ksm", }; |
2ffd8679c
|
2215 |
#endif /* CONFIG_SYSFS */ |
31dbd01f3
|
2216 2217 2218 2219 2220 2221 2222 2223 2224 |
static int __init ksm_init(void) { struct task_struct *ksm_thread; int err; err = ksm_slab_init(); if (err) goto out; |
31dbd01f3
|
2225 2226 |
ksm_thread = kthread_run(ksm_scan_thread, NULL, "ksmd"); if (IS_ERR(ksm_thread)) { |
25acde317
|
2227 2228 |
pr_err("ksm: creating kthread failed "); |
31dbd01f3
|
2229 |
err = PTR_ERR(ksm_thread); |
d9f8984c2
|
2230 |
goto out_free; |
31dbd01f3
|
2231 |
} |
2ffd8679c
|
2232 |
#ifdef CONFIG_SYSFS |
31dbd01f3
|
2233 2234 |
err = sysfs_create_group(mm_kobj, &ksm_attr_group); if (err) { |
25acde317
|
2235 2236 |
pr_err("ksm: register sysfs failed "); |
2ffd8679c
|
2237 |
kthread_stop(ksm_thread); |
d9f8984c2
|
2238 |
goto out_free; |
31dbd01f3
|
2239 |
} |
c73602ad3
|
2240 2241 |
#else ksm_run = KSM_RUN_MERGE; /* no way for user to start it */ |
2ffd8679c
|
2242 |
#endif /* CONFIG_SYSFS */ |
31dbd01f3
|
2243 |
|
62b61f611
|
2244 |
#ifdef CONFIG_MEMORY_HOTREMOVE |
ef4d43a80
|
2245 |
/* There is no significance to this priority 100 */ |
62b61f611
|
2246 2247 |
hotplug_memory_notifier(ksm_memory_callback, 100); #endif |
31dbd01f3
|
2248 |
return 0; |
d9f8984c2
|
2249 |
out_free: |
31dbd01f3
|
2250 2251 2252 |
ksm_slab_free(); out: return err; |
f8af4da3b
|
2253 |
} |
a64fb3cd6
|
2254 |
subsys_initcall(ksm_init); |