Blame view
mm/ksm.c
88.2 KB
f8af4da3b ksm: the mm inter... |
1 |
/* |
31dbd01f3 ksm: Kernel SameP... |
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 ksm: change copyr... |
7 |
* Copyright (C) 2008-2009 Red Hat, Inc. |
31dbd01f3 ksm: Kernel SameP... |
8 9 10 11 |
* Authors: * Izik Eidus * Andrea Arcangeli * Chris Wright |
36b2528dc ksm: change copyr... |
12 |
* Hugh Dickins |
31dbd01f3 ksm: Kernel SameP... |
13 14 |
* * This work is licensed under the terms of the GNU GPL, version 2. |
f8af4da3b ksm: the mm inter... |
15 16 17 |
*/ #include <linux/errno.h> |
31dbd01f3 ksm: Kernel SameP... |
18 19 |
#include <linux/mm.h> #include <linux/fs.h> |
f8af4da3b ksm: the mm inter... |
20 |
#include <linux/mman.h> |
31dbd01f3 ksm: Kernel SameP... |
21 |
#include <linux/sched.h> |
6e84f3152 sched/headers: Pr... |
22 |
#include <linux/sched/mm.h> |
f7ccbae45 sched/headers: Pr... |
23 |
#include <linux/sched/coredump.h> |
31dbd01f3 ksm: Kernel SameP... |
24 25 26 27 28 29 30 31 32 33 |
#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 ksm: memory hotre... |
34 |
#include <linux/memory.h> |
31dbd01f3 ksm: Kernel SameP... |
35 |
#include <linux/mmu_notifier.h> |
2c6854fda ksm: change defau... |
36 |
#include <linux/swap.h> |
f8af4da3b ksm: the mm inter... |
37 |
#include <linux/ksm.h> |
4ca3a69bc mm/ksm.c: use new... |
38 |
#include <linux/hashtable.h> |
878aee7d6 thp: freeze khuge... |
39 |
#include <linux/freezer.h> |
72788c385 oom: replace PF_O... |
40 |
#include <linux/oom.h> |
90bd6fd31 ksm: allow trees ... |
41 |
#include <linux/numa.h> |
f8af4da3b ksm: the mm inter... |
42 |
|
31dbd01f3 ksm: Kernel SameP... |
43 |
#include <asm/tlbflush.h> |
73848b468 ksm: fix mlockfre... |
44 |
#include "internal.h" |
31dbd01f3 ksm: Kernel SameP... |
45 |
|
e850dcf53 ksm: trivial tidyups |
46 47 48 49 50 51 52 |
#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 |
5a2ca3efe mm/ksm: docs: ext... |
53 54 55 |
/** * DOC: Overview * |
31dbd01f3 ksm: Kernel SameP... |
56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 |
* 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. * |
5a2ca3efe mm/ksm: docs: ext... |
71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 |
* The stable tree node includes information required for reverse * mapping from a KSM page to virtual addresses that map this page. * * In order to avoid large latencies of the rmap walks on KSM pages, * KSM maintains two types of nodes in the stable tree: * * * the regular nodes that keep the reverse mapping structures in a * linked list * * the "chains" that link nodes ("dups") that represent the same * write protected memory content, but each "dup" corresponds to a * different KSM page copy of that content * * Internally, the regular nodes, "dups" and "chains" are represented * using the same :c:type:`struct stable_node` structure. * |
31dbd01f3 ksm: Kernel SameP... |
86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 |
* 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 ksm: add some com... |
108 109 110 |
* * 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 ksm: Kernel SameP... |
111 112 113 114 115 116 |
*/ /** * 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 ksm: singly-linke... |
117 |
* @rmap_list: head for this mm_slot's singly-linked list of rmap_items |
31dbd01f3 ksm: Kernel SameP... |
118 119 120 121 122 |
* @mm: the mm that this information is valid for */ struct mm_slot { struct hlist_node link; struct list_head mm_list; |
6514d511d ksm: singly-linke... |
123 |
struct rmap_item *rmap_list; |
31dbd01f3 ksm: Kernel SameP... |
124 125 126 127 128 129 130 |
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 ksm: singly-linke... |
131 |
* @rmap_list: link to the next rmap to be scanned in the rmap_list |
31dbd01f3 ksm: Kernel SameP... |
132 133 134 135 136 137 138 |
* @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 ksm: singly-linke... |
139 |
struct rmap_item **rmap_list; |
31dbd01f3 ksm: Kernel SameP... |
140 141 142 143 |
unsigned long seqnr; }; /** |
7b6ba2c7d ksm: separate sta... |
144 145 |
* struct stable_node - node of the stable rbtree * @node: rb node of this ksm page in the stable tree |
4146d2d67 ksm: make !merge_... |
146 |
* @head: (overlaying parent) &migrate_nodes indicates temporarily on that list |
2c653d0ee ksm: introduce ks... |
147 |
* @hlist_dup: linked into the stable_node->hlist with a stable_node chain |
4146d2d67 ksm: make !merge_... |
148 |
* @list: linked into migrate_nodes, pending placement in the proper node tree |
7b6ba2c7d ksm: separate sta... |
149 |
* @hlist: hlist head of rmap_items using this ksm page |
4146d2d67 ksm: make !merge_... |
150 |
* @kpfn: page frame number of this ksm page (perhaps temporarily on wrong nid) |
2c653d0ee ksm: introduce ks... |
151 152 |
* @chain_prune_time: time of the last full garbage collection * @rmap_hlist_len: number of rmap_item entries in hlist or STABLE_NODE_CHAIN |
4146d2d67 ksm: make !merge_... |
153 |
* @nid: NUMA node id of stable tree in which linked (may not match kpfn) |
7b6ba2c7d ksm: separate sta... |
154 155 |
*/ struct stable_node { |
4146d2d67 ksm: make !merge_... |
156 157 158 159 |
union { struct rb_node node; /* when node of stable tree */ struct { /* when listed for migration */ struct list_head *head; |
2c653d0ee ksm: introduce ks... |
160 161 162 163 |
struct { struct hlist_node hlist_dup; struct list_head list; }; |
4146d2d67 ksm: make !merge_... |
164 165 |
}; }; |
7b6ba2c7d ksm: separate sta... |
166 |
struct hlist_head hlist; |
2c653d0ee ksm: introduce ks... |
167 168 169 170 171 172 173 174 175 176 177 |
union { unsigned long kpfn; unsigned long chain_prune_time; }; /* * STABLE_NODE_CHAIN can be any negative number in * rmap_hlist_len negative range, but better not -1 to be able * to reliably detect underflows. */ #define STABLE_NODE_CHAIN -1024 int rmap_hlist_len; |
4146d2d67 ksm: make !merge_... |
178 179 180 |
#ifdef CONFIG_NUMA int nid; #endif |
7b6ba2c7d ksm: separate sta... |
181 182 183 |
}; /** |
31dbd01f3 ksm: Kernel SameP... |
184 |
* struct rmap_item - reverse mapping item for virtual addresses |
6514d511d ksm: singly-linke... |
185 |
* @rmap_list: next rmap_item in mm_slot's singly-linked rmap_list |
db114b83a ksm: hold anon_vm... |
186 |
* @anon_vma: pointer to anon_vma for this mm,address, when in stable tree |
bc56620b4 ksm: shrink 32-bi... |
187 |
* @nid: NUMA node id of unstable tree in which linked (may not match page) |
31dbd01f3 ksm: Kernel SameP... |
188 189 190 |
* @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 ksm: separate sta... |
191 192 193 |
* @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 ksm: Kernel SameP... |
194 195 |
*/ struct rmap_item { |
6514d511d ksm: singly-linke... |
196 |
struct rmap_item *rmap_list; |
bc56620b4 ksm: shrink 32-bi... |
197 198 199 200 201 202 |
union { struct anon_vma *anon_vma; /* when stable */ #ifdef CONFIG_NUMA int nid; /* when node of unstable tree */ #endif }; |
31dbd01f3 ksm: Kernel SameP... |
203 204 |
struct mm_struct *mm; unsigned long address; /* + low bits used for flags below */ |
7b6ba2c7d ksm: separate sta... |
205 |
unsigned int oldchecksum; /* when unstable */ |
31dbd01f3 ksm: Kernel SameP... |
206 |
union { |
7b6ba2c7d ksm: separate sta... |
207 208 209 210 211 |
struct rb_node node; /* when node of unstable tree */ struct { /* when listed from stable tree */ struct stable_node *head; struct hlist_node hlist; }; |
31dbd01f3 ksm: Kernel SameP... |
212 213 214 215 |
}; }; #define SEQNR_MASK 0x0ff /* low bits of unstable tree seqnr */ |
7b6ba2c7d ksm: separate sta... |
216 217 |
#define UNSTABLE_FLAG 0x100 /* is a node of the unstable tree */ #define STABLE_FLAG 0x200 /* is listed from the stable tree */ |
1105a2fc0 mm/ksm.c: ignore ... |
218 219 |
#define KSM_FLAG_MASK (SEQNR_MASK|UNSTABLE_FLAG|STABLE_FLAG) /* to mask all the flags */ |
31dbd01f3 ksm: Kernel SameP... |
220 221 |
/* The stable and unstable tree heads */ |
ef53d16cd ksm: allocate roo... |
222 223 224 225 |
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 ksm: Kernel SameP... |
226 |
|
4146d2d67 ksm: make !merge_... |
227 228 |
/* Recently migrated nodes of stable tree, pending proper placement */ static LIST_HEAD(migrate_nodes); |
2c653d0ee ksm: introduce ks... |
229 |
#define STABLE_NODE_DUP_HEAD ((struct list_head *)&migrate_nodes.prev) |
4146d2d67 ksm: make !merge_... |
230 |
|
4ca3a69bc mm/ksm.c: use new... |
231 232 |
#define MM_SLOTS_HASH_BITS 10 static DEFINE_HASHTABLE(mm_slots_hash, MM_SLOTS_HASH_BITS); |
31dbd01f3 ksm: Kernel SameP... |
233 234 235 236 237 238 239 240 241 |
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 ksm: separate sta... |
242 |
static struct kmem_cache *stable_node_cache; |
31dbd01f3 ksm: Kernel SameP... |
243 244 245 |
static struct kmem_cache *mm_slot_cache; /* The number of nodes in the stable tree */ |
b40282603 ksm: rename kerne... |
246 |
static unsigned long ksm_pages_shared; |
31dbd01f3 ksm: Kernel SameP... |
247 |
|
e178dfde3 ksm: move pages_s... |
248 |
/* The number of page slots additionally sharing those nodes */ |
b40282603 ksm: rename kerne... |
249 |
static unsigned long ksm_pages_sharing; |
31dbd01f3 ksm: Kernel SameP... |
250 |
|
473b0ce4d ksm: pages_unshar... |
251 252 253 254 255 |
/* 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; |
2c653d0ee ksm: introduce ks... |
256 257 258 259 260 261 262 263 264 265 266 |
/* The number of stable_node chains */ static unsigned long ksm_stable_node_chains; /* The number of stable_node dups linked to the stable_node chains */ static unsigned long ksm_stable_node_dups; /* Delay in pruning stale stable_node_dups in the stable_node_chains */ static int ksm_stable_node_chains_prune_millisecs = 2000; /* Maximum number of page slots sharing a stable node */ static int ksm_max_page_sharing = 256; |
31dbd01f3 ksm: Kernel SameP... |
267 |
/* Number of pages ksmd should scan in one batch */ |
2c6854fda ksm: change defau... |
268 |
static unsigned int ksm_thread_pages_to_scan = 100; |
31dbd01f3 ksm: Kernel SameP... |
269 270 |
/* Milliseconds ksmd should sleep between batches */ |
2ffd8679c ksm: sysfs and de... |
271 |
static unsigned int ksm_thread_sleep_millisecs = 20; |
31dbd01f3 ksm: Kernel SameP... |
272 |
|
e86c59b1b mm/ksm: improve d... |
273 274 275 276 277 |
/* Checksum of an empty (zeroed) page */ static unsigned int zero_checksum __read_mostly; /* Whether to merge empty (zeroed) pages with actual zero pages */ static bool ksm_use_zero_pages __read_mostly; |
e850dcf53 ksm: trivial tidyups |
278 |
#ifdef CONFIG_NUMA |
90bd6fd31 ksm: allow trees ... |
279 280 |
/* Zeroed when merging across nodes is not allowed */ static unsigned int ksm_merge_across_nodes = 1; |
ef53d16cd ksm: allocate roo... |
281 |
static int ksm_nr_node_ids = 1; |
e850dcf53 ksm: trivial tidyups |
282 283 |
#else #define ksm_merge_across_nodes 1U |
ef53d16cd ksm: allocate roo... |
284 |
#define ksm_nr_node_ids 1 |
e850dcf53 ksm: trivial tidyups |
285 |
#endif |
90bd6fd31 ksm: allow trees ... |
286 |
|
31dbd01f3 ksm: Kernel SameP... |
287 288 289 |
#define KSM_RUN_STOP 0 #define KSM_RUN_MERGE 1 #define KSM_RUN_UNMERGE 2 |
ef4d43a80 ksm: stop hotremo... |
290 291 292 |
#define KSM_RUN_OFFLINE 4 static unsigned long ksm_run = KSM_RUN_STOP; static void wait_while_offlining(void); |
31dbd01f3 ksm: Kernel SameP... |
293 294 295 296 297 298 299 300 301 302 303 304 305 306 |
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 ksm: separate sta... |
307 308 309 |
stable_node_cache = KSM_KMEM_CACHE(stable_node, 0); if (!stable_node_cache) goto out_free1; |
31dbd01f3 ksm: Kernel SameP... |
310 311 |
mm_slot_cache = KSM_KMEM_CACHE(mm_slot, 0); if (!mm_slot_cache) |
7b6ba2c7d ksm: separate sta... |
312 |
goto out_free2; |
31dbd01f3 ksm: Kernel SameP... |
313 314 |
return 0; |
7b6ba2c7d ksm: separate sta... |
315 316 317 |
out_free2: kmem_cache_destroy(stable_node_cache); out_free1: |
31dbd01f3 ksm: Kernel SameP... |
318 319 320 321 322 323 324 325 |
kmem_cache_destroy(rmap_item_cache); out: return -ENOMEM; } static void __init ksm_slab_free(void) { kmem_cache_destroy(mm_slot_cache); |
7b6ba2c7d ksm: separate sta... |
326 |
kmem_cache_destroy(stable_node_cache); |
31dbd01f3 ksm: Kernel SameP... |
327 328 329 |
kmem_cache_destroy(rmap_item_cache); mm_slot_cache = NULL; } |
2c653d0ee ksm: introduce ks... |
330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 |
static __always_inline bool is_stable_node_chain(struct stable_node *chain) { return chain->rmap_hlist_len == STABLE_NODE_CHAIN; } static __always_inline bool is_stable_node_dup(struct stable_node *dup) { return dup->head == STABLE_NODE_DUP_HEAD; } static inline void stable_node_chain_add_dup(struct stable_node *dup, struct stable_node *chain) { VM_BUG_ON(is_stable_node_dup(dup)); dup->head = STABLE_NODE_DUP_HEAD; VM_BUG_ON(!is_stable_node_chain(chain)); hlist_add_head(&dup->hlist_dup, &chain->hlist); ksm_stable_node_dups++; } static inline void __stable_node_dup_del(struct stable_node *dup) { |
b4fecc67c ksm: fix use afte... |
352 |
VM_BUG_ON(!is_stable_node_dup(dup)); |
2c653d0ee ksm: introduce ks... |
353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 |
hlist_del(&dup->hlist_dup); ksm_stable_node_dups--; } static inline void stable_node_dup_del(struct stable_node *dup) { VM_BUG_ON(is_stable_node_chain(dup)); if (is_stable_node_dup(dup)) __stable_node_dup_del(dup); else rb_erase(&dup->node, root_stable_tree + NUMA(dup->nid)); #ifdef CONFIG_DEBUG_VM dup->head = NULL; #endif } |
31dbd01f3 ksm: Kernel SameP... |
368 369 |
static inline struct rmap_item *alloc_rmap_item(void) { |
473b0ce4d ksm: pages_unshar... |
370 |
struct rmap_item *rmap_item; |
5b398e416 mm,ksm: fix endle... |
371 372 |
rmap_item = kmem_cache_zalloc(rmap_item_cache, GFP_KERNEL | __GFP_NORETRY | __GFP_NOWARN); |
473b0ce4d ksm: pages_unshar... |
373 374 375 |
if (rmap_item) ksm_rmap_items++; return rmap_item; |
31dbd01f3 ksm: Kernel SameP... |
376 377 378 379 |
} static inline void free_rmap_item(struct rmap_item *rmap_item) { |
473b0ce4d ksm: pages_unshar... |
380 |
ksm_rmap_items--; |
31dbd01f3 ksm: Kernel SameP... |
381 382 383 |
rmap_item->mm = NULL; /* debug safety */ kmem_cache_free(rmap_item_cache, rmap_item); } |
7b6ba2c7d ksm: separate sta... |
384 385 |
static inline struct stable_node *alloc_stable_node(void) { |
6213055f2 mm,ksm: add __GFP... |
386 387 388 389 390 391 |
/* * 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 ksm: separate sta... |
392 393 394 395 |
} static inline void free_stable_node(struct stable_node *stable_node) { |
2c653d0ee ksm: introduce ks... |
396 397 |
VM_BUG_ON(stable_node->rmap_hlist_len && !is_stable_node_chain(stable_node)); |
7b6ba2c7d ksm: separate sta... |
398 399 |
kmem_cache_free(stable_node_cache, stable_node); } |
31dbd01f3 ksm: Kernel SameP... |
400 401 402 403 404 405 406 407 408 409 410 |
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 ksm: Kernel SameP... |
411 412 |
static struct mm_slot *get_mm_slot(struct mm_struct *mm) { |
4ca3a69bc mm/ksm.c: use new... |
413 |
struct mm_slot *slot; |
b67bfe0d4 hlist: drop the n... |
414 |
hash_for_each_possible(mm_slots_hash, slot, link, (unsigned long)mm) |
4ca3a69bc mm/ksm.c: use new... |
415 416 |
if (slot->mm == mm) return slot; |
31dbd01f3 ksm: Kernel SameP... |
417 |
|
31dbd01f3 ksm: Kernel SameP... |
418 419 420 421 422 423 |
return NULL; } static void insert_to_mm_slots_hash(struct mm_struct *mm, struct mm_slot *mm_slot) { |
31dbd01f3 ksm: Kernel SameP... |
424 |
mm_slot->mm = mm; |
4ca3a69bc mm/ksm.c: use new... |
425 |
hash_add(mm_slots_hash, &mm_slot->link, (unsigned long)mm); |
31dbd01f3 ksm: Kernel SameP... |
426 |
} |
31dbd01f3 ksm: Kernel SameP... |
427 |
/* |
a913e182a ksm: clean up obs... |
428 429 430 431 432 433 434 435 436 437 438 439 440 |
* 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 ksm: Kernel SameP... |
441 442 |
* We use break_ksm to break COW on a ksm page: it's a stripped down * |
d4edcf0d5 mm/gup: Switch al... |
443 |
* if (get_user_pages(addr, 1, 1, 1, &page, NULL) == 1) |
31dbd01f3 ksm: Kernel SameP... |
444 445 446 447 448 449 |
* 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 mm/core: Do not e... |
450 451 452 453 |
* * 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 ksm: Kernel SameP... |
454 |
*/ |
d952b7913 ksm: fix endless ... |
455 |
static int break_ksm(struct vm_area_struct *vma, unsigned long addr) |
31dbd01f3 ksm: Kernel SameP... |
456 457 |
{ struct page *page; |
50a7ca3c6 mm: convert retur... |
458 |
vm_fault_t ret = 0; |
31dbd01f3 ksm: Kernel SameP... |
459 460 461 |
do { cond_resched(); |
1b2ee1266 mm/core: Do not e... |
462 463 |
page = follow_page(vma, addr, FOLL_GET | FOLL_MIGRATION | FOLL_REMOTE); |
22eccdd7d ksm: check for ER... |
464 |
if (IS_ERR_OR_NULL(page)) |
31dbd01f3 ksm: Kernel SameP... |
465 466 |
break; if (PageKsm(page)) |
dcddffd41 mm: do not pass m... |
467 468 |
ret = handle_mm_fault(vma, addr, FAULT_FLAG_WRITE | FAULT_FLAG_REMOTE); |
31dbd01f3 ksm: Kernel SameP... |
469 470 471 |
else ret = VM_FAULT_WRITE; put_page(page); |
33692f275 vm: add VM_FAULT_... |
472 |
} while (!(ret & (VM_FAULT_WRITE | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | VM_FAULT_OOM))); |
d952b7913 ksm: fix endless ... |
473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 |
/* * 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 ksm: Kernel SameP... |
502 |
} |
ef6942224 ksm: cleanup: int... |
503 504 505 506 507 508 509 510 511 512 513 514 515 |
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 ksm: cleanup some... |
516 |
static void break_cow(struct rmap_item *rmap_item) |
31dbd01f3 ksm: Kernel SameP... |
517 |
{ |
8dd3557a5 ksm: cleanup some... |
518 519 |
struct mm_struct *mm = rmap_item->mm; unsigned long addr = rmap_item->address; |
31dbd01f3 ksm: Kernel SameP... |
520 |
struct vm_area_struct *vma; |
4035c07a8 ksm: take keyhole... |
521 522 523 524 |
/* * 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 mm: rename drop_a... |
525 |
put_anon_vma(rmap_item->anon_vma); |
4035c07a8 ksm: take keyhole... |
526 |
|
81464e306 ksm: five little ... |
527 |
down_read(&mm->mmap_sem); |
ef6942224 ksm: cleanup: int... |
528 529 530 |
vma = find_mergeable_vma(mm, addr); if (vma) break_ksm(vma, addr); |
31dbd01f3 ksm: Kernel SameP... |
531 532 533 534 535 536 537 538 539 540 541 |
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 ksm: cleanup: int... |
542 543 |
vma = find_mergeable_vma(mm, addr); if (!vma) |
31dbd01f3 ksm: Kernel SameP... |
544 545 546 |
goto out; page = follow_page(vma, addr, FOLL_GET); |
22eccdd7d ksm: check for ER... |
547 |
if (IS_ERR_OR_NULL(page)) |
31dbd01f3 ksm: Kernel SameP... |
548 |
goto out; |
f765f5405 ksm: prepare to n... |
549 |
if (PageAnon(page)) { |
31dbd01f3 ksm: Kernel SameP... |
550 551 552 553 |
flush_anon_page(vma, page, addr); flush_dcache_page(page); } else { put_page(page); |
c8f95ed1a ksm: unstable_tre... |
554 555 |
out: page = NULL; |
31dbd01f3 ksm: Kernel SameP... |
556 557 558 559 |
} up_read(&mm->mmap_sem); return page; } |
90bd6fd31 ksm: allow trees ... |
560 561 562 563 564 565 566 567 |
/* * 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 ksm: fix m68k bui... |
568 |
return ksm_merge_across_nodes ? 0 : NUMA(pfn_to_nid(kpfn)); |
90bd6fd31 ksm: allow trees ... |
569 |
} |
2c653d0ee ksm: introduce ks... |
570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 |
static struct stable_node *alloc_stable_node_chain(struct stable_node *dup, struct rb_root *root) { struct stable_node *chain = alloc_stable_node(); VM_BUG_ON(is_stable_node_chain(dup)); if (likely(chain)) { INIT_HLIST_HEAD(&chain->hlist); chain->chain_prune_time = jiffies; chain->rmap_hlist_len = STABLE_NODE_CHAIN; #if defined (CONFIG_DEBUG_VM) && defined(CONFIG_NUMA) chain->nid = -1; /* debug */ #endif ksm_stable_node_chains++; /* * Put the stable node chain in the first dimension of * the stable tree and at the same time remove the old * stable node. */ rb_replace_node(&dup->node, &chain->node, root); /* * Move the old stable node to the second dimension * queued in the hlist_dup. The invariant is that all * dup stable_nodes in the chain->hlist point to pages * that are wrprotected and have the exact same * content. */ stable_node_chain_add_dup(dup, chain); } return chain; } static inline void free_stable_node_chain(struct stable_node *chain, struct rb_root *root) { rb_erase(&chain->node, root); free_stable_node(chain); ksm_stable_node_chains--; } |
4035c07a8 ksm: take keyhole... |
610 611 612 |
static void remove_node_from_stable_tree(struct stable_node *stable_node) { struct rmap_item *rmap_item; |
4035c07a8 ksm: take keyhole... |
613 |
|
2c653d0ee ksm: introduce ks... |
614 615 |
/* check it's not STABLE_NODE_CHAIN or negative */ BUG_ON(stable_node->rmap_hlist_len < 0); |
b67bfe0d4 hlist: drop the n... |
616 |
hlist_for_each_entry(rmap_item, &stable_node->hlist, hlist) { |
4035c07a8 ksm: take keyhole... |
617 618 619 620 |
if (rmap_item->hlist.next) ksm_pages_sharing--; else ksm_pages_shared--; |
2c653d0ee ksm: introduce ks... |
621 622 |
VM_BUG_ON(stable_node->rmap_hlist_len <= 0); stable_node->rmap_hlist_len--; |
9e60109f1 mm: rename drop_a... |
623 |
put_anon_vma(rmap_item->anon_vma); |
4035c07a8 ksm: take keyhole... |
624 625 626 |
rmap_item->address &= PAGE_MASK; cond_resched(); } |
2c653d0ee ksm: introduce ks... |
627 628 629 630 631 |
/* * We need the second aligned pointer of the migrate_nodes * list_head to stay clear from the rb_parent_color union * (aligned and different than any node) and also different * from &migrate_nodes. This will verify that future list.h changes |
815f0ddb3 include/linux/com... |
632 |
* don't break STABLE_NODE_DUP_HEAD. Only recent gcc can handle it. |
2c653d0ee ksm: introduce ks... |
633 |
*/ |
815f0ddb3 include/linux/com... |
634 |
#if defined(GCC_VERSION) && GCC_VERSION >= 40903 |
2c653d0ee ksm: introduce ks... |
635 636 637 |
BUILD_BUG_ON(STABLE_NODE_DUP_HEAD <= &migrate_nodes); BUILD_BUG_ON(STABLE_NODE_DUP_HEAD >= &migrate_nodes + 1); #endif |
4146d2d67 ksm: make !merge_... |
638 639 640 |
if (stable_node->head == &migrate_nodes) list_del(&stable_node->list); else |
2c653d0ee ksm: introduce ks... |
641 |
stable_node_dup_del(stable_node); |
4035c07a8 ksm: take keyhole... |
642 643 644 645 646 647 648 649 650 |
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 ksm: make KSM pag... |
651 |
* But beware, the stable node's page might be being migrated. |
4035c07a8 ksm: take keyhole... |
652 653 654 655 656 657 658 659 660 661 |
* * 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 ksm: take keyhole... |
662 663 |
* is on its way to being freed; but it is an anomaly to bear in mind. */ |
8fdb3dbf0 ksm: add some com... |
664 |
static struct page *get_ksm_page(struct stable_node *stable_node, bool lock_it) |
4035c07a8 ksm: take keyhole... |
665 666 667 |
{ struct page *page; void *expected_mapping; |
c8d6553b9 ksm: make KSM pag... |
668 |
unsigned long kpfn; |
4035c07a8 ksm: take keyhole... |
669 |
|
bda807d44 mm: migrate: supp... |
670 671 |
expected_mapping = (void *)((unsigned long)stable_node | PAGE_MAPPING_KSM); |
c8d6553b9 ksm: make KSM pag... |
672 |
again: |
08df47743 mm/ksm: Remove no... |
673 |
kpfn = READ_ONCE(stable_node->kpfn); /* Address dependency. */ |
c8d6553b9 ksm: make KSM pag... |
674 |
page = pfn_to_page(kpfn); |
4db0c3c29 mm: remove rest o... |
675 |
if (READ_ONCE(page->mapping) != expected_mapping) |
4035c07a8 ksm: take keyhole... |
676 |
goto stale; |
c8d6553b9 ksm: make KSM pag... |
677 678 679 680 681 |
/* * 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; |
1c4c3b99c mm: fix page_free... |
682 |
* however, it might mean that the page is under page_ref_freeze(). |
c8d6553b9 ksm: make KSM pag... |
683 684 685 686 687 688 689 690 691 692 |
* 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) |
1c4c3b99c mm: fix page_free... |
693 |
* in the ref_freeze section of __remove_mapping(); but Anon |
c8d6553b9 ksm: make KSM pag... |
694 695 696 697 698 699 |
* page->mapping reset to NULL later, in free_pages_prepare(). */ if (!PageSwapCache(page)) goto stale; cpu_relax(); } |
4db0c3c29 mm: remove rest o... |
700 |
if (READ_ONCE(page->mapping) != expected_mapping) { |
4035c07a8 ksm: take keyhole... |
701 702 703 |
put_page(page); goto stale; } |
c8d6553b9 ksm: make KSM pag... |
704 |
|
8fdb3dbf0 ksm: add some com... |
705 |
if (lock_it) { |
8aafa6a48 ksm: get_ksm_page... |
706 |
lock_page(page); |
4db0c3c29 mm: remove rest o... |
707 |
if (READ_ONCE(page->mapping) != expected_mapping) { |
8aafa6a48 ksm: get_ksm_page... |
708 709 710 711 712 |
unlock_page(page); put_page(page); goto stale; } } |
4035c07a8 ksm: take keyhole... |
713 |
return page; |
c8d6553b9 ksm: make KSM pag... |
714 |
|
4035c07a8 ksm: take keyhole... |
715 |
stale: |
c8d6553b9 ksm: make KSM pag... |
716 717 718 719 720 721 722 |
/* * 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 mm: remove rest o... |
723 |
if (READ_ONCE(stable_node->kpfn) != kpfn) |
c8d6553b9 ksm: make KSM pag... |
724 |
goto again; |
4035c07a8 ksm: take keyhole... |
725 726 727 |
remove_node_from_stable_tree(stable_node); return NULL; } |
31dbd01f3 ksm: Kernel SameP... |
728 |
/* |
31dbd01f3 ksm: Kernel SameP... |
729 730 731 732 733 |
* 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 ksm: separate sta... |
734 735 |
if (rmap_item->address & STABLE_FLAG) { struct stable_node *stable_node; |
5ad646880 ksm: let shared p... |
736 |
struct page *page; |
31dbd01f3 ksm: Kernel SameP... |
737 |
|
7b6ba2c7d ksm: separate sta... |
738 |
stable_node = rmap_item->head; |
8aafa6a48 ksm: get_ksm_page... |
739 |
page = get_ksm_page(stable_node, true); |
4035c07a8 ksm: take keyhole... |
740 741 |
if (!page) goto out; |
5ad646880 ksm: let shared p... |
742 |
|
7b6ba2c7d ksm: separate sta... |
743 |
hlist_del(&rmap_item->hlist); |
4035c07a8 ksm: take keyhole... |
744 745 |
unlock_page(page); put_page(page); |
08beca44d ksm: stable_node ... |
746 |
|
98666f8a2 ksm: use the help... |
747 |
if (!hlist_empty(&stable_node->hlist)) |
4035c07a8 ksm: take keyhole... |
748 749 |
ksm_pages_sharing--; else |
7b6ba2c7d ksm: separate sta... |
750 |
ksm_pages_shared--; |
2c653d0ee ksm: introduce ks... |
751 752 |
VM_BUG_ON(stable_node->rmap_hlist_len <= 0); stable_node->rmap_hlist_len--; |
31dbd01f3 ksm: Kernel SameP... |
753 |
|
9e60109f1 mm: rename drop_a... |
754 |
put_anon_vma(rmap_item->anon_vma); |
93d17715a ksm: three remove... |
755 |
rmap_item->address &= PAGE_MASK; |
31dbd01f3 ksm: Kernel SameP... |
756 |
|
7b6ba2c7d ksm: separate sta... |
757 |
} else if (rmap_item->address & UNSTABLE_FLAG) { |
31dbd01f3 ksm: Kernel SameP... |
758 759 |
unsigned char age; /* |
9ba692948 ksm: fix oom dead... |
760 |
* Usually ksmd can and must skip the rb_erase, because |
31dbd01f3 ksm: Kernel SameP... |
761 |
* root_unstable_tree was already reset to RB_ROOT. |
9ba692948 ksm: fix oom dead... |
762 763 764 |
* 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 ksm: Kernel SameP... |
765 766 |
*/ age = (unsigned char)(ksm_scan.seqnr - rmap_item->address); |
cd551f975 ksm: distribute r... |
767 |
BUG_ON(age > 1); |
31dbd01f3 ksm: Kernel SameP... |
768 |
if (!age) |
90bd6fd31 ksm: allow trees ... |
769 |
rb_erase(&rmap_item->node, |
ef53d16cd ksm: allocate roo... |
770 |
root_unstable_tree + NUMA(rmap_item->nid)); |
473b0ce4d ksm: pages_unshar... |
771 |
ksm_pages_unshared--; |
93d17715a ksm: three remove... |
772 |
rmap_item->address &= PAGE_MASK; |
31dbd01f3 ksm: Kernel SameP... |
773 |
} |
4035c07a8 ksm: take keyhole... |
774 |
out: |
31dbd01f3 ksm: Kernel SameP... |
775 776 |
cond_resched(); /* we're called from many long loops */ } |
31dbd01f3 ksm: Kernel SameP... |
777 |
static void remove_trailing_rmap_items(struct mm_slot *mm_slot, |
6514d511d ksm: singly-linke... |
778 |
struct rmap_item **rmap_list) |
31dbd01f3 ksm: Kernel SameP... |
779 |
{ |
6514d511d ksm: singly-linke... |
780 781 782 |
while (*rmap_list) { struct rmap_item *rmap_item = *rmap_list; *rmap_list = rmap_item->rmap_list; |
31dbd01f3 ksm: Kernel SameP... |
783 |
remove_rmap_item_from_tree(rmap_item); |
31dbd01f3 ksm: Kernel SameP... |
784 785 786 787 788 |
free_rmap_item(rmap_item); } } /* |
e850dcf53 ksm: trivial tidyups |
789 |
* Though it's very tempting to unmerge rmap_items from stable tree rather |
31dbd01f3 ksm: Kernel SameP... |
790 791 792 793 794 |
* 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 ksm: five little ... |
795 796 797 798 799 |
* * 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 ksm: Kernel SameP... |
800 |
*/ |
d952b7913 ksm: fix endless ... |
801 802 |
static int unmerge_ksm_pages(struct vm_area_struct *vma, unsigned long start, unsigned long end) |
31dbd01f3 ksm: Kernel SameP... |
803 804 |
{ unsigned long addr; |
d952b7913 ksm: fix endless ... |
805 |
int err = 0; |
31dbd01f3 ksm: Kernel SameP... |
806 |
|
d952b7913 ksm: fix endless ... |
807 |
for (addr = start; addr < end && !err; addr += PAGE_SIZE) { |
9ba692948 ksm: fix oom dead... |
808 809 |
if (ksm_test_exit(vma->vm_mm)) break; |
d952b7913 ksm: fix endless ... |
810 811 812 813 814 815 |
if (signal_pending(current)) err = -ERESTARTSYS; else err = break_ksm(vma, addr); } return err; |
31dbd01f3 ksm: Kernel SameP... |
816 |
} |
88484826b mm/ksm: move [set... |
817 818 819 820 821 822 823 824 825 826 |
static inline struct stable_node *page_stable_node(struct page *page) { return PageKsm(page) ? page_rmapping(page) : NULL; } static inline void set_page_stable_node(struct page *page, struct stable_node *stable_node) { page->mapping = (void *)((unsigned long)stable_node | PAGE_MAPPING_KSM); } |
2ffd8679c ksm: sysfs and de... |
827 828 829 830 |
#ifdef CONFIG_SYSFS /* * Only called through the sysfs control interface: */ |
cbf86cfe0 ksm: remove old s... |
831 832 833 834 835 836 837 838 839 840 841 842 |
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 ksm: add some com... |
843 844 845 846 847 |
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 ksm: remove old s... |
848 |
err = -EBUSY; |
8fdb3dbf0 ksm: add some com... |
849 |
} else { |
cbf86cfe0 ksm: remove old s... |
850 |
/* |
8fdb3dbf0 ksm: add some com... |
851 852 853 |
* 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 ksm: remove old s... |
854 855 856 857 858 859 860 861 862 863 864 865 866 |
* 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; } |
2c653d0ee ksm: introduce ks... |
867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 |
static int remove_stable_node_chain(struct stable_node *stable_node, struct rb_root *root) { struct stable_node *dup; struct hlist_node *hlist_safe; if (!is_stable_node_chain(stable_node)) { VM_BUG_ON(is_stable_node_dup(stable_node)); if (remove_stable_node(stable_node)) return true; else return false; } hlist_for_each_entry_safe(dup, hlist_safe, &stable_node->hlist, hlist_dup) { VM_BUG_ON(!is_stable_node_dup(dup)); if (remove_stable_node(dup)) return true; } BUG_ON(!hlist_empty(&stable_node->hlist)); free_stable_node_chain(stable_node, root); return false; } |
cbf86cfe0 ksm: remove old s... |
891 892 |
static int remove_all_stable_nodes(void) { |
036404183 mm/ksm.c: use lis... |
893 |
struct stable_node *stable_node, *next; |
cbf86cfe0 ksm: remove old s... |
894 895 |
int nid; int err = 0; |
ef53d16cd ksm: allocate roo... |
896 |
for (nid = 0; nid < ksm_nr_node_ids; nid++) { |
cbf86cfe0 ksm: remove old s... |
897 898 899 |
while (root_stable_tree[nid].rb_node) { stable_node = rb_entry(root_stable_tree[nid].rb_node, struct stable_node, node); |
2c653d0ee ksm: introduce ks... |
900 901 |
if (remove_stable_node_chain(stable_node, root_stable_tree + nid)) { |
cbf86cfe0 ksm: remove old s... |
902 903 904 905 906 907 |
err = -EBUSY; break; /* proceed to next nid */ } cond_resched(); } } |
036404183 mm/ksm.c: use lis... |
908 |
list_for_each_entry_safe(stable_node, next, &migrate_nodes, list) { |
4146d2d67 ksm: make !merge_... |
909 910 911 912 |
if (remove_stable_node(stable_node)) err = -EBUSY; cond_resched(); } |
cbf86cfe0 ksm: remove old s... |
913 914 |
return err; } |
d952b7913 ksm: fix endless ... |
915 |
static int unmerge_and_remove_all_rmap_items(void) |
31dbd01f3 ksm: Kernel SameP... |
916 917 918 919 |
{ struct mm_slot *mm_slot; struct mm_struct *mm; struct vm_area_struct *vma; |
d952b7913 ksm: fix endless ... |
920 921 922 |
int err = 0; spin_lock(&ksm_mmlist_lock); |
9ba692948 ksm: fix oom dead... |
923 |
ksm_scan.mm_slot = list_entry(ksm_mm_head.mm_list.next, |
d952b7913 ksm: fix endless ... |
924 925 |
struct mm_slot, mm_list); spin_unlock(&ksm_mmlist_lock); |
31dbd01f3 ksm: Kernel SameP... |
926 |
|
9ba692948 ksm: fix oom dead... |
927 928 |
for (mm_slot = ksm_scan.mm_slot; mm_slot != &ksm_mm_head; mm_slot = ksm_scan.mm_slot) { |
31dbd01f3 ksm: Kernel SameP... |
929 930 931 |
mm = mm_slot->mm; down_read(&mm->mmap_sem); for (vma = mm->mmap; vma; vma = vma->vm_next) { |
9ba692948 ksm: fix oom dead... |
932 933 |
if (ksm_test_exit(mm)) break; |
31dbd01f3 ksm: Kernel SameP... |
934 935 |
if (!(vma->vm_flags & VM_MERGEABLE) || !vma->anon_vma) continue; |
d952b7913 ksm: fix endless ... |
936 937 |
err = unmerge_ksm_pages(vma, vma->vm_start, vma->vm_end); |
9ba692948 ksm: fix oom dead... |
938 939 |
if (err) goto error; |
31dbd01f3 ksm: Kernel SameP... |
940 |
} |
9ba692948 ksm: fix oom dead... |
941 |
|
6514d511d ksm: singly-linke... |
942 |
remove_trailing_rmap_items(mm_slot, &mm_slot->rmap_list); |
7496fea9a ksm: fix conflict... |
943 |
up_read(&mm->mmap_sem); |
d952b7913 ksm: fix endless ... |
944 945 |
spin_lock(&ksm_mmlist_lock); |
9ba692948 ksm: fix oom dead... |
946 |
ksm_scan.mm_slot = list_entry(mm_slot->mm_list.next, |
d952b7913 ksm: fix endless ... |
947 |
struct mm_slot, mm_list); |
9ba692948 ksm: fix oom dead... |
948 |
if (ksm_test_exit(mm)) { |
4ca3a69bc mm/ksm.c: use new... |
949 |
hash_del(&mm_slot->link); |
9ba692948 ksm: fix oom dead... |
950 951 952 953 954 |
list_del(&mm_slot->mm_list); spin_unlock(&ksm_mmlist_lock); free_mm_slot(mm_slot); clear_bit(MMF_VM_MERGEABLE, &mm->flags); |
9ba692948 ksm: fix oom dead... |
955 |
mmdrop(mm); |
7496fea9a ksm: fix conflict... |
956 |
} else |
9ba692948 ksm: fix oom dead... |
957 |
spin_unlock(&ksm_mmlist_lock); |
31dbd01f3 ksm: Kernel SameP... |
958 |
} |
cbf86cfe0 ksm: remove old s... |
959 960 |
/* Clean up stable nodes, but don't worry if some are still busy */ remove_all_stable_nodes(); |
d952b7913 ksm: fix endless ... |
961 |
ksm_scan.seqnr = 0; |
9ba692948 ksm: fix oom dead... |
962 963 964 965 |
return 0; error: up_read(&mm->mmap_sem); |
31dbd01f3 ksm: Kernel SameP... |
966 |
spin_lock(&ksm_mmlist_lock); |
d952b7913 ksm: fix endless ... |
967 |
ksm_scan.mm_slot = &ksm_mm_head; |
31dbd01f3 ksm: Kernel SameP... |
968 |
spin_unlock(&ksm_mmlist_lock); |
d952b7913 ksm: fix endless ... |
969 |
return err; |
31dbd01f3 ksm: Kernel SameP... |
970 |
} |
2ffd8679c ksm: sysfs and de... |
971 |
#endif /* CONFIG_SYSFS */ |
31dbd01f3 ksm: Kernel SameP... |
972 |
|
31dbd01f3 ksm: Kernel SameP... |
973 974 975 |
static u32 calc_checksum(struct page *page) { u32 checksum; |
9b04c5fec mm: remove the se... |
976 |
void *addr = kmap_atomic(page); |
31dbd01f3 ksm: Kernel SameP... |
977 |
checksum = jhash2(addr, PAGE_SIZE / 4, 17); |
9b04c5fec mm: remove the se... |
978 |
kunmap_atomic(addr); |
31dbd01f3 ksm: Kernel SameP... |
979 980 981 982 983 984 985 |
return checksum; } static int memcmp_pages(struct page *page1, struct page *page2) { char *addr1, *addr2; int ret; |
9b04c5fec mm: remove the se... |
986 987 |
addr1 = kmap_atomic(page1); addr2 = kmap_atomic(page2); |
31dbd01f3 ksm: Kernel SameP... |
988 |
ret = memcmp(addr1, addr2, PAGE_SIZE); |
9b04c5fec mm: remove the se... |
989 990 |
kunmap_atomic(addr2); kunmap_atomic(addr1); |
31dbd01f3 ksm: Kernel SameP... |
991 992 993 994 995 996 997 998 999 1000 1001 1002 |
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; |
36eaff336 mm, ksm: convert ... |
1003 1004 1005 1006 |
struct page_vma_mapped_walk pvmw = { .page = page, .vma = vma, }; |
31dbd01f3 ksm: Kernel SameP... |
1007 1008 |
int swapped; int err = -EFAULT; |
6bdb913f0 mm: wrap calls to... |
1009 1010 |
unsigned long mmun_start; /* For mmu_notifiers */ unsigned long mmun_end; /* For mmu_notifiers */ |
31dbd01f3 ksm: Kernel SameP... |
1011 |
|
36eaff336 mm, ksm: convert ... |
1012 1013 |
pvmw.address = page_address_in_vma(page, vma); if (pvmw.address == -EFAULT) |
31dbd01f3 ksm: Kernel SameP... |
1014 |
goto out; |
29ad768cf thp: KSM on THP |
1015 |
BUG_ON(PageTransCompound(page)); |
6bdb913f0 mm: wrap calls to... |
1016 |
|
36eaff336 mm, ksm: convert ... |
1017 1018 |
mmun_start = pvmw.address; mmun_end = pvmw.address + PAGE_SIZE; |
6bdb913f0 mm: wrap calls to... |
1019 |
mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); |
36eaff336 mm, ksm: convert ... |
1020 |
if (!page_vma_mapped_walk(&pvmw)) |
6bdb913f0 mm: wrap calls to... |
1021 |
goto out_mn; |
36eaff336 mm, ksm: convert ... |
1022 1023 |
if (WARN_ONCE(!pvmw.pte, "Unexpected PMD mapping?")) goto out_unlock; |
31dbd01f3 ksm: Kernel SameP... |
1024 |
|
595cd8f25 mm/ksm: handle pr... |
1025 |
if (pte_write(*pvmw.pte) || pte_dirty(*pvmw.pte) || |
b3a81d084 mm: fix KSM data ... |
1026 1027 |
(pte_protnone(*pvmw.pte) && pte_savedwrite(*pvmw.pte)) || mm_tlb_flush_pending(mm)) { |
31dbd01f3 ksm: Kernel SameP... |
1028 1029 1030 |
pte_t entry; swapped = PageSwapCache(page); |
36eaff336 mm, ksm: convert ... |
1031 |
flush_cache_page(vma, pvmw.address, page_to_pfn(page)); |
31dbd01f3 ksm: Kernel SameP... |
1032 |
/* |
25985edce Fix common misspe... |
1033 |
* Ok this is tricky, when get_user_pages_fast() run it doesn't |
31dbd01f3 ksm: Kernel SameP... |
1034 1035 1036 1037 1038 1039 |
* 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. |
0f10851ea mm/mmu_notifier: ... |
1040 1041 1042 1043 |
* * No need to notify as we are downgrading page table to read * only not changing it to point to a new page. * |
ad56b738c docs/vm: rename d... |
1044 |
* See Documentation/vm/mmu_notifier.rst |
31dbd01f3 ksm: Kernel SameP... |
1045 |
*/ |
0f10851ea mm/mmu_notifier: ... |
1046 |
entry = ptep_clear_flush(vma, pvmw.address, pvmw.pte); |
31dbd01f3 ksm: Kernel SameP... |
1047 1048 1049 1050 |
/* * Check that no O_DIRECT or similar I/O is in progress on the * page */ |
31e855ea7 ksm: remove redun... |
1051 |
if (page_mapcount(page) + 1 + swapped != page_count(page)) { |
36eaff336 mm, ksm: convert ... |
1052 |
set_pte_at(mm, pvmw.address, pvmw.pte, entry); |
31dbd01f3 ksm: Kernel SameP... |
1053 1054 |
goto out_unlock; } |
4e31635c3 ksm: fix bad user... |
1055 1056 |
if (pte_dirty(entry)) set_page_dirty(page); |
595cd8f25 mm/ksm: handle pr... |
1057 1058 1059 1060 1061 |
if (pte_protnone(entry)) entry = pte_mkclean(pte_clear_savedwrite(entry)); else entry = pte_mkclean(pte_wrprotect(entry)); |
36eaff336 mm, ksm: convert ... |
1062 |
set_pte_at_notify(mm, pvmw.address, pvmw.pte, entry); |
31dbd01f3 ksm: Kernel SameP... |
1063 |
} |
36eaff336 mm, ksm: convert ... |
1064 |
*orig_pte = *pvmw.pte; |
31dbd01f3 ksm: Kernel SameP... |
1065 1066 1067 |
err = 0; out_unlock: |
36eaff336 mm, ksm: convert ... |
1068 |
page_vma_mapped_walk_done(&pvmw); |
6bdb913f0 mm: wrap calls to... |
1069 1070 |
out_mn: mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
31dbd01f3 ksm: Kernel SameP... |
1071 1072 1073 1074 1075 1076 |
out: return err; } /** * replace_page - replace page in vma by new ksm page |
8dd3557a5 ksm: cleanup some... |
1077 1078 1079 |
* @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 ksm: Kernel SameP... |
1080 1081 1082 1083 |
* @orig_pte: the original value of the pte * * Returns 0 on success, -EFAULT on failure. */ |
8dd3557a5 ksm: cleanup some... |
1084 1085 |
static int replace_page(struct vm_area_struct *vma, struct page *page, struct page *kpage, pte_t orig_pte) |
31dbd01f3 ksm: Kernel SameP... |
1086 1087 |
{ struct mm_struct *mm = vma->vm_mm; |
31dbd01f3 ksm: Kernel SameP... |
1088 1089 |
pmd_t *pmd; pte_t *ptep; |
e86c59b1b mm/ksm: improve d... |
1090 |
pte_t newpte; |
31dbd01f3 ksm: Kernel SameP... |
1091 1092 |
spinlock_t *ptl; unsigned long addr; |
31dbd01f3 ksm: Kernel SameP... |
1093 |
int err = -EFAULT; |
6bdb913f0 mm: wrap calls to... |
1094 1095 |
unsigned long mmun_start; /* For mmu_notifiers */ unsigned long mmun_end; /* For mmu_notifiers */ |
31dbd01f3 ksm: Kernel SameP... |
1096 |
|
8dd3557a5 ksm: cleanup some... |
1097 |
addr = page_address_in_vma(page, vma); |
31dbd01f3 ksm: Kernel SameP... |
1098 1099 |
if (addr == -EFAULT) goto out; |
6219049ae mm: introduce mm_... |
1100 1101 |
pmd = mm_find_pmd(mm, addr); if (!pmd) |
31dbd01f3 ksm: Kernel SameP... |
1102 |
goto out; |
31dbd01f3 ksm: Kernel SameP... |
1103 |
|
6bdb913f0 mm: wrap calls to... |
1104 1105 1106 |
mmun_start = addr; mmun_end = addr + PAGE_SIZE; mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); |
31dbd01f3 ksm: Kernel SameP... |
1107 1108 1109 |
ptep = pte_offset_map_lock(mm, pmd, addr, &ptl); if (!pte_same(*ptep, orig_pte)) { pte_unmap_unlock(ptep, ptl); |
6bdb913f0 mm: wrap calls to... |
1110 |
goto out_mn; |
31dbd01f3 ksm: Kernel SameP... |
1111 |
} |
e86c59b1b mm/ksm: improve d... |
1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 |
/* * No need to check ksm_use_zero_pages here: we can only have a * zero_page here if ksm_use_zero_pages was enabled alreaady. */ if (!is_zero_pfn(page_to_pfn(kpage))) { get_page(kpage); page_add_anon_rmap(kpage, vma, addr, false); newpte = mk_pte(kpage, vma->vm_page_prot); } else { newpte = pte_mkspecial(pfn_pte(page_to_pfn(kpage), vma->vm_page_prot)); |
a38c015f3 mm/ksm.c: fix inc... |
1123 1124 1125 1126 1127 1128 1129 |
/* * We're replacing an anonymous page with a zero page, which is * not anonymous. We need to do proper accounting otherwise we * will get wrong values in /proc, and a BUG message in dmesg * when tearing down the mm. */ dec_mm_counter(mm, MM_ANONPAGES); |
e86c59b1b mm/ksm: improve d... |
1130 |
} |
31dbd01f3 ksm: Kernel SameP... |
1131 1132 |
flush_cache_page(vma, addr, pte_pfn(*ptep)); |
0f10851ea mm/mmu_notifier: ... |
1133 1134 1135 1136 |
/* * No need to notify as we are replacing a read only page with another * read only page with the same content. * |
ad56b738c docs/vm: rename d... |
1137 |
* See Documentation/vm/mmu_notifier.rst |
0f10851ea mm/mmu_notifier: ... |
1138 1139 |
*/ ptep_clear_flush(vma, addr, ptep); |
e86c59b1b mm/ksm: improve d... |
1140 |
set_pte_at_notify(mm, addr, ptep, newpte); |
31dbd01f3 ksm: Kernel SameP... |
1141 |
|
d281ee614 rmap: add argumen... |
1142 |
page_remove_rmap(page, false); |
ae52a2adb thp: ksm: free sw... |
1143 1144 |
if (!page_mapped(page)) try_to_free_swap(page); |
8dd3557a5 ksm: cleanup some... |
1145 |
put_page(page); |
31dbd01f3 ksm: Kernel SameP... |
1146 1147 1148 |
pte_unmap_unlock(ptep, ptl); err = 0; |
6bdb913f0 mm: wrap calls to... |
1149 1150 |
out_mn: mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
31dbd01f3 ksm: Kernel SameP... |
1151 1152 1153 1154 1155 1156 |
out: return err; } /* * try_to_merge_one_page - take two pages and merge them into one |
8dd3557a5 ksm: cleanup some... |
1157 1158 |
* @vma: the vma that holds the pte pointing to page * @page: the PageAnon page that we want to replace with kpage |
80e148226 ksm: share anon p... |
1159 1160 |
* @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 ksm: Kernel SameP... |
1161 1162 1163 1164 |
* * This function returns 0 if the pages were merged, -EFAULT otherwise. */ static int try_to_merge_one_page(struct vm_area_struct *vma, |
8dd3557a5 ksm: cleanup some... |
1165 |
struct page *page, struct page *kpage) |
31dbd01f3 ksm: Kernel SameP... |
1166 1167 1168 |
{ pte_t orig_pte = __pte(0); int err = -EFAULT; |
db114b83a ksm: hold anon_vm... |
1169 1170 |
if (page == kpage) /* ksm page forked */ return 0; |
8dd3557a5 ksm: cleanup some... |
1171 |
if (!PageAnon(page)) |
31dbd01f3 ksm: Kernel SameP... |
1172 |
goto out; |
31dbd01f3 ksm: Kernel SameP... |
1173 1174 1175 1176 1177 1178 1179 |
/* * 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 ksm: cleanup some... |
1180 |
if (!trylock_page(page)) |
31e855ea7 ksm: remove redun... |
1181 |
goto out; |
f765f5405 ksm: prepare to n... |
1182 1183 |
if (PageTransCompound(page)) { |
a7306c343 ksm: prevent cras... |
1184 |
if (split_huge_page(page)) |
f765f5405 ksm: prepare to n... |
1185 1186 |
goto out_unlock; } |
31dbd01f3 ksm: Kernel SameP... |
1187 1188 1189 1190 1191 1192 |
/* * 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 ksm: share anon p... |
1193 1194 1195 1196 1197 1198 1199 1200 1201 |
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 mm/ksm.c: mark st... |
1202 1203 1204 1205 1206 1207 |
/* * 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 ksm: share anon p... |
1208 1209 1210 1211 |
err = 0; } else if (pages_identical(page, kpage)) err = replace_page(vma, page, kpage, orig_pte); } |
31dbd01f3 ksm: Kernel SameP... |
1212 |
|
80e148226 ksm: share anon p... |
1213 |
if ((vma->vm_flags & VM_LOCKED) && kpage && !err) { |
73848b468 ksm: fix mlockfre... |
1214 |
munlock_vma_page(page); |
5ad646880 ksm: let shared p... |
1215 1216 |
if (!PageMlocked(kpage)) { unlock_page(page); |
5ad646880 ksm: let shared p... |
1217 1218 1219 1220 1221 |
lock_page(kpage); mlock_vma_page(kpage); page = kpage; /* for final unlock */ } } |
73848b468 ksm: fix mlockfre... |
1222 |
|
f765f5405 ksm: prepare to n... |
1223 |
out_unlock: |
8dd3557a5 ksm: cleanup some... |
1224 |
unlock_page(page); |
31dbd01f3 ksm: Kernel SameP... |
1225 1226 1227 1228 1229 |
out: return err; } /* |
81464e306 ksm: five little ... |
1230 1231 |
* 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 ksm: cleanup some... |
1232 1233 |
* * This function returns 0 if the pages were merged, -EFAULT otherwise. |
81464e306 ksm: five little ... |
1234 |
*/ |
8dd3557a5 ksm: cleanup some... |
1235 1236 |
static int try_to_merge_with_ksm_page(struct rmap_item *rmap_item, struct page *page, struct page *kpage) |
81464e306 ksm: five little ... |
1237 |
{ |
8dd3557a5 ksm: cleanup some... |
1238 |
struct mm_struct *mm = rmap_item->mm; |
81464e306 ksm: five little ... |
1239 1240 |
struct vm_area_struct *vma; int err = -EFAULT; |
8dd3557a5 ksm: cleanup some... |
1241 |
down_read(&mm->mmap_sem); |
85c6e8dd2 ksm: use find_mer... |
1242 1243 |
vma = find_mergeable_vma(mm, rmap_item->address); if (!vma) |
81464e306 ksm: five little ... |
1244 |
goto out; |
8dd3557a5 ksm: cleanup some... |
1245 |
err = try_to_merge_one_page(vma, page, kpage); |
db114b83a ksm: hold anon_vm... |
1246 1247 |
if (err) goto out; |
bc56620b4 ksm: shrink 32-bi... |
1248 1249 |
/* Unstable nid is in union with stable anon_vma: remove first */ remove_rmap_item_from_tree(rmap_item); |
db114b83a ksm: hold anon_vm... |
1250 |
/* Must get reference to anon_vma while still holding mmap_sem */ |
9e60109f1 mm: rename drop_a... |
1251 1252 |
rmap_item->anon_vma = vma->anon_vma; get_anon_vma(vma->anon_vma); |
81464e306 ksm: five little ... |
1253 |
out: |
8dd3557a5 ksm: cleanup some... |
1254 |
up_read(&mm->mmap_sem); |
81464e306 ksm: five little ... |
1255 1256 1257 1258 |
return err; } /* |
31dbd01f3 ksm: Kernel SameP... |
1259 1260 1261 |
* try_to_merge_two_pages - take two identical pages and prepare them * to be merged into one page. * |
8dd3557a5 ksm: cleanup some... |
1262 1263 |
* This function returns the kpage if we successfully merged two identical * pages into one ksm page, NULL otherwise. |
31dbd01f3 ksm: Kernel SameP... |
1264 |
* |
80e148226 ksm: share anon p... |
1265 |
* Note that this function upgrades page to ksm page: if one of the pages |
31dbd01f3 ksm: Kernel SameP... |
1266 1267 |
* is already a ksm page, try_to_merge_with_ksm_page should be used. */ |
8dd3557a5 ksm: cleanup some... |
1268 1269 1270 1271 |
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 ksm: Kernel SameP... |
1272 |
{ |
80e148226 ksm: share anon p... |
1273 |
int err; |
31dbd01f3 ksm: Kernel SameP... |
1274 |
|
80e148226 ksm: share anon p... |
1275 |
err = try_to_merge_with_ksm_page(rmap_item, page, NULL); |
31dbd01f3 ksm: Kernel SameP... |
1276 |
if (!err) { |
8dd3557a5 ksm: cleanup some... |
1277 |
err = try_to_merge_with_ksm_page(tree_rmap_item, |
80e148226 ksm: share anon p... |
1278 |
tree_page, page); |
31dbd01f3 ksm: Kernel SameP... |
1279 |
/* |
81464e306 ksm: five little ... |
1280 1281 |
* If that fails, we have a ksm page with only one pte * pointing to it: so break it. |
31dbd01f3 ksm: Kernel SameP... |
1282 |
*/ |
4035c07a8 ksm: take keyhole... |
1283 |
if (err) |
8dd3557a5 ksm: cleanup some... |
1284 |
break_cow(rmap_item); |
31dbd01f3 ksm: Kernel SameP... |
1285 |
} |
80e148226 ksm: share anon p... |
1286 |
return err ? NULL : page; |
31dbd01f3 ksm: Kernel SameP... |
1287 |
} |
2c653d0ee ksm: introduce ks... |
1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 |
static __always_inline bool __is_page_sharing_candidate(struct stable_node *stable_node, int offset) { VM_BUG_ON(stable_node->rmap_hlist_len < 0); /* * Check that at least one mapping still exists, otherwise * there's no much point to merge and share with this * stable_node, as the underlying tree_page of the other * sharer is going to be freed soon. */ return stable_node->rmap_hlist_len && stable_node->rmap_hlist_len + offset < ksm_max_page_sharing; } static __always_inline bool is_page_sharing_candidate(struct stable_node *stable_node) { return __is_page_sharing_candidate(stable_node, 0); } |
c01f0b54e mm/ksm.c: make st... |
1307 1308 1309 1310 |
static struct page *stable_node_dup(struct stable_node **_stable_node_dup, struct stable_node **_stable_node, struct rb_root *root, bool prune_stale_stable_nodes) |
2c653d0ee ksm: introduce ks... |
1311 |
{ |
b4fecc67c ksm: fix use afte... |
1312 |
struct stable_node *dup, *found = NULL, *stable_node = *_stable_node; |
2c653d0ee ksm: introduce ks... |
1313 |
struct hlist_node *hlist_safe; |
8dc5ffcd5 ksm: swap the two... |
1314 |
struct page *_tree_page, *tree_page = NULL; |
2c653d0ee ksm: introduce ks... |
1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 |
int nr = 0; int found_rmap_hlist_len; if (!prune_stale_stable_nodes || time_before(jiffies, stable_node->chain_prune_time + msecs_to_jiffies( ksm_stable_node_chains_prune_millisecs))) prune_stale_stable_nodes = false; else stable_node->chain_prune_time = jiffies; hlist_for_each_entry_safe(dup, hlist_safe, &stable_node->hlist, hlist_dup) { cond_resched(); /* * We must walk all stable_node_dup to prune the stale * stable nodes during lookup. * * get_ksm_page can drop the nodes from the * stable_node->hlist if they point to freed pages * (that's why we do a _safe walk). The "dup" * stable_node parameter itself will be freed from * under us if it returns NULL. */ _tree_page = get_ksm_page(dup, false); if (!_tree_page) continue; nr += 1; if (is_page_sharing_candidate(dup)) { if (!found || dup->rmap_hlist_len > found_rmap_hlist_len) { if (found) |
8dc5ffcd5 ksm: swap the two... |
1347 |
put_page(tree_page); |
2c653d0ee ksm: introduce ks... |
1348 1349 |
found = dup; found_rmap_hlist_len = found->rmap_hlist_len; |
8dc5ffcd5 ksm: swap the two... |
1350 |
tree_page = _tree_page; |
2c653d0ee ksm: introduce ks... |
1351 |
|
8dc5ffcd5 ksm: swap the two... |
1352 |
/* skip put_page for found dup */ |
2c653d0ee ksm: introduce ks... |
1353 1354 |
if (!prune_stale_stable_nodes) break; |
2c653d0ee ksm: introduce ks... |
1355 1356 1357 1358 1359 |
continue; } } put_page(_tree_page); } |
80b18dfa5 ksm: optimize ref... |
1360 1361 1362 1363 1364 1365 1366 1367 |
if (found) { /* * nr is counting all dups in the chain only if * prune_stale_stable_nodes is true, otherwise we may * break the loop at nr == 1 even if there are * multiple entries. */ if (prune_stale_stable_nodes && nr == 1) { |
2c653d0ee ksm: introduce ks... |
1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 |
/* * If there's not just one entry it would * corrupt memory, better BUG_ON. In KSM * context with no lock held it's not even * fatal. */ BUG_ON(stable_node->hlist.first->next); /* * There's just one entry and it is below the * deduplication limit so drop the chain. */ rb_replace_node(&stable_node->node, &found->node, root); free_stable_node(stable_node); ksm_stable_node_chains--; ksm_stable_node_dups--; |
b4fecc67c ksm: fix use afte... |
1385 |
/* |
0ba1d0f7c ksm: cleanup stab... |
1386 1387 1388 |
* NOTE: the caller depends on the stable_node * to be equal to stable_node_dup if the chain * was collapsed. |
b4fecc67c ksm: fix use afte... |
1389 |
*/ |
0ba1d0f7c ksm: cleanup stab... |
1390 1391 1392 1393 1394 1395 1396 1397 |
*_stable_node = found; /* * Just for robustneess as stable_node is * otherwise left as a stable pointer, the * compiler shall optimize it away at build * time. */ stable_node = NULL; |
80b18dfa5 ksm: optimize ref... |
1398 1399 |
} else if (stable_node->hlist.first != &found->hlist_dup && __is_page_sharing_candidate(found, 1)) { |
2c653d0ee ksm: introduce ks... |
1400 |
/* |
80b18dfa5 ksm: optimize ref... |
1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 |
* If the found stable_node dup can accept one * more future merge (in addition to the one * that is underway) and is not at the head of * the chain, put it there so next search will * be quicker in the !prune_stale_stable_nodes * case. * * NOTE: it would be inaccurate to use nr > 1 * instead of checking the hlist.first pointer * directly, because in the * prune_stale_stable_nodes case "nr" isn't * the position of the found dup in the chain, * but the total number of dups in the chain. |
2c653d0ee ksm: introduce ks... |
1414 1415 1416 1417 1418 1419 |
*/ hlist_del(&found->hlist_dup); hlist_add_head(&found->hlist_dup, &stable_node->hlist); } } |
8dc5ffcd5 ksm: swap the two... |
1420 1421 |
*_stable_node_dup = found; return tree_page; |
2c653d0ee ksm: introduce ks... |
1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 |
} static struct stable_node *stable_node_dup_any(struct stable_node *stable_node, struct rb_root *root) { if (!is_stable_node_chain(stable_node)) return stable_node; if (hlist_empty(&stable_node->hlist)) { free_stable_node_chain(stable_node, root); return NULL; } return hlist_entry(stable_node->hlist.first, typeof(*stable_node), hlist_dup); } |
8dc5ffcd5 ksm: swap the two... |
1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 |
/* * Like for get_ksm_page, this function can free the *_stable_node and * *_stable_node_dup if the returned tree_page is NULL. * * It can also free and overwrite *_stable_node with the found * stable_node_dup if the chain is collapsed (in which case * *_stable_node will be equal to *_stable_node_dup like if the chain * never existed). It's up to the caller to verify tree_page is not * NULL before dereferencing *_stable_node or *_stable_node_dup. * * *_stable_node_dup is really a second output parameter of this * function and will be overwritten in all cases, the caller doesn't * need to initialize it. */ static struct page *__stable_node_chain(struct stable_node **_stable_node_dup, struct stable_node **_stable_node, struct rb_root *root, bool prune_stale_stable_nodes) |
2c653d0ee ksm: introduce ks... |
1454 |
{ |
b4fecc67c ksm: fix use afte... |
1455 |
struct stable_node *stable_node = *_stable_node; |
2c653d0ee ksm: introduce ks... |
1456 1457 |
if (!is_stable_node_chain(stable_node)) { if (is_page_sharing_candidate(stable_node)) { |
8dc5ffcd5 ksm: swap the two... |
1458 1459 |
*_stable_node_dup = stable_node; return get_ksm_page(stable_node, false); |
2c653d0ee ksm: introduce ks... |
1460 |
} |
8dc5ffcd5 ksm: swap the two... |
1461 1462 1463 1464 1465 |
/* * _stable_node_dup set to NULL means the stable_node * reached the ksm_max_page_sharing limit. */ *_stable_node_dup = NULL; |
2c653d0ee ksm: introduce ks... |
1466 1467 |
return NULL; } |
8dc5ffcd5 ksm: swap the two... |
1468 |
return stable_node_dup(_stable_node_dup, _stable_node, root, |
2c653d0ee ksm: introduce ks... |
1469 1470 |
prune_stale_stable_nodes); } |
8dc5ffcd5 ksm: swap the two... |
1471 1472 1473 |
static __always_inline struct page *chain_prune(struct stable_node **s_n_d, struct stable_node **s_n, struct rb_root *root) |
2c653d0ee ksm: introduce ks... |
1474 |
{ |
8dc5ffcd5 ksm: swap the two... |
1475 |
return __stable_node_chain(s_n_d, s_n, root, true); |
2c653d0ee ksm: introduce ks... |
1476 |
} |
8dc5ffcd5 ksm: swap the two... |
1477 1478 1479 |
static __always_inline struct page *chain(struct stable_node **s_n_d, struct stable_node *s_n, struct rb_root *root) |
2c653d0ee ksm: introduce ks... |
1480 |
{ |
8dc5ffcd5 ksm: swap the two... |
1481 1482 1483 1484 1485 1486 1487 |
struct stable_node *old_stable_node = s_n; struct page *tree_page; tree_page = __stable_node_chain(s_n_d, &s_n, root, false); /* not pruning dups so s_n cannot have changed */ VM_BUG_ON(s_n != old_stable_node); return tree_page; |
2c653d0ee ksm: introduce ks... |
1488 |
} |
31dbd01f3 ksm: Kernel SameP... |
1489 |
/* |
8dd3557a5 ksm: cleanup some... |
1490 |
* stable_tree_search - search for page inside the stable tree |
31dbd01f3 ksm: Kernel SameP... |
1491 1492 1493 1494 |
* * 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 ksm: separate sta... |
1495 |
* This function returns the stable tree node of identical content if found, |
31dbd01f3 ksm: Kernel SameP... |
1496 1497 |
* NULL otherwise. */ |
62b61f611 ksm: memory hotre... |
1498 |
static struct page *stable_tree_search(struct page *page) |
31dbd01f3 ksm: Kernel SameP... |
1499 |
{ |
90bd6fd31 ksm: allow trees ... |
1500 |
int nid; |
ef53d16cd ksm: allocate roo... |
1501 |
struct rb_root *root; |
4146d2d67 ksm: make !merge_... |
1502 1503 |
struct rb_node **new; struct rb_node *parent; |
2c653d0ee ksm: introduce ks... |
1504 |
struct stable_node *stable_node, *stable_node_dup, *stable_node_any; |
4146d2d67 ksm: make !merge_... |
1505 |
struct stable_node *page_node; |
31dbd01f3 ksm: Kernel SameP... |
1506 |
|
4146d2d67 ksm: make !merge_... |
1507 1508 1509 |
page_node = page_stable_node(page); if (page_node && page_node->head != &migrate_nodes) { /* ksm page forked */ |
08beca44d ksm: stable_node ... |
1510 |
get_page(page); |
62b61f611 ksm: memory hotre... |
1511 |
return page; |
08beca44d ksm: stable_node ... |
1512 |
} |
90bd6fd31 ksm: allow trees ... |
1513 |
nid = get_kpfn_nid(page_to_pfn(page)); |
ef53d16cd ksm: allocate roo... |
1514 |
root = root_stable_tree + nid; |
4146d2d67 ksm: make !merge_... |
1515 |
again: |
ef53d16cd ksm: allocate roo... |
1516 |
new = &root->rb_node; |
4146d2d67 ksm: make !merge_... |
1517 |
parent = NULL; |
90bd6fd31 ksm: allow trees ... |
1518 |
|
4146d2d67 ksm: make !merge_... |
1519 |
while (*new) { |
4035c07a8 ksm: take keyhole... |
1520 |
struct page *tree_page; |
31dbd01f3 ksm: Kernel SameP... |
1521 |
int ret; |
08beca44d ksm: stable_node ... |
1522 |
cond_resched(); |
4146d2d67 ksm: make !merge_... |
1523 |
stable_node = rb_entry(*new, struct stable_node, node); |
2c653d0ee ksm: introduce ks... |
1524 |
stable_node_any = NULL; |
8dc5ffcd5 ksm: swap the two... |
1525 |
tree_page = chain_prune(&stable_node_dup, &stable_node, root); |
b4fecc67c ksm: fix use afte... |
1526 1527 1528 1529 1530 1531 |
/* * NOTE: stable_node may have been freed by * chain_prune() if the returned stable_node_dup is * not NULL. stable_node_dup may have been inserted in * the rbtree instead as a regular stable_node (in * order to collapse the stable_node chain if a single |
0ba1d0f7c ksm: cleanup stab... |
1532 1533 1534 1535 1536 |
* stable_node dup was found in it). In such case the * stable_node is overwritten by the calleee to point * to the stable_node_dup that was collapsed in the * stable rbtree and stable_node will be equal to * stable_node_dup like if the chain never existed. |
b4fecc67c ksm: fix use afte... |
1537 |
*/ |
2c653d0ee ksm: introduce ks... |
1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 |
if (!stable_node_dup) { /* * Either all stable_node dups were full in * this stable_node chain, or this chain was * empty and should be rb_erased. */ stable_node_any = stable_node_dup_any(stable_node, root); if (!stable_node_any) { /* rb_erase just run */ goto again; } /* * Take any of the stable_node dups page of * this stable_node chain to let the tree walk * continue. All KSM pages belonging to the * stable_node dups in a stable_node chain * have the same content and they're * wrprotected at all times. Any will work * fine to continue the walk. */ tree_page = get_ksm_page(stable_node_any, false); } VM_BUG_ON(!stable_node_dup ^ !!stable_node_any); |
f2e5ff85e ksm: don't fail s... |
1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 |
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 ksm: Kernel SameP... |
1574 |
|
4035c07a8 ksm: take keyhole... |
1575 |
ret = memcmp_pages(page, tree_page); |
c8d6553b9 ksm: make KSM pag... |
1576 |
put_page(tree_page); |
31dbd01f3 ksm: Kernel SameP... |
1577 |
|
4146d2d67 ksm: make !merge_... |
1578 |
parent = *new; |
c8d6553b9 ksm: make KSM pag... |
1579 |
if (ret < 0) |
4146d2d67 ksm: make !merge_... |
1580 |
new = &parent->rb_left; |
c8d6553b9 ksm: make KSM pag... |
1581 |
else if (ret > 0) |
4146d2d67 ksm: make !merge_... |
1582 |
new = &parent->rb_right; |
c8d6553b9 ksm: make KSM pag... |
1583 |
else { |
2c653d0ee ksm: introduce ks... |
1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 |
if (page_node) { VM_BUG_ON(page_node->head != &migrate_nodes); /* * Test if the migrated page should be merged * into a stable node dup. If the mapcount is * 1 we can migrate it with another KSM page * without adding it to the chain. */ if (page_mapcount(page) > 1) goto chain_append; } if (!stable_node_dup) { /* * If the stable_node is a chain and * we got a payload match in memcmp * but we cannot merge the scanned * page in any of the existing * stable_node dups because they're * all full, we need to wait the * scanned page to find itself a match * in the unstable tree to create a * brand new KSM page to add later to * the dups of this stable_node. */ return NULL; } |
c8d6553b9 ksm: make KSM pag... |
1611 1612 1613 1614 1615 1616 1617 |
/* * 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. */ |
2c653d0ee ksm: introduce ks... |
1618 1619 1620 1621 1622 1623 |
tree_page = get_ksm_page(stable_node_dup, true); if (unlikely(!tree_page)) /* * The tree may have been rebalanced, * so re-evaluate parent and new. */ |
4146d2d67 ksm: make !merge_... |
1624 |
goto again; |
2c653d0ee ksm: introduce ks... |
1625 1626 1627 1628 1629 1630 1631 1632 |
unlock_page(tree_page); if (get_kpfn_nid(stable_node_dup->kpfn) != NUMA(stable_node_dup->nid)) { put_page(tree_page); goto replace; } return tree_page; |
c8d6553b9 ksm: make KSM pag... |
1633 |
} |
31dbd01f3 ksm: Kernel SameP... |
1634 |
} |
4146d2d67 ksm: make !merge_... |
1635 1636 1637 1638 1639 1640 |
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 ksm: allocate roo... |
1641 |
rb_insert_color(&page_node->node, root); |
2c653d0ee ksm: introduce ks... |
1642 1643 1644 1645 1646 1647 |
out: if (is_page_sharing_candidate(page_node)) { get_page(page); return page; } else return NULL; |
4146d2d67 ksm: make !merge_... |
1648 1649 |
replace: |
b4fecc67c ksm: fix use afte... |
1650 1651 |
/* * If stable_node was a chain and chain_prune collapsed it, |
0ba1d0f7c ksm: cleanup stab... |
1652 1653 1654 1655 1656 |
* stable_node has been updated to be the new regular * stable_node. A collapse of the chain is indistinguishable * from the case there was no chain in the stable * rbtree. Otherwise stable_node is the chain and * stable_node_dup is the dup to replace. |
b4fecc67c ksm: fix use afte... |
1657 |
*/ |
0ba1d0f7c ksm: cleanup stab... |
1658 |
if (stable_node_dup == stable_node) { |
b4fecc67c ksm: fix use afte... |
1659 1660 |
VM_BUG_ON(is_stable_node_chain(stable_node_dup)); VM_BUG_ON(is_stable_node_dup(stable_node_dup)); |
2c653d0ee ksm: introduce ks... |
1661 1662 1663 1664 1665 |
/* there is no chain */ if (page_node) { VM_BUG_ON(page_node->head != &migrate_nodes); list_del(&page_node->list); DO_NUMA(page_node->nid = nid); |
b4fecc67c ksm: fix use afte... |
1666 1667 |
rb_replace_node(&stable_node_dup->node, &page_node->node, |
2c653d0ee ksm: introduce ks... |
1668 1669 1670 1671 1672 1673 |
root); if (is_page_sharing_candidate(page_node)) get_page(page); else page = NULL; } else { |
b4fecc67c ksm: fix use afte... |
1674 |
rb_erase(&stable_node_dup->node, root); |
2c653d0ee ksm: introduce ks... |
1675 1676 |
page = NULL; } |
4146d2d67 ksm: make !merge_... |
1677 |
} else { |
2c653d0ee ksm: introduce ks... |
1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 |
VM_BUG_ON(!is_stable_node_chain(stable_node)); __stable_node_dup_del(stable_node_dup); if (page_node) { VM_BUG_ON(page_node->head != &migrate_nodes); list_del(&page_node->list); DO_NUMA(page_node->nid = nid); stable_node_chain_add_dup(page_node, stable_node); if (is_page_sharing_candidate(page_node)) get_page(page); else page = NULL; } else { page = NULL; } |
4146d2d67 ksm: make !merge_... |
1692 |
} |
2c653d0ee ksm: introduce ks... |
1693 1694 |
stable_node_dup->head = &migrate_nodes; list_add(&stable_node_dup->list, stable_node_dup->head); |
4146d2d67 ksm: make !merge_... |
1695 |
return page; |
2c653d0ee ksm: introduce ks... |
1696 1697 1698 1699 1700 |
chain_append: /* stable_node_dup could be null if it reached the limit */ if (!stable_node_dup) stable_node_dup = stable_node_any; |
b4fecc67c ksm: fix use afte... |
1701 1702 |
/* * If stable_node was a chain and chain_prune collapsed it, |
0ba1d0f7c ksm: cleanup stab... |
1703 1704 1705 1706 1707 |
* stable_node has been updated to be the new regular * stable_node. A collapse of the chain is indistinguishable * from the case there was no chain in the stable * rbtree. Otherwise stable_node is the chain and * stable_node_dup is the dup to replace. |
b4fecc67c ksm: fix use afte... |
1708 |
*/ |
0ba1d0f7c ksm: cleanup stab... |
1709 |
if (stable_node_dup == stable_node) { |
b4fecc67c ksm: fix use afte... |
1710 1711 |
VM_BUG_ON(is_stable_node_chain(stable_node_dup)); VM_BUG_ON(is_stable_node_dup(stable_node_dup)); |
2c653d0ee ksm: introduce ks... |
1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 |
/* chain is missing so create it */ stable_node = alloc_stable_node_chain(stable_node_dup, root); if (!stable_node) return NULL; } /* * Add this stable_node dup that was * migrated to the stable_node chain * of the current nid for this page * content. */ |
b4fecc67c ksm: fix use afte... |
1724 1725 |
VM_BUG_ON(!is_stable_node_chain(stable_node)); VM_BUG_ON(!is_stable_node_dup(stable_node_dup)); |
2c653d0ee ksm: introduce ks... |
1726 1727 1728 1729 1730 |
VM_BUG_ON(page_node->head != &migrate_nodes); list_del(&page_node->list); DO_NUMA(page_node->nid = nid); stable_node_chain_add_dup(page_node, stable_node); goto out; |
31dbd01f3 ksm: Kernel SameP... |
1731 1732 1733 |
} /* |
e850dcf53 ksm: trivial tidyups |
1734 |
* stable_tree_insert - insert stable tree node pointing to new ksm page |
31dbd01f3 ksm: Kernel SameP... |
1735 1736 |
* into the stable tree. * |
7b6ba2c7d ksm: separate sta... |
1737 1738 |
* This function returns the stable tree node just allocated on success, * NULL otherwise. |
31dbd01f3 ksm: Kernel SameP... |
1739 |
*/ |
7b6ba2c7d ksm: separate sta... |
1740 |
static struct stable_node *stable_tree_insert(struct page *kpage) |
31dbd01f3 ksm: Kernel SameP... |
1741 |
{ |
90bd6fd31 ksm: allow trees ... |
1742 1743 |
int nid; unsigned long kpfn; |
ef53d16cd ksm: allocate roo... |
1744 |
struct rb_root *root; |
90bd6fd31 ksm: allow trees ... |
1745 |
struct rb_node **new; |
f2e5ff85e ksm: don't fail s... |
1746 |
struct rb_node *parent; |
2c653d0ee ksm: introduce ks... |
1747 1748 |
struct stable_node *stable_node, *stable_node_dup, *stable_node_any; bool need_chain = false; |
31dbd01f3 ksm: Kernel SameP... |
1749 |
|
90bd6fd31 ksm: allow trees ... |
1750 1751 |
kpfn = page_to_pfn(kpage); nid = get_kpfn_nid(kpfn); |
ef53d16cd ksm: allocate roo... |
1752 |
root = root_stable_tree + nid; |
f2e5ff85e ksm: don't fail s... |
1753 1754 |
again: parent = NULL; |
ef53d16cd ksm: allocate roo... |
1755 |
new = &root->rb_node; |
90bd6fd31 ksm: allow trees ... |
1756 |
|
31dbd01f3 ksm: Kernel SameP... |
1757 |
while (*new) { |
4035c07a8 ksm: take keyhole... |
1758 |
struct page *tree_page; |
31dbd01f3 ksm: Kernel SameP... |
1759 |
int ret; |
08beca44d ksm: stable_node ... |
1760 |
cond_resched(); |
7b6ba2c7d ksm: separate sta... |
1761 |
stable_node = rb_entry(*new, struct stable_node, node); |
2c653d0ee ksm: introduce ks... |
1762 |
stable_node_any = NULL; |
8dc5ffcd5 ksm: swap the two... |
1763 |
tree_page = chain(&stable_node_dup, stable_node, root); |
2c653d0ee ksm: introduce ks... |
1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 |
if (!stable_node_dup) { /* * Either all stable_node dups were full in * this stable_node chain, or this chain was * empty and should be rb_erased. */ stable_node_any = stable_node_dup_any(stable_node, root); if (!stable_node_any) { /* rb_erase just run */ goto again; } /* * Take any of the stable_node dups page of * this stable_node chain to let the tree walk * continue. All KSM pages belonging to the * stable_node dups in a stable_node chain * have the same content and they're * wrprotected at all times. Any will work * fine to continue the walk. */ tree_page = get_ksm_page(stable_node_any, false); } VM_BUG_ON(!stable_node_dup ^ !!stable_node_any); |
f2e5ff85e ksm: don't fail s... |
1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 |
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 ksm: Kernel SameP... |
1800 |
|
4035c07a8 ksm: take keyhole... |
1801 1802 |
ret = memcmp_pages(kpage, tree_page); put_page(tree_page); |
31dbd01f3 ksm: Kernel SameP... |
1803 1804 1805 1806 1807 1808 1809 |
parent = *new; if (ret < 0) new = &parent->rb_left; else if (ret > 0) new = &parent->rb_right; else { |
2c653d0ee ksm: introduce ks... |
1810 1811 |
need_chain = true; break; |
31dbd01f3 ksm: Kernel SameP... |
1812 1813 |
} } |
2c653d0ee ksm: introduce ks... |
1814 1815 |
stable_node_dup = alloc_stable_node(); if (!stable_node_dup) |
7b6ba2c7d ksm: separate sta... |
1816 |
return NULL; |
31dbd01f3 ksm: Kernel SameP... |
1817 |
|
2c653d0ee ksm: introduce ks... |
1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 |
INIT_HLIST_HEAD(&stable_node_dup->hlist); stable_node_dup->kpfn = kpfn; set_page_stable_node(kpage, stable_node_dup); stable_node_dup->rmap_hlist_len = 0; DO_NUMA(stable_node_dup->nid = nid); if (!need_chain) { rb_link_node(&stable_node_dup->node, parent, new); rb_insert_color(&stable_node_dup->node, root); } else { if (!is_stable_node_chain(stable_node)) { struct stable_node *orig = stable_node; /* chain is missing so create it */ stable_node = alloc_stable_node_chain(orig, root); if (!stable_node) { free_stable_node(stable_node_dup); return NULL; } } stable_node_chain_add_dup(stable_node_dup, stable_node); } |
08beca44d ksm: stable_node ... |
1838 |
|
2c653d0ee ksm: introduce ks... |
1839 |
return stable_node_dup; |
31dbd01f3 ksm: Kernel SameP... |
1840 1841 1842 |
} /* |
8dd3557a5 ksm: cleanup some... |
1843 1844 |
* unstable_tree_search_insert - search for identical page, * else insert rmap_item into the unstable tree. |
31dbd01f3 ksm: Kernel SameP... |
1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 |
* * 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 ksm: cleanup some... |
1856 1857 1858 1859 |
static struct rmap_item *unstable_tree_search_insert(struct rmap_item *rmap_item, struct page *page, struct page **tree_pagep) |
31dbd01f3 ksm: Kernel SameP... |
1860 |
{ |
90bd6fd31 ksm: allow trees ... |
1861 1862 |
struct rb_node **new; struct rb_root *root; |
31dbd01f3 ksm: Kernel SameP... |
1863 |
struct rb_node *parent = NULL; |
90bd6fd31 ksm: allow trees ... |
1864 1865 1866 |
int nid; nid = get_kpfn_nid(page_to_pfn(page)); |
ef53d16cd ksm: allocate roo... |
1867 |
root = root_unstable_tree + nid; |
90bd6fd31 ksm: allow trees ... |
1868 |
new = &root->rb_node; |
31dbd01f3 ksm: Kernel SameP... |
1869 1870 1871 |
while (*new) { struct rmap_item *tree_rmap_item; |
8dd3557a5 ksm: cleanup some... |
1872 |
struct page *tree_page; |
31dbd01f3 ksm: Kernel SameP... |
1873 |
int ret; |
d178f27fc ksm: cond_resched... |
1874 |
cond_resched(); |
31dbd01f3 ksm: Kernel SameP... |
1875 |
tree_rmap_item = rb_entry(*new, struct rmap_item, node); |
8dd3557a5 ksm: cleanup some... |
1876 |
tree_page = get_mergeable_page(tree_rmap_item); |
c8f95ed1a ksm: unstable_tre... |
1877 |
if (!tree_page) |
31dbd01f3 ksm: Kernel SameP... |
1878 1879 1880 |
return NULL; /* |
8dd3557a5 ksm: cleanup some... |
1881 |
* Don't substitute a ksm page for a forked page. |
31dbd01f3 ksm: Kernel SameP... |
1882 |
*/ |
8dd3557a5 ksm: cleanup some... |
1883 1884 |
if (page == tree_page) { put_page(tree_page); |
31dbd01f3 ksm: Kernel SameP... |
1885 1886 |
return NULL; } |
8dd3557a5 ksm: cleanup some... |
1887 |
ret = memcmp_pages(page, tree_page); |
31dbd01f3 ksm: Kernel SameP... |
1888 1889 1890 |
parent = *new; if (ret < 0) { |
8dd3557a5 ksm: cleanup some... |
1891 |
put_page(tree_page); |
31dbd01f3 ksm: Kernel SameP... |
1892 1893 |
new = &parent->rb_left; } else if (ret > 0) { |
8dd3557a5 ksm: cleanup some... |
1894 |
put_page(tree_page); |
31dbd01f3 ksm: Kernel SameP... |
1895 |
new = &parent->rb_right; |
b599cbdf1 ksm: treat unstab... |
1896 1897 1898 1899 1900 1901 1902 1903 1904 |
} 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 ksm: Kernel SameP... |
1905 |
} else { |
8dd3557a5 ksm: cleanup some... |
1906 |
*tree_pagep = tree_page; |
31dbd01f3 ksm: Kernel SameP... |
1907 1908 1909 |
return tree_rmap_item; } } |
7b6ba2c7d ksm: separate sta... |
1910 |
rmap_item->address |= UNSTABLE_FLAG; |
31dbd01f3 ksm: Kernel SameP... |
1911 |
rmap_item->address |= (ksm_scan.seqnr & SEQNR_MASK); |
e850dcf53 ksm: trivial tidyups |
1912 |
DO_NUMA(rmap_item->nid = nid); |
31dbd01f3 ksm: Kernel SameP... |
1913 |
rb_link_node(&rmap_item->node, parent, new); |
90bd6fd31 ksm: allow trees ... |
1914 |
rb_insert_color(&rmap_item->node, root); |
31dbd01f3 ksm: Kernel SameP... |
1915 |
|
473b0ce4d ksm: pages_unshar... |
1916 |
ksm_pages_unshared++; |
31dbd01f3 ksm: Kernel SameP... |
1917 1918 1919 1920 1921 1922 1923 1924 1925 |
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, |
2c653d0ee ksm: introduce ks... |
1926 1927 |
struct stable_node *stable_node, bool max_page_sharing_bypass) |
31dbd01f3 ksm: Kernel SameP... |
1928 |
{ |
2c653d0ee ksm: introduce ks... |
1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 |
/* * rmap won't find this mapping if we don't insert the * rmap_item in the right stable_node * duplicate. page_migration could break later if rmap breaks, * so we can as well crash here. We really need to check for * rmap_hlist_len == STABLE_NODE_CHAIN, but we can as well check * for other negative values as an undeflow if detected here * for the first time (and not when decreasing rmap_hlist_len) * would be sign of memory corruption in the stable_node. */ BUG_ON(stable_node->rmap_hlist_len < 0); stable_node->rmap_hlist_len++; if (!max_page_sharing_bypass) /* possibly non fatal but unexpected overflow, only warn */ WARN_ON_ONCE(stable_node->rmap_hlist_len > ksm_max_page_sharing); |
7b6ba2c7d ksm: separate sta... |
1946 |
rmap_item->head = stable_node; |
31dbd01f3 ksm: Kernel SameP... |
1947 |
rmap_item->address |= STABLE_FLAG; |
7b6ba2c7d ksm: separate sta... |
1948 |
hlist_add_head(&rmap_item->hlist, &stable_node->hlist); |
e178dfde3 ksm: move pages_s... |
1949 |
|
7b6ba2c7d ksm: separate sta... |
1950 1951 1952 1953 |
if (rmap_item->hlist.next) ksm_pages_sharing++; else ksm_pages_shared++; |
31dbd01f3 ksm: Kernel SameP... |
1954 1955 1956 |
} /* |
81464e306 ksm: five little ... |
1957 1958 1959 1960 |
* 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 ksm: Kernel SameP... |
1961 1962 1963 1964 1965 1966 |
* * @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) { |
4b22927f0 ksm: fix unlocked... |
1967 |
struct mm_struct *mm = rmap_item->mm; |
31dbd01f3 ksm: Kernel SameP... |
1968 |
struct rmap_item *tree_rmap_item; |
8dd3557a5 ksm: cleanup some... |
1969 |
struct page *tree_page = NULL; |
7b6ba2c7d ksm: separate sta... |
1970 |
struct stable_node *stable_node; |
8dd3557a5 ksm: cleanup some... |
1971 |
struct page *kpage; |
31dbd01f3 ksm: Kernel SameP... |
1972 1973 |
unsigned int checksum; int err; |
2c653d0ee ksm: introduce ks... |
1974 |
bool max_page_sharing_bypass = false; |
31dbd01f3 ksm: Kernel SameP... |
1975 |
|
4146d2d67 ksm: make !merge_... |
1976 1977 1978 |
stable_node = page_stable_node(page); if (stable_node) { if (stable_node->head != &migrate_nodes && |
2c653d0ee ksm: introduce ks... |
1979 1980 1981 |
get_kpfn_nid(READ_ONCE(stable_node->kpfn)) != NUMA(stable_node->nid)) { stable_node_dup_del(stable_node); |
4146d2d67 ksm: make !merge_... |
1982 1983 1984 1985 1986 1987 |
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; |
2c653d0ee ksm: introduce ks... |
1988 1989 1990 1991 1992 1993 |
/* * If it's a KSM fork, allow it to go over the sharing limit * without warnings. */ if (!is_page_sharing_candidate(stable_node)) max_page_sharing_bypass = true; |
4146d2d67 ksm: make !merge_... |
1994 |
} |
31dbd01f3 ksm: Kernel SameP... |
1995 1996 |
/* We first start with searching the page inside the stable tree */ |
62b61f611 ksm: memory hotre... |
1997 |
kpage = stable_tree_search(page); |
4146d2d67 ksm: make !merge_... |
1998 1999 2000 2001 2002 2003 |
if (kpage == page && rmap_item->head == stable_node) { put_page(kpage); return; } remove_rmap_item_from_tree(rmap_item); |
62b61f611 ksm: memory hotre... |
2004 |
if (kpage) { |
08beca44d ksm: stable_node ... |
2005 |
err = try_to_merge_with_ksm_page(rmap_item, page, kpage); |
31dbd01f3 ksm: Kernel SameP... |
2006 2007 2008 2009 2010 |
if (!err) { /* * The page was successfully merged: * add its rmap_item to the stable tree. */ |
5ad646880 ksm: let shared p... |
2011 |
lock_page(kpage); |
2c653d0ee ksm: introduce ks... |
2012 2013 |
stable_tree_append(rmap_item, page_stable_node(kpage), max_page_sharing_bypass); |
5ad646880 ksm: let shared p... |
2014 |
unlock_page(kpage); |
31dbd01f3 ksm: Kernel SameP... |
2015 |
} |
8dd3557a5 ksm: cleanup some... |
2016 |
put_page(kpage); |
31dbd01f3 ksm: Kernel SameP... |
2017 2018 2019 2020 |
return; } /* |
4035c07a8 ksm: take keyhole... |
2021 2022 2023 2024 |
* 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 ksm: Kernel SameP... |
2025 2026 2027 2028 2029 2030 |
*/ checksum = calc_checksum(page); if (rmap_item->oldchecksum != checksum) { rmap_item->oldchecksum = checksum; return; } |
e86c59b1b mm/ksm: improve d... |
2031 2032 2033 2034 2035 2036 |
/* * Same checksum as an empty page. We attempt to merge it with the * appropriate zero page if the user enabled this via sysfs. */ if (ksm_use_zero_pages && (checksum == zero_checksum)) { struct vm_area_struct *vma; |
4b22927f0 ksm: fix unlocked... |
2037 2038 |
down_read(&mm->mmap_sem); vma = find_mergeable_vma(mm, rmap_item->address); |
e86c59b1b mm/ksm: improve d... |
2039 2040 |
err = try_to_merge_one_page(vma, page, ZERO_PAGE(rmap_item->address)); |
4b22927f0 ksm: fix unlocked... |
2041 |
up_read(&mm->mmap_sem); |
e86c59b1b mm/ksm: improve d... |
2042 2043 2044 2045 2046 2047 2048 |
/* * In case of failure, the page was not really empty, so we * need to continue. Otherwise we're done. */ if (!err) return; } |
8dd3557a5 ksm: cleanup some... |
2049 2050 |
tree_rmap_item = unstable_tree_search_insert(rmap_item, page, &tree_page); |
31dbd01f3 ksm: Kernel SameP... |
2051 |
if (tree_rmap_item) { |
77da2ba06 mm/ksm: fix inter... |
2052 |
bool split; |
8dd3557a5 ksm: cleanup some... |
2053 2054 |
kpage = try_to_merge_two_pages(rmap_item, page, tree_rmap_item, tree_page); |
77da2ba06 mm/ksm: fix inter... |
2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 |
/* * If both pages we tried to merge belong to the same compound * page, then we actually ended up increasing the reference * count of the same compound page twice, and split_huge_page * failed. * Here we set a flag if that happened, and we use it later to * try split_huge_page again. Since we call put_page right * afterwards, the reference count will be correct and * split_huge_page should succeed. */ split = PageTransCompound(page) && compound_head(page) == compound_head(tree_page); |
8dd3557a5 ksm: cleanup some... |
2067 |
put_page(tree_page); |
8dd3557a5 ksm: cleanup some... |
2068 |
if (kpage) { |
bc56620b4 ksm: shrink 32-bi... |
2069 2070 2071 2072 |
/* * The pages were successfully merged: insert new * node in the stable tree and add both rmap_items. */ |
5ad646880 ksm: let shared p... |
2073 |
lock_page(kpage); |
7b6ba2c7d ksm: separate sta... |
2074 2075 |
stable_node = stable_tree_insert(kpage); if (stable_node) { |
2c653d0ee ksm: introduce ks... |
2076 2077 2078 2079 |
stable_tree_append(tree_rmap_item, stable_node, false); stable_tree_append(rmap_item, stable_node, false); |
7b6ba2c7d ksm: separate sta... |
2080 |
} |
5ad646880 ksm: let shared p... |
2081 |
unlock_page(kpage); |
7b6ba2c7d ksm: separate sta... |
2082 |
|
31dbd01f3 ksm: Kernel SameP... |
2083 2084 2085 2086 2087 2088 |
/* * 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 ksm: separate sta... |
2089 |
if (!stable_node) { |
8dd3557a5 ksm: cleanup some... |
2090 2091 |
break_cow(tree_rmap_item); break_cow(rmap_item); |
31dbd01f3 ksm: Kernel SameP... |
2092 |
} |
77da2ba06 mm/ksm: fix inter... |
2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 |
} else if (split) { /* * We are here if we tried to merge two pages and * failed because they both belonged to the same * compound page. We will split the page now, but no * merging will take place. * We do not want to add the cost of a full lock; if * the page is locked, it is better to skip it and * perhaps try again later. */ if (!trylock_page(page)) return; split_huge_page(page); unlock_page(page); |
31dbd01f3 ksm: Kernel SameP... |
2107 |
} |
31dbd01f3 ksm: Kernel SameP... |
2108 2109 2110 2111 |
} } static struct rmap_item *get_next_rmap_item(struct mm_slot *mm_slot, |
6514d511d ksm: singly-linke... |
2112 |
struct rmap_item **rmap_list, |
31dbd01f3 ksm: Kernel SameP... |
2113 2114 2115 |
unsigned long addr) { struct rmap_item *rmap_item; |
6514d511d ksm: singly-linke... |
2116 2117 |
while (*rmap_list) { rmap_item = *rmap_list; |
93d17715a ksm: three remove... |
2118 |
if ((rmap_item->address & PAGE_MASK) == addr) |
31dbd01f3 ksm: Kernel SameP... |
2119 |
return rmap_item; |
31dbd01f3 ksm: Kernel SameP... |
2120 2121 |
if (rmap_item->address > addr) break; |
6514d511d ksm: singly-linke... |
2122 |
*rmap_list = rmap_item->rmap_list; |
31dbd01f3 ksm: Kernel SameP... |
2123 |
remove_rmap_item_from_tree(rmap_item); |
31dbd01f3 ksm: Kernel SameP... |
2124 2125 2126 2127 2128 2129 2130 2131 |
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 ksm: singly-linke... |
2132 2133 |
rmap_item->rmap_list = *rmap_list; *rmap_list = rmap_item; |
31dbd01f3 ksm: Kernel SameP... |
2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 |
} 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 ksm: allow trees ... |
2144 |
int nid; |
31dbd01f3 ksm: Kernel SameP... |
2145 2146 2147 2148 2149 2150 |
if (list_empty(&ksm_mm_head.mm_list)) return NULL; slot = ksm_scan.mm_slot; if (slot == &ksm_mm_head) { |
2919bfd07 ksm: drain pageve... |
2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 |
/* * 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 ksm: make !merge_... |
2162 2163 2164 2165 2166 2167 2168 |
/* * 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 mm/ksm.c: use lis... |
2169 |
struct stable_node *stable_node, *next; |
4146d2d67 ksm: make !merge_... |
2170 |
struct page *page; |
036404183 mm/ksm.c: use lis... |
2171 2172 |
list_for_each_entry_safe(stable_node, next, &migrate_nodes, list) { |
4146d2d67 ksm: make !merge_... |
2173 2174 2175 2176 2177 2178 |
page = get_ksm_page(stable_node, false); if (page) put_page(page); cond_resched(); } } |
ef53d16cd ksm: allocate roo... |
2179 |
for (nid = 0; nid < ksm_nr_node_ids; nid++) |
90bd6fd31 ksm: allow trees ... |
2180 |
root_unstable_tree[nid] = RB_ROOT; |
31dbd01f3 ksm: Kernel SameP... |
2181 2182 2183 2184 2185 |
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 ksm: fix NULL poi... |
2186 2187 2188 2189 2190 2191 |
/* * 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 ksm: Kernel SameP... |
2192 2193 |
next_mm: ksm_scan.address = 0; |
6514d511d ksm: singly-linke... |
2194 |
ksm_scan.rmap_list = &slot->rmap_list; |
31dbd01f3 ksm: Kernel SameP... |
2195 2196 2197 2198 |
} mm = slot->mm; down_read(&mm->mmap_sem); |
9ba692948 ksm: fix oom dead... |
2199 2200 2201 2202 2203 2204 |
if (ksm_test_exit(mm)) vma = NULL; else vma = find_vma(mm, ksm_scan.address); for (; vma; vma = vma->vm_next) { |
31dbd01f3 ksm: Kernel SameP... |
2205 2206 2207 2208 2209 2210 2211 2212 |
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 ksm: fix oom dead... |
2213 2214 |
if (ksm_test_exit(mm)) break; |
31dbd01f3 ksm: Kernel SameP... |
2215 |
*page = follow_page(vma, ksm_scan.address, FOLL_GET); |
21ae5b017 thp: skip transhu... |
2216 2217 2218 2219 2220 |
if (IS_ERR_OR_NULL(*page)) { ksm_scan.address += PAGE_SIZE; cond_resched(); continue; } |
f765f5405 ksm: prepare to n... |
2221 |
if (PageAnon(*page)) { |
31dbd01f3 ksm: Kernel SameP... |
2222 2223 2224 |
flush_anon_page(vma, *page, ksm_scan.address); flush_dcache_page(*page); rmap_item = get_next_rmap_item(slot, |
6514d511d ksm: singly-linke... |
2225 |
ksm_scan.rmap_list, ksm_scan.address); |
31dbd01f3 ksm: Kernel SameP... |
2226 |
if (rmap_item) { |
6514d511d ksm: singly-linke... |
2227 2228 |
ksm_scan.rmap_list = &rmap_item->rmap_list; |
31dbd01f3 ksm: Kernel SameP... |
2229 2230 2231 2232 2233 2234 |
ksm_scan.address += PAGE_SIZE; } else put_page(*page); up_read(&mm->mmap_sem); return rmap_item; } |
21ae5b017 thp: skip transhu... |
2235 |
put_page(*page); |
31dbd01f3 ksm: Kernel SameP... |
2236 2237 2238 2239 |
ksm_scan.address += PAGE_SIZE; cond_resched(); } } |
9ba692948 ksm: fix oom dead... |
2240 2241 |
if (ksm_test_exit(mm)) { ksm_scan.address = 0; |
6514d511d ksm: singly-linke... |
2242 |
ksm_scan.rmap_list = &slot->rmap_list; |
9ba692948 ksm: fix oom dead... |
2243 |
} |
31dbd01f3 ksm: Kernel SameP... |
2244 2245 2246 2247 |
/* * Nuke all the rmap_items that are above this current rmap: * because there were no VM_MERGEABLE vmas with such addresses. */ |
6514d511d ksm: singly-linke... |
2248 |
remove_trailing_rmap_items(slot, ksm_scan.rmap_list); |
31dbd01f3 ksm: Kernel SameP... |
2249 2250 |
spin_lock(&ksm_mmlist_lock); |
cd551f975 ksm: distribute r... |
2251 2252 2253 2254 2255 2256 2257 |
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 ksm: fix oom dead... |
2258 2259 2260 2261 |
* 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 ksm: distribute r... |
2262 |
*/ |
4ca3a69bc mm/ksm.c: use new... |
2263 |
hash_del(&slot->link); |
cd551f975 ksm: distribute r... |
2264 |
list_del(&slot->mm_list); |
9ba692948 ksm: fix oom dead... |
2265 |
spin_unlock(&ksm_mmlist_lock); |
cd551f975 ksm: distribute r... |
2266 2267 |
free_mm_slot(slot); clear_bit(MMF_VM_MERGEABLE, &mm->flags); |
9ba692948 ksm: fix oom dead... |
2268 2269 2270 |
up_read(&mm->mmap_sem); mmdrop(mm); } else { |
9ba692948 ksm: fix oom dead... |
2271 |
up_read(&mm->mmap_sem); |
7496fea9a ksm: fix conflict... |
2272 2273 2274 2275 2276 2277 2278 2279 |
/* * 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 ksm: distribute r... |
2280 |
} |
31dbd01f3 ksm: Kernel SameP... |
2281 2282 |
/* Repeat until we've completed scanning the whole list */ |
cd551f975 ksm: distribute r... |
2283 |
slot = ksm_scan.mm_slot; |
31dbd01f3 ksm: Kernel SameP... |
2284 2285 |
if (slot != &ksm_mm_head) goto next_mm; |
31dbd01f3 ksm: Kernel SameP... |
2286 2287 2288 2289 2290 2291 |
ksm_scan.seqnr++; return NULL; } /** * ksm_do_scan - the ksm scanner main worker function. |
b7701a5f2 mm: docs: fixup p... |
2292 |
* @scan_npages: number of pages we want to scan before we return. |
31dbd01f3 ksm: Kernel SameP... |
2293 2294 2295 2296 |
*/ static void ksm_do_scan(unsigned int scan_npages) { struct rmap_item *rmap_item; |
22eccdd7d ksm: check for ER... |
2297 |
struct page *uninitialized_var(page); |
31dbd01f3 ksm: Kernel SameP... |
2298 |
|
878aee7d6 thp: freeze khuge... |
2299 |
while (scan_npages-- && likely(!freezing(current))) { |
31dbd01f3 ksm: Kernel SameP... |
2300 2301 2302 2303 |
cond_resched(); rmap_item = scan_get_next_rmap_item(&page); if (!rmap_item) return; |
4146d2d67 ksm: make !merge_... |
2304 |
cmp_and_merge_page(page, rmap_item); |
31dbd01f3 ksm: Kernel SameP... |
2305 2306 2307 |
put_page(page); } } |
6e1583842 ksm: keep quiet w... |
2308 2309 2310 2311 |
static int ksmd_should_run(void) { return (ksm_run & KSM_RUN_MERGE) && !list_empty(&ksm_mm_head.mm_list); } |
31dbd01f3 ksm: Kernel SameP... |
2312 2313 |
static int ksm_scan_thread(void *nothing) { |
878aee7d6 thp: freeze khuge... |
2314 |
set_freezable(); |
339aa6246 ksm: change ksm n... |
2315 |
set_user_nice(current, 5); |
31dbd01f3 ksm: Kernel SameP... |
2316 2317 |
while (!kthread_should_stop()) { |
6e1583842 ksm: keep quiet w... |
2318 |
mutex_lock(&ksm_thread_mutex); |
ef4d43a80 ksm: stop hotremo... |
2319 |
wait_while_offlining(); |
6e1583842 ksm: keep quiet w... |
2320 |
if (ksmd_should_run()) |
31dbd01f3 ksm: Kernel SameP... |
2321 |
ksm_do_scan(ksm_thread_pages_to_scan); |
6e1583842 ksm: keep quiet w... |
2322 |
mutex_unlock(&ksm_thread_mutex); |
878aee7d6 thp: freeze khuge... |
2323 |
try_to_freeze(); |
6e1583842 ksm: keep quiet w... |
2324 |
if (ksmd_should_run()) { |
31dbd01f3 ksm: Kernel SameP... |
2325 2326 2327 |
schedule_timeout_interruptible( msecs_to_jiffies(ksm_thread_sleep_millisecs)); } else { |
878aee7d6 thp: freeze khuge... |
2328 |
wait_event_freezable(ksm_thread_wait, |
6e1583842 ksm: keep quiet w... |
2329 |
ksmd_should_run() || kthread_should_stop()); |
31dbd01f3 ksm: Kernel SameP... |
2330 2331 2332 2333 |
} } return 0; } |
f8af4da3b ksm: the mm inter... |
2334 2335 2336 2337 |
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 ksm: fix endless ... |
2338 |
int err; |
f8af4da3b ksm: the mm inter... |
2339 2340 2341 2342 2343 2344 2345 2346 |
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 mm: remove rest u... |
2347 |
VM_HUGETLB | VM_MIXEDMAP)) |
f8af4da3b ksm: the mm inter... |
2348 |
return 0; /* just ignore the advice */ |
e1fb4a086 dax: remove VM_MI... |
2349 2350 |
if (vma_is_dax(vma)) return 0; |
cc2383ec0 mm: introduce arc... |
2351 2352 2353 2354 |
#ifdef VM_SAO if (*vm_flags & VM_SAO) return 0; #endif |
74a049674 sparc64: Add supp... |
2355 2356 2357 2358 |
#ifdef VM_SPARC_ADI if (*vm_flags & VM_SPARC_ADI) return 0; #endif |
cc2383ec0 mm: introduce arc... |
2359 |
|
d952b7913 ksm: fix endless ... |
2360 2361 2362 2363 2364 |
if (!test_bit(MMF_VM_MERGEABLE, &mm->flags)) { err = __ksm_enter(mm); if (err) return err; } |
f8af4da3b ksm: the mm inter... |
2365 2366 2367 2368 2369 2370 2371 |
*vm_flags |= VM_MERGEABLE; break; case MADV_UNMERGEABLE: if (!(*vm_flags & VM_MERGEABLE)) return 0; /* just ignore the advice */ |
d952b7913 ksm: fix endless ... |
2372 2373 2374 2375 2376 |
if (vma->anon_vma) { err = unmerge_ksm_pages(vma, start, end); if (err) return err; } |
f8af4da3b ksm: the mm inter... |
2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 |
*vm_flags &= ~VM_MERGEABLE; break; } return 0; } int __ksm_enter(struct mm_struct *mm) { |
6e1583842 ksm: keep quiet w... |
2387 2388 2389 2390 |
struct mm_slot *mm_slot; int needs_wakeup; mm_slot = alloc_mm_slot(); |
31dbd01f3 ksm: Kernel SameP... |
2391 2392 |
if (!mm_slot) return -ENOMEM; |
6e1583842 ksm: keep quiet w... |
2393 2394 |
/* Check ksm_run too? Would need tighter locking */ needs_wakeup = list_empty(&ksm_mm_head.mm_list); |
31dbd01f3 ksm: Kernel SameP... |
2395 2396 2397 |
spin_lock(&ksm_mmlist_lock); insert_to_mm_slots_hash(mm, mm_slot); /* |
cbf86cfe0 ksm: remove old s... |
2398 2399 |
* When KSM_RUN_MERGE (or KSM_RUN_STOP), * insert just behind the scanning cursor, to let the area settle |
31dbd01f3 ksm: Kernel SameP... |
2400 2401 |
* 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 ksm: remove old s... |
2402 2403 2404 2405 |
* * 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 ksm: Kernel SameP... |
2406 |
*/ |
cbf86cfe0 ksm: remove old s... |
2407 2408 2409 2410 |
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 ksm: Kernel SameP... |
2411 |
spin_unlock(&ksm_mmlist_lock); |
f8af4da3b ksm: the mm inter... |
2412 |
set_bit(MMF_VM_MERGEABLE, &mm->flags); |
f1f100764 mm: add new mmgra... |
2413 |
mmgrab(mm); |
6e1583842 ksm: keep quiet w... |
2414 2415 2416 |
if (needs_wakeup) wake_up_interruptible(&ksm_thread_wait); |
f8af4da3b ksm: the mm inter... |
2417 2418 |
return 0; } |
1c2fb7a4c ksm: fix deadlock... |
2419 |
void __ksm_exit(struct mm_struct *mm) |
f8af4da3b ksm: the mm inter... |
2420 |
{ |
cd551f975 ksm: distribute r... |
2421 |
struct mm_slot *mm_slot; |
9ba692948 ksm: fix oom dead... |
2422 |
int easy_to_free = 0; |
cd551f975 ksm: distribute r... |
2423 |
|
31dbd01f3 ksm: Kernel SameP... |
2424 |
/* |
9ba692948 ksm: fix oom dead... |
2425 2426 2427 2428 2429 2430 |
* 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 ksm: Kernel SameP... |
2431 |
*/ |
9ba692948 ksm: fix oom dead... |
2432 |
|
cd551f975 ksm: distribute r... |
2433 2434 |
spin_lock(&ksm_mmlist_lock); mm_slot = get_mm_slot(mm); |
9ba692948 ksm: fix oom dead... |
2435 |
if (mm_slot && ksm_scan.mm_slot != mm_slot) { |
6514d511d ksm: singly-linke... |
2436 |
if (!mm_slot->rmap_list) { |
4ca3a69bc mm/ksm.c: use new... |
2437 |
hash_del(&mm_slot->link); |
9ba692948 ksm: fix oom dead... |
2438 2439 2440 2441 2442 2443 |
list_del(&mm_slot->mm_list); easy_to_free = 1; } else { list_move(&mm_slot->mm_list, &ksm_scan.mm_slot->mm_list); } |
cd551f975 ksm: distribute r... |
2444 |
} |
cd551f975 ksm: distribute r... |
2445 |
spin_unlock(&ksm_mmlist_lock); |
9ba692948 ksm: fix oom dead... |
2446 2447 2448 2449 2450 |
if (easy_to_free) { free_mm_slot(mm_slot); clear_bit(MMF_VM_MERGEABLE, &mm->flags); mmdrop(mm); } else if (mm_slot) { |
9ba692948 ksm: fix oom dead... |
2451 2452 |
down_write(&mm->mmap_sem); up_write(&mm->mmap_sem); |
9ba692948 ksm: fix oom dead... |
2453 |
} |
31dbd01f3 ksm: Kernel SameP... |
2454 |
} |
cbf86cfe0 ksm: remove old s... |
2455 |
struct page *ksm_might_need_to_copy(struct page *page, |
5ad646880 ksm: let shared p... |
2456 2457 |
struct vm_area_struct *vma, unsigned long address) { |
cbf86cfe0 ksm: remove old s... |
2458 |
struct anon_vma *anon_vma = page_anon_vma(page); |
5ad646880 ksm: let shared p... |
2459 |
struct page *new_page; |
cbf86cfe0 ksm: remove old s... |
2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 |
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 ksm: let shared p... |
2472 2473 2474 2475 2476 2477 |
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 page-flags: defin... |
2478 |
__SetPageLocked(new_page); |
5ad646880 ksm: let shared p... |
2479 |
} |
5ad646880 ksm: let shared p... |
2480 2481 |
return new_page; } |
1df631ae1 mm: make rmap_wal... |
2482 |
void rmap_walk_ksm(struct page *page, struct rmap_walk_control *rwc) |
e9995ef97 ksm: rmap_walk to... |
2483 2484 |
{ struct stable_node *stable_node; |
e9995ef97 ksm: rmap_walk to... |
2485 |
struct rmap_item *rmap_item; |
e9995ef97 ksm: rmap_walk to... |
2486 |
int search_new_forks = 0; |
309381fea mm: dump page whe... |
2487 |
VM_BUG_ON_PAGE(!PageKsm(page), page); |
9f32624be mm/rmap: use rmap... |
2488 2489 2490 2491 2492 |
/* * Rely on the page lock to protect against concurrent modifications * to that page's node of the stable tree. */ |
309381fea mm: dump page whe... |
2493 |
VM_BUG_ON_PAGE(!PageLocked(page), page); |
e9995ef97 ksm: rmap_walk to... |
2494 2495 2496 |
stable_node = page_stable_node(page); if (!stable_node) |
1df631ae1 mm: make rmap_wal... |
2497 |
return; |
e9995ef97 ksm: rmap_walk to... |
2498 |
again: |
b67bfe0d4 hlist: drop the n... |
2499 |
hlist_for_each_entry(rmap_item, &stable_node->hlist, hlist) { |
e9995ef97 ksm: rmap_walk to... |
2500 |
struct anon_vma *anon_vma = rmap_item->anon_vma; |
5beb49305 mm: change anon_v... |
2501 |
struct anon_vma_chain *vmac; |
e9995ef97 ksm: rmap_walk to... |
2502 |
struct vm_area_struct *vma; |
ad12695f1 ksm: add cond_res... |
2503 |
cond_resched(); |
b6b19f25f ksm: make rmap wa... |
2504 |
anon_vma_lock_read(anon_vma); |
bf181b9f9 mm anon rmap: rep... |
2505 2506 |
anon_vma_interval_tree_foreach(vmac, &anon_vma->rb_root, 0, ULONG_MAX) { |
1105a2fc0 mm/ksm.c: ignore ... |
2507 |
unsigned long addr; |
ad12695f1 ksm: add cond_res... |
2508 |
cond_resched(); |
5beb49305 mm: change anon_v... |
2509 |
vma = vmac->vma; |
1105a2fc0 mm/ksm.c: ignore ... |
2510 2511 2512 2513 2514 |
/* Ignore the stable/unstable/sqnr flags */ addr = rmap_item->address & ~KSM_FLAG_MASK; if (addr < vma->vm_start || addr >= vma->vm_end) |
e9995ef97 ksm: rmap_walk to... |
2515 2516 2517 2518 2519 2520 2521 2522 2523 |
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 mm/rmap: extend r... |
2524 2525 |
if (rwc->invalid_vma && rwc->invalid_vma(vma, rwc->arg)) continue; |
1105a2fc0 mm/ksm.c: ignore ... |
2526 |
if (!rwc->rmap_one(page, vma, addr, rwc->arg)) { |
b6b19f25f ksm: make rmap wa... |
2527 |
anon_vma_unlock_read(anon_vma); |
1df631ae1 mm: make rmap_wal... |
2528 |
return; |
e9995ef97 ksm: rmap_walk to... |
2529 |
} |
0dd1c7bbc mm/rmap: extend r... |
2530 2531 |
if (rwc->done && rwc->done(page)) { anon_vma_unlock_read(anon_vma); |
1df631ae1 mm: make rmap_wal... |
2532 |
return; |
0dd1c7bbc mm/rmap: extend r... |
2533 |
} |
e9995ef97 ksm: rmap_walk to... |
2534 |
} |
b6b19f25f ksm: make rmap wa... |
2535 |
anon_vma_unlock_read(anon_vma); |
e9995ef97 ksm: rmap_walk to... |
2536 2537 2538 |
} if (!search_new_forks++) goto again; |
e9995ef97 ksm: rmap_walk to... |
2539 |
} |
526295064 mm/rmap: use rmap... |
2540 |
#ifdef CONFIG_MIGRATION |
e9995ef97 ksm: rmap_walk to... |
2541 2542 2543 |
void ksm_migrate_page(struct page *newpage, struct page *oldpage) { struct stable_node *stable_node; |
309381fea mm: dump page whe... |
2544 2545 2546 |
VM_BUG_ON_PAGE(!PageLocked(oldpage), oldpage); VM_BUG_ON_PAGE(!PageLocked(newpage), newpage); VM_BUG_ON_PAGE(newpage->mapping != oldpage->mapping, newpage); |
e9995ef97 ksm: rmap_walk to... |
2547 2548 2549 |
stable_node = page_stable_node(newpage); if (stable_node) { |
309381fea mm: dump page whe... |
2550 |
VM_BUG_ON_PAGE(stable_node->kpfn != page_to_pfn(oldpage), oldpage); |
62b61f611 ksm: memory hotre... |
2551 |
stable_node->kpfn = page_to_pfn(newpage); |
c8d6553b9 ksm: make KSM pag... |
2552 2553 2554 2555 2556 2557 2558 2559 |
/* * 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 ksm: rmap_walk to... |
2560 2561 2562 |
} } #endif /* CONFIG_MIGRATION */ |
62b61f611 ksm: memory hotre... |
2563 |
#ifdef CONFIG_MEMORY_HOTREMOVE |
ef4d43a80 ksm: stop hotremo... |
2564 2565 2566 2567 2568 |
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 sched: Remove pro... |
2569 |
TASK_UNINTERRUPTIBLE); |
ef4d43a80 ksm: stop hotremo... |
2570 2571 2572 |
mutex_lock(&ksm_thread_mutex); } } |
2c653d0ee ksm: introduce ks... |
2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 |
static bool stable_node_dup_remove_range(struct stable_node *stable_node, unsigned long start_pfn, unsigned long end_pfn) { if (stable_node->kpfn >= start_pfn && 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); return true; } return false; } static bool stable_node_chain_remove_range(struct stable_node *stable_node, unsigned long start_pfn, unsigned long end_pfn, struct rb_root *root) { struct stable_node *dup; struct hlist_node *hlist_safe; if (!is_stable_node_chain(stable_node)) { VM_BUG_ON(is_stable_node_dup(stable_node)); return stable_node_dup_remove_range(stable_node, start_pfn, end_pfn); } hlist_for_each_entry_safe(dup, hlist_safe, &stable_node->hlist, hlist_dup) { VM_BUG_ON(!is_stable_node_dup(dup)); stable_node_dup_remove_range(dup, start_pfn, end_pfn); } if (hlist_empty(&stable_node->hlist)) { free_stable_node_chain(stable_node, root); return true; /* notify caller that tree was rebalanced */ } else return false; } |
ee0ea59cf ksm: reorganize k... |
2614 2615 |
static void ksm_check_stable_tree(unsigned long start_pfn, unsigned long end_pfn) |
62b61f611 ksm: memory hotre... |
2616 |
{ |
036404183 mm/ksm.c: use lis... |
2617 |
struct stable_node *stable_node, *next; |
62b61f611 ksm: memory hotre... |
2618 |
struct rb_node *node; |
90bd6fd31 ksm: allow trees ... |
2619 |
int nid; |
62b61f611 ksm: memory hotre... |
2620 |
|
ef53d16cd ksm: allocate roo... |
2621 2622 |
for (nid = 0; nid < ksm_nr_node_ids; nid++) { node = rb_first(root_stable_tree + nid); |
ee0ea59cf ksm: reorganize k... |
2623 |
while (node) { |
90bd6fd31 ksm: allow trees ... |
2624 |
stable_node = rb_entry(node, struct stable_node, node); |
2c653d0ee ksm: introduce ks... |
2625 2626 2627 2628 |
if (stable_node_chain_remove_range(stable_node, start_pfn, end_pfn, root_stable_tree + nid)) |
ef53d16cd ksm: allocate roo... |
2629 |
node = rb_first(root_stable_tree + nid); |
2c653d0ee ksm: introduce ks... |
2630 |
else |
ee0ea59cf ksm: reorganize k... |
2631 2632 |
node = rb_next(node); cond_resched(); |
90bd6fd31 ksm: allow trees ... |
2633 |
} |
ee0ea59cf ksm: reorganize k... |
2634 |
} |
036404183 mm/ksm.c: use lis... |
2635 |
list_for_each_entry_safe(stable_node, next, &migrate_nodes, list) { |
4146d2d67 ksm: make !merge_... |
2636 2637 2638 2639 2640 |
if (stable_node->kpfn >= start_pfn && stable_node->kpfn < end_pfn) remove_node_from_stable_tree(stable_node); cond_resched(); } |
62b61f611 ksm: memory hotre... |
2641 2642 2643 2644 2645 2646 |
} static int ksm_memory_callback(struct notifier_block *self, unsigned long action, void *arg) { struct memory_notify *mn = arg; |
62b61f611 ksm: memory hotre... |
2647 2648 2649 2650 |
switch (action) { case MEM_GOING_OFFLINE: /* |
ef4d43a80 ksm: stop hotremo... |
2651 2652 2653 2654 2655 |
* 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 ksm: memory hotre... |
2656 |
*/ |
ef4d43a80 ksm: stop hotremo... |
2657 2658 2659 |
mutex_lock(&ksm_thread_mutex); ksm_run |= KSM_RUN_OFFLINE; mutex_unlock(&ksm_thread_mutex); |
62b61f611 ksm: memory hotre... |
2660 2661 2662 2663 2664 2665 |
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 ksm: reorganize k... |
2666 2667 2668 |
* 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 ksm: memory hotre... |
2669 |
*/ |
ee0ea59cf ksm: reorganize k... |
2670 2671 |
ksm_check_stable_tree(mn->start_pfn, mn->start_pfn + mn->nr_pages); |
62b61f611 ksm: memory hotre... |
2672 2673 2674 |
/* fallthrough */ case MEM_CANCEL_OFFLINE: |
ef4d43a80 ksm: stop hotremo... |
2675 2676 |
mutex_lock(&ksm_thread_mutex); ksm_run &= ~KSM_RUN_OFFLINE; |
62b61f611 ksm: memory hotre... |
2677 |
mutex_unlock(&ksm_thread_mutex); |
ef4d43a80 ksm: stop hotremo... |
2678 2679 2680 |
smp_mb(); /* wake_up_bit advises this */ wake_up_bit(&ksm_run, ilog2(KSM_RUN_OFFLINE)); |
62b61f611 ksm: memory hotre... |
2681 2682 2683 2684 |
break; } return NOTIFY_OK; } |
ef4d43a80 ksm: stop hotremo... |
2685 2686 2687 2688 |
#else static void wait_while_offlining(void) { } |
62b61f611 ksm: memory hotre... |
2689 |
#endif /* CONFIG_MEMORY_HOTREMOVE */ |
2ffd8679c ksm: sysfs and de... |
2690 2691 2692 2693 |
#ifdef CONFIG_SYSFS /* * This all compiles without CONFIG_SYSFS, but is a waste of space. */ |
31dbd01f3 ksm: Kernel SameP... |
2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 |
#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 mm: replace stric... |
2713 |
err = kstrtoul(buf, 10, &msecs); |
31dbd01f3 ksm: Kernel SameP... |
2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 |
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 mm: replace stric... |
2736 |
err = kstrtoul(buf, 10, &nr_pages); |
31dbd01f3 ksm: Kernel SameP... |
2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 |
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 ksm: stop hotremo... |
2749 2750 |
return sprintf(buf, "%lu ", ksm_run); |
31dbd01f3 ksm: Kernel SameP... |
2751 2752 2753 2754 2755 2756 2757 |
} static ssize_t run_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { int err; unsigned long flags; |
3dbb95f78 mm: replace stric... |
2758 |
err = kstrtoul(buf, 10, &flags); |
31dbd01f3 ksm: Kernel SameP... |
2759 2760 2761 2762 2763 2764 2765 2766 |
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 ksm: remove unswa... |
2767 2768 |
* breaking COW to free the pages_shared (but leaves mm_slots * on the list for when ksmd may be set running again). |
31dbd01f3 ksm: Kernel SameP... |
2769 2770 2771 |
*/ mutex_lock(&ksm_thread_mutex); |
ef4d43a80 ksm: stop hotremo... |
2772 |
wait_while_offlining(); |
31dbd01f3 ksm: Kernel SameP... |
2773 2774 |
if (ksm_run != flags) { ksm_run = flags; |
d952b7913 ksm: fix endless ... |
2775 |
if (flags & KSM_RUN_UNMERGE) { |
e1e12d2f3 mm, oom: fix race... |
2776 |
set_current_oom_origin(); |
d952b7913 ksm: fix endless ... |
2777 |
err = unmerge_and_remove_all_rmap_items(); |
e1e12d2f3 mm, oom: fix race... |
2778 |
clear_current_oom_origin(); |
d952b7913 ksm: fix endless ... |
2779 2780 2781 2782 2783 |
if (err) { ksm_run = KSM_RUN_STOP; count = err; } } |
31dbd01f3 ksm: Kernel SameP... |
2784 2785 2786 2787 2788 2789 2790 2791 2792 |
} mutex_unlock(&ksm_thread_mutex); if (flags & KSM_RUN_MERGE) wake_up_interruptible(&ksm_thread_wait); return count; } KSM_ATTR(run); |
90bd6fd31 ksm: allow trees ... |
2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 |
#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 ksm: stop hotremo... |
2815 |
wait_while_offlining(); |
90bd6fd31 ksm: allow trees ... |
2816 |
if (ksm_merge_across_nodes != knob) { |
cbf86cfe0 ksm: remove old s... |
2817 |
if (ksm_pages_shared || remove_all_stable_nodes()) |
90bd6fd31 ksm: allow trees ... |
2818 |
err = -EBUSY; |
ef53d16cd ksm: allocate roo... |
2819 2820 2821 2822 2823 2824 2825 2826 2827 |
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 ksm: remove redun... |
2828 2829 |
buf = kcalloc(nr_node_ids + nr_node_ids, sizeof(*buf), GFP_KERNEL); |
ef53d16cd ksm: allocate roo... |
2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 |
/* 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 ksm: allow trees ... |
2841 |
ksm_merge_across_nodes = knob; |
ef53d16cd ksm: allocate roo... |
2842 2843 |
ksm_nr_node_ids = knob ? 1 : nr_node_ids; } |
90bd6fd31 ksm: allow trees ... |
2844 2845 2846 2847 2848 2849 2850 |
} mutex_unlock(&ksm_thread_mutex); return err ? err : count; } KSM_ATTR(merge_across_nodes); #endif |
e86c59b1b mm/ksm: improve d... |
2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 |
static ssize_t use_zero_pages_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return sprintf(buf, "%u ", ksm_use_zero_pages); } static ssize_t use_zero_pages_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { int err; bool value; err = kstrtobool(buf, &value); if (err) return -EINVAL; ksm_use_zero_pages = value; return count; } KSM_ATTR(use_zero_pages); |
2c653d0ee ksm: introduce ks... |
2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 |
static ssize_t max_page_sharing_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return sprintf(buf, "%u ", ksm_max_page_sharing); } static ssize_t max_page_sharing_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { int err; int knob; err = kstrtoint(buf, 10, &knob); if (err) return err; /* * When a KSM page is created it is shared by 2 mappings. This * being a signed comparison, it implicitly verifies it's not * negative. */ if (knob < 2) return -EINVAL; if (READ_ONCE(ksm_max_page_sharing) == knob) return count; mutex_lock(&ksm_thread_mutex); wait_while_offlining(); if (ksm_max_page_sharing != knob) { if (ksm_pages_shared || remove_all_stable_nodes()) err = -EBUSY; else ksm_max_page_sharing = knob; } mutex_unlock(&ksm_thread_mutex); return err ? err : count; } KSM_ATTR(max_page_sharing); |
b40282603 ksm: rename kerne... |
2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 |
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 ksm: move pages_s... |
2925 2926 |
return sprintf(buf, "%lu ", ksm_pages_sharing); |
b40282603 ksm: rename kerne... |
2927 2928 |
} KSM_ATTR_RO(pages_sharing); |
473b0ce4d ksm: pages_unshar... |
2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 |
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); |
2c653d0ee ksm: introduce ks... |
2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 |
static ssize_t stable_node_dups_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return sprintf(buf, "%lu ", ksm_stable_node_dups); } KSM_ATTR_RO(stable_node_dups); static ssize_t stable_node_chains_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return sprintf(buf, "%lu ", ksm_stable_node_chains); } KSM_ATTR_RO(stable_node_chains); static ssize_t stable_node_chains_prune_millisecs_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return sprintf(buf, "%u ", ksm_stable_node_chains_prune_millisecs); } static ssize_t stable_node_chains_prune_millisecs_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { unsigned long msecs; int err; err = kstrtoul(buf, 10, &msecs); if (err || msecs > UINT_MAX) return -EINVAL; ksm_stable_node_chains_prune_millisecs = msecs; return count; } KSM_ATTR(stable_node_chains_prune_millisecs); |
473b0ce4d ksm: pages_unshar... |
2996 2997 2998 2999 3000 3001 3002 |
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 ksm: Kernel SameP... |
3003 3004 3005 3006 |
static struct attribute *ksm_attrs[] = { &sleep_millisecs_attr.attr, &pages_to_scan_attr.attr, &run_attr.attr, |
b40282603 ksm: rename kerne... |
3007 3008 |
&pages_shared_attr.attr, &pages_sharing_attr.attr, |
473b0ce4d ksm: pages_unshar... |
3009 3010 3011 |
&pages_unshared_attr.attr, &pages_volatile_attr.attr, &full_scans_attr.attr, |
90bd6fd31 ksm: allow trees ... |
3012 3013 3014 |
#ifdef CONFIG_NUMA &merge_across_nodes_attr.attr, #endif |
2c653d0ee ksm: introduce ks... |
3015 3016 3017 3018 |
&max_page_sharing_attr.attr, &stable_node_chains_attr.attr, &stable_node_dups_attr.attr, &stable_node_chains_prune_millisecs_attr.attr, |
e86c59b1b mm/ksm: improve d... |
3019 |
&use_zero_pages_attr.attr, |
31dbd01f3 ksm: Kernel SameP... |
3020 3021 |
NULL, }; |
f907c26a9 mm/ksm.c: constif... |
3022 |
static const struct attribute_group ksm_attr_group = { |
31dbd01f3 ksm: Kernel SameP... |
3023 3024 3025 |
.attrs = ksm_attrs, .name = "ksm", }; |
2ffd8679c ksm: sysfs and de... |
3026 |
#endif /* CONFIG_SYSFS */ |
31dbd01f3 ksm: Kernel SameP... |
3027 3028 3029 3030 3031 |
static int __init ksm_init(void) { struct task_struct *ksm_thread; int err; |
e86c59b1b mm/ksm: improve d... |
3032 3033 3034 3035 |
/* The correct value depends on page size and endianness */ zero_checksum = calc_checksum(ZERO_PAGE(0)); /* Default to false for backwards compatibility */ ksm_use_zero_pages = false; |
31dbd01f3 ksm: Kernel SameP... |
3036 3037 3038 |
err = ksm_slab_init(); if (err) goto out; |
31dbd01f3 ksm: Kernel SameP... |
3039 3040 |
ksm_thread = kthread_run(ksm_scan_thread, NULL, "ksmd"); if (IS_ERR(ksm_thread)) { |
25acde317 mm: ksm use pr_er... |
3041 3042 |
pr_err("ksm: creating kthread failed "); |
31dbd01f3 ksm: Kernel SameP... |
3043 |
err = PTR_ERR(ksm_thread); |
d9f8984c2 ksm: cleanup for ... |
3044 |
goto out_free; |
31dbd01f3 ksm: Kernel SameP... |
3045 |
} |
2ffd8679c ksm: sysfs and de... |
3046 |
#ifdef CONFIG_SYSFS |
31dbd01f3 ksm: Kernel SameP... |
3047 3048 |
err = sysfs_create_group(mm_kobj, &ksm_attr_group); if (err) { |
25acde317 mm: ksm use pr_er... |
3049 3050 |
pr_err("ksm: register sysfs failed "); |
2ffd8679c ksm: sysfs and de... |
3051 |
kthread_stop(ksm_thread); |
d9f8984c2 ksm: cleanup for ... |
3052 |
goto out_free; |
31dbd01f3 ksm: Kernel SameP... |
3053 |
} |
c73602ad3 ksm: more on defa... |
3054 3055 |
#else ksm_run = KSM_RUN_MERGE; /* no way for user to start it */ |
2ffd8679c ksm: sysfs and de... |
3056 |
#endif /* CONFIG_SYSFS */ |
31dbd01f3 ksm: Kernel SameP... |
3057 |
|
62b61f611 ksm: memory hotre... |
3058 |
#ifdef CONFIG_MEMORY_HOTREMOVE |
ef4d43a80 ksm: stop hotremo... |
3059 |
/* There is no significance to this priority 100 */ |
62b61f611 ksm: memory hotre... |
3060 3061 |
hotplug_memory_notifier(ksm_memory_callback, 100); #endif |
31dbd01f3 ksm: Kernel SameP... |
3062 |
return 0; |
d9f8984c2 ksm: cleanup for ... |
3063 |
out_free: |
31dbd01f3 ksm: Kernel SameP... |
3064 3065 3066 |
ksm_slab_free(); out: return err; |
f8af4da3b ksm: the mm inter... |
3067 |
} |
a64fb3cd6 mm: audit/fix non... |
3068 |
subsys_initcall(ksm_init); |