Blame view
mm/huge_memory.c
73.4 KB
71e3aac07 thp: transparent ... |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 |
/* * Copyright (C) 2009 Red Hat, Inc. * * This work is licensed under the terms of the GNU GPL, version 2. See * the COPYING file in the top-level directory. */ #include <linux/mm.h> #include <linux/sched.h> #include <linux/highmem.h> #include <linux/hugetlb.h> #include <linux/mmu_notifier.h> #include <linux/rmap.h> #include <linux/swap.h> |
97ae17497 thp: implement re... |
15 |
#include <linux/shrinker.h> |
ba76149f4 thp: khugepaged |
16 17 18 |
#include <linux/mm_inline.h> #include <linux/kthread.h> #include <linux/khugepaged.h> |
878aee7d6 thp: freeze khuge... |
19 |
#include <linux/freezer.h> |
a664b2d85 thp: madvise(MADV... |
20 |
#include <linux/mman.h> |
325adeb55 mm: huge_memory: ... |
21 |
#include <linux/pagemap.h> |
4daae3b4b mm: mempolicy: Us... |
22 |
#include <linux/migrate.h> |
43b5fbbd2 mm/huge_memory.c:... |
23 |
#include <linux/hashtable.h> |
97ae17497 thp: implement re... |
24 |
|
71e3aac07 thp: transparent ... |
25 26 27 |
#include <asm/tlb.h> #include <asm/pgalloc.h> #include "internal.h" |
ba76149f4 thp: khugepaged |
28 29 30 31 32 33 34 |
/* * By default transparent hugepage support is enabled for all mappings * and khugepaged scans all mappings. Defrag is only invoked by * khugepaged hugepage allocations and by page faults inside * MADV_HUGEPAGE regions to avoid the risk of slowing down short lived * allocations. */ |
71e3aac07 thp: transparent ... |
35 |
unsigned long transparent_hugepage_flags __read_mostly = |
13ece886d thp: transparent ... |
36 |
#ifdef CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS |
ba76149f4 thp: khugepaged |
37 |
(1<<TRANSPARENT_HUGEPAGE_FLAG)| |
13ece886d thp: transparent ... |
38 39 40 41 |
#endif #ifdef CONFIG_TRANSPARENT_HUGEPAGE_MADVISE (1<<TRANSPARENT_HUGEPAGE_REQ_MADV_FLAG)| #endif |
d39d33c33 thp: enable direc... |
42 |
(1<<TRANSPARENT_HUGEPAGE_DEFRAG_FLAG)| |
79da5407e thp: introduce sy... |
43 44 |
(1<<TRANSPARENT_HUGEPAGE_DEFRAG_KHUGEPAGED_FLAG)| (1<<TRANSPARENT_HUGEPAGE_USE_ZERO_PAGE_FLAG); |
ba76149f4 thp: khugepaged |
45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 |
/* default scan 8*512 pte (or vmas) every 30 second */ static unsigned int khugepaged_pages_to_scan __read_mostly = HPAGE_PMD_NR*8; static unsigned int khugepaged_pages_collapsed; static unsigned int khugepaged_full_scans; static unsigned int khugepaged_scan_sleep_millisecs __read_mostly = 10000; /* during fragmentation poll the hugepage allocator once every minute */ static unsigned int khugepaged_alloc_sleep_millisecs __read_mostly = 60000; static struct task_struct *khugepaged_thread __read_mostly; static DEFINE_MUTEX(khugepaged_mutex); static DEFINE_SPINLOCK(khugepaged_mm_lock); static DECLARE_WAIT_QUEUE_HEAD(khugepaged_wait); /* * default collapse hugepages if there is at least one pte mapped like * it would have happened if the vma was large enough during page * fault. */ static unsigned int khugepaged_max_ptes_none __read_mostly = HPAGE_PMD_NR-1; static int khugepaged(void *none); |
ba76149f4 thp: khugepaged |
65 |
static int khugepaged_slab_init(void); |
ba76149f4 thp: khugepaged |
66 |
|
43b5fbbd2 mm/huge_memory.c:... |
67 68 |
#define MM_SLOTS_HASH_BITS 10 static __read_mostly DEFINE_HASHTABLE(mm_slots_hash, MM_SLOTS_HASH_BITS); |
ba76149f4 thp: khugepaged |
69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 |
static struct kmem_cache *mm_slot_cache __read_mostly; /** * struct mm_slot - hash lookup from mm to mm_slot * @hash: hash collision list * @mm_node: khugepaged scan list headed in khugepaged_scan.mm_head * @mm: the mm that this information is valid for */ struct mm_slot { struct hlist_node hash; struct list_head mm_node; struct mm_struct *mm; }; /** * struct khugepaged_scan - cursor for scanning * @mm_head: the head of the mm list to scan * @mm_slot: the current mm_slot we are scanning * @address: the next address inside that to be scanned * * There is only the one khugepaged_scan instance of this cursor structure. */ struct khugepaged_scan { struct list_head mm_head; struct mm_slot *mm_slot; unsigned long address; |
2f1da6421 mm/huge_memory.c:... |
95 96 |
}; static struct khugepaged_scan khugepaged_scan = { |
ba76149f4 thp: khugepaged |
97 98 |
.mm_head = LIST_HEAD_INIT(khugepaged_scan.mm_head), }; |
f000565ad thp: set recommen... |
99 100 101 102 103 104 |
static int set_recommended_min_free_kbytes(void) { struct zone *zone; int nr_zones = 0; unsigned long recommended_min; |
f000565ad thp: set recommen... |
105 |
|
17c230afa thp: use khugepag... |
106 |
if (!khugepaged_enabled()) |
f000565ad thp: set recommen... |
107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 |
return 0; for_each_populated_zone(zone) nr_zones++; /* Make sure at least 2 hugepages are free for MIGRATE_RESERVE */ recommended_min = pageblock_nr_pages * nr_zones * 2; /* * Make sure that on average at least two pageblocks are almost free * of another type, one for a migratetype to fall back to and a * second to avoid subsequent fallbacks of other types There are 3 * MIGRATE_TYPES we care about. */ recommended_min += pageblock_nr_pages * nr_zones * MIGRATE_PCPTYPES * MIGRATE_PCPTYPES; /* don't ever allow to reserve more than 5% of the lowmem */ recommended_min = min(recommended_min, (unsigned long) nr_free_buffer_pages() / 20); recommended_min <<= (PAGE_SHIFT-10); if (recommended_min > min_free_kbytes) min_free_kbytes = recommended_min; setup_per_zone_wmarks(); return 0; } late_initcall(set_recommended_min_free_kbytes); |
ba76149f4 thp: khugepaged |
135 136 137 138 |
static int start_khugepaged(void) { int err = 0; if (khugepaged_enabled()) { |
ba76149f4 thp: khugepaged |
139 140 141 142 143 144 145 146 147 148 |
if (!khugepaged_thread) khugepaged_thread = kthread_run(khugepaged, NULL, "khugepaged"); if (unlikely(IS_ERR(khugepaged_thread))) { printk(KERN_ERR "khugepaged: kthread_run(khugepaged) failed "); err = PTR_ERR(khugepaged_thread); khugepaged_thread = NULL; } |
911891afe thp: move khugepa... |
149 150 |
if (!list_empty(&khugepaged_scan.mm_head)) |
ba76149f4 thp: khugepaged |
151 |
wake_up_interruptible(&khugepaged_wait); |
f000565ad thp: set recommen... |
152 153 |
set_recommended_min_free_kbytes(); |
911891afe thp: move khugepa... |
154 |
} else if (khugepaged_thread) { |
911891afe thp: move khugepa... |
155 156 157 |
kthread_stop(khugepaged_thread); khugepaged_thread = NULL; } |
637e3a27e thp: remove unnec... |
158 |
|
ba76149f4 thp: khugepaged |
159 160 |
return err; } |
71e3aac07 thp: transparent ... |
161 |
|
97ae17497 thp: implement re... |
162 |
static atomic_t huge_zero_refcount; |
5918d10a4 thp: fix huge zer... |
163 |
static struct page *huge_zero_page __read_mostly; |
97ae17497 thp: implement re... |
164 |
|
5918d10a4 thp: fix huge zer... |
165 |
static inline bool is_huge_zero_page(struct page *page) |
4a6c12972 thp: huge zero pa... |
166 |
{ |
5918d10a4 thp: fix huge zer... |
167 |
return ACCESS_ONCE(huge_zero_page) == page; |
97ae17497 thp: implement re... |
168 |
} |
4a6c12972 thp: huge zero pa... |
169 |
|
97ae17497 thp: implement re... |
170 171 |
static inline bool is_huge_zero_pmd(pmd_t pmd) { |
5918d10a4 thp: fix huge zer... |
172 |
return is_huge_zero_page(pmd_page(pmd)); |
97ae17497 thp: implement re... |
173 |
} |
5918d10a4 thp: fix huge zer... |
174 |
static struct page *get_huge_zero_page(void) |
97ae17497 thp: implement re... |
175 176 177 178 |
{ struct page *zero_page; retry: if (likely(atomic_inc_not_zero(&huge_zero_refcount))) |
5918d10a4 thp: fix huge zer... |
179 |
return ACCESS_ONCE(huge_zero_page); |
97ae17497 thp: implement re... |
180 181 |
zero_page = alloc_pages((GFP_TRANSHUGE | __GFP_ZERO) & ~__GFP_MOVABLE, |
4a6c12972 thp: huge zero pa... |
182 |
HPAGE_PMD_ORDER); |
d8a8e1f0d thp, vmstat: impl... |
183 184 |
if (!zero_page) { count_vm_event(THP_ZERO_PAGE_ALLOC_FAILED); |
5918d10a4 thp: fix huge zer... |
185 |
return NULL; |
d8a8e1f0d thp, vmstat: impl... |
186 187 |
} count_vm_event(THP_ZERO_PAGE_ALLOC); |
97ae17497 thp: implement re... |
188 |
preempt_disable(); |
5918d10a4 thp: fix huge zer... |
189 |
if (cmpxchg(&huge_zero_page, NULL, zero_page)) { |
97ae17497 thp: implement re... |
190 191 192 193 194 195 196 197 |
preempt_enable(); __free_page(zero_page); goto retry; } /* We take additional reference here. It will be put back by shrinker */ atomic_set(&huge_zero_refcount, 2); preempt_enable(); |
5918d10a4 thp: fix huge zer... |
198 |
return ACCESS_ONCE(huge_zero_page); |
4a6c12972 thp: huge zero pa... |
199 |
} |
97ae17497 thp: implement re... |
200 |
static void put_huge_zero_page(void) |
4a6c12972 thp: huge zero pa... |
201 |
{ |
97ae17497 thp: implement re... |
202 203 204 205 206 |
/* * Counter should never go to zero here. Only shrinker can put * last reference. */ BUG_ON(atomic_dec_and_test(&huge_zero_refcount)); |
4a6c12972 thp: huge zero pa... |
207 |
} |
97ae17497 thp: implement re... |
208 209 |
static int shrink_huge_zero_page(struct shrinker *shrink, struct shrink_control *sc) |
4a6c12972 thp: huge zero pa... |
210 |
{ |
97ae17497 thp: implement re... |
211 212 213 214 215 |
if (!sc->nr_to_scan) /* we can free zero page only if last reference remains */ return atomic_read(&huge_zero_refcount) == 1 ? HPAGE_PMD_NR : 0; if (atomic_cmpxchg(&huge_zero_refcount, 1, 0) == 1) { |
5918d10a4 thp: fix huge zer... |
216 217 218 |
struct page *zero_page = xchg(&huge_zero_page, NULL); BUG_ON(zero_page == NULL); __free_page(zero_page); |
97ae17497 thp: implement re... |
219 220 221 |
} return 0; |
4a6c12972 thp: huge zero pa... |
222 |
} |
97ae17497 thp: implement re... |
223 224 225 226 |
static struct shrinker huge_zero_page_shrinker = { .shrink = shrink_huge_zero_page, .seeks = DEFAULT_SEEKS, }; |
71e3aac07 thp: transparent ... |
227 |
#ifdef CONFIG_SYSFS |
ba76149f4 thp: khugepaged |
228 |
|
71e3aac07 thp: transparent ... |
229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 |
static ssize_t double_flag_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf, enum transparent_hugepage_flag enabled, enum transparent_hugepage_flag req_madv) { if (test_bit(enabled, &transparent_hugepage_flags)) { VM_BUG_ON(test_bit(req_madv, &transparent_hugepage_flags)); return sprintf(buf, "[always] madvise never "); } else if (test_bit(req_madv, &transparent_hugepage_flags)) return sprintf(buf, "always [madvise] never "); else return sprintf(buf, "always madvise [never] "); } static ssize_t double_flag_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count, enum transparent_hugepage_flag enabled, enum transparent_hugepage_flag req_madv) { if (!memcmp("always", buf, min(sizeof("always")-1, count))) { set_bit(enabled, &transparent_hugepage_flags); clear_bit(req_madv, &transparent_hugepage_flags); } else if (!memcmp("madvise", buf, min(sizeof("madvise")-1, count))) { clear_bit(enabled, &transparent_hugepage_flags); set_bit(req_madv, &transparent_hugepage_flags); } else if (!memcmp("never", buf, min(sizeof("never")-1, count))) { clear_bit(enabled, &transparent_hugepage_flags); clear_bit(req_madv, &transparent_hugepage_flags); } else return -EINVAL; return count; } static ssize_t enabled_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return double_flag_show(kobj, attr, buf, TRANSPARENT_HUGEPAGE_FLAG, TRANSPARENT_HUGEPAGE_REQ_MADV_FLAG); } static ssize_t enabled_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { |
ba76149f4 thp: khugepaged |
280 281 282 283 284 285 286 |
ssize_t ret; ret = double_flag_store(kobj, attr, buf, count, TRANSPARENT_HUGEPAGE_FLAG, TRANSPARENT_HUGEPAGE_REQ_MADV_FLAG); if (ret > 0) { |
911891afe thp: move khugepa... |
287 288 289 290 291 |
int err; mutex_lock(&khugepaged_mutex); err = start_khugepaged(); mutex_unlock(&khugepaged_mutex); |
ba76149f4 thp: khugepaged |
292 293 294 295 296 |
if (err) ret = err; } return ret; |
71e3aac07 thp: transparent ... |
297 298 299 300 301 302 303 304 |
} static struct kobj_attribute enabled_attr = __ATTR(enabled, 0644, enabled_show, enabled_store); static ssize_t single_flag_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf, enum transparent_hugepage_flag flag) { |
e27e6151b mm/thp: use conve... |
305 306 307 |
return sprintf(buf, "%d ", !!test_bit(flag, &transparent_hugepage_flags)); |
71e3aac07 thp: transparent ... |
308 |
} |
e27e6151b mm/thp: use conve... |
309 |
|
71e3aac07 thp: transparent ... |
310 311 312 313 314 |
static ssize_t single_flag_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count, enum transparent_hugepage_flag flag) { |
e27e6151b mm/thp: use conve... |
315 316 317 318 319 320 321 322 323 324 |
unsigned long value; int ret; ret = kstrtoul(buf, 10, &value); if (ret < 0) return ret; if (value > 1) return -EINVAL; if (value) |
71e3aac07 thp: transparent ... |
325 |
set_bit(flag, &transparent_hugepage_flags); |
e27e6151b mm/thp: use conve... |
326 |
else |
71e3aac07 thp: transparent ... |
327 |
clear_bit(flag, &transparent_hugepage_flags); |
71e3aac07 thp: transparent ... |
328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 |
return count; } /* * Currently defrag only disables __GFP_NOWAIT for allocation. A blind * __GFP_REPEAT is too aggressive, it's never worth swapping tons of * memory just to allocate one more hugepage. */ static ssize_t defrag_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return double_flag_show(kobj, attr, buf, TRANSPARENT_HUGEPAGE_DEFRAG_FLAG, TRANSPARENT_HUGEPAGE_DEFRAG_REQ_MADV_FLAG); } static ssize_t defrag_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { return double_flag_store(kobj, attr, buf, count, TRANSPARENT_HUGEPAGE_DEFRAG_FLAG, TRANSPARENT_HUGEPAGE_DEFRAG_REQ_MADV_FLAG); } static struct kobj_attribute defrag_attr = __ATTR(defrag, 0644, defrag_show, defrag_store); |
79da5407e thp: introduce sy... |
354 355 356 357 358 359 360 361 362 363 364 365 366 367 |
static ssize_t use_zero_page_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return single_flag_show(kobj, attr, buf, TRANSPARENT_HUGEPAGE_USE_ZERO_PAGE_FLAG); } static ssize_t use_zero_page_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { return single_flag_store(kobj, attr, buf, count, TRANSPARENT_HUGEPAGE_USE_ZERO_PAGE_FLAG); } static struct kobj_attribute use_zero_page_attr = __ATTR(use_zero_page, 0644, use_zero_page_show, use_zero_page_store); |
71e3aac07 thp: transparent ... |
368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 |
#ifdef CONFIG_DEBUG_VM static ssize_t debug_cow_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return single_flag_show(kobj, attr, buf, TRANSPARENT_HUGEPAGE_DEBUG_COW_FLAG); } static ssize_t debug_cow_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { return single_flag_store(kobj, attr, buf, count, TRANSPARENT_HUGEPAGE_DEBUG_COW_FLAG); } static struct kobj_attribute debug_cow_attr = __ATTR(debug_cow, 0644, debug_cow_show, debug_cow_store); #endif /* CONFIG_DEBUG_VM */ static struct attribute *hugepage_attr[] = { &enabled_attr.attr, &defrag_attr.attr, |
79da5407e thp: introduce sy... |
389 |
&use_zero_page_attr.attr, |
71e3aac07 thp: transparent ... |
390 391 392 393 394 395 396 397 |
#ifdef CONFIG_DEBUG_VM &debug_cow_attr.attr, #endif NULL, }; static struct attribute_group hugepage_attr_group = { .attrs = hugepage_attr, |
ba76149f4 thp: khugepaged |
398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 |
}; static ssize_t scan_sleep_millisecs_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return sprintf(buf, "%u ", khugepaged_scan_sleep_millisecs); } static ssize_t scan_sleep_millisecs_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { unsigned long msecs; int err; err = strict_strtoul(buf, 10, &msecs); if (err || msecs > UINT_MAX) return -EINVAL; khugepaged_scan_sleep_millisecs = msecs; wake_up_interruptible(&khugepaged_wait); return count; } static struct kobj_attribute scan_sleep_millisecs_attr = __ATTR(scan_sleep_millisecs, 0644, scan_sleep_millisecs_show, scan_sleep_millisecs_store); static ssize_t alloc_sleep_millisecs_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return sprintf(buf, "%u ", khugepaged_alloc_sleep_millisecs); } static ssize_t alloc_sleep_millisecs_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { unsigned long msecs; int err; err = strict_strtoul(buf, 10, &msecs); if (err || msecs > UINT_MAX) return -EINVAL; khugepaged_alloc_sleep_millisecs = msecs; wake_up_interruptible(&khugepaged_wait); return count; } static struct kobj_attribute alloc_sleep_millisecs_attr = __ATTR(alloc_sleep_millisecs, 0644, alloc_sleep_millisecs_show, alloc_sleep_millisecs_store); static ssize_t pages_to_scan_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return sprintf(buf, "%u ", khugepaged_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 pages; err = strict_strtoul(buf, 10, &pages); if (err || !pages || pages > UINT_MAX) return -EINVAL; khugepaged_pages_to_scan = pages; return count; } static struct kobj_attribute pages_to_scan_attr = __ATTR(pages_to_scan, 0644, pages_to_scan_show, pages_to_scan_store); static ssize_t pages_collapsed_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return sprintf(buf, "%u ", khugepaged_pages_collapsed); } static struct kobj_attribute pages_collapsed_attr = __ATTR_RO(pages_collapsed); static ssize_t full_scans_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return sprintf(buf, "%u ", khugepaged_full_scans); } static struct kobj_attribute full_scans_attr = __ATTR_RO(full_scans); static ssize_t khugepaged_defrag_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return single_flag_show(kobj, attr, buf, TRANSPARENT_HUGEPAGE_DEFRAG_KHUGEPAGED_FLAG); } static ssize_t khugepaged_defrag_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { return single_flag_store(kobj, attr, buf, count, TRANSPARENT_HUGEPAGE_DEFRAG_KHUGEPAGED_FLAG); } static struct kobj_attribute khugepaged_defrag_attr = __ATTR(defrag, 0644, khugepaged_defrag_show, khugepaged_defrag_store); /* * max_ptes_none controls if khugepaged should collapse hugepages over * any unmapped ptes in turn potentially increasing the memory * footprint of the vmas. When max_ptes_none is 0 khugepaged will not * reduce the available free memory in the system as it * runs. Increasing max_ptes_none will instead potentially reduce the * free memory in the system during the khugepaged scan. */ static ssize_t khugepaged_max_ptes_none_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return sprintf(buf, "%u ", khugepaged_max_ptes_none); } static ssize_t khugepaged_max_ptes_none_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { int err; unsigned long max_ptes_none; err = strict_strtoul(buf, 10, &max_ptes_none); if (err || max_ptes_none > HPAGE_PMD_NR-1) return -EINVAL; khugepaged_max_ptes_none = max_ptes_none; return count; } static struct kobj_attribute khugepaged_max_ptes_none_attr = __ATTR(max_ptes_none, 0644, khugepaged_max_ptes_none_show, khugepaged_max_ptes_none_store); static struct attribute *khugepaged_attr[] = { &khugepaged_defrag_attr.attr, &khugepaged_max_ptes_none_attr.attr, &pages_to_scan_attr.attr, &pages_collapsed_attr.attr, &full_scans_attr.attr, &scan_sleep_millisecs_attr.attr, &alloc_sleep_millisecs_attr.attr, NULL, }; static struct attribute_group khugepaged_attr_group = { .attrs = khugepaged_attr, .name = "khugepaged", |
71e3aac07 thp: transparent ... |
567 |
}; |
71e3aac07 thp: transparent ... |
568 |
|
569e55900 thp: improve the ... |
569 |
static int __init hugepage_init_sysfs(struct kobject **hugepage_kobj) |
71e3aac07 thp: transparent ... |
570 |
{ |
71e3aac07 thp: transparent ... |
571 |
int err; |
569e55900 thp: improve the ... |
572 573 |
*hugepage_kobj = kobject_create_and_add("transparent_hugepage", mm_kobj); if (unlikely(!*hugepage_kobj)) { |
2c79737af mm: clean up tran... |
574 575 |
printk(KERN_ERR "hugepage: failed to create transparent hugepage kobject "); |
569e55900 thp: improve the ... |
576 |
return -ENOMEM; |
ba76149f4 thp: khugepaged |
577 |
} |
569e55900 thp: improve the ... |
578 |
err = sysfs_create_group(*hugepage_kobj, &hugepage_attr_group); |
ba76149f4 thp: khugepaged |
579 |
if (err) { |
2c79737af mm: clean up tran... |
580 581 |
printk(KERN_ERR "hugepage: failed to register transparent hugepage group "); |
569e55900 thp: improve the ... |
582 |
goto delete_obj; |
ba76149f4 thp: khugepaged |
583 |
} |
569e55900 thp: improve the ... |
584 |
err = sysfs_create_group(*hugepage_kobj, &khugepaged_attr_group); |
ba76149f4 thp: khugepaged |
585 |
if (err) { |
2c79737af mm: clean up tran... |
586 587 |
printk(KERN_ERR "hugepage: failed to register transparent hugepage group "); |
569e55900 thp: improve the ... |
588 |
goto remove_hp_group; |
ba76149f4 thp: khugepaged |
589 |
} |
569e55900 thp: improve the ... |
590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 |
return 0; remove_hp_group: sysfs_remove_group(*hugepage_kobj, &hugepage_attr_group); delete_obj: kobject_put(*hugepage_kobj); return err; } static void __init hugepage_exit_sysfs(struct kobject *hugepage_kobj) { sysfs_remove_group(hugepage_kobj, &khugepaged_attr_group); sysfs_remove_group(hugepage_kobj, &hugepage_attr_group); kobject_put(hugepage_kobj); } #else static inline int hugepage_init_sysfs(struct kobject **hugepage_kobj) { return 0; } static inline void hugepage_exit_sysfs(struct kobject *hugepage_kobj) { } #endif /* CONFIG_SYSFS */ static int __init hugepage_init(void) { int err; struct kobject *hugepage_kobj; if (!has_transparent_hugepage()) { transparent_hugepage_flags = 0; return -EINVAL; } err = hugepage_init_sysfs(&hugepage_kobj); if (err) return err; |
ba76149f4 thp: khugepaged |
630 631 632 633 |
err = khugepaged_slab_init(); if (err) goto out; |
97ae17497 thp: implement re... |
634 |
register_shrinker(&huge_zero_page_shrinker); |
97562cd24 thp: disable tran... |
635 636 637 638 639 640 641 |
/* * By default disable transparent hugepages on smaller systems, * where the extra memory used could hurt more than TLB overhead * is likely to save. The admin can still enable it through /sys. */ if (totalram_pages < (512 << (20 - PAGE_SHIFT))) transparent_hugepage_flags = 0; |
ba76149f4 thp: khugepaged |
642 |
start_khugepaged(); |
569e55900 thp: improve the ... |
643 |
return 0; |
ba76149f4 thp: khugepaged |
644 |
out: |
569e55900 thp: improve the ... |
645 |
hugepage_exit_sysfs(hugepage_kobj); |
ba76149f4 thp: khugepaged |
646 |
return err; |
71e3aac07 thp: transparent ... |
647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 |
} module_init(hugepage_init) static int __init setup_transparent_hugepage(char *str) { int ret = 0; if (!str) goto out; if (!strcmp(str, "always")) { set_bit(TRANSPARENT_HUGEPAGE_FLAG, &transparent_hugepage_flags); clear_bit(TRANSPARENT_HUGEPAGE_REQ_MADV_FLAG, &transparent_hugepage_flags); ret = 1; } else if (!strcmp(str, "madvise")) { clear_bit(TRANSPARENT_HUGEPAGE_FLAG, &transparent_hugepage_flags); set_bit(TRANSPARENT_HUGEPAGE_REQ_MADV_FLAG, &transparent_hugepage_flags); ret = 1; } else if (!strcmp(str, "never")) { clear_bit(TRANSPARENT_HUGEPAGE_FLAG, &transparent_hugepage_flags); clear_bit(TRANSPARENT_HUGEPAGE_REQ_MADV_FLAG, &transparent_hugepage_flags); ret = 1; } out: if (!ret) printk(KERN_WARNING "transparent_hugepage= cannot parse, ignored "); return ret; } __setup("transparent_hugepage=", setup_transparent_hugepage); |
b32967ff1 mm: numa: Add THP... |
682 |
pmd_t maybe_pmd_mkwrite(pmd_t pmd, struct vm_area_struct *vma) |
71e3aac07 thp: transparent ... |
683 684 685 686 687 |
{ if (likely(vma->vm_flags & VM_WRITE)) pmd = pmd_mkwrite(pmd); return pmd; } |
b3092b3b7 thp: cleanup: int... |
688 689 690 691 692 693 694 695 |
static inline pmd_t mk_huge_pmd(struct page *page, struct vm_area_struct *vma) { pmd_t entry; entry = mk_pmd(page, vma->vm_page_prot); entry = maybe_pmd_mkwrite(pmd_mkdirty(entry), vma); entry = pmd_mkhuge(entry); return entry; } |
71e3aac07 thp: transparent ... |
696 697 698 699 700 |
static int __do_huge_pmd_anonymous_page(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long haddr, pmd_t *pmd, struct page *page) { |
71e3aac07 thp: transparent ... |
701 702 703 704 |
pgtable_t pgtable; VM_BUG_ON(!PageCompound(page)); pgtable = pte_alloc_one(mm, haddr); |
edad9d2c3 mm, thp: allow fa... |
705 |
if (unlikely(!pgtable)) |
71e3aac07 thp: transparent ... |
706 |
return VM_FAULT_OOM; |
71e3aac07 thp: transparent ... |
707 708 |
clear_huge_page(page, haddr, HPAGE_PMD_NR); |
52f37629f THP: fix comment ... |
709 710 711 712 713 |
/* * The memory barrier inside __SetPageUptodate makes sure that * clear_huge_page writes become visible before the set_pmd_at() * write. */ |
71e3aac07 thp: transparent ... |
714 715 716 717 718 |
__SetPageUptodate(page); spin_lock(&mm->page_table_lock); if (unlikely(!pmd_none(*pmd))) { spin_unlock(&mm->page_table_lock); |
b9bbfbe30 thp: memcg huge m... |
719 |
mem_cgroup_uncharge_page(page); |
71e3aac07 thp: transparent ... |
720 721 722 723 |
put_page(page); pte_free(mm, pgtable); } else { pmd_t entry; |
b3092b3b7 thp: cleanup: int... |
724 |
entry = mk_huge_pmd(page, vma); |
71e3aac07 thp: transparent ... |
725 726 |
page_add_new_anon_rmap(page, vma, haddr); set_pmd_at(mm, haddr, pmd, entry); |
e3ebcf643 thp: remove assum... |
727 |
pgtable_trans_huge_deposit(mm, pgtable); |
71e3aac07 thp: transparent ... |
728 |
add_mm_counter(mm, MM_ANONPAGES, HPAGE_PMD_NR); |
1c641e847 mm: thp: fix BUG ... |
729 |
mm->nr_ptes++; |
71e3aac07 thp: transparent ... |
730 731 |
spin_unlock(&mm->page_table_lock); } |
aa2e878ef mm, thp: remove u... |
732 |
return 0; |
71e3aac07 thp: transparent ... |
733 |
} |
cc5d462f7 mm: use __GFP_OTH... |
734 |
static inline gfp_t alloc_hugepage_gfpmask(int defrag, gfp_t extra_gfp) |
0bbbc0b33 thp: add numa awa... |
735 |
{ |
cc5d462f7 mm: use __GFP_OTH... |
736 |
return (GFP_TRANSHUGE & ~(defrag ? 0 : __GFP_WAIT)) | extra_gfp; |
0bbbc0b33 thp: add numa awa... |
737 738 739 740 |
} static inline struct page *alloc_hugepage_vma(int defrag, struct vm_area_struct *vma, |
cc5d462f7 mm: use __GFP_OTH... |
741 742 |
unsigned long haddr, int nd, gfp_t extra_gfp) |
0bbbc0b33 thp: add numa awa... |
743 |
{ |
cc5d462f7 mm: use __GFP_OTH... |
744 |
return alloc_pages_vma(alloc_hugepage_gfpmask(defrag, extra_gfp), |
5c4b4be3b mm: use correct n... |
745 |
HPAGE_PMD_ORDER, vma, haddr, nd); |
0bbbc0b33 thp: add numa awa... |
746 747 748 |
} #ifndef CONFIG_NUMA |
71e3aac07 thp: transparent ... |
749 750 |
static inline struct page *alloc_hugepage(int defrag) { |
cc5d462f7 mm: use __GFP_OTH... |
751 |
return alloc_pages(alloc_hugepage_gfpmask(defrag, 0), |
71e3aac07 thp: transparent ... |
752 753 |
HPAGE_PMD_ORDER); } |
0bbbc0b33 thp: add numa awa... |
754 |
#endif |
71e3aac07 thp: transparent ... |
755 |
|
3ea41e621 thp: avoid race o... |
756 |
static bool set_huge_zero_page(pgtable_t pgtable, struct mm_struct *mm, |
97ae17497 thp: implement re... |
757 |
struct vm_area_struct *vma, unsigned long haddr, pmd_t *pmd, |
5918d10a4 thp: fix huge zer... |
758 |
struct page *zero_page) |
fc9fe822f thp: copy_huge_pm... |
759 760 |
{ pmd_t entry; |
3ea41e621 thp: avoid race o... |
761 762 |
if (!pmd_none(*pmd)) return false; |
5918d10a4 thp: fix huge zer... |
763 |
entry = mk_pmd(zero_page, vma->vm_page_prot); |
fc9fe822f thp: copy_huge_pm... |
764 765 766 767 768 |
entry = pmd_wrprotect(entry); entry = pmd_mkhuge(entry); set_pmd_at(mm, haddr, pmd, entry); pgtable_trans_huge_deposit(mm, pgtable); mm->nr_ptes++; |
3ea41e621 thp: avoid race o... |
769 |
return true; |
fc9fe822f thp: copy_huge_pm... |
770 |
} |
71e3aac07 thp: transparent ... |
771 772 773 774 775 776 777 778 779 780 781 |
int do_huge_pmd_anonymous_page(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long address, pmd_t *pmd, unsigned int flags) { struct page *page; unsigned long haddr = address & HPAGE_PMD_MASK; pte_t *pte; if (haddr >= vma->vm_start && haddr + HPAGE_PMD_SIZE <= vma->vm_end) { if (unlikely(anon_vma_prepare(vma))) return VM_FAULT_OOM; |
ba76149f4 thp: khugepaged |
782 783 |
if (unlikely(khugepaged_enter(vma))) return VM_FAULT_OOM; |
79da5407e thp: introduce sy... |
784 785 |
if (!(flags & FAULT_FLAG_WRITE) && transparent_hugepage_use_zero_page()) { |
80371957f thp: setup huge z... |
786 |
pgtable_t pgtable; |
5918d10a4 thp: fix huge zer... |
787 |
struct page *zero_page; |
3ea41e621 thp: avoid race o... |
788 |
bool set; |
80371957f thp: setup huge z... |
789 790 791 |
pgtable = pte_alloc_one(mm, haddr); if (unlikely(!pgtable)) return VM_FAULT_OOM; |
5918d10a4 thp: fix huge zer... |
792 793 |
zero_page = get_huge_zero_page(); if (unlikely(!zero_page)) { |
97ae17497 thp: implement re... |
794 795 796 797 |
pte_free(mm, pgtable); count_vm_event(THP_FAULT_FALLBACK); goto out; } |
80371957f thp: setup huge z... |
798 |
spin_lock(&mm->page_table_lock); |
3ea41e621 thp: avoid race o... |
799 |
set = set_huge_zero_page(pgtable, mm, vma, haddr, pmd, |
5918d10a4 thp: fix huge zer... |
800 |
zero_page); |
80371957f thp: setup huge z... |
801 |
spin_unlock(&mm->page_table_lock); |
3ea41e621 thp: avoid race o... |
802 803 804 805 |
if (!set) { pte_free(mm, pgtable); put_huge_zero_page(); } |
80371957f thp: setup huge z... |
806 807 |
return 0; } |
0bbbc0b33 thp: add numa awa... |
808 |
page = alloc_hugepage_vma(transparent_hugepage_defrag(vma), |
cc5d462f7 mm: use __GFP_OTH... |
809 |
vma, haddr, numa_node_id(), 0); |
81ab4201f mm: add VM counte... |
810 811 |
if (unlikely(!page)) { count_vm_event(THP_FAULT_FALLBACK); |
71e3aac07 thp: transparent ... |
812 |
goto out; |
81ab4201f mm: add VM counte... |
813 814 |
} count_vm_event(THP_FAULT_ALLOC); |
b9bbfbe30 thp: memcg huge m... |
815 816 817 818 |
if (unlikely(mem_cgroup_newpage_charge(page, mm, GFP_KERNEL))) { put_page(page); goto out; } |
edad9d2c3 mm, thp: allow fa... |
819 820 821 822 823 824 |
if (unlikely(__do_huge_pmd_anonymous_page(mm, vma, haddr, pmd, page))) { mem_cgroup_uncharge_page(page); put_page(page); goto out; } |
71e3aac07 thp: transparent ... |
825 |
|
edad9d2c3 mm, thp: allow fa... |
826 |
return 0; |
71e3aac07 thp: transparent ... |
827 828 829 830 831 832 833 |
} out: /* * Use __pte_alloc instead of pte_alloc_map, because we can't * run pte_offset_map on the pmd, if an huge pmd could * materialize from under us from a different thread. */ |
4fd017708 mm: Check if PTE ... |
834 835 |
if (unlikely(pmd_none(*pmd)) && unlikely(__pte_alloc(mm, vma, pmd, address))) |
71e3aac07 thp: transparent ... |
836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 |
return VM_FAULT_OOM; /* if an huge pmd materialized from under us just retry later */ if (unlikely(pmd_trans_huge(*pmd))) return 0; /* * A regular pmd is established and it can't morph into a huge pmd * from under us anymore at this point because we hold the mmap_sem * read mode and khugepaged takes it in write mode. So now it's * safe to run pte_offset_map(). */ pte = pte_offset_map(pmd, address); return handle_pte_fault(mm, vma, address, pte, pmd, flags); } int copy_huge_pmd(struct mm_struct *dst_mm, struct mm_struct *src_mm, pmd_t *dst_pmd, pmd_t *src_pmd, unsigned long addr, struct vm_area_struct *vma) { struct page *src_page; pmd_t pmd; pgtable_t pgtable; int ret; ret = -ENOMEM; pgtable = pte_alloc_one(dst_mm, addr); if (unlikely(!pgtable)) goto out; spin_lock(&dst_mm->page_table_lock); spin_lock_nested(&src_mm->page_table_lock, SINGLE_DEPTH_NESTING); ret = -EAGAIN; pmd = *src_pmd; if (unlikely(!pmd_trans_huge(pmd))) { pte_free(dst_mm, pgtable); goto out_unlock; } |
fc9fe822f thp: copy_huge_pm... |
873 874 875 876 877 878 |
/* * mm->page_table_lock is enough to be sure that huge zero pmd is not * under splitting since we don't split the page itself, only pmd to * a page table. */ if (is_huge_zero_pmd(pmd)) { |
5918d10a4 thp: fix huge zer... |
879 |
struct page *zero_page; |
3ea41e621 thp: avoid race o... |
880 |
bool set; |
97ae17497 thp: implement re... |
881 882 883 884 885 |
/* * get_huge_zero_page() will never allocate a new page here, * since we already have a zero page to copy. It just takes a * reference. */ |
5918d10a4 thp: fix huge zer... |
886 |
zero_page = get_huge_zero_page(); |
3ea41e621 thp: avoid race o... |
887 |
set = set_huge_zero_page(pgtable, dst_mm, vma, addr, dst_pmd, |
5918d10a4 thp: fix huge zer... |
888 |
zero_page); |
3ea41e621 thp: avoid race o... |
889 |
BUG_ON(!set); /* unexpected !pmd_none(dst_pmd) */ |
fc9fe822f thp: copy_huge_pm... |
890 891 892 |
ret = 0; goto out_unlock; } |
71e3aac07 thp: transparent ... |
893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 |
if (unlikely(pmd_trans_splitting(pmd))) { /* split huge page running from under us */ spin_unlock(&src_mm->page_table_lock); spin_unlock(&dst_mm->page_table_lock); pte_free(dst_mm, pgtable); wait_split_huge_page(vma->anon_vma, src_pmd); /* src_vma */ goto out; } src_page = pmd_page(pmd); VM_BUG_ON(!PageHead(src_page)); get_page(src_page); page_dup_rmap(src_page); add_mm_counter(dst_mm, MM_ANONPAGES, HPAGE_PMD_NR); pmdp_set_wrprotect(src_mm, addr, src_pmd); pmd = pmd_mkold(pmd_wrprotect(pmd)); set_pmd_at(dst_mm, addr, dst_pmd, pmd); |
e3ebcf643 thp: remove assum... |
911 |
pgtable_trans_huge_deposit(dst_mm, pgtable); |
1c641e847 mm: thp: fix BUG ... |
912 |
dst_mm->nr_ptes++; |
71e3aac07 thp: transparent ... |
913 914 915 916 917 918 919 920 |
ret = 0; out_unlock: spin_unlock(&src_mm->page_table_lock); spin_unlock(&dst_mm->page_table_lock); out: return ret; } |
a1dd450bc mm: thp: set the ... |
921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 |
void huge_pmd_set_accessed(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long address, pmd_t *pmd, pmd_t orig_pmd, int dirty) { pmd_t entry; unsigned long haddr; spin_lock(&mm->page_table_lock); if (unlikely(!pmd_same(*pmd, orig_pmd))) goto unlock; entry = pmd_mkyoung(orig_pmd); haddr = address & HPAGE_PMD_MASK; if (pmdp_set_access_flags(vma, haddr, pmd, entry, dirty)) update_mmu_cache_pmd(vma, address, pmd); unlock: spin_unlock(&mm->page_table_lock); } |
93b4796de thp: do_huge_pmd_... |
942 943 |
static int do_huge_pmd_wp_zero_page_fallback(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long address, |
3ea41e621 thp: avoid race o... |
944 |
pmd_t *pmd, pmd_t orig_pmd, unsigned long haddr) |
93b4796de thp: do_huge_pmd_... |
945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 |
{ pgtable_t pgtable; pmd_t _pmd; struct page *page; int i, ret = 0; unsigned long mmun_start; /* For mmu_notifiers */ unsigned long mmun_end; /* For mmu_notifiers */ page = alloc_page_vma(GFP_HIGHUSER_MOVABLE, vma, address); if (!page) { ret |= VM_FAULT_OOM; goto out; } if (mem_cgroup_newpage_charge(page, mm, GFP_KERNEL)) { put_page(page); ret |= VM_FAULT_OOM; goto out; } clear_user_highpage(page, address); __SetPageUptodate(page); mmun_start = haddr; mmun_end = haddr + HPAGE_PMD_SIZE; mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); spin_lock(&mm->page_table_lock); |
3ea41e621 thp: avoid race o... |
973 974 |
if (unlikely(!pmd_same(*pmd, orig_pmd))) goto out_free_page; |
93b4796de thp: do_huge_pmd_... |
975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 |
pmdp_clear_flush(vma, haddr, pmd); /* leave pmd empty until pte is filled */ pgtable = pgtable_trans_huge_withdraw(mm); pmd_populate(mm, &_pmd, pgtable); for (i = 0; i < HPAGE_PMD_NR; i++, haddr += PAGE_SIZE) { pte_t *pte, entry; if (haddr == (address & PAGE_MASK)) { entry = mk_pte(page, vma->vm_page_prot); entry = maybe_mkwrite(pte_mkdirty(entry), vma); page_add_new_anon_rmap(page, vma, haddr); } else { entry = pfn_pte(my_zero_pfn(haddr), vma->vm_page_prot); entry = pte_mkspecial(entry); } pte = pte_offset_map(&_pmd, haddr); VM_BUG_ON(!pte_none(*pte)); set_pte_at(mm, haddr, pte, entry); pte_unmap(pte); } smp_wmb(); /* make pte visible before pmd */ pmd_populate(mm, pmd, pgtable); spin_unlock(&mm->page_table_lock); |
97ae17497 thp: implement re... |
999 |
put_huge_zero_page(); |
93b4796de thp: do_huge_pmd_... |
1000 1001 1002 1003 1004 1005 1006 |
inc_mm_counter(mm, MM_ANONPAGES); mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); ret |= VM_FAULT_WRITE; out: return ret; |
3ea41e621 thp: avoid race o... |
1007 1008 1009 1010 1011 1012 |
out_free_page: spin_unlock(&mm->page_table_lock); mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); mem_cgroup_uncharge_page(page); put_page(page); goto out; |
93b4796de thp: do_huge_pmd_... |
1013 |
} |
71e3aac07 thp: transparent ... |
1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 |
static int do_huge_pmd_wp_page_fallback(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long address, pmd_t *pmd, pmd_t orig_pmd, struct page *page, unsigned long haddr) { pgtable_t pgtable; pmd_t _pmd; int ret = 0, i; struct page **pages; |
2ec74c3ef mm: move all mmu ... |
1025 1026 |
unsigned long mmun_start; /* For mmu_notifiers */ unsigned long mmun_end; /* For mmu_notifiers */ |
71e3aac07 thp: transparent ... |
1027 1028 1029 1030 1031 1032 1033 1034 1035 |
pages = kmalloc(sizeof(struct page *) * HPAGE_PMD_NR, GFP_KERNEL); if (unlikely(!pages)) { ret |= VM_FAULT_OOM; goto out; } for (i = 0; i < HPAGE_PMD_NR; i++) { |
cc5d462f7 mm: use __GFP_OTH... |
1036 1037 |
pages[i] = alloc_page_vma_node(GFP_HIGHUSER_MOVABLE | __GFP_OTHER_NODE, |
19ee151e1 mm: preserve orig... |
1038 |
vma, address, page_to_nid(page)); |
b9bbfbe30 thp: memcg huge m... |
1039 1040 1041 1042 |
if (unlikely(!pages[i] || mem_cgroup_newpage_charge(pages[i], mm, GFP_KERNEL))) { if (pages[i]) |
71e3aac07 thp: transparent ... |
1043 |
put_page(pages[i]); |
b9bbfbe30 thp: memcg huge m... |
1044 1045 1046 1047 1048 1049 |
mem_cgroup_uncharge_start(); while (--i >= 0) { mem_cgroup_uncharge_page(pages[i]); put_page(pages[i]); } mem_cgroup_uncharge_end(); |
71e3aac07 thp: transparent ... |
1050 1051 1052 1053 1054 1055 1056 1057 |
kfree(pages); ret |= VM_FAULT_OOM; goto out; } } for (i = 0; i < HPAGE_PMD_NR; i++) { copy_user_highpage(pages[i], page + i, |
0089e4853 mm/huge_memory: f... |
1058 |
haddr + PAGE_SIZE * i, vma); |
71e3aac07 thp: transparent ... |
1059 1060 1061 |
__SetPageUptodate(pages[i]); cond_resched(); } |
2ec74c3ef mm: move all mmu ... |
1062 1063 1064 |
mmun_start = haddr; mmun_end = haddr + HPAGE_PMD_SIZE; mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); |
71e3aac07 thp: transparent ... |
1065 1066 1067 1068 |
spin_lock(&mm->page_table_lock); if (unlikely(!pmd_same(*pmd, orig_pmd))) goto out_free_pages; VM_BUG_ON(!PageHead(page)); |
2ec74c3ef mm: move all mmu ... |
1069 |
pmdp_clear_flush(vma, haddr, pmd); |
71e3aac07 thp: transparent ... |
1070 |
/* leave pmd empty until pte is filled */ |
e3ebcf643 thp: remove assum... |
1071 |
pgtable = pgtable_trans_huge_withdraw(mm); |
71e3aac07 thp: transparent ... |
1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 |
pmd_populate(mm, &_pmd, pgtable); for (i = 0; i < HPAGE_PMD_NR; i++, haddr += PAGE_SIZE) { pte_t *pte, entry; entry = mk_pte(pages[i], vma->vm_page_prot); entry = maybe_mkwrite(pte_mkdirty(entry), vma); page_add_new_anon_rmap(pages[i], vma, haddr); pte = pte_offset_map(&_pmd, haddr); VM_BUG_ON(!pte_none(*pte)); set_pte_at(mm, haddr, pte, entry); pte_unmap(pte); } kfree(pages); |
71e3aac07 thp: transparent ... |
1085 1086 1087 1088 |
smp_wmb(); /* make pte visible before pmd */ pmd_populate(mm, pmd, pgtable); page_remove_rmap(page); spin_unlock(&mm->page_table_lock); |
2ec74c3ef mm: move all mmu ... |
1089 |
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
71e3aac07 thp: transparent ... |
1090 1091 1092 1093 1094 1095 1096 1097 |
ret |= VM_FAULT_WRITE; put_page(page); out: return ret; out_free_pages: spin_unlock(&mm->page_table_lock); |
2ec74c3ef mm: move all mmu ... |
1098 |
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
b9bbfbe30 thp: memcg huge m... |
1099 1100 1101 |
mem_cgroup_uncharge_start(); for (i = 0; i < HPAGE_PMD_NR; i++) { mem_cgroup_uncharge_page(pages[i]); |
71e3aac07 thp: transparent ... |
1102 |
put_page(pages[i]); |
b9bbfbe30 thp: memcg huge m... |
1103 1104 |
} mem_cgroup_uncharge_end(); |
71e3aac07 thp: transparent ... |
1105 1106 1107 1108 1109 1110 1111 1112 |
kfree(pages); goto out; } int do_huge_pmd_wp_page(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long address, pmd_t *pmd, pmd_t orig_pmd) { int ret = 0; |
93b4796de thp: do_huge_pmd_... |
1113 |
struct page *page = NULL, *new_page; |
71e3aac07 thp: transparent ... |
1114 |
unsigned long haddr; |
2ec74c3ef mm: move all mmu ... |
1115 1116 |
unsigned long mmun_start; /* For mmu_notifiers */ unsigned long mmun_end; /* For mmu_notifiers */ |
71e3aac07 thp: transparent ... |
1117 1118 |
VM_BUG_ON(!vma->anon_vma); |
93b4796de thp: do_huge_pmd_... |
1119 1120 1121 |
haddr = address & HPAGE_PMD_MASK; if (is_huge_zero_pmd(orig_pmd)) goto alloc; |
71e3aac07 thp: transparent ... |
1122 1123 1124 1125 1126 1127 |
spin_lock(&mm->page_table_lock); if (unlikely(!pmd_same(*pmd, orig_pmd))) goto out_unlock; page = pmd_page(orig_pmd); VM_BUG_ON(!PageCompound(page) || !PageHead(page)); |
71e3aac07 thp: transparent ... |
1128 1129 1130 1131 1132 |
if (page_mapcount(page) == 1) { pmd_t entry; entry = pmd_mkyoung(orig_pmd); entry = maybe_pmd_mkwrite(pmd_mkdirty(entry), vma); if (pmdp_set_access_flags(vma, haddr, pmd, entry, 1)) |
b113da657 mm: Add and use u... |
1133 |
update_mmu_cache_pmd(vma, address, pmd); |
71e3aac07 thp: transparent ... |
1134 1135 1136 1137 1138 |
ret |= VM_FAULT_WRITE; goto out_unlock; } get_page(page); spin_unlock(&mm->page_table_lock); |
93b4796de thp: do_huge_pmd_... |
1139 |
alloc: |
71e3aac07 thp: transparent ... |
1140 1141 |
if (transparent_hugepage_enabled(vma) && !transparent_hugepage_debug_cow()) |
0bbbc0b33 thp: add numa awa... |
1142 |
new_page = alloc_hugepage_vma(transparent_hugepage_defrag(vma), |
cc5d462f7 mm: use __GFP_OTH... |
1143 |
vma, haddr, numa_node_id(), 0); |
71e3aac07 thp: transparent ... |
1144 1145 1146 1147 |
else new_page = NULL; if (unlikely(!new_page)) { |
81ab4201f mm: add VM counte... |
1148 |
count_vm_event(THP_FAULT_FALLBACK); |
93b4796de thp: do_huge_pmd_... |
1149 1150 |
if (is_huge_zero_pmd(orig_pmd)) { ret = do_huge_pmd_wp_zero_page_fallback(mm, vma, |
3ea41e621 thp: avoid race o... |
1151 |
address, pmd, orig_pmd, haddr); |
93b4796de thp: do_huge_pmd_... |
1152 1153 1154 1155 1156 1157 1158 |
} else { ret = do_huge_pmd_wp_page_fallback(mm, vma, address, pmd, orig_pmd, page, haddr); if (ret & VM_FAULT_OOM) split_huge_page(page); put_page(page); } |
71e3aac07 thp: transparent ... |
1159 1160 |
goto out; } |
81ab4201f mm: add VM counte... |
1161 |
count_vm_event(THP_FAULT_ALLOC); |
71e3aac07 thp: transparent ... |
1162 |
|
b9bbfbe30 thp: memcg huge m... |
1163 1164 |
if (unlikely(mem_cgroup_newpage_charge(new_page, mm, GFP_KERNEL))) { put_page(new_page); |
93b4796de thp: do_huge_pmd_... |
1165 1166 1167 1168 |
if (page) { split_huge_page(page); put_page(page); } |
b9bbfbe30 thp: memcg huge m... |
1169 1170 1171 |
ret |= VM_FAULT_OOM; goto out; } |
93b4796de thp: do_huge_pmd_... |
1172 1173 1174 1175 |
if (is_huge_zero_pmd(orig_pmd)) clear_huge_page(new_page, haddr, HPAGE_PMD_NR); else copy_user_huge_page(new_page, page, haddr, vma, HPAGE_PMD_NR); |
71e3aac07 thp: transparent ... |
1176 |
__SetPageUptodate(new_page); |
2ec74c3ef mm: move all mmu ... |
1177 1178 1179 |
mmun_start = haddr; mmun_end = haddr + HPAGE_PMD_SIZE; mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); |
71e3aac07 thp: transparent ... |
1180 |
spin_lock(&mm->page_table_lock); |
93b4796de thp: do_huge_pmd_... |
1181 1182 |
if (page) put_page(page); |
b9bbfbe30 thp: memcg huge m... |
1183 |
if (unlikely(!pmd_same(*pmd, orig_pmd))) { |
6f60b69d8 mm, thp: drop pag... |
1184 |
spin_unlock(&mm->page_table_lock); |
b9bbfbe30 thp: memcg huge m... |
1185 |
mem_cgroup_uncharge_page(new_page); |
71e3aac07 thp: transparent ... |
1186 |
put_page(new_page); |
2ec74c3ef mm: move all mmu ... |
1187 |
goto out_mn; |
b9bbfbe30 thp: memcg huge m... |
1188 |
} else { |
71e3aac07 thp: transparent ... |
1189 |
pmd_t entry; |
b3092b3b7 thp: cleanup: int... |
1190 |
entry = mk_huge_pmd(new_page, vma); |
2ec74c3ef mm: move all mmu ... |
1191 |
pmdp_clear_flush(vma, haddr, pmd); |
71e3aac07 thp: transparent ... |
1192 1193 |
page_add_new_anon_rmap(new_page, vma, haddr); set_pmd_at(mm, haddr, pmd, entry); |
b113da657 mm: Add and use u... |
1194 |
update_mmu_cache_pmd(vma, address, pmd); |
97ae17497 thp: implement re... |
1195 |
if (is_huge_zero_pmd(orig_pmd)) { |
93b4796de thp: do_huge_pmd_... |
1196 |
add_mm_counter(mm, MM_ANONPAGES, HPAGE_PMD_NR); |
97ae17497 thp: implement re... |
1197 1198 |
put_huge_zero_page(); } else { |
93b4796de thp: do_huge_pmd_... |
1199 1200 1201 1202 |
VM_BUG_ON(!PageHead(page)); page_remove_rmap(page); put_page(page); } |
71e3aac07 thp: transparent ... |
1203 1204 |
ret |= VM_FAULT_WRITE; } |
71e3aac07 thp: transparent ... |
1205 |
spin_unlock(&mm->page_table_lock); |
2ec74c3ef mm: move all mmu ... |
1206 1207 |
out_mn: mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
71e3aac07 thp: transparent ... |
1208 1209 |
out: return ret; |
2ec74c3ef mm: move all mmu ... |
1210 1211 1212 |
out_unlock: spin_unlock(&mm->page_table_lock); return ret; |
71e3aac07 thp: transparent ... |
1213 |
} |
b676b293f mm, thp: fix mapp... |
1214 |
struct page *follow_trans_huge_pmd(struct vm_area_struct *vma, |
71e3aac07 thp: transparent ... |
1215 1216 1217 1218 |
unsigned long addr, pmd_t *pmd, unsigned int flags) { |
b676b293f mm, thp: fix mapp... |
1219 |
struct mm_struct *mm = vma->vm_mm; |
71e3aac07 thp: transparent ... |
1220 1221 1222 1223 1224 1225 |
struct page *page = NULL; assert_spin_locked(&mm->page_table_lock); if (flags & FOLL_WRITE && !pmd_write(*pmd)) goto out; |
85facf257 thp: avoid dumpin... |
1226 1227 1228 |
/* Avoid dumping huge zero page */ if ((flags & FOLL_DUMP) && is_huge_zero_pmd(*pmd)) return ERR_PTR(-EFAULT); |
71e3aac07 thp: transparent ... |
1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 |
page = pmd_page(*pmd); VM_BUG_ON(!PageHead(page)); if (flags & FOLL_TOUCH) { pmd_t _pmd; /* * We should set the dirty bit only for FOLL_WRITE but * for now the dirty bit in the pmd is meaningless. * And if the dirty bit will become meaningful and * we'll only set it with FOLL_WRITE, an atomic * set_bit will be required on the pmd to set the * young bit, instead of the current set_pmd_at. */ _pmd = pmd_mkyoung(pmd_mkdirty(*pmd)); set_pmd_at(mm, addr & HPAGE_PMD_MASK, pmd, _pmd); } |
b676b293f mm, thp: fix mapp... |
1244 1245 1246 1247 1248 1249 1250 1251 |
if ((flags & FOLL_MLOCK) && (vma->vm_flags & VM_LOCKED)) { if (page->mapping && trylock_page(page)) { lru_add_drain(); if (page->mapping) mlock_vma_page(page); unlock_page(page); } } |
71e3aac07 thp: transparent ... |
1252 1253 1254 |
page += (addr & ~HPAGE_PMD_MASK) >> PAGE_SHIFT; VM_BUG_ON(!PageCompound(page)); if (flags & FOLL_GET) |
70b50f94f mm: thp: tail pag... |
1255 |
get_page_foll(page); |
71e3aac07 thp: transparent ... |
1256 1257 1258 1259 |
out: return page; } |
d10e63f29 mm: numa: Create ... |
1260 |
/* NUMA hinting page fault entry point for trans huge pmds */ |
4daae3b4b mm: mempolicy: Us... |
1261 1262 |
int do_huge_pmd_numa_page(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, pmd_t pmd, pmd_t *pmdp) |
d10e63f29 mm: numa: Create ... |
1263 |
{ |
b32967ff1 mm: numa: Add THP... |
1264 |
struct page *page; |
d10e63f29 mm: numa: Create ... |
1265 |
unsigned long haddr = addr & HPAGE_PMD_MASK; |
4daae3b4b mm: mempolicy: Us... |
1266 |
int target_nid; |
03c5a6e16 mm: numa: Add pte... |
1267 |
int current_nid = -1; |
b32967ff1 mm: numa: Add THP... |
1268 |
bool migrated; |
d10e63f29 mm: numa: Create ... |
1269 1270 1271 1272 1273 1274 |
spin_lock(&mm->page_table_lock); if (unlikely(!pmd_same(pmd, *pmdp))) goto out_unlock; page = pmd_page(pmd); |
4daae3b4b mm: mempolicy: Us... |
1275 |
get_page(page); |
03c5a6e16 mm: numa: Add pte... |
1276 1277 1278 1279 |
current_nid = page_to_nid(page); count_vm_numa_event(NUMA_HINT_FAULTS); if (current_nid == numa_node_id()) count_vm_numa_event(NUMA_HINT_FAULTS_LOCAL); |
4daae3b4b mm: mempolicy: Us... |
1280 1281 |
target_nid = mpol_misplaced(page, vma, haddr); |
b32967ff1 mm: numa: Add THP... |
1282 1283 |
if (target_nid == -1) { put_page(page); |
4daae3b4b mm: mempolicy: Us... |
1284 |
goto clear_pmdnuma; |
b32967ff1 mm: numa: Add THP... |
1285 |
} |
4daae3b4b mm: mempolicy: Us... |
1286 |
|
b32967ff1 mm: numa: Add THP... |
1287 1288 1289 |
/* Acquire the page lock to serialise THP migrations */ spin_unlock(&mm->page_table_lock); lock_page(page); |
4daae3b4b mm: mempolicy: Us... |
1290 |
|
b32967ff1 mm: numa: Add THP... |
1291 |
/* Confirm the PTE did not while locked */ |
4daae3b4b mm: mempolicy: Us... |
1292 |
spin_lock(&mm->page_table_lock); |
b32967ff1 mm: numa: Add THP... |
1293 1294 1295 |
if (unlikely(!pmd_same(pmd, *pmdp))) { unlock_page(page); put_page(page); |
4daae3b4b mm: mempolicy: Us... |
1296 |
goto out_unlock; |
b32967ff1 mm: numa: Add THP... |
1297 1298 |
} spin_unlock(&mm->page_table_lock); |
4daae3b4b mm: mempolicy: Us... |
1299 |
|
b32967ff1 mm: numa: Add THP... |
1300 1301 |
/* Migrate the THP to the requested node */ migrated = migrate_misplaced_transhuge_page(mm, vma, |
340ef3902 mm: numa: cleanup... |
1302 1303 1304 |
pmdp, pmd, addr, page, target_nid); if (!migrated) goto check_same; |
b32967ff1 mm: numa: Add THP... |
1305 |
|
340ef3902 mm: numa: cleanup... |
1306 |
task_numa_fault(target_nid, HPAGE_PMD_NR, true); |
b32967ff1 mm: numa: Add THP... |
1307 |
return 0; |
340ef3902 mm: numa: cleanup... |
1308 1309 1310 1311 |
check_same: spin_lock(&mm->page_table_lock); if (unlikely(!pmd_same(pmd, *pmdp))) goto out_unlock; |
b32967ff1 mm: numa: Add THP... |
1312 |
clear_pmdnuma: |
d10e63f29 mm: numa: Create ... |
1313 1314 1315 1316 |
pmd = pmd_mknonnuma(pmd); set_pmd_at(mm, haddr, pmdp, pmd); VM_BUG_ON(pmd_numa(*pmdp)); update_mmu_cache_pmd(vma, addr, pmdp); |
d10e63f29 mm: numa: Create ... |
1317 1318 |
out_unlock: spin_unlock(&mm->page_table_lock); |
b32967ff1 mm: numa: Add THP... |
1319 |
if (current_nid != -1) |
340ef3902 mm: numa: cleanup... |
1320 |
task_numa_fault(current_nid, HPAGE_PMD_NR, false); |
d10e63f29 mm: numa: Create ... |
1321 1322 |
return 0; } |
71e3aac07 thp: transparent ... |
1323 |
int zap_huge_pmd(struct mmu_gather *tlb, struct vm_area_struct *vma, |
f21760b15 thp: add tlb_remo... |
1324 |
pmd_t *pmd, unsigned long addr) |
71e3aac07 thp: transparent ... |
1325 1326 |
{ int ret = 0; |
025c5b245 thp: optimize awa... |
1327 1328 1329 |
if (__pmd_trans_huge_lock(pmd, vma) == 1) { struct page *page; pgtable_t pgtable; |
f5c8ad472 mm: thp: Use more... |
1330 |
pmd_t orig_pmd; |
e3ebcf643 thp: remove assum... |
1331 |
pgtable = pgtable_trans_huge_withdraw(tlb->mm); |
f5c8ad472 mm: thp: Use more... |
1332 |
orig_pmd = pmdp_get_and_clear(tlb->mm, addr, pmd); |
025c5b245 thp: optimize awa... |
1333 |
tlb_remove_pmd_tlb_entry(tlb, pmd, addr); |
479f0abbf thp: zap_huge_pmd... |
1334 1335 1336 |
if (is_huge_zero_pmd(orig_pmd)) { tlb->mm->nr_ptes--; spin_unlock(&tlb->mm->page_table_lock); |
97ae17497 thp: implement re... |
1337 |
put_huge_zero_page(); |
479f0abbf thp: zap_huge_pmd... |
1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 |
} else { page = pmd_page(orig_pmd); page_remove_rmap(page); VM_BUG_ON(page_mapcount(page) < 0); add_mm_counter(tlb->mm, MM_ANONPAGES, -HPAGE_PMD_NR); VM_BUG_ON(!PageHead(page)); tlb->mm->nr_ptes--; spin_unlock(&tlb->mm->page_table_lock); tlb_remove_page(tlb, page); } |
025c5b245 thp: optimize awa... |
1348 1349 1350 |
pte_free(tlb->mm, pgtable); ret = 1; } |
71e3aac07 thp: transparent ... |
1351 1352 |
return ret; } |
0ca1634d4 thp: mincore tran... |
1353 1354 1355 1356 1357 |
int mincore_huge_pmd(struct vm_area_struct *vma, pmd_t *pmd, unsigned long addr, unsigned long end, unsigned char *vec) { int ret = 0; |
025c5b245 thp: optimize awa... |
1358 1359 1360 1361 1362 |
if (__pmd_trans_huge_lock(pmd, vma) == 1) { /* * All logical pages in the range are present * if backed by a huge page. */ |
0ca1634d4 thp: mincore tran... |
1363 |
spin_unlock(&vma->vm_mm->page_table_lock); |
025c5b245 thp: optimize awa... |
1364 1365 1366 |
memset(vec, 1, (end - addr) >> PAGE_SHIFT); ret = 1; } |
0ca1634d4 thp: mincore tran... |
1367 1368 1369 |
return ret; } |
37a1c49a9 thp: mremap suppo... |
1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 |
int move_huge_pmd(struct vm_area_struct *vma, struct vm_area_struct *new_vma, unsigned long old_addr, unsigned long new_addr, unsigned long old_end, pmd_t *old_pmd, pmd_t *new_pmd) { int ret = 0; pmd_t pmd; struct mm_struct *mm = vma->vm_mm; if ((old_addr & ~HPAGE_PMD_MASK) || (new_addr & ~HPAGE_PMD_MASK) || old_end - old_addr < HPAGE_PMD_SIZE || (new_vma->vm_flags & VM_NOHUGEPAGE)) goto out; /* * The destination pmd shouldn't be established, free_pgtables() * should have release it. */ if (WARN_ON(!pmd_none(*new_pmd))) { VM_BUG_ON(pmd_trans_huge(*new_pmd)); goto out; } |
025c5b245 thp: optimize awa... |
1394 1395 1396 1397 1398 |
ret = __pmd_trans_huge_lock(old_pmd, vma); if (ret == 1) { pmd = pmdp_get_and_clear(mm, old_addr, old_pmd); VM_BUG_ON(!pmd_none(*new_pmd)); set_pmd_at(mm, new_addr, new_pmd, pmd); |
37a1c49a9 thp: mremap suppo... |
1399 1400 1401 1402 1403 |
spin_unlock(&mm->page_table_lock); } out: return ret; } |
cd7548ab3 thp: mprotect: tr... |
1404 |
int change_huge_pmd(struct vm_area_struct *vma, pmd_t *pmd, |
4b10e7d56 mm: mempolicy: Im... |
1405 |
unsigned long addr, pgprot_t newprot, int prot_numa) |
cd7548ab3 thp: mprotect: tr... |
1406 1407 1408 |
{ struct mm_struct *mm = vma->vm_mm; int ret = 0; |
025c5b245 thp: optimize awa... |
1409 1410 1411 |
if (__pmd_trans_huge_lock(pmd, vma) == 1) { pmd_t entry; entry = pmdp_get_and_clear(mm, addr, pmd); |
a4f1de176 mm: fix kernel BU... |
1412 |
if (!prot_numa) { |
4b10e7d56 mm: mempolicy: Im... |
1413 |
entry = pmd_modify(entry, newprot); |
a4f1de176 mm: fix kernel BU... |
1414 1415 |
BUG_ON(pmd_write(entry)); } else { |
4b10e7d56 mm: mempolicy: Im... |
1416 1417 1418 1419 1420 1421 1422 1423 |
struct page *page = pmd_page(*pmd); /* only check non-shared pages */ if (page_mapcount(page) == 1 && !pmd_numa(*pmd)) { entry = pmd_mknuma(entry); } } |
025c5b245 thp: optimize awa... |
1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 |
set_pmd_at(mm, addr, pmd, entry); spin_unlock(&vma->vm_mm->page_table_lock); ret = 1; } return ret; } /* * Returns 1 if a given pmd maps a stable (not under splitting) thp. * Returns -1 if it maps a thp under splitting. Returns 0 otherwise. * * Note that if it returns 1, this routine returns without unlocking page * table locks. So callers must unlock them. */ int __pmd_trans_huge_lock(pmd_t *pmd, struct vm_area_struct *vma) { spin_lock(&vma->vm_mm->page_table_lock); |
cd7548ab3 thp: mprotect: tr... |
1442 1443 |
if (likely(pmd_trans_huge(*pmd))) { if (unlikely(pmd_trans_splitting(*pmd))) { |
025c5b245 thp: optimize awa... |
1444 |
spin_unlock(&vma->vm_mm->page_table_lock); |
cd7548ab3 thp: mprotect: tr... |
1445 |
wait_split_huge_page(vma->anon_vma, pmd); |
025c5b245 thp: optimize awa... |
1446 |
return -1; |
cd7548ab3 thp: mprotect: tr... |
1447 |
} else { |
025c5b245 thp: optimize awa... |
1448 1449 1450 |
/* Thp mapped by 'pmd' is stable, so we can * handle it as it is. */ return 1; |
cd7548ab3 thp: mprotect: tr... |
1451 |
} |
025c5b245 thp: optimize awa... |
1452 1453 1454 |
} spin_unlock(&vma->vm_mm->page_table_lock); return 0; |
cd7548ab3 thp: mprotect: tr... |
1455 |
} |
71e3aac07 thp: transparent ... |
1456 1457 1458 1459 1460 |
pmd_t *page_check_address_pmd(struct page *page, struct mm_struct *mm, unsigned long address, enum page_check_address_pmd_flag flag) { |
71e3aac07 thp: transparent ... |
1461 1462 1463 1464 |
pmd_t *pmd, *ret = NULL; if (address & ~HPAGE_PMD_MASK) goto out; |
6219049ae mm: introduce mm_... |
1465 1466 |
pmd = mm_find_pmd(mm, address); if (!pmd) |
71e3aac07 thp: transparent ... |
1467 |
goto out; |
71e3aac07 thp: transparent ... |
1468 1469 1470 1471 |
if (pmd_none(*pmd)) goto out; if (pmd_page(*pmd) != page) goto out; |
94fcc585f thp: avoid breaki... |
1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 |
/* * split_vma() may create temporary aliased mappings. There is * no risk as long as all huge pmd are found and have their * splitting bit set before __split_huge_page_refcount * runs. Finding the same huge pmd more than once during the * same rmap walk is not a problem. */ if (flag == PAGE_CHECK_ADDRESS_PMD_NOTSPLITTING_FLAG && pmd_trans_splitting(*pmd)) goto out; |
71e3aac07 thp: transparent ... |
1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 |
if (pmd_trans_huge(*pmd)) { VM_BUG_ON(flag == PAGE_CHECK_ADDRESS_PMD_SPLITTING_FLAG && !pmd_trans_splitting(*pmd)); ret = pmd; } out: return ret; } static int __split_huge_page_splitting(struct page *page, struct vm_area_struct *vma, unsigned long address) { struct mm_struct *mm = vma->vm_mm; pmd_t *pmd; int ret = 0; |
2ec74c3ef mm: move all mmu ... |
1498 1499 1500 |
/* For mmu_notifiers */ const unsigned long mmun_start = address; const unsigned long mmun_end = address + HPAGE_PMD_SIZE; |
71e3aac07 thp: transparent ... |
1501 |
|
2ec74c3ef mm: move all mmu ... |
1502 |
mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); |
71e3aac07 thp: transparent ... |
1503 1504 1505 1506 1507 1508 1509 1510 |
spin_lock(&mm->page_table_lock); pmd = page_check_address_pmd(page, mm, address, PAGE_CHECK_ADDRESS_PMD_NOTSPLITTING_FLAG); if (pmd) { /* * We can't temporarily set the pmd to null in order * to split it, the pmd must remain marked huge at all * times or the VM won't take the pmd_trans_huge paths |
5a505085f mm/rmap: Convert ... |
1511 |
* and it won't wait on the anon_vma->root->rwsem to |
71e3aac07 thp: transparent ... |
1512 1513 |
* serialize against split_huge_page*. */ |
2ec74c3ef mm: move all mmu ... |
1514 |
pmdp_splitting_flush(vma, address, pmd); |
71e3aac07 thp: transparent ... |
1515 1516 1517 |
ret = 1; } spin_unlock(&mm->page_table_lock); |
2ec74c3ef mm: move all mmu ... |
1518 |
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
71e3aac07 thp: transparent ... |
1519 1520 1521 |
return ret; } |
5bc7b8aca mm: thp: add spli... |
1522 1523 |
static void __split_huge_page_refcount(struct page *page, struct list_head *list) |
71e3aac07 thp: transparent ... |
1524 1525 |
{ int i; |
71e3aac07 thp: transparent ... |
1526 |
struct zone *zone = page_zone(page); |
fa9add641 mm/memcg: apply a... |
1527 |
struct lruvec *lruvec; |
70b50f94f mm: thp: tail pag... |
1528 |
int tail_count = 0; |
71e3aac07 thp: transparent ... |
1529 1530 1531 |
/* prevent PageLRU to go away from under us, and freeze lru stats */ spin_lock_irq(&zone->lru_lock); |
fa9add641 mm/memcg: apply a... |
1532 |
lruvec = mem_cgroup_page_lruvec(page, zone); |
71e3aac07 thp: transparent ... |
1533 |
compound_lock(page); |
e94c8a9cb memcg: make mem_c... |
1534 1535 |
/* complete memcg works before add pages to LRU */ mem_cgroup_split_huge_fixup(page); |
71e3aac07 thp: transparent ... |
1536 |
|
45676885b thp: improve orde... |
1537 |
for (i = HPAGE_PMD_NR - 1; i >= 1; i--) { |
71e3aac07 thp: transparent ... |
1538 |
struct page *page_tail = page + i; |
70b50f94f mm: thp: tail pag... |
1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 |
/* tail_page->_mapcount cannot change */ BUG_ON(page_mapcount(page_tail) < 0); tail_count += page_mapcount(page_tail); /* check for overflow */ BUG_ON(tail_count < 0); BUG_ON(atomic_read(&page_tail->_count) != 0); /* * tail_page->_count is zero and not changing from * under us. But get_page_unless_zero() may be running * from under us on the tail_page. If we used * atomic_set() below instead of atomic_add(), we * would then run atomic_set() concurrently with * get_page_unless_zero(), and atomic_set() is * implemented in C not using locked ops. spin_unlock * on x86 sometime uses locked ops because of PPro * errata 66, 92, so unless somebody can guarantee * atomic_set() here would be safe on all archs (and * not only on x86), it's safer to use atomic_add(). */ atomic_add(page_mapcount(page) + page_mapcount(page_tail) + 1, &page_tail->_count); |
71e3aac07 thp: transparent ... |
1560 1561 1562 |
/* after clearing PageTail the gup refcount can be released */ smp_mb(); |
a6d30ddda thp: fix the wron... |
1563 1564 1565 1566 1567 1568 |
/* * retain hwpoison flag of the poisoned tail page: * fix for the unsuitable process killed on Guest Machine(KVM) * by the memory-failure. */ page_tail->flags &= ~PAGE_FLAGS_CHECK_AT_PREP | __PG_HWPOISON; |
71e3aac07 thp: transparent ... |
1569 1570 1571 1572 1573 1574 |
page_tail->flags |= (page->flags & ((1L << PG_referenced) | (1L << PG_swapbacked) | (1L << PG_mlocked) | (1L << PG_uptodate))); page_tail->flags |= (1L << PG_dirty); |
70b50f94f mm: thp: tail pag... |
1575 |
/* clear PageTail before overwriting first_page */ |
71e3aac07 thp: transparent ... |
1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 |
smp_wmb(); /* * __split_huge_page_splitting() already set the * splitting bit in all pmd that could map this * hugepage, that will ensure no CPU can alter the * mapcount on the head page. The mapcount is only * accounted in the head page and it has to be * transferred to all tail pages in the below code. So * for this code to be safe, the split the mapcount * can't change. But that doesn't mean userland can't * keep changing and reading the page contents while * we transfer the mapcount, so the pmd splitting * status is achieved setting a reserved bit in the * pmd, not by clearing the present bit. */ |
71e3aac07 thp: transparent ... |
1592 1593 1594 1595 |
page_tail->_mapcount = page->_mapcount; BUG_ON(page_tail->mapping); page_tail->mapping = page->mapping; |
45676885b thp: improve orde... |
1596 |
page_tail->index = page->index + i; |
22b751c3d mm: rename page s... |
1597 |
page_nid_xchg_last(page_tail, page_nid_last(page)); |
71e3aac07 thp: transparent ... |
1598 1599 1600 1601 1602 |
BUG_ON(!PageAnon(page_tail)); BUG_ON(!PageUptodate(page_tail)); BUG_ON(!PageDirty(page_tail)); BUG_ON(!PageSwapBacked(page_tail)); |
5bc7b8aca mm: thp: add spli... |
1603 |
lru_add_page_tail(page, page_tail, lruvec, list); |
71e3aac07 thp: transparent ... |
1604 |
} |
70b50f94f mm: thp: tail pag... |
1605 1606 |
atomic_sub(tail_count, &page->_count); BUG_ON(atomic_read(&page->_count) <= 0); |
71e3aac07 thp: transparent ... |
1607 |
|
fa9add641 mm/memcg: apply a... |
1608 |
__mod_zone_page_state(zone, NR_ANON_TRANSPARENT_HUGEPAGES, -1); |
79134171d thp: transparent ... |
1609 |
__mod_zone_page_state(zone, NR_ANON_PAGES, HPAGE_PMD_NR); |
71e3aac07 thp: transparent ... |
1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 |
ClearPageCompound(page); compound_unlock(page); spin_unlock_irq(&zone->lru_lock); for (i = 1; i < HPAGE_PMD_NR; i++) { struct page *page_tail = page + i; BUG_ON(page_count(page_tail) <= 0); /* * Tail pages may be freed if there wasn't any mapping * like if add_to_swap() is running on a lru page that * had its mapping zapped. And freeing these pages * requires taking the lru_lock so we do the put_page * of the tail pages after the split is complete. */ put_page(page_tail); } /* * Only the head page (now become a regular page) is required * to be pinned by the caller. */ BUG_ON(page_count(page) <= 0); } static int __split_huge_page_map(struct page *page, struct vm_area_struct *vma, unsigned long address) { struct mm_struct *mm = vma->vm_mm; pmd_t *pmd, _pmd; int ret = 0, i; pgtable_t pgtable; unsigned long haddr; spin_lock(&mm->page_table_lock); pmd = page_check_address_pmd(page, mm, address, PAGE_CHECK_ADDRESS_PMD_SPLITTING_FLAG); if (pmd) { |
e3ebcf643 thp: remove assum... |
1648 |
pgtable = pgtable_trans_huge_withdraw(mm); |
71e3aac07 thp: transparent ... |
1649 |
pmd_populate(mm, &_pmd, pgtable); |
e3ebcf643 thp: remove assum... |
1650 1651 |
haddr = address; for (i = 0; i < HPAGE_PMD_NR; i++, haddr += PAGE_SIZE) { |
71e3aac07 thp: transparent ... |
1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 |
pte_t *pte, entry; BUG_ON(PageCompound(page+i)); entry = mk_pte(page + i, vma->vm_page_prot); entry = maybe_mkwrite(pte_mkdirty(entry), vma); if (!pmd_write(*pmd)) entry = pte_wrprotect(entry); else BUG_ON(page_mapcount(page) != 1); if (!pmd_young(*pmd)) entry = pte_mkold(entry); |
1ba6e0b50 mm: numa: split_h... |
1662 1663 |
if (pmd_numa(*pmd)) entry = pte_mknuma(entry); |
71e3aac07 thp: transparent ... |
1664 1665 1666 1667 1668 |
pte = pte_offset_map(&_pmd, haddr); BUG_ON(!pte_none(*pte)); set_pte_at(mm, haddr, pte, entry); pte_unmap(pte); } |
71e3aac07 thp: transparent ... |
1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 |
smp_wmb(); /* make pte visible before pmd */ /* * Up to this point the pmd is present and huge and * userland has the whole access to the hugepage * during the split (which happens in place). If we * overwrite the pmd with the not-huge version * pointing to the pte here (which of course we could * if all CPUs were bug free), userland could trigger * a small page size TLB miss on the small sized TLB * while the hugepage TLB entry is still established * in the huge TLB. Some CPU doesn't like that. See * http://support.amd.com/us/Processor_TechDocs/41322.pdf, * Erratum 383 on page 93. Intel should be safe but is * also warns that it's only safe if the permission * and cache attributes of the two entries loaded in * the two TLB is identical (which should be the case * here). But it is generally safer to never allow * small and huge TLB entries for the same virtual * address to be loaded simultaneously. So instead of * doing "pmd_populate(); flush_tlb_range();" we first * mark the current pmd notpresent (atomically because * here the pmd_trans_huge and pmd_trans_splitting * must remain set at all times on the pmd until the * split is complete for this pmd), then we flush the * SMP TLB and finally we write the non-huge version * of the pmd entry with pmd_populate. */ |
46dcde735 thp: introduce pm... |
1696 |
pmdp_invalidate(vma, address, pmd); |
71e3aac07 thp: transparent ... |
1697 1698 1699 1700 1701 1702 1703 |
pmd_populate(mm, pmd, pgtable); ret = 1; } spin_unlock(&mm->page_table_lock); return ret; } |
5a505085f mm/rmap: Convert ... |
1704 |
/* must be called with anon_vma->root->rwsem held */ |
71e3aac07 thp: transparent ... |
1705 |
static void __split_huge_page(struct page *page, |
5bc7b8aca mm: thp: add spli... |
1706 1707 |
struct anon_vma *anon_vma, struct list_head *list) |
71e3aac07 thp: transparent ... |
1708 1709 |
{ int mapcount, mapcount2; |
bf181b9f9 mm anon rmap: rep... |
1710 |
pgoff_t pgoff = page->index << (PAGE_CACHE_SHIFT - PAGE_SHIFT); |
71e3aac07 thp: transparent ... |
1711 1712 1713 1714 1715 1716 |
struct anon_vma_chain *avc; BUG_ON(!PageHead(page)); BUG_ON(PageTail(page)); mapcount = 0; |
bf181b9f9 mm anon rmap: rep... |
1717 |
anon_vma_interval_tree_foreach(avc, &anon_vma->rb_root, pgoff, pgoff) { |
71e3aac07 thp: transparent ... |
1718 1719 1720 |
struct vm_area_struct *vma = avc->vma; unsigned long addr = vma_address(page, vma); BUG_ON(is_vma_temporary_stack(vma)); |
71e3aac07 thp: transparent ... |
1721 1722 |
mapcount += __split_huge_page_splitting(page, vma, addr); } |
05759d380 thp: split_huge_p... |
1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 |
/* * It is critical that new vmas are added to the tail of the * anon_vma list. This guarantes that if copy_huge_pmd() runs * and establishes a child pmd before * __split_huge_page_splitting() freezes the parent pmd (so if * we fail to prevent copy_huge_pmd() from running until the * whole __split_huge_page() is complete), we will still see * the newly established pmd of the child later during the * walk, to be able to set it as pmd_trans_splitting too. */ if (mapcount != page_mapcount(page)) printk(KERN_ERR "mapcount %d page_mapcount %d ", mapcount, page_mapcount(page)); |
71e3aac07 thp: transparent ... |
1737 |
BUG_ON(mapcount != page_mapcount(page)); |
5bc7b8aca mm: thp: add spli... |
1738 |
__split_huge_page_refcount(page, list); |
71e3aac07 thp: transparent ... |
1739 1740 |
mapcount2 = 0; |
bf181b9f9 mm anon rmap: rep... |
1741 |
anon_vma_interval_tree_foreach(avc, &anon_vma->rb_root, pgoff, pgoff) { |
71e3aac07 thp: transparent ... |
1742 1743 1744 |
struct vm_area_struct *vma = avc->vma; unsigned long addr = vma_address(page, vma); BUG_ON(is_vma_temporary_stack(vma)); |
71e3aac07 thp: transparent ... |
1745 1746 |
mapcount2 += __split_huge_page_map(page, vma, addr); } |
05759d380 thp: split_huge_p... |
1747 1748 1749 1750 |
if (mapcount != mapcount2) printk(KERN_ERR "mapcount %d mapcount2 %d page_mapcount %d ", mapcount, mapcount2, page_mapcount(page)); |
71e3aac07 thp: transparent ... |
1751 1752 |
BUG_ON(mapcount != mapcount2); } |
5bc7b8aca mm: thp: add spli... |
1753 1754 1755 1756 1757 1758 1759 1760 |
/* * Split a hugepage into normal pages. This doesn't change the position of head * page. If @list is null, tail pages will be added to LRU list, otherwise, to * @list. Both head page and tail pages will inherit mapping, flags, and so on * from the hugepage. * Return 0 if the hugepage is split successfully otherwise return 1. */ int split_huge_page_to_list(struct page *page, struct list_head *list) |
71e3aac07 thp: transparent ... |
1761 1762 1763 |
{ struct anon_vma *anon_vma; int ret = 1; |
5918d10a4 thp: fix huge zer... |
1764 |
BUG_ON(is_huge_zero_page(page)); |
71e3aac07 thp: transparent ... |
1765 |
BUG_ON(!PageAnon(page)); |
062f1af21 mm: thp: acquire ... |
1766 1767 1768 1769 1770 1771 1772 1773 1774 |
/* * The caller does not necessarily hold an mmap_sem that would prevent * the anon_vma disappearing so we first we take a reference to it * and then lock the anon_vma for write. This is similar to * page_lock_anon_vma_read except the write lock is taken to serialise * against parallel split or collapse operations. */ anon_vma = page_get_anon_vma(page); |
71e3aac07 thp: transparent ... |
1775 1776 |
if (!anon_vma) goto out; |
062f1af21 mm: thp: acquire ... |
1777 |
anon_vma_lock_write(anon_vma); |
71e3aac07 thp: transparent ... |
1778 1779 1780 1781 1782 |
ret = 0; if (!PageCompound(page)) goto out_unlock; BUG_ON(!PageSwapBacked(page)); |
5bc7b8aca mm: thp: add spli... |
1783 |
__split_huge_page(page, anon_vma, list); |
81ab4201f mm: add VM counte... |
1784 |
count_vm_event(THP_SPLIT); |
71e3aac07 thp: transparent ... |
1785 1786 1787 |
BUG_ON(PageCompound(page)); out_unlock: |
08b52706d mm/rmap: rename a... |
1788 |
anon_vma_unlock_write(anon_vma); |
062f1af21 mm: thp: acquire ... |
1789 |
put_anon_vma(anon_vma); |
71e3aac07 thp: transparent ... |
1790 1791 1792 |
out: return ret; } |
4b6e1e370 mm: kill vma flag... |
1793 |
#define VM_NO_THP (VM_SPECIAL|VM_MIXEDMAP|VM_HUGETLB|VM_SHARED|VM_MAYSHARE) |
78f11a255 mm: thp: fix /dev... |
1794 |
|
60ab3244e thp: khugepaged: ... |
1795 1796 |
int hugepage_madvise(struct vm_area_struct *vma, unsigned long *vm_flags, int advice) |
0af4e98b6 thp: madvise(MADV... |
1797 |
{ |
8e72033f2 thp: make MADV_HU... |
1798 |
struct mm_struct *mm = vma->vm_mm; |
a664b2d85 thp: madvise(MADV... |
1799 1800 1801 1802 1803 |
switch (advice) { case MADV_HUGEPAGE: /* * Be somewhat over-protective like KSM for now! */ |
78f11a255 mm: thp: fix /dev... |
1804 |
if (*vm_flags & (VM_HUGEPAGE | VM_NO_THP)) |
a664b2d85 thp: madvise(MADV... |
1805 |
return -EINVAL; |
8e72033f2 thp: make MADV_HU... |
1806 1807 |
if (mm->def_flags & VM_NOHUGEPAGE) return -EINVAL; |
a664b2d85 thp: madvise(MADV... |
1808 1809 |
*vm_flags &= ~VM_NOHUGEPAGE; *vm_flags |= VM_HUGEPAGE; |
60ab3244e thp: khugepaged: ... |
1810 1811 1812 1813 1814 1815 1816 |
/* * If the vma become good for khugepaged to scan, * register it here without waiting a page fault that * may not happen any time soon. */ if (unlikely(khugepaged_enter_vma_merge(vma))) return -ENOMEM; |
a664b2d85 thp: madvise(MADV... |
1817 1818 1819 1820 1821 |
break; case MADV_NOHUGEPAGE: /* * Be somewhat over-protective like KSM for now! */ |
78f11a255 mm: thp: fix /dev... |
1822 |
if (*vm_flags & (VM_NOHUGEPAGE | VM_NO_THP)) |
a664b2d85 thp: madvise(MADV... |
1823 1824 1825 |
return -EINVAL; *vm_flags &= ~VM_HUGEPAGE; *vm_flags |= VM_NOHUGEPAGE; |
60ab3244e thp: khugepaged: ... |
1826 1827 1828 1829 1830 |
/* * Setting VM_NOHUGEPAGE will prevent khugepaged from scanning * this vma even if we leave the mm registered in khugepaged if * it got registered before VM_NOHUGEPAGE was set. */ |
a664b2d85 thp: madvise(MADV... |
1831 1832 |
break; } |
0af4e98b6 thp: madvise(MADV... |
1833 1834 1835 |
return 0; } |
ba76149f4 thp: khugepaged |
1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 |
static int __init khugepaged_slab_init(void) { mm_slot_cache = kmem_cache_create("khugepaged_mm_slot", sizeof(struct mm_slot), __alignof__(struct mm_slot), 0, NULL); if (!mm_slot_cache) return -ENOMEM; return 0; } |
ba76149f4 thp: khugepaged |
1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 |
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); } |
ba76149f4 thp: khugepaged |
1857 1858 1859 |
static struct mm_slot *get_mm_slot(struct mm_struct *mm) { struct mm_slot *mm_slot; |
ba76149f4 thp: khugepaged |
1860 |
|
b67bfe0d4 hlist: drop the n... |
1861 |
hash_for_each_possible(mm_slots_hash, mm_slot, hash, (unsigned long)mm) |
ba76149f4 thp: khugepaged |
1862 1863 |
if (mm == mm_slot->mm) return mm_slot; |
43b5fbbd2 mm/huge_memory.c:... |
1864 |
|
ba76149f4 thp: khugepaged |
1865 1866 1867 1868 1869 1870 |
return NULL; } static void insert_to_mm_slots_hash(struct mm_struct *mm, struct mm_slot *mm_slot) { |
ba76149f4 thp: khugepaged |
1871 |
mm_slot->mm = mm; |
43b5fbbd2 mm/huge_memory.c:... |
1872 |
hash_add(mm_slots_hash, &mm_slot->hash, (long)mm); |
ba76149f4 thp: khugepaged |
1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 |
} static inline int khugepaged_test_exit(struct mm_struct *mm) { return atomic_read(&mm->mm_users) == 0; } int __khugepaged_enter(struct mm_struct *mm) { struct mm_slot *mm_slot; int wakeup; mm_slot = alloc_mm_slot(); if (!mm_slot) return -ENOMEM; /* __khugepaged_exit() must not run from under us */ VM_BUG_ON(khugepaged_test_exit(mm)); if (unlikely(test_and_set_bit(MMF_VM_HUGEPAGE, &mm->flags))) { free_mm_slot(mm_slot); return 0; } spin_lock(&khugepaged_mm_lock); insert_to_mm_slots_hash(mm, mm_slot); /* * Insert just behind the scanning cursor, to let the area settle * down a little. */ wakeup = list_empty(&khugepaged_scan.mm_head); list_add_tail(&mm_slot->mm_node, &khugepaged_scan.mm_head); spin_unlock(&khugepaged_mm_lock); atomic_inc(&mm->mm_count); if (wakeup) wake_up_interruptible(&khugepaged_wait); return 0; } int khugepaged_enter_vma_merge(struct vm_area_struct *vma) { unsigned long hstart, hend; if (!vma->anon_vma) /* * Not yet faulted in so we will register later in the * page fault if needed. */ return 0; |
78f11a255 mm: thp: fix /dev... |
1922 |
if (vma->vm_ops) |
ba76149f4 thp: khugepaged |
1923 1924 |
/* khugepaged not yet working on file or special mappings */ return 0; |
b3b9c2932 mm, x86, pat: rew... |
1925 |
VM_BUG_ON(vma->vm_flags & VM_NO_THP); |
ba76149f4 thp: khugepaged |
1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 |
hstart = (vma->vm_start + ~HPAGE_PMD_MASK) & HPAGE_PMD_MASK; hend = vma->vm_end & HPAGE_PMD_MASK; if (hstart < hend) return khugepaged_enter(vma); return 0; } void __khugepaged_exit(struct mm_struct *mm) { struct mm_slot *mm_slot; int free = 0; spin_lock(&khugepaged_mm_lock); mm_slot = get_mm_slot(mm); if (mm_slot && khugepaged_scan.mm_slot != mm_slot) { |
43b5fbbd2 mm/huge_memory.c:... |
1941 |
hash_del(&mm_slot->hash); |
ba76149f4 thp: khugepaged |
1942 1943 1944 |
list_del(&mm_slot->mm_node); free = 1; } |
d788e80a8 mm/huge_memory.c:... |
1945 |
spin_unlock(&khugepaged_mm_lock); |
ba76149f4 thp: khugepaged |
1946 1947 |
if (free) { |
ba76149f4 thp: khugepaged |
1948 1949 1950 1951 |
clear_bit(MMF_VM_HUGEPAGE, &mm->flags); free_mm_slot(mm_slot); mmdrop(mm); } else if (mm_slot) { |
ba76149f4 thp: khugepaged |
1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 |
/* * This is required to serialize against * khugepaged_test_exit() (which is guaranteed to run * under mmap sem read mode). Stop here (after we * return all pagetables will be destroyed) until * khugepaged has finished working on the pagetables * under the mmap_sem. */ down_write(&mm->mmap_sem); up_write(&mm->mmap_sem); |
d788e80a8 mm/huge_memory.c:... |
1962 |
} |
ba76149f4 thp: khugepaged |
1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 |
} static void release_pte_page(struct page *page) { /* 0 stands for page_is_file_cache(page) == false */ dec_zone_page_state(page, NR_ISOLATED_ANON + 0); unlock_page(page); putback_lru_page(page); } static void release_pte_pages(pte_t *pte, pte_t *_pte) { while (--_pte >= pte) { pte_t pteval = *_pte; if (!pte_none(pteval)) release_pte_page(pte_page(pteval)); } } |
ba76149f4 thp: khugepaged |
1981 1982 1983 1984 1985 1986 |
static int __collapse_huge_page_isolate(struct vm_area_struct *vma, unsigned long address, pte_t *pte) { struct page *page; pte_t *_pte; |
344aa35c2 thp: clean up __c... |
1987 |
int referenced = 0, none = 0; |
ba76149f4 thp: khugepaged |
1988 1989 1990 1991 1992 1993 |
for (_pte = pte; _pte < pte+HPAGE_PMD_NR; _pte++, address += PAGE_SIZE) { pte_t pteval = *_pte; if (pte_none(pteval)) { if (++none <= khugepaged_max_ptes_none) continue; |
344aa35c2 thp: clean up __c... |
1994 |
else |
ba76149f4 thp: khugepaged |
1995 |
goto out; |
ba76149f4 thp: khugepaged |
1996 |
} |
344aa35c2 thp: clean up __c... |
1997 |
if (!pte_present(pteval) || !pte_write(pteval)) |
ba76149f4 thp: khugepaged |
1998 |
goto out; |
ba76149f4 thp: khugepaged |
1999 |
page = vm_normal_page(vma, address, pteval); |
344aa35c2 thp: clean up __c... |
2000 |
if (unlikely(!page)) |
ba76149f4 thp: khugepaged |
2001 |
goto out; |
344aa35c2 thp: clean up __c... |
2002 |
|
ba76149f4 thp: khugepaged |
2003 2004 2005 2006 2007 |
VM_BUG_ON(PageCompound(page)); BUG_ON(!PageAnon(page)); VM_BUG_ON(!PageSwapBacked(page)); /* cannot use mapcount: can't collapse if there's a gup pin */ |
344aa35c2 thp: clean up __c... |
2008 |
if (page_count(page) != 1) |
ba76149f4 thp: khugepaged |
2009 |
goto out; |
ba76149f4 thp: khugepaged |
2010 2011 2012 2013 2014 2015 |
/* * We can do it before isolate_lru_page because the * page can't be freed from under us. NOTE: PG_lock * is needed to serialize against split_huge_page * when invoked from the VM. */ |
344aa35c2 thp: clean up __c... |
2016 |
if (!trylock_page(page)) |
ba76149f4 thp: khugepaged |
2017 |
goto out; |
ba76149f4 thp: khugepaged |
2018 2019 2020 2021 2022 2023 |
/* * Isolate the page to avoid collapsing an hugepage * currently in use by the VM. */ if (isolate_lru_page(page)) { unlock_page(page); |
ba76149f4 thp: khugepaged |
2024 2025 2026 2027 2028 2029 2030 2031 |
goto out; } /* 0 stands for page_is_file_cache(page) == false */ inc_zone_page_state(page, NR_ISOLATED_ANON + 0); VM_BUG_ON(!PageLocked(page)); VM_BUG_ON(PageLRU(page)); /* If there is no mapped pte young don't collapse the page */ |
8ee53820e thp: mmu_notifier... |
2032 2033 |
if (pte_young(pteval) || PageReferenced(page) || mmu_notifier_test_young(vma->vm_mm, address)) |
ba76149f4 thp: khugepaged |
2034 2035 |
referenced = 1; } |
344aa35c2 thp: clean up __c... |
2036 2037 |
if (likely(referenced)) return 1; |
ba76149f4 thp: khugepaged |
2038 |
out: |
344aa35c2 thp: clean up __c... |
2039 2040 |
release_pte_pages(pte, _pte); return 0; |
ba76149f4 thp: khugepaged |
2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 |
} static void __collapse_huge_page_copy(pte_t *pte, struct page *page, struct vm_area_struct *vma, unsigned long address, spinlock_t *ptl) { pte_t *_pte; for (_pte = pte; _pte < pte+HPAGE_PMD_NR; _pte++) { pte_t pteval = *_pte; struct page *src_page; if (pte_none(pteval)) { clear_user_highpage(page, address); add_mm_counter(vma->vm_mm, MM_ANONPAGES, 1); } else { src_page = pte_page(pteval); copy_user_highpage(page, src_page, address, vma); VM_BUG_ON(page_mapcount(src_page) != 1); |
ba76149f4 thp: khugepaged |
2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 |
release_pte_page(src_page); /* * ptl mostly unnecessary, but preempt has to * be disabled to update the per-cpu stats * inside page_remove_rmap(). */ spin_lock(ptl); /* * paravirt calls inside pte_clear here are * superfluous. */ pte_clear(vma->vm_mm, address, _pte); page_remove_rmap(src_page); spin_unlock(ptl); free_page_and_swap_cache(src_page); } address += PAGE_SIZE; page++; } } |
26234f36e thp: introduce kh... |
2081 |
static void khugepaged_alloc_sleep(void) |
ba76149f4 thp: khugepaged |
2082 |
{ |
26234f36e thp: introduce kh... |
2083 2084 2085 |
wait_event_freezable_timeout(khugepaged_wait, false, msecs_to_jiffies(khugepaged_alloc_sleep_millisecs)); } |
ba76149f4 thp: khugepaged |
2086 |
|
26234f36e thp: introduce kh... |
2087 2088 2089 2090 2091 2092 2093 2094 |
#ifdef CONFIG_NUMA static bool khugepaged_prealloc_page(struct page **hpage, bool *wait) { if (IS_ERR(*hpage)) { if (!*wait) return false; *wait = false; |
e3b4126c5 thp: khugepaged_p... |
2095 |
*hpage = NULL; |
26234f36e thp: introduce kh... |
2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 |
khugepaged_alloc_sleep(); } else if (*hpage) { put_page(*hpage); *hpage = NULL; } return true; } static struct page *khugepaged_alloc_page(struct page **hpage, struct mm_struct *mm, struct vm_area_struct *vma, unsigned long address, int node) { |
0bbbc0b33 thp: add numa awa... |
2110 |
VM_BUG_ON(*hpage); |
ce83d2174 thp: allocate mem... |
2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 |
/* * Allocate the page while the vma is still valid and under * the mmap_sem read mode so there is no memory allocation * later when we take the mmap_sem in write mode. This is more * friendly behavior (OTOH it may actually hide bugs) to * filesystems in userland with daemons allocating memory in * the userland I/O paths. Allocating memory with the * mmap_sem in read mode is good idea also to allow greater * scalability. */ |
26234f36e thp: introduce kh... |
2121 |
*hpage = alloc_hugepage_vma(khugepaged_defrag(), vma, address, |
cc5d462f7 mm: use __GFP_OTH... |
2122 |
node, __GFP_OTHER_NODE); |
692e0b354 mm: thp: optimize... |
2123 2124 2125 2126 2127 2128 |
/* * After allocating the hugepage, release the mmap_sem read lock in * preparation for taking it in write mode. */ up_read(&mm->mmap_sem); |
26234f36e thp: introduce kh... |
2129 |
if (unlikely(!*hpage)) { |
81ab4201f mm: add VM counte... |
2130 |
count_vm_event(THP_COLLAPSE_ALLOC_FAILED); |
ce83d2174 thp: allocate mem... |
2131 |
*hpage = ERR_PTR(-ENOMEM); |
26234f36e thp: introduce kh... |
2132 |
return NULL; |
ce83d2174 thp: allocate mem... |
2133 |
} |
26234f36e thp: introduce kh... |
2134 |
|
65b3c07b4 thp: fix the coun... |
2135 |
count_vm_event(THP_COLLAPSE_ALLOC); |
26234f36e thp: introduce kh... |
2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 |
return *hpage; } #else static struct page *khugepaged_alloc_hugepage(bool *wait) { struct page *hpage; do { hpage = alloc_hugepage(khugepaged_defrag()); if (!hpage) { count_vm_event(THP_COLLAPSE_ALLOC_FAILED); if (!*wait) return NULL; *wait = false; khugepaged_alloc_sleep(); } else count_vm_event(THP_COLLAPSE_ALLOC); } while (unlikely(!hpage) && likely(khugepaged_enabled())); return hpage; } static bool khugepaged_prealloc_page(struct page **hpage, bool *wait) { if (!*hpage) *hpage = khugepaged_alloc_hugepage(wait); if (unlikely(!*hpage)) return false; return true; } static struct page *khugepaged_alloc_page(struct page **hpage, struct mm_struct *mm, struct vm_area_struct *vma, unsigned long address, int node) { up_read(&mm->mmap_sem); VM_BUG_ON(!*hpage); return *hpage; } |
692e0b354 mm: thp: optimize... |
2179 |
#endif |
fa475e517 thp: introduce hu... |
2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 |
static bool hugepage_vma_check(struct vm_area_struct *vma) { if ((!(vma->vm_flags & VM_HUGEPAGE) && !khugepaged_always()) || (vma->vm_flags & VM_NOHUGEPAGE)) return false; if (!vma->anon_vma || vma->vm_ops) return false; if (is_vma_temporary_stack(vma)) return false; VM_BUG_ON(vma->vm_flags & VM_NO_THP); return true; } |
26234f36e thp: introduce kh... |
2193 2194 2195 2196 2197 2198 |
static void collapse_huge_page(struct mm_struct *mm, unsigned long address, struct page **hpage, struct vm_area_struct *vma, int node) { |
26234f36e thp: introduce kh... |
2199 2200 2201 2202 2203 2204 2205 |
pmd_t *pmd, _pmd; pte_t *pte; pgtable_t pgtable; struct page *new_page; spinlock_t *ptl; int isolated; unsigned long hstart, hend; |
2ec74c3ef mm: move all mmu ... |
2206 2207 |
unsigned long mmun_start; /* For mmu_notifiers */ unsigned long mmun_end; /* For mmu_notifiers */ |
26234f36e thp: introduce kh... |
2208 2209 2210 2211 2212 2213 2214 |
VM_BUG_ON(address & ~HPAGE_PMD_MASK); /* release the mmap_sem read lock. */ new_page = khugepaged_alloc_page(hpage, mm, vma, address, node); if (!new_page) return; |
420256ef0 thp: release page... |
2215 |
if (unlikely(mem_cgroup_newpage_charge(new_page, mm, GFP_KERNEL))) |
ce83d2174 thp: allocate mem... |
2216 |
return; |
ba76149f4 thp: khugepaged |
2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 |
/* * Prevent all access to pagetables with the exception of * gup_fast later hanlded by the ptep_clear_flush and the VM * handled by the anon_vma lock + PG_lock. */ down_write(&mm->mmap_sem); if (unlikely(khugepaged_test_exit(mm))) goto out; vma = find_vma(mm, address); |
8b89ae8a4 mm/huge_memory.c:... |
2228 2229 |
if (!vma) goto out; |
ba76149f4 thp: khugepaged |
2230 2231 2232 2233 |
hstart = (vma->vm_start + ~HPAGE_PMD_MASK) & HPAGE_PMD_MASK; hend = vma->vm_end & HPAGE_PMD_MASK; if (address < hstart || address + HPAGE_PMD_SIZE > hend) goto out; |
fa475e517 thp: introduce hu... |
2234 |
if (!hugepage_vma_check(vma)) |
a7d6e4ecd thp: prevent huge... |
2235 |
goto out; |
6219049ae mm: introduce mm_... |
2236 2237 |
pmd = mm_find_pmd(mm, address); if (!pmd) |
ba76149f4 thp: khugepaged |
2238 |
goto out; |
6219049ae mm: introduce mm_... |
2239 |
if (pmd_trans_huge(*pmd)) |
ba76149f4 thp: khugepaged |
2240 |
goto out; |
4fc3f1d66 mm/rmap, migratio... |
2241 |
anon_vma_lock_write(vma->anon_vma); |
ba76149f4 thp: khugepaged |
2242 2243 2244 |
pte = pte_offset_map(pmd, address); ptl = pte_lockptr(mm, pmd); |
2ec74c3ef mm: move all mmu ... |
2245 2246 2247 |
mmun_start = address; mmun_end = address + HPAGE_PMD_SIZE; mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); |
ba76149f4 thp: khugepaged |
2248 2249 2250 2251 2252 2253 2254 |
spin_lock(&mm->page_table_lock); /* probably unnecessary */ /* * After this gup_fast can't run anymore. This also removes * any huge TLB entry from the CPU so we won't allow * huge and small TLB entries for the same virtual address * to avoid the risk of CPU bugs in that area. */ |
2ec74c3ef mm: move all mmu ... |
2255 |
_pmd = pmdp_clear_flush(vma, address, pmd); |
ba76149f4 thp: khugepaged |
2256 |
spin_unlock(&mm->page_table_lock); |
2ec74c3ef mm: move all mmu ... |
2257 |
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
ba76149f4 thp: khugepaged |
2258 2259 2260 2261 |
spin_lock(ptl); isolated = __collapse_huge_page_isolate(vma, address, pte); spin_unlock(ptl); |
ba76149f4 thp: khugepaged |
2262 2263 |
if (unlikely(!isolated)) { |
453c71926 thp: keep highpte... |
2264 |
pte_unmap(pte); |
ba76149f4 thp: khugepaged |
2265 2266 |
spin_lock(&mm->page_table_lock); BUG_ON(!pmd_none(*pmd)); |
7c3425123 mm/THP: use pmd_p... |
2267 2268 2269 2270 2271 2272 |
/* * We can only use set_pmd_at when establishing * hugepmds and never for establishing regular pmds that * points to regular pagetables. Use pmd_populate for that */ pmd_populate(mm, pmd, pmd_pgtable(_pmd)); |
ba76149f4 thp: khugepaged |
2273 |
spin_unlock(&mm->page_table_lock); |
08b52706d mm/rmap: rename a... |
2274 |
anon_vma_unlock_write(vma->anon_vma); |
ce83d2174 thp: allocate mem... |
2275 |
goto out; |
ba76149f4 thp: khugepaged |
2276 2277 2278 2279 2280 2281 |
} /* * All pages are isolated and locked so anon_vma rmap * can't run anymore. */ |
08b52706d mm/rmap: rename a... |
2282 |
anon_vma_unlock_write(vma->anon_vma); |
ba76149f4 thp: khugepaged |
2283 2284 |
__collapse_huge_page_copy(pte, new_page, vma, address, ptl); |
453c71926 thp: keep highpte... |
2285 |
pte_unmap(pte); |
ba76149f4 thp: khugepaged |
2286 2287 |
__SetPageUptodate(new_page); pgtable = pmd_pgtable(_pmd); |
ba76149f4 thp: khugepaged |
2288 |
|
b3092b3b7 thp: cleanup: int... |
2289 |
_pmd = mk_huge_pmd(new_page, vma); |
ba76149f4 thp: khugepaged |
2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 |
/* * spin_lock() below is not the equivalent of smp_wmb(), so * this is needed to avoid the copy_huge_page writes to become * visible after the set_pmd_at() write. */ smp_wmb(); spin_lock(&mm->page_table_lock); BUG_ON(!pmd_none(*pmd)); page_add_new_anon_rmap(new_page, vma, address); set_pmd_at(mm, address, pmd, _pmd); |
b113da657 mm: Add and use u... |
2302 |
update_mmu_cache_pmd(vma, address, pmd); |
e3ebcf643 thp: remove assum... |
2303 |
pgtable_trans_huge_deposit(mm, pgtable); |
ba76149f4 thp: khugepaged |
2304 2305 2306 |
spin_unlock(&mm->page_table_lock); *hpage = NULL; |
420256ef0 thp: release page... |
2307 |
|
ba76149f4 thp: khugepaged |
2308 |
khugepaged_pages_collapsed++; |
ce83d2174 thp: allocate mem... |
2309 |
out_up_write: |
ba76149f4 thp: khugepaged |
2310 |
up_write(&mm->mmap_sem); |
0bbbc0b33 thp: add numa awa... |
2311 |
return; |
ce83d2174 thp: allocate mem... |
2312 |
out: |
678ff896a memcg: fix leak o... |
2313 |
mem_cgroup_uncharge_page(new_page); |
ce83d2174 thp: allocate mem... |
2314 |
goto out_up_write; |
ba76149f4 thp: khugepaged |
2315 2316 2317 2318 2319 2320 2321 |
} static int khugepaged_scan_pmd(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long address, struct page **hpage) { |
ba76149f4 thp: khugepaged |
2322 2323 2324 2325 2326 2327 |
pmd_t *pmd; pte_t *pte, *_pte; int ret = 0, referenced = 0, none = 0; struct page *page; unsigned long _address; spinlock_t *ptl; |
00ef2d2f8 mm: use NUMA_NO_NODE |
2328 |
int node = NUMA_NO_NODE; |
ba76149f4 thp: khugepaged |
2329 2330 |
VM_BUG_ON(address & ~HPAGE_PMD_MASK); |
6219049ae mm: introduce mm_... |
2331 2332 |
pmd = mm_find_pmd(mm, address); if (!pmd) |
ba76149f4 thp: khugepaged |
2333 |
goto out; |
6219049ae mm: introduce mm_... |
2334 |
if (pmd_trans_huge(*pmd)) |
ba76149f4 thp: khugepaged |
2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 |
goto out; pte = pte_offset_map_lock(mm, pmd, address, &ptl); for (_address = address, _pte = pte; _pte < pte+HPAGE_PMD_NR; _pte++, _address += PAGE_SIZE) { pte_t pteval = *_pte; if (pte_none(pteval)) { if (++none <= khugepaged_max_ptes_none) continue; else goto out_unmap; } if (!pte_present(pteval) || !pte_write(pteval)) goto out_unmap; page = vm_normal_page(vma, _address, pteval); if (unlikely(!page)) goto out_unmap; |
5c4b4be3b mm: use correct n... |
2352 2353 2354 2355 2356 |
/* * Chose the node of the first page. This could * be more sophisticated and look at more pages, * but isn't for now. */ |
00ef2d2f8 mm: use NUMA_NO_NODE |
2357 |
if (node == NUMA_NO_NODE) |
5c4b4be3b mm: use correct n... |
2358 |
node = page_to_nid(page); |
ba76149f4 thp: khugepaged |
2359 2360 2361 2362 2363 2364 |
VM_BUG_ON(PageCompound(page)); if (!PageLRU(page) || PageLocked(page) || !PageAnon(page)) goto out_unmap; /* cannot use mapcount: can't collapse if there's a gup pin */ if (page_count(page) != 1) goto out_unmap; |
8ee53820e thp: mmu_notifier... |
2365 2366 |
if (pte_young(pteval) || PageReferenced(page) || mmu_notifier_test_young(vma->vm_mm, address)) |
ba76149f4 thp: khugepaged |
2367 2368 2369 2370 2371 2372 |
referenced = 1; } if (referenced) ret = 1; out_unmap: pte_unmap_unlock(pte, ptl); |
ce83d2174 thp: allocate mem... |
2373 2374 |
if (ret) /* collapse_huge_page will return with the mmap_sem released */ |
5c4b4be3b mm: use correct n... |
2375 |
collapse_huge_page(mm, address, hpage, vma, node); |
ba76149f4 thp: khugepaged |
2376 2377 2378 2379 2380 2381 2382 |
out: return ret; } static void collect_mm_slot(struct mm_slot *mm_slot) { struct mm_struct *mm = mm_slot->mm; |
b9980cdcf mm: fix UP THP sp... |
2383 |
VM_BUG_ON(NR_CPUS != 1 && !spin_is_locked(&khugepaged_mm_lock)); |
ba76149f4 thp: khugepaged |
2384 2385 2386 |
if (khugepaged_test_exit(mm)) { /* free mm_slot */ |
43b5fbbd2 mm/huge_memory.c:... |
2387 |
hash_del(&mm_slot->hash); |
ba76149f4 thp: khugepaged |
2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 |
list_del(&mm_slot->mm_node); /* * Not strictly needed because the mm exited already. * * clear_bit(MMF_VM_HUGEPAGE, &mm->flags); */ /* khugepaged_mm_lock actually not necessary for the below */ free_mm_slot(mm_slot); mmdrop(mm); } } static unsigned int khugepaged_scan_mm_slot(unsigned int pages, struct page **hpage) |
2f1da6421 mm/huge_memory.c:... |
2404 2405 |
__releases(&khugepaged_mm_lock) __acquires(&khugepaged_mm_lock) |
ba76149f4 thp: khugepaged |
2406 2407 2408 2409 2410 2411 2412 |
{ struct mm_slot *mm_slot; struct mm_struct *mm; struct vm_area_struct *vma; int progress = 0; VM_BUG_ON(!pages); |
b9980cdcf mm: fix UP THP sp... |
2413 |
VM_BUG_ON(NR_CPUS != 1 && !spin_is_locked(&khugepaged_mm_lock)); |
ba76149f4 thp: khugepaged |
2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 |
if (khugepaged_scan.mm_slot) mm_slot = khugepaged_scan.mm_slot; else { mm_slot = list_entry(khugepaged_scan.mm_head.next, struct mm_slot, mm_node); khugepaged_scan.address = 0; khugepaged_scan.mm_slot = mm_slot; } spin_unlock(&khugepaged_mm_lock); mm = mm_slot->mm; down_read(&mm->mmap_sem); if (unlikely(khugepaged_test_exit(mm))) vma = NULL; else vma = find_vma(mm, khugepaged_scan.address); progress++; for (; vma; vma = vma->vm_next) { unsigned long hstart, hend; cond_resched(); if (unlikely(khugepaged_test_exit(mm))) { progress++; break; } |
fa475e517 thp: introduce hu... |
2441 2442 |
if (!hugepage_vma_check(vma)) { skip: |
ba76149f4 thp: khugepaged |
2443 2444 2445 |
progress++; continue; } |
ba76149f4 thp: khugepaged |
2446 2447 |
hstart = (vma->vm_start + ~HPAGE_PMD_MASK) & HPAGE_PMD_MASK; hend = vma->vm_end & HPAGE_PMD_MASK; |
a7d6e4ecd thp: prevent huge... |
2448 2449 2450 2451 |
if (hstart >= hend) goto skip; if (khugepaged_scan.address > hend) goto skip; |
ba76149f4 thp: khugepaged |
2452 2453 |
if (khugepaged_scan.address < hstart) khugepaged_scan.address = hstart; |
a7d6e4ecd thp: prevent huge... |
2454 |
VM_BUG_ON(khugepaged_scan.address & ~HPAGE_PMD_MASK); |
ba76149f4 thp: khugepaged |
2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 |
while (khugepaged_scan.address < hend) { int ret; cond_resched(); if (unlikely(khugepaged_test_exit(mm))) goto breakouterloop; VM_BUG_ON(khugepaged_scan.address < hstart || khugepaged_scan.address + HPAGE_PMD_SIZE > hend); ret = khugepaged_scan_pmd(mm, vma, khugepaged_scan.address, hpage); /* move to next address */ khugepaged_scan.address += HPAGE_PMD_SIZE; progress += HPAGE_PMD_NR; if (ret) /* we released mmap_sem so break loop */ goto breakouterloop_mmap_sem; if (progress >= pages) goto breakouterloop; } } breakouterloop: up_read(&mm->mmap_sem); /* exit_mmap will destroy ptes after this */ breakouterloop_mmap_sem: spin_lock(&khugepaged_mm_lock); |
a7d6e4ecd thp: prevent huge... |
2483 |
VM_BUG_ON(khugepaged_scan.mm_slot != mm_slot); |
ba76149f4 thp: khugepaged |
2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 |
/* * Release the current mm_slot if this mm is about to die, or * if we scanned all vmas of this mm. */ if (khugepaged_test_exit(mm) || !vma) { /* * Make sure that if mm_users is reaching zero while * khugepaged runs here, khugepaged_exit will find * mm_slot not pointing to the exiting mm. */ if (mm_slot->mm_node.next != &khugepaged_scan.mm_head) { khugepaged_scan.mm_slot = list_entry( mm_slot->mm_node.next, struct mm_slot, mm_node); khugepaged_scan.address = 0; } else { khugepaged_scan.mm_slot = NULL; khugepaged_full_scans++; } collect_mm_slot(mm_slot); } return progress; } static int khugepaged_has_work(void) { return !list_empty(&khugepaged_scan.mm_head) && khugepaged_enabled(); } static int khugepaged_wait_event(void) { return !list_empty(&khugepaged_scan.mm_head) || |
2017c0bff thp: remove wake_... |
2519 |
kthread_should_stop(); |
ba76149f4 thp: khugepaged |
2520 |
} |
d516904bd thp: merge page p... |
2521 |
static void khugepaged_do_scan(void) |
ba76149f4 thp: khugepaged |
2522 |
{ |
d516904bd thp: merge page p... |
2523 |
struct page *hpage = NULL; |
ba76149f4 thp: khugepaged |
2524 2525 |
unsigned int progress = 0, pass_through_head = 0; unsigned int pages = khugepaged_pages_to_scan; |
d516904bd thp: merge page p... |
2526 |
bool wait = true; |
ba76149f4 thp: khugepaged |
2527 2528 2529 2530 |
barrier(); /* write khugepaged_pages_to_scan to local stack */ while (progress < pages) { |
26234f36e thp: introduce kh... |
2531 |
if (!khugepaged_prealloc_page(&hpage, &wait)) |
d516904bd thp: merge page p... |
2532 |
break; |
26234f36e thp: introduce kh... |
2533 |
|
420256ef0 thp: release page... |
2534 |
cond_resched(); |
ba76149f4 thp: khugepaged |
2535 |
|
878aee7d6 thp: freeze khuge... |
2536 2537 |
if (unlikely(kthread_should_stop() || freezing(current))) break; |
ba76149f4 thp: khugepaged |
2538 2539 2540 2541 2542 2543 |
spin_lock(&khugepaged_mm_lock); if (!khugepaged_scan.mm_slot) pass_through_head++; if (khugepaged_has_work() && pass_through_head < 2) progress += khugepaged_scan_mm_slot(pages - progress, |
d516904bd thp: merge page p... |
2544 |
&hpage); |
ba76149f4 thp: khugepaged |
2545 2546 2547 2548 |
else progress = pages; spin_unlock(&khugepaged_mm_lock); } |
ba76149f4 thp: khugepaged |
2549 |
|
d516904bd thp: merge page p... |
2550 2551 |
if (!IS_ERR_OR_NULL(hpage)) put_page(hpage); |
0bbbc0b33 thp: add numa awa... |
2552 |
} |
2017c0bff thp: remove wake_... |
2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 |
static void khugepaged_wait_work(void) { try_to_freeze(); if (khugepaged_has_work()) { if (!khugepaged_scan_sleep_millisecs) return; wait_event_freezable_timeout(khugepaged_wait, kthread_should_stop(), msecs_to_jiffies(khugepaged_scan_sleep_millisecs)); return; } if (khugepaged_enabled()) wait_event_freezable(khugepaged_wait, khugepaged_wait_event()); } |
ba76149f4 thp: khugepaged |
2570 2571 2572 |
static int khugepaged(void *none) { struct mm_slot *mm_slot; |
878aee7d6 thp: freeze khuge... |
2573 |
set_freezable(); |
ba76149f4 thp: khugepaged |
2574 |
set_user_nice(current, 19); |
b7231789b thp: remove khuge... |
2575 2576 2577 2578 |
while (!kthread_should_stop()) { khugepaged_do_scan(); khugepaged_wait_work(); } |
ba76149f4 thp: khugepaged |
2579 2580 2581 2582 2583 2584 2585 |
spin_lock(&khugepaged_mm_lock); mm_slot = khugepaged_scan.mm_slot; khugepaged_scan.mm_slot = NULL; if (mm_slot) collect_mm_slot(mm_slot); spin_unlock(&khugepaged_mm_lock); |
ba76149f4 thp: khugepaged |
2586 2587 |
return 0; } |
c5a647d09 thp: implement sp... |
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 |
static void __split_huge_zero_page_pmd(struct vm_area_struct *vma, unsigned long haddr, pmd_t *pmd) { struct mm_struct *mm = vma->vm_mm; pgtable_t pgtable; pmd_t _pmd; int i; pmdp_clear_flush(vma, haddr, pmd); /* leave pmd empty until pte is filled */ pgtable = pgtable_trans_huge_withdraw(mm); pmd_populate(mm, &_pmd, pgtable); for (i = 0; i < HPAGE_PMD_NR; i++, haddr += PAGE_SIZE) { pte_t *pte, entry; entry = pfn_pte(my_zero_pfn(haddr), vma->vm_page_prot); entry = pte_mkspecial(entry); pte = pte_offset_map(&_pmd, haddr); VM_BUG_ON(!pte_none(*pte)); set_pte_at(mm, haddr, pte, entry); pte_unmap(pte); } smp_wmb(); /* make pte visible before pmd */ pmd_populate(mm, pmd, pgtable); |
97ae17497 thp: implement re... |
2613 |
put_huge_zero_page(); |
c5a647d09 thp: implement sp... |
2614 |
} |
e180377f1 thp: change split... |
2615 2616 |
void __split_huge_page_pmd(struct vm_area_struct *vma, unsigned long address, pmd_t *pmd) |
71e3aac07 thp: transparent ... |
2617 2618 |
{ struct page *page; |
e180377f1 thp: change split... |
2619 |
struct mm_struct *mm = vma->vm_mm; |
c5a647d09 thp: implement sp... |
2620 2621 2622 |
unsigned long haddr = address & HPAGE_PMD_MASK; unsigned long mmun_start; /* For mmu_notifiers */ unsigned long mmun_end; /* For mmu_notifiers */ |
e180377f1 thp: change split... |
2623 2624 |
BUG_ON(vma->vm_start > haddr || vma->vm_end < haddr + HPAGE_PMD_SIZE); |
71e3aac07 thp: transparent ... |
2625 |
|
c5a647d09 thp: implement sp... |
2626 2627 2628 |
mmun_start = haddr; mmun_end = haddr + HPAGE_PMD_SIZE; mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); |
71e3aac07 thp: transparent ... |
2629 2630 2631 |
spin_lock(&mm->page_table_lock); if (unlikely(!pmd_trans_huge(*pmd))) { spin_unlock(&mm->page_table_lock); |
c5a647d09 thp: implement sp... |
2632 2633 2634 2635 2636 2637 2638 |
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); return; } if (is_huge_zero_pmd(*pmd)) { __split_huge_zero_page_pmd(vma, haddr, pmd); spin_unlock(&mm->page_table_lock); mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
71e3aac07 thp: transparent ... |
2639 2640 2641 2642 2643 2644 |
return; } page = pmd_page(*pmd); VM_BUG_ON(!page_count(page)); get_page(page); spin_unlock(&mm->page_table_lock); |
c5a647d09 thp: implement sp... |
2645 |
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
71e3aac07 thp: transparent ... |
2646 2647 2648 2649 2650 2651 |
split_huge_page(page); put_page(page); BUG_ON(pmd_trans_huge(*pmd)); } |
94fcc585f thp: avoid breaki... |
2652 |
|
e180377f1 thp: change split... |
2653 2654 2655 2656 2657 2658 2659 2660 2661 |
void split_huge_page_pmd_mm(struct mm_struct *mm, unsigned long address, pmd_t *pmd) { struct vm_area_struct *vma; vma = find_vma(mm, address); BUG_ON(vma == NULL); split_huge_page_pmd(vma, address, pmd); } |
94fcc585f thp: avoid breaki... |
2662 2663 2664 |
static void split_huge_page_address(struct mm_struct *mm, unsigned long address) { |
94fcc585f thp: avoid breaki... |
2665 2666 2667 |
pmd_t *pmd; VM_BUG_ON(!(address & ~HPAGE_PMD_MASK)); |
6219049ae mm: introduce mm_... |
2668 2669 |
pmd = mm_find_pmd(mm, address); if (!pmd) |
94fcc585f thp: avoid breaki... |
2670 2671 2672 2673 2674 |
return; /* * Caller holds the mmap_sem write mode, so a huge pmd cannot * materialize from under us. */ |
e180377f1 thp: change split... |
2675 |
split_huge_page_pmd_mm(mm, address, pmd); |
94fcc585f thp: avoid breaki... |
2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 |
} void __vma_adjust_trans_huge(struct vm_area_struct *vma, unsigned long start, unsigned long end, long adjust_next) { /* * If the new start address isn't hpage aligned and it could * previously contain an hugepage: check if we need to split * an huge pmd. */ if (start & ~HPAGE_PMD_MASK && (start & HPAGE_PMD_MASK) >= vma->vm_start && (start & HPAGE_PMD_MASK) + HPAGE_PMD_SIZE <= vma->vm_end) split_huge_page_address(vma->vm_mm, start); /* * If the new end address isn't hpage aligned and it could * previously contain an hugepage: check if we need to split * an huge pmd. */ if (end & ~HPAGE_PMD_MASK && (end & HPAGE_PMD_MASK) >= vma->vm_start && (end & HPAGE_PMD_MASK) + HPAGE_PMD_SIZE <= vma->vm_end) split_huge_page_address(vma->vm_mm, end); /* * If we're also updating the vma->vm_next->vm_start, if the new * vm_next->vm_start isn't page aligned and it could previously * contain an hugepage: check if we need to split an huge pmd. */ if (adjust_next > 0) { struct vm_area_struct *next = vma->vm_next; unsigned long nstart = next->vm_start; nstart += adjust_next << PAGE_SHIFT; if (nstart & ~HPAGE_PMD_MASK && (nstart & HPAGE_PMD_MASK) >= next->vm_start && (nstart & HPAGE_PMD_MASK) + HPAGE_PMD_SIZE <= next->vm_end) split_huge_page_address(next->vm_mm, nstart); } } |