Blame view
mm/migrate.c
48.6 KB
b20a35035
|
1 2 3 4 5 6 7 8 9 10 11 |
/* * Memory Migration functionality - linux/mm/migration.c * * Copyright (C) 2006 Silicon Graphics, Inc., Christoph Lameter * * Page migration was first developed in the context of the memory hotplug * project. The main authors of the migration code are: * * IWAMOTO Toshihiro <iwamoto@valinux.co.jp> * Hirokazu Takahashi <taka@valinux.co.jp> * Dave Hansen <haveblue@us.ibm.com> |
cde535359
|
12 |
* Christoph Lameter |
b20a35035
|
13 14 15 |
*/ #include <linux/migrate.h> |
b95f1b31b
|
16 |
#include <linux/export.h> |
b20a35035
|
17 |
#include <linux/swap.h> |
0697212a4
|
18 |
#include <linux/swapops.h> |
b20a35035
|
19 |
#include <linux/pagemap.h> |
e23ca00bf
|
20 |
#include <linux/buffer_head.h> |
b20a35035
|
21 |
#include <linux/mm_inline.h> |
b488893a3
|
22 |
#include <linux/nsproxy.h> |
b20a35035
|
23 |
#include <linux/pagevec.h> |
e9995ef97
|
24 |
#include <linux/ksm.h> |
b20a35035
|
25 26 27 28 |
#include <linux/rmap.h> #include <linux/topology.h> #include <linux/cpu.h> #include <linux/cpuset.h> |
04e62a29b
|
29 |
#include <linux/writeback.h> |
742755a1d
|
30 31 |
#include <linux/mempolicy.h> #include <linux/vmalloc.h> |
86c3a7645
|
32 |
#include <linux/security.h> |
8a9f3ccd2
|
33 |
#include <linux/memcontrol.h> |
4f5ca2657
|
34 |
#include <linux/syscalls.h> |
290408d4a
|
35 |
#include <linux/hugetlb.h> |
8e6ac7fab
|
36 |
#include <linux/hugetlb_cgroup.h> |
5a0e3ad6a
|
37 |
#include <linux/gfp.h> |
bf6bddf19
|
38 |
#include <linux/balloon_compaction.h> |
f714f4f20
|
39 |
#include <linux/mmu_notifier.h> |
b20a35035
|
40 |
|
0d1836c36
|
41 |
#include <asm/tlbflush.h> |
7b2a2d4a1
|
42 43 |
#define CREATE_TRACE_POINTS #include <trace/events/migrate.h> |
b20a35035
|
44 |
#include "internal.h" |
b20a35035
|
45 |
/* |
742755a1d
|
46 |
* migrate_prep() needs to be called before we start compiling a list of pages |
748446bb6
|
47 48 |
* to be migrated using isolate_lru_page(). If scheduling work on other CPUs is * undesirable, use migrate_prep_local() |
b20a35035
|
49 50 51 |
*/ int migrate_prep(void) { |
b20a35035
|
52 53 54 55 56 57 58 59 60 61 |
/* * Clear the LRU lists so pages can be isolated. * Note that pages may be moved off the LRU after we have * drained them. Those pages will fail to migrate like other * pages that may be busy. */ lru_add_drain_all(); return 0; } |
748446bb6
|
62 63 64 65 66 67 68 |
/* Do the necessary work of migrate_prep but not if it involves other CPUs */ int migrate_prep_local(void) { lru_add_drain(); return 0; } |
b20a35035
|
69 |
/* |
5733c7d11
|
70 71 72 |
* Put previously isolated pages back onto the appropriate lists * from where they were once taken off for compaction/migration. * |
59c82b70d
|
73 74 75 |
* This function shall be used whenever the isolated pageset has been * built from lru, balloon, hugetlbfs page. See isolate_migratepages_range() * and isolate_huge_page(). |
5733c7d11
|
76 77 78 79 80 81 82 |
*/ void putback_movable_pages(struct list_head *l) { struct page *page; struct page *page2; list_for_each_entry_safe(page, page2, l, lru) { |
31caf665e
|
83 84 85 86 |
if (unlikely(PageHuge(page))) { putback_active_hugepage(page); continue; } |
5733c7d11
|
87 88 89 |
list_del(&page->lru); dec_zone_page_state(page, NR_ISOLATED_ANON + page_is_file_cache(page)); |
117aad1e9
|
90 |
if (unlikely(isolated_balloon_page(page))) |
bf6bddf19
|
91 92 93 |
balloon_page_putback(page); else putback_lru_page(page); |
b20a35035
|
94 |
} |
b20a35035
|
95 |
} |
0697212a4
|
96 97 98 |
/* * Restore a potential migration pte to a working pte entry */ |
e9995ef97
|
99 100 |
static int remove_migration_pte(struct page *new, struct vm_area_struct *vma, unsigned long addr, void *old) |
0697212a4
|
101 102 103 |
{ struct mm_struct *mm = vma->vm_mm; swp_entry_t entry; |
0697212a4
|
104 105 106 |
pmd_t *pmd; pte_t *ptep, pte; spinlock_t *ptl; |
290408d4a
|
107 108 109 110 |
if (unlikely(PageHuge(new))) { ptep = huge_pte_offset(mm, addr); if (!ptep) goto out; |
cb900f412
|
111 |
ptl = huge_pte_lockptr(hstate_vma(vma), mm, ptep); |
290408d4a
|
112 |
} else { |
6219049ae
|
113 114 |
pmd = mm_find_pmd(mm, addr); if (!pmd) |
290408d4a
|
115 |
goto out; |
0697212a4
|
116 |
|
290408d4a
|
117 |
ptep = pte_offset_map(pmd, addr); |
0697212a4
|
118 |
|
486cf46f3
|
119 120 121 122 |
/* * Peek to check is_swap_pte() before taking ptlock? No, we * can race mremap's move_ptes(), which skips anon_vma lock. */ |
290408d4a
|
123 124 125 |
ptl = pte_lockptr(mm, pmd); } |
0697212a4
|
126 |
|
0697212a4
|
127 128 129 |
spin_lock(ptl); pte = *ptep; if (!is_swap_pte(pte)) |
e9995ef97
|
130 |
goto unlock; |
0697212a4
|
131 132 |
entry = pte_to_swp_entry(pte); |
e9995ef97
|
133 134 135 |
if (!is_migration_entry(entry) || migration_entry_to_page(entry) != old) goto unlock; |
0697212a4
|
136 |
|
0697212a4
|
137 138 |
get_page(new); pte = pte_mkold(mk_pte(new, vma->vm_page_prot)); |
c3d16e165
|
139 140 |
if (pte_swp_soft_dirty(*ptep)) pte = pte_mksoft_dirty(pte); |
d3cb8bf60
|
141 142 |
/* Recheck VMA as permissions can change since migration started */ |
0697212a4
|
143 |
if (is_write_migration_entry(entry)) |
d3cb8bf60
|
144 |
pte = maybe_mkwrite(pte, vma); |
3ef8fd7f7
|
145 |
#ifdef CONFIG_HUGETLB_PAGE |
be7517d6a
|
146 |
if (PageHuge(new)) { |
290408d4a
|
147 |
pte = pte_mkhuge(pte); |
be7517d6a
|
148 149 |
pte = arch_make_huge_pte(pte, vma, new, 0); } |
3ef8fd7f7
|
150 |
#endif |
c2cc499c5
|
151 |
flush_dcache_page(new); |
0697212a4
|
152 |
set_pte_at(mm, addr, ptep, pte); |
04e62a29b
|
153 |
|
290408d4a
|
154 155 156 157 158 159 |
if (PageHuge(new)) { if (PageAnon(new)) hugepage_add_anon_rmap(new, vma, addr); else page_dup_rmap(new); } else if (PageAnon(new)) |
04e62a29b
|
160 161 162 163 164 |
page_add_anon_rmap(new, vma, addr); else page_add_file_rmap(new); /* No need to invalidate - it was non-present before */ |
4b3073e1c
|
165 |
update_mmu_cache(vma, addr, ptep); |
e9995ef97
|
166 |
unlock: |
0697212a4
|
167 |
pte_unmap_unlock(ptep, ptl); |
e9995ef97
|
168 169 |
out: return SWAP_AGAIN; |
0697212a4
|
170 171 172 |
} /* |
7e09e738a
|
173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 |
* Congratulations to trinity for discovering this bug. * mm/fremap.c's remap_file_pages() accepts any range within a single vma to * convert that vma to VM_NONLINEAR; and generic_file_remap_pages() will then * replace the specified range by file ptes throughout (maybe populated after). * If page migration finds a page within that range, while it's still located * by vma_interval_tree rather than lost to i_mmap_nonlinear list, no problem: * zap_pte() clears the temporary migration entry before mmap_sem is dropped. * But if the migrating page is in a part of the vma outside the range to be * remapped, then it will not be cleared, and remove_migration_ptes() needs to * deal with it. Fortunately, this part of the vma is of course still linear, * so we just need to use linear location on the nonlinear list. */ static int remove_linear_migration_ptes_from_nonlinear(struct page *page, struct address_space *mapping, void *arg) { struct vm_area_struct *vma; /* hugetlbfs does not support remap_pages, so no huge pgoff worries */ pgoff_t pgoff = page->index << (PAGE_CACHE_SHIFT - PAGE_SHIFT); unsigned long addr; list_for_each_entry(vma, &mapping->i_mmap_nonlinear, shared.nonlinear) { addr = vma->vm_start + ((pgoff - vma->vm_pgoff) << PAGE_SHIFT); if (addr >= vma->vm_start && addr < vma->vm_end) remove_migration_pte(page, vma, addr, arg); } return SWAP_AGAIN; } /* |
04e62a29b
|
204 205 206 207 208 |
* Get rid of all migration entries and replace them by * references to the indicated page. */ static void remove_migration_ptes(struct page *old, struct page *new) { |
051ac83ad
|
209 210 211 |
struct rmap_walk_control rwc = { .rmap_one = remove_migration_pte, .arg = old, |
7e09e738a
|
212 |
.file_nonlinear = remove_linear_migration_ptes_from_nonlinear, |
051ac83ad
|
213 214 215 |
}; rmap_walk(new, &rwc); |
04e62a29b
|
216 217 218 |
} /* |
0697212a4
|
219 220 221 |
* Something used the pte of a page under migration. We need to * get to the page and wait until migration is finished. * When we return from this function the fault will be retried. |
0697212a4
|
222 |
*/ |
30dad3092
|
223 224 |
static void __migration_entry_wait(struct mm_struct *mm, pte_t *ptep, spinlock_t *ptl) |
0697212a4
|
225 |
{ |
30dad3092
|
226 |
pte_t pte; |
0697212a4
|
227 228 |
swp_entry_t entry; struct page *page; |
30dad3092
|
229 |
spin_lock(ptl); |
0697212a4
|
230 231 232 233 234 235 236 237 238 |
pte = *ptep; if (!is_swap_pte(pte)) goto out; entry = pte_to_swp_entry(pte); if (!is_migration_entry(entry)) goto out; page = migration_entry_to_page(entry); |
e286781d5
|
239 240 241 242 243 244 245 246 247 |
/* * Once radix-tree replacement of page migration started, page_count * *must* be zero. And, we don't want to call wait_on_page_locked() * against a page without get_page(). * So, we use get_page_unless_zero(), here. Even failed, page fault * will occur again. */ if (!get_page_unless_zero(page)) goto out; |
0697212a4
|
248 249 250 251 252 253 254 |
pte_unmap_unlock(ptep, ptl); wait_on_page_locked(page); put_page(page); return; out: pte_unmap_unlock(ptep, ptl); } |
30dad3092
|
255 256 257 258 259 260 261 |
void migration_entry_wait(struct mm_struct *mm, pmd_t *pmd, unsigned long address) { spinlock_t *ptl = pte_lockptr(mm, pmd); pte_t *ptep = pte_offset_map(pmd, address); __migration_entry_wait(mm, ptep, ptl); } |
cb900f412
|
262 263 |
void migration_entry_wait_huge(struct vm_area_struct *vma, struct mm_struct *mm, pte_t *pte) |
30dad3092
|
264 |
{ |
cb900f412
|
265 |
spinlock_t *ptl = huge_pte_lockptr(hstate_vma(vma), mm, pte); |
30dad3092
|
266 267 |
__migration_entry_wait(mm, pte, ptl); } |
b969c4ab9
|
268 269 |
#ifdef CONFIG_BLOCK /* Returns true if all buffers are successfully locked */ |
a6bc32b89
|
270 271 |
static bool buffer_migrate_lock_buffers(struct buffer_head *head, enum migrate_mode mode) |
b969c4ab9
|
272 273 274 275 |
{ struct buffer_head *bh = head; /* Simple case, sync compaction */ |
a6bc32b89
|
276 |
if (mode != MIGRATE_ASYNC) { |
b969c4ab9
|
277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 |
do { get_bh(bh); lock_buffer(bh); bh = bh->b_this_page; } while (bh != head); return true; } /* async case, we cannot block on lock_buffer so use trylock_buffer */ do { get_bh(bh); if (!trylock_buffer(bh)) { /* * We failed to lock the buffer and cannot stall in * async migration. Release the taken locks */ struct buffer_head *failed_bh = bh; put_bh(failed_bh); bh = head; while (bh != failed_bh) { unlock_buffer(bh); put_bh(bh); bh = bh->b_this_page; } return false; } bh = bh->b_this_page; } while (bh != head); return true; } #else static inline bool buffer_migrate_lock_buffers(struct buffer_head *head, |
a6bc32b89
|
312 |
enum migrate_mode mode) |
b969c4ab9
|
313 314 315 316 |
{ return true; } #endif /* CONFIG_BLOCK */ |
b20a35035
|
317 |
/* |
c3fcf8a5d
|
318 |
* Replace the page in the mapping. |
5b5c7120e
|
319 320 321 322 |
* * The number of remaining references must be: * 1 for anonymous pages without a mapping * 2 for pages with a mapping |
266cf658e
|
323 |
* 3 for pages with a mapping and PagePrivate/PagePrivate2 set. |
b20a35035
|
324 |
*/ |
36bc08cc0
|
325 |
int migrate_page_move_mapping(struct address_space *mapping, |
b969c4ab9
|
326 |
struct page *newpage, struct page *page, |
8e321fefb
|
327 328 |
struct buffer_head *head, enum migrate_mode mode, int extra_count) |
b20a35035
|
329 |
{ |
8e321fefb
|
330 |
int expected_count = 1 + extra_count; |
7cf9c2c76
|
331 |
void **pslot; |
b20a35035
|
332 |
|
6c5240ae7
|
333 |
if (!mapping) { |
0e8c7d0fd
|
334 |
/* Anonymous page without mapping */ |
8e321fefb
|
335 |
if (page_count(page) != expected_count) |
6c5240ae7
|
336 |
return -EAGAIN; |
78bd52097
|
337 |
return MIGRATEPAGE_SUCCESS; |
6c5240ae7
|
338 |
} |
19fd62312
|
339 |
spin_lock_irq(&mapping->tree_lock); |
b20a35035
|
340 |
|
7cf9c2c76
|
341 342 |
pslot = radix_tree_lookup_slot(&mapping->page_tree, page_index(page)); |
b20a35035
|
343 |
|
8e321fefb
|
344 |
expected_count += 1 + page_has_private(page); |
e286781d5
|
345 |
if (page_count(page) != expected_count || |
29c1f677d
|
346 |
radix_tree_deref_slot_protected(pslot, &mapping->tree_lock) != page) { |
19fd62312
|
347 |
spin_unlock_irq(&mapping->tree_lock); |
e23ca00bf
|
348 |
return -EAGAIN; |
b20a35035
|
349 |
} |
e286781d5
|
350 |
if (!page_freeze_refs(page, expected_count)) { |
19fd62312
|
351 |
spin_unlock_irq(&mapping->tree_lock); |
e286781d5
|
352 353 |
return -EAGAIN; } |
b20a35035
|
354 |
/* |
b969c4ab9
|
355 356 357 358 359 360 |
* In the async migration case of moving a page with buffers, lock the * buffers using trylock before the mapping is moved. If the mapping * was moved, we later failed to lock the buffers and could not move * the mapping back due to an elevated page count, we would have to * block waiting on other references to be dropped. */ |
a6bc32b89
|
361 362 |
if (mode == MIGRATE_ASYNC && head && !buffer_migrate_lock_buffers(head, mode)) { |
b969c4ab9
|
363 364 365 366 367 368 |
page_unfreeze_refs(page, expected_count); spin_unlock_irq(&mapping->tree_lock); return -EAGAIN; } /* |
b20a35035
|
369 |
* Now we know that no one else is looking at the page. |
b20a35035
|
370 |
*/ |
7cf9c2c76
|
371 |
get_page(newpage); /* add cache reference */ |
b20a35035
|
372 373 374 375 |
if (PageSwapCache(page)) { SetPageSwapCache(newpage); set_page_private(newpage, page_private(page)); } |
7cf9c2c76
|
376 377 378 |
radix_tree_replace_slot(pslot, newpage); /* |
937a94c9d
|
379 380 |
* Drop cache reference from old page by unfreezing * to one less reference. |
7cf9c2c76
|
381 382 |
* We know this isn't the last reference. */ |
937a94c9d
|
383 |
page_unfreeze_refs(page, expected_count - 1); |
7cf9c2c76
|
384 |
|
0e8c7d0fd
|
385 386 387 388 389 390 391 392 393 394 395 396 |
/* * If moved to a different zone then also account * the page for that zone. Other VM counters will be * taken care of when we establish references to the * new page and drop references to the old page. * * Note that anonymous pages are accounted for * via NR_FILE_PAGES and NR_ANON_PAGES if they * are mapped to swap space. */ __dec_zone_page_state(page, NR_FILE_PAGES); __inc_zone_page_state(newpage, NR_FILE_PAGES); |
99a15e21d
|
397 |
if (!PageSwapCache(page) && PageSwapBacked(page)) { |
4b02108ac
|
398 399 400 |
__dec_zone_page_state(page, NR_SHMEM); __inc_zone_page_state(newpage, NR_SHMEM); } |
19fd62312
|
401 |
spin_unlock_irq(&mapping->tree_lock); |
b20a35035
|
402 |
|
78bd52097
|
403 |
return MIGRATEPAGE_SUCCESS; |
b20a35035
|
404 |
} |
b20a35035
|
405 406 |
/* |
290408d4a
|
407 408 409 410 411 412 413 414 415 416 417 418 |
* The expected number of remaining references is the same as that * of migrate_page_move_mapping(). */ int migrate_huge_page_move_mapping(struct address_space *mapping, struct page *newpage, struct page *page) { int expected_count; void **pslot; if (!mapping) { if (page_count(page) != 1) return -EAGAIN; |
78bd52097
|
419 |
return MIGRATEPAGE_SUCCESS; |
290408d4a
|
420 421 422 423 424 425 426 427 428 |
} spin_lock_irq(&mapping->tree_lock); pslot = radix_tree_lookup_slot(&mapping->page_tree, page_index(page)); expected_count = 2 + page_has_private(page); if (page_count(page) != expected_count || |
29c1f677d
|
429 |
radix_tree_deref_slot_protected(pslot, &mapping->tree_lock) != page) { |
290408d4a
|
430 431 432 433 434 435 436 437 438 439 440 441 |
spin_unlock_irq(&mapping->tree_lock); return -EAGAIN; } if (!page_freeze_refs(page, expected_count)) { spin_unlock_irq(&mapping->tree_lock); return -EAGAIN; } get_page(newpage); radix_tree_replace_slot(pslot, newpage); |
937a94c9d
|
442 |
page_unfreeze_refs(page, expected_count - 1); |
290408d4a
|
443 444 |
spin_unlock_irq(&mapping->tree_lock); |
78bd52097
|
445 |
return MIGRATEPAGE_SUCCESS; |
290408d4a
|
446 447 448 |
} /* |
30b0a105d
|
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 |
* Gigantic pages are so large that we do not guarantee that page++ pointer * arithmetic will work across the entire page. We need something more * specialized. */ static void __copy_gigantic_page(struct page *dst, struct page *src, int nr_pages) { int i; struct page *dst_base = dst; struct page *src_base = src; for (i = 0; i < nr_pages; ) { cond_resched(); copy_highpage(dst, src); i++; dst = mem_map_next(dst, dst_base, i); src = mem_map_next(src, src_base, i); } } static void copy_huge_page(struct page *dst, struct page *src) { int i; int nr_pages; if (PageHuge(src)) { /* hugetlbfs page */ struct hstate *h = page_hstate(src); nr_pages = pages_per_huge_page(h); if (unlikely(nr_pages > MAX_ORDER_NR_PAGES)) { __copy_gigantic_page(dst, src, nr_pages); return; } } else { /* thp page */ BUG_ON(!PageTransHuge(src)); nr_pages = hpage_nr_pages(src); } for (i = 0; i < nr_pages; i++) { cond_resched(); copy_highpage(dst + i, src + i); } } /* |
b20a35035
|
497 498 |
* Copy the page to its new location */ |
290408d4a
|
499 |
void migrate_page_copy(struct page *newpage, struct page *page) |
b20a35035
|
500 |
{ |
7851a45cd
|
501 |
int cpupid; |
b32967ff1
|
502 |
if (PageHuge(page) || PageTransHuge(page)) |
290408d4a
|
503 504 505 |
copy_huge_page(newpage, page); else copy_highpage(newpage, page); |
b20a35035
|
506 507 508 509 510 511 512 |
if (PageError(page)) SetPageError(newpage); if (PageReferenced(page)) SetPageReferenced(newpage); if (PageUptodate(page)) SetPageUptodate(newpage); |
894bc3104
|
513 |
if (TestClearPageActive(page)) { |
309381fea
|
514 |
VM_BUG_ON_PAGE(PageUnevictable(page), page); |
b20a35035
|
515 |
SetPageActive(newpage); |
418b27ef5
|
516 517 |
} else if (TestClearPageUnevictable(page)) SetPageUnevictable(newpage); |
b20a35035
|
518 519 520 521 522 523 524 |
if (PageChecked(page)) SetPageChecked(newpage); if (PageMappedToDisk(page)) SetPageMappedToDisk(newpage); if (PageDirty(page)) { clear_page_dirty_for_io(page); |
3a902c5f6
|
525 526 527 528 529 |
/* * Want to mark the page and the radix tree as dirty, and * redo the accounting that clear_page_dirty_for_io undid, * but we can't use set_page_dirty because that function * is actually a signal that all of the page has become dirty. |
25985edce
|
530 |
* Whereas only part of our page may be dirty. |
3a902c5f6
|
531 |
*/ |
752dc185d
|
532 533 534 535 |
if (PageSwapBacked(page)) SetPageDirty(newpage); else __set_page_dirty_nobuffers(newpage); |
b20a35035
|
536 |
} |
7851a45cd
|
537 538 539 540 541 542 |
/* * Copy NUMA information to the new page, to prevent over-eager * future migrations of this same page. */ cpupid = page_cpupid_xchg_last(page, -1); page_cpupid_xchg_last(newpage, cpupid); |
b291f0003
|
543 |
mlock_migrate_page(newpage, page); |
e9995ef97
|
544 |
ksm_migrate_page(newpage, page); |
c8d6553b9
|
545 546 547 548 |
/* * Please do not reorder this without considering how mm/ksm.c's * get_ksm_page() depends upon ksm_migrate_page() and PageSwapCache(). */ |
b20a35035
|
549 |
ClearPageSwapCache(page); |
b20a35035
|
550 551 |
ClearPagePrivate(page); set_page_private(page, 0); |
b20a35035
|
552 553 554 555 556 557 558 559 |
/* * If any waiters have accumulated on the new page then * wake them up. */ if (PageWriteback(newpage)) end_page_writeback(newpage); } |
b20a35035
|
560 |
|
1d8b85ccf
|
561 562 563 |
/************************************************************ * Migration functions ***********************************************************/ |
b20a35035
|
564 565 |
/* * Common logic to directly migrate a single page suitable for |
266cf658e
|
566 |
* pages that do not use PagePrivate/PagePrivate2. |
b20a35035
|
567 568 569 |
* * Pages are locked upon entry and exit. */ |
2d1db3b11
|
570 |
int migrate_page(struct address_space *mapping, |
a6bc32b89
|
571 572 |
struct page *newpage, struct page *page, enum migrate_mode mode) |
b20a35035
|
573 574 575 576 |
{ int rc; BUG_ON(PageWriteback(page)); /* Writeback must be complete */ |
8e321fefb
|
577 |
rc = migrate_page_move_mapping(mapping, newpage, page, NULL, mode, 0); |
b20a35035
|
578 |
|
78bd52097
|
579 |
if (rc != MIGRATEPAGE_SUCCESS) |
b20a35035
|
580 581 582 |
return rc; migrate_page_copy(newpage, page); |
78bd52097
|
583 |
return MIGRATEPAGE_SUCCESS; |
b20a35035
|
584 585 |
} EXPORT_SYMBOL(migrate_page); |
9361401eb
|
586 |
#ifdef CONFIG_BLOCK |
b20a35035
|
587 |
/* |
1d8b85ccf
|
588 589 590 591 |
* Migration function for pages with buffers. This function can only be used * if the underlying filesystem guarantees that no other references to "page" * exist. */ |
2d1db3b11
|
592 |
int buffer_migrate_page(struct address_space *mapping, |
a6bc32b89
|
593 |
struct page *newpage, struct page *page, enum migrate_mode mode) |
1d8b85ccf
|
594 |
{ |
1d8b85ccf
|
595 596 |
struct buffer_head *bh, *head; int rc; |
1d8b85ccf
|
597 |
if (!page_has_buffers(page)) |
a6bc32b89
|
598 |
return migrate_page(mapping, newpage, page, mode); |
1d8b85ccf
|
599 600 |
head = page_buffers(page); |
8e321fefb
|
601 |
rc = migrate_page_move_mapping(mapping, newpage, page, head, mode, 0); |
1d8b85ccf
|
602 |
|
78bd52097
|
603 |
if (rc != MIGRATEPAGE_SUCCESS) |
1d8b85ccf
|
604 |
return rc; |
b969c4ab9
|
605 606 607 608 609 |
/* * In the async case, migrate_page_move_mapping locked the buffers * with an IRQ-safe spinlock held. In the sync case, the buffers * need to be locked now */ |
a6bc32b89
|
610 611 |
if (mode != MIGRATE_ASYNC) BUG_ON(!buffer_migrate_lock_buffers(head, mode)); |
1d8b85ccf
|
612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 |
ClearPagePrivate(page); set_page_private(newpage, page_private(page)); set_page_private(page, 0); put_page(page); get_page(newpage); bh = head; do { set_bh_page(bh, newpage, bh_offset(bh)); bh = bh->b_this_page; } while (bh != head); SetPagePrivate(newpage); migrate_page_copy(newpage, page); bh = head; do { unlock_buffer(bh); put_bh(bh); bh = bh->b_this_page; } while (bh != head); |
78bd52097
|
637 |
return MIGRATEPAGE_SUCCESS; |
1d8b85ccf
|
638 639 |
} EXPORT_SYMBOL(buffer_migrate_page); |
9361401eb
|
640 |
#endif |
1d8b85ccf
|
641 |
|
04e62a29b
|
642 643 644 645 |
/* * Writeback a page to clean the dirty state */ static int writeout(struct address_space *mapping, struct page *page) |
8351a6e47
|
646 |
{ |
04e62a29b
|
647 648 649 650 651 |
struct writeback_control wbc = { .sync_mode = WB_SYNC_NONE, .nr_to_write = 1, .range_start = 0, .range_end = LLONG_MAX, |
04e62a29b
|
652 653 654 655 656 657 658 659 660 661 662 |
.for_reclaim = 1 }; int rc; if (!mapping->a_ops->writepage) /* No write method for the address space */ return -EINVAL; if (!clear_page_dirty_for_io(page)) /* Someone else already triggered a write */ return -EAGAIN; |
8351a6e47
|
663 |
/* |
04e62a29b
|
664 665 666 667 668 669 |
* A dirty page may imply that the underlying filesystem has * the page on some queue. So the page must be clean for * migration. Writeout may mean we loose the lock and the * page state is no longer what we checked for earlier. * At this point we know that the migration attempt cannot * be successful. |
8351a6e47
|
670 |
*/ |
04e62a29b
|
671 |
remove_migration_ptes(page, page); |
8351a6e47
|
672 |
|
04e62a29b
|
673 |
rc = mapping->a_ops->writepage(page, &wbc); |
8351a6e47
|
674 |
|
04e62a29b
|
675 676 677 |
if (rc != AOP_WRITEPAGE_ACTIVATE) /* unlocked. Relock */ lock_page(page); |
bda8550de
|
678 |
return (rc < 0) ? -EIO : -EAGAIN; |
04e62a29b
|
679 680 681 682 683 684 |
} /* * Default handling if a filesystem does not provide a migration function. */ static int fallback_migrate_page(struct address_space *mapping, |
a6bc32b89
|
685 |
struct page *newpage, struct page *page, enum migrate_mode mode) |
04e62a29b
|
686 |
{ |
b969c4ab9
|
687 |
if (PageDirty(page)) { |
a6bc32b89
|
688 689 |
/* Only writeback pages in full synchronous migration */ if (mode != MIGRATE_SYNC) |
b969c4ab9
|
690 |
return -EBUSY; |
04e62a29b
|
691 |
return writeout(mapping, page); |
b969c4ab9
|
692 |
} |
8351a6e47
|
693 694 695 696 697 |
/* * Buffers may be managed in a filesystem specific way. * We must have no buffers or drop them. */ |
266cf658e
|
698 |
if (page_has_private(page) && |
8351a6e47
|
699 700 |
!try_to_release_page(page, GFP_KERNEL)) return -EAGAIN; |
a6bc32b89
|
701 |
return migrate_page(mapping, newpage, page, mode); |
8351a6e47
|
702 |
} |
1d8b85ccf
|
703 |
/* |
e24f0b8f7
|
704 705 706 707 708 |
* Move a page to a newly allocated page * The page is locked and all ptes have been successfully removed. * * The new page will have replaced the old page if this function * is successful. |
894bc3104
|
709 710 711 |
* * Return value: * < 0 - error code |
78bd52097
|
712 |
* MIGRATEPAGE_SUCCESS - success |
e24f0b8f7
|
713 |
*/ |
3fe2011ff
|
714 |
static int move_to_new_page(struct page *newpage, struct page *page, |
2ebba6b7e
|
715 |
int page_was_mapped, enum migrate_mode mode) |
e24f0b8f7
|
716 717 718 719 720 721 722 723 724 |
{ struct address_space *mapping; int rc; /* * Block others from accessing the page when we get around to * establishing additional references. We are the only one * holding a reference to the new page at this point. */ |
529ae9aaa
|
725 |
if (!trylock_page(newpage)) |
e24f0b8f7
|
726 727 728 729 730 |
BUG(); /* Prepare mapping for the new page.*/ newpage->index = page->index; newpage->mapping = page->mapping; |
b2e185384
|
731 732 |
if (PageSwapBacked(page)) SetPageSwapBacked(newpage); |
e24f0b8f7
|
733 734 735 |
mapping = page_mapping(page); if (!mapping) |
a6bc32b89
|
736 |
rc = migrate_page(mapping, newpage, page, mode); |
b969c4ab9
|
737 |
else if (mapping->a_ops->migratepage) |
e24f0b8f7
|
738 |
/* |
b969c4ab9
|
739 740 741 742 |
* Most pages have a mapping and most filesystems provide a * migratepage callback. Anonymous pages are part of swap * space which also has its own migratepage callback. This * is the most common path for page migration. |
e24f0b8f7
|
743 |
*/ |
b969c4ab9
|
744 |
rc = mapping->a_ops->migratepage(mapping, |
a6bc32b89
|
745 |
newpage, page, mode); |
b969c4ab9
|
746 |
else |
a6bc32b89
|
747 |
rc = fallback_migrate_page(mapping, newpage, page, mode); |
e24f0b8f7
|
748 |
|
78bd52097
|
749 |
if (rc != MIGRATEPAGE_SUCCESS) { |
e24f0b8f7
|
750 |
newpage->mapping = NULL; |
3fe2011ff
|
751 |
} else { |
0a31bc97c
|
752 |
mem_cgroup_migrate(page, newpage, false); |
2ebba6b7e
|
753 |
if (page_was_mapped) |
3fe2011ff
|
754 |
remove_migration_ptes(page, newpage); |
35512ecae
|
755 |
page->mapping = NULL; |
3fe2011ff
|
756 |
} |
e24f0b8f7
|
757 758 759 760 761 |
unlock_page(newpage); return rc; } |
0dabec93d
|
762 |
static int __unmap_and_move(struct page *page, struct page *newpage, |
9c620e2bc
|
763 |
int force, enum migrate_mode mode) |
e24f0b8f7
|
764 |
{ |
0dabec93d
|
765 |
int rc = -EAGAIN; |
2ebba6b7e
|
766 |
int page_was_mapped = 0; |
3f6c82728
|
767 |
struct anon_vma *anon_vma = NULL; |
95a402c38
|
768 |
|
529ae9aaa
|
769 |
if (!trylock_page(page)) { |
a6bc32b89
|
770 |
if (!force || mode == MIGRATE_ASYNC) |
0dabec93d
|
771 |
goto out; |
3e7d34497
|
772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 |
/* * It's not safe for direct compaction to call lock_page. * For example, during page readahead pages are added locked * to the LRU. Later, when the IO completes the pages are * marked uptodate and unlocked. However, the queueing * could be merging multiple pages for one bio (e.g. * mpage_readpages). If an allocation happens for the * second or third page, the process can end up locking * the same page twice and deadlocking. Rather than * trying to be clever about what pages can be locked, * avoid the use of lock_page for direct compaction * altogether. */ if (current->flags & PF_MEMALLOC) |
0dabec93d
|
787 |
goto out; |
3e7d34497
|
788 |
|
e24f0b8f7
|
789 790 791 792 |
lock_page(page); } if (PageWriteback(page)) { |
11bc82d67
|
793 |
/* |
fed5b64a9
|
794 |
* Only in the case of a full synchronous migration is it |
a6bc32b89
|
795 796 797 |
* necessary to wait for PageWriteback. In the async case, * the retry loop is too short and in the sync-light case, * the overhead of stalling is too much |
11bc82d67
|
798 |
*/ |
a6bc32b89
|
799 |
if (mode != MIGRATE_SYNC) { |
11bc82d67
|
800 |
rc = -EBUSY; |
0a31bc97c
|
801 |
goto out_unlock; |
11bc82d67
|
802 803 |
} if (!force) |
0a31bc97c
|
804 |
goto out_unlock; |
e24f0b8f7
|
805 806 |
wait_on_page_writeback(page); } |
e24f0b8f7
|
807 |
/* |
dc386d4d1
|
808 809 |
* By try_to_unmap(), page->mapcount goes down to 0 here. In this case, * we cannot notice that anon_vma is freed while we migrates a page. |
1ce82b69e
|
810 |
* This get_anon_vma() delays freeing anon_vma pointer until the end |
dc386d4d1
|
811 |
* of migration. File cache pages are no problem because of page_lock() |
989f89c57
|
812 813 |
* File Caches may use write_page() or lock_page() in migration, then, * just care Anon page here. |
dc386d4d1
|
814 |
*/ |
b79bc0a0c
|
815 |
if (PageAnon(page) && !PageKsm(page)) { |
1ce82b69e
|
816 |
/* |
4fc3f1d66
|
817 |
* Only page_lock_anon_vma_read() understands the subtleties of |
1ce82b69e
|
818 819 |
* getting a hold on an anon_vma from outside one of its mms. */ |
746b18d42
|
820 |
anon_vma = page_get_anon_vma(page); |
1ce82b69e
|
821 822 |
if (anon_vma) { /* |
746b18d42
|
823 |
* Anon page |
1ce82b69e
|
824 |
*/ |
1ce82b69e
|
825 |
} else if (PageSwapCache(page)) { |
3fe2011ff
|
826 827 828 829 830 831 832 833 834 835 836 837 |
/* * We cannot be sure that the anon_vma of an unmapped * swapcache page is safe to use because we don't * know in advance if the VMA that this page belonged * to still exists. If the VMA and others sharing the * data have been freed, then the anon_vma could * already be invalid. * * To avoid this possibility, swapcache pages get * migrated but are not remapped when migration * completes */ |
3fe2011ff
|
838 |
} else { |
0a31bc97c
|
839 |
goto out_unlock; |
3fe2011ff
|
840 |
} |
989f89c57
|
841 |
} |
62e1c5530
|
842 |
|
d6d86c0a7
|
843 |
if (unlikely(isolated_balloon_page(page))) { |
bf6bddf19
|
844 845 846 847 848 849 850 851 |
/* * A ballooned page does not need any special attention from * physical to virtual reverse mapping procedures. * Skip any attempt to unmap PTEs or to remap swap cache, * in order to avoid burning cycles at rmap level, and perform * the page migration right away (proteced by page lock). */ rc = balloon_page_migrate(newpage, page, mode); |
0a31bc97c
|
852 |
goto out_unlock; |
bf6bddf19
|
853 |
} |
dc386d4d1
|
854 |
/* |
62e1c5530
|
855 856 857 858 859 860 861 862 863 864 |
* Corner case handling: * 1. When a new swap-cache page is read into, it is added to the LRU * and treated as swapcache but it has no rmap yet. * Calling try_to_unmap() against a page->mapping==NULL page will * trigger a BUG. So handle it here. * 2. An orphaned page (see truncate_complete_page) might have * fs-private metadata. The page can be picked up due to memory * offlining. Everywhere else except page reclaim, the page is * invisible to the vm, so the page can not be migrated. So try to * free the metadata, so the page can be freed. |
e24f0b8f7
|
865 |
*/ |
62e1c5530
|
866 |
if (!page->mapping) { |
309381fea
|
867 |
VM_BUG_ON_PAGE(PageAnon(page), page); |
1ce82b69e
|
868 |
if (page_has_private(page)) { |
62e1c5530
|
869 |
try_to_free_buffers(page); |
0a31bc97c
|
870 |
goto out_unlock; |
62e1c5530
|
871 |
} |
abfc34881
|
872 |
goto skip_unmap; |
62e1c5530
|
873 |
} |
dc386d4d1
|
874 |
/* Establish migration ptes or remove ptes */ |
2ebba6b7e
|
875 876 877 878 879 |
if (page_mapped(page)) { try_to_unmap(page, TTU_MIGRATION|TTU_IGNORE_MLOCK|TTU_IGNORE_ACCESS); page_was_mapped = 1; } |
dc386d4d1
|
880 |
|
abfc34881
|
881 |
skip_unmap: |
e6a1530d6
|
882 |
if (!page_mapped(page)) |
2ebba6b7e
|
883 |
rc = move_to_new_page(newpage, page, page_was_mapped, mode); |
e24f0b8f7
|
884 |
|
2ebba6b7e
|
885 |
if (rc && page_was_mapped) |
e24f0b8f7
|
886 |
remove_migration_ptes(page, page); |
3f6c82728
|
887 888 |
/* Drop an anon_vma reference if we took one */ |
76545066c
|
889 |
if (anon_vma) |
9e60109f1
|
890 |
put_anon_vma(anon_vma); |
3f6c82728
|
891 |
|
0a31bc97c
|
892 |
out_unlock: |
e24f0b8f7
|
893 |
unlock_page(page); |
0dabec93d
|
894 895 896 |
out: return rc; } |
95a402c38
|
897 |
|
0dabec93d
|
898 899 900 901 |
/* * Obtain the lock on page, remove all ptes and migrate the page * to the newly allocated page in newpage. */ |
68711a746
|
902 903 904 |
static int unmap_and_move(new_page_t get_new_page, free_page_t put_new_page, unsigned long private, struct page *page, int force, enum migrate_mode mode) |
0dabec93d
|
905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 |
{ int rc = 0; int *result = NULL; struct page *newpage = get_new_page(page, private, &result); if (!newpage) return -ENOMEM; if (page_count(page) == 1) { /* page was freed from under us. So we are done. */ goto out; } if (unlikely(PageTransHuge(page))) if (unlikely(split_huge_page(page))) goto out; |
9c620e2bc
|
921 |
rc = __unmap_and_move(page, newpage, force, mode); |
bf6bddf19
|
922 |
|
0dabec93d
|
923 |
out: |
e24f0b8f7
|
924 |
if (rc != -EAGAIN) { |
0dabec93d
|
925 926 927 928 929 930 931 |
/* * A page that has been migrated has all references * removed and will be freed. A page that has not been * migrated will have kepts its references and be * restored. */ list_del(&page->lru); |
a731286de
|
932 |
dec_zone_page_state(page, NR_ISOLATED_ANON + |
6c0b13519
|
933 |
page_is_file_cache(page)); |
894bc3104
|
934 |
putback_lru_page(page); |
e24f0b8f7
|
935 |
} |
68711a746
|
936 |
|
95a402c38
|
937 |
/* |
68711a746
|
938 939 940 |
* If migration was not successful and there's a freeing callback, use * it. Otherwise, putback_lru_page() will drop the reference grabbed * during isolation. |
95a402c38
|
941 |
*/ |
8bdd63809
|
942 943 |
if (rc != MIGRATEPAGE_SUCCESS && put_new_page) { ClearPageSwapBacked(newpage); |
68711a746
|
944 |
put_new_page(newpage, private); |
d6d86c0a7
|
945 946 947 |
} else if (unlikely(__is_movable_balloon_page(newpage))) { /* drop our reference, page already in the balloon */ put_page(newpage); |
8bdd63809
|
948 |
} else |
68711a746
|
949 |
putback_lru_page(newpage); |
742755a1d
|
950 951 952 953 954 955 |
if (result) { if (rc) *result = rc; else *result = page_to_nid(newpage); } |
e24f0b8f7
|
956 957 958 959 |
return rc; } /* |
290408d4a
|
960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 |
* Counterpart of unmap_and_move_page() for hugepage migration. * * This function doesn't wait the completion of hugepage I/O * because there is no race between I/O and migration for hugepage. * Note that currently hugepage I/O occurs only in direct I/O * where no lock is held and PG_writeback is irrelevant, * and writeback status of all subpages are counted in the reference * count of the head page (i.e. if all subpages of a 2MB hugepage are * under direct I/O, the reference of the head page is 512 and a bit more.) * This means that when we try to migrate hugepage whose subpages are * doing direct I/O, some references remain after try_to_unmap() and * hugepage migration fails without data corruption. * * There is also no race when direct I/O is issued on the page under migration, * because then pte is replaced with migration swap entry and direct I/O code * will wait in the page fault for migration to complete. */ static int unmap_and_move_huge_page(new_page_t get_new_page, |
68711a746
|
978 979 980 |
free_page_t put_new_page, unsigned long private, struct page *hpage, int force, enum migrate_mode mode) |
290408d4a
|
981 982 983 |
{ int rc = 0; int *result = NULL; |
2ebba6b7e
|
984 |
int page_was_mapped = 0; |
32665f2bb
|
985 |
struct page *new_hpage; |
290408d4a
|
986 |
struct anon_vma *anon_vma = NULL; |
83467efbd
|
987 988 989 990 991 992 993 |
/* * Movability of hugepages depends on architectures and hugepage size. * This check is necessary because some callers of hugepage migration * like soft offline and memory hotremove don't walk through page * tables or check whether the hugepage is pmd-based or not before * kicking migration. */ |
100873d7a
|
994 |
if (!hugepage_migration_supported(page_hstate(hpage))) { |
32665f2bb
|
995 |
putback_active_hugepage(hpage); |
83467efbd
|
996 |
return -ENOSYS; |
32665f2bb
|
997 |
} |
83467efbd
|
998 |
|
32665f2bb
|
999 |
new_hpage = get_new_page(hpage, private, &result); |
290408d4a
|
1000 1001 1002 1003 1004 1005 |
if (!new_hpage) return -ENOMEM; rc = -EAGAIN; if (!trylock_page(hpage)) { |
a6bc32b89
|
1006 |
if (!force || mode != MIGRATE_SYNC) |
290408d4a
|
1007 1008 1009 |
goto out; lock_page(hpage); } |
746b18d42
|
1010 1011 |
if (PageAnon(hpage)) anon_vma = page_get_anon_vma(hpage); |
290408d4a
|
1012 |
|
2ebba6b7e
|
1013 1014 1015 1016 1017 |
if (page_mapped(hpage)) { try_to_unmap(hpage, TTU_MIGRATION|TTU_IGNORE_MLOCK|TTU_IGNORE_ACCESS); page_was_mapped = 1; } |
290408d4a
|
1018 1019 |
if (!page_mapped(hpage)) |
2ebba6b7e
|
1020 |
rc = move_to_new_page(new_hpage, hpage, page_was_mapped, mode); |
290408d4a
|
1021 |
|
2ebba6b7e
|
1022 |
if (rc != MIGRATEPAGE_SUCCESS && page_was_mapped) |
290408d4a
|
1023 |
remove_migration_ptes(hpage, hpage); |
fd4a4663d
|
1024 |
if (anon_vma) |
9e60109f1
|
1025 |
put_anon_vma(anon_vma); |
8e6ac7fab
|
1026 |
|
68711a746
|
1027 |
if (rc == MIGRATEPAGE_SUCCESS) |
8e6ac7fab
|
1028 |
hugetlb_cgroup_migrate(hpage, new_hpage); |
290408d4a
|
1029 |
unlock_page(hpage); |
09761333e
|
1030 |
out: |
b8ec1cee5
|
1031 1032 |
if (rc != -EAGAIN) putback_active_hugepage(hpage); |
68711a746
|
1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 |
/* * If migration was not successful and there's a freeing callback, use * it. Otherwise, put_page() will drop the reference grabbed during * isolation. */ if (rc != MIGRATEPAGE_SUCCESS && put_new_page) put_new_page(new_hpage, private); else put_page(new_hpage); |
290408d4a
|
1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 |
if (result) { if (rc) *result = rc; else *result = page_to_nid(new_hpage); } return rc; } /* |
c73e5c9c5
|
1053 1054 |
* migrate_pages - migrate the pages specified in a list, to the free pages * supplied as the target for the page migration |
b20a35035
|
1055 |
* |
c73e5c9c5
|
1056 1057 1058 |
* @from: The list of pages to be migrated. * @get_new_page: The function used to allocate free pages to be used * as the target of the page migration. |
68711a746
|
1059 1060 |
* @put_new_page: The function used to free target pages if migration * fails, or NULL if no special handling is necessary. |
c73e5c9c5
|
1061 1062 1063 1064 |
* @private: Private data to be passed on to get_new_page() * @mode: The migration mode that specifies the constraints for * page migration, if any. * @reason: The reason for page migration. |
b20a35035
|
1065 |
* |
c73e5c9c5
|
1066 1067 1068 |
* The function returns after 10 attempts or if no pages are movable any more * because the list has become empty or no retryable pages exist any more. * The caller should call putback_lru_pages() to return pages to the LRU |
28bd65781
|
1069 |
* or free list only if ret != 0. |
b20a35035
|
1070 |
* |
c73e5c9c5
|
1071 |
* Returns the number of pages that were not migrated, or an error code. |
b20a35035
|
1072 |
*/ |
9c620e2bc
|
1073 |
int migrate_pages(struct list_head *from, new_page_t get_new_page, |
68711a746
|
1074 1075 |
free_page_t put_new_page, unsigned long private, enum migrate_mode mode, int reason) |
b20a35035
|
1076 |
{ |
e24f0b8f7
|
1077 |
int retry = 1; |
b20a35035
|
1078 |
int nr_failed = 0; |
5647bc293
|
1079 |
int nr_succeeded = 0; |
b20a35035
|
1080 1081 1082 1083 1084 1085 1086 1087 |
int pass = 0; struct page *page; struct page *page2; int swapwrite = current->flags & PF_SWAPWRITE; int rc; if (!swapwrite) current->flags |= PF_SWAPWRITE; |
e24f0b8f7
|
1088 1089 |
for(pass = 0; pass < 10 && retry; pass++) { retry = 0; |
b20a35035
|
1090 |
|
e24f0b8f7
|
1091 |
list_for_each_entry_safe(page, page2, from, lru) { |
e24f0b8f7
|
1092 |
cond_resched(); |
2d1db3b11
|
1093 |
|
31caf665e
|
1094 1095 |
if (PageHuge(page)) rc = unmap_and_move_huge_page(get_new_page, |
68711a746
|
1096 1097 |
put_new_page, private, page, pass > 2, mode); |
31caf665e
|
1098 |
else |
68711a746
|
1099 1100 |
rc = unmap_and_move(get_new_page, put_new_page, private, page, pass > 2, mode); |
2d1db3b11
|
1101 |
|
e24f0b8f7
|
1102 |
switch(rc) { |
95a402c38
|
1103 1104 |
case -ENOMEM: goto out; |
e24f0b8f7
|
1105 |
case -EAGAIN: |
2d1db3b11
|
1106 |
retry++; |
e24f0b8f7
|
1107 |
break; |
78bd52097
|
1108 |
case MIGRATEPAGE_SUCCESS: |
5647bc293
|
1109 |
nr_succeeded++; |
e24f0b8f7
|
1110 1111 |
break; default: |
354a33633
|
1112 1113 1114 1115 1116 1117 |
/* * Permanent failure (-EBUSY, -ENOSYS, etc.): * unlike -EAGAIN case, the failed page is * removed from migration page list and not * retried in the next outer loop. */ |
2d1db3b11
|
1118 |
nr_failed++; |
e24f0b8f7
|
1119 |
break; |
2d1db3b11
|
1120 |
} |
b20a35035
|
1121 1122 |
} } |
78bd52097
|
1123 |
rc = nr_failed + retry; |
95a402c38
|
1124 |
out: |
5647bc293
|
1125 1126 1127 1128 |
if (nr_succeeded) count_vm_events(PGMIGRATE_SUCCESS, nr_succeeded); if (nr_failed) count_vm_events(PGMIGRATE_FAIL, nr_failed); |
7b2a2d4a1
|
1129 |
trace_mm_migrate_pages(nr_succeeded, nr_failed, mode, reason); |
b20a35035
|
1130 1131 |
if (!swapwrite) current->flags &= ~PF_SWAPWRITE; |
78bd52097
|
1132 |
return rc; |
b20a35035
|
1133 |
} |
95a402c38
|
1134 |
|
742755a1d
|
1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 |
#ifdef CONFIG_NUMA /* * Move a list of individual pages */ struct page_to_node { unsigned long addr; struct page *page; int node; int status; }; static struct page *new_page_node(struct page *p, unsigned long private, int **result) { struct page_to_node *pm = (struct page_to_node *)private; while (pm->node != MAX_NUMNODES && pm->page != p) pm++; if (pm->node == MAX_NUMNODES) return NULL; *result = &pm->status; |
e632a938d
|
1158 1159 1160 1161 1162 |
if (PageHuge(p)) return alloc_huge_page_node(page_hstate(compound_head(p)), pm->node); else return alloc_pages_exact_node(pm->node, |
e97ca8e5b
|
1163 |
GFP_HIGHUSER_MOVABLE | __GFP_THISNODE, 0); |
742755a1d
|
1164 1165 1166 1167 1168 1169 |
} /* * Move a set of pages as indicated in the pm array. The addr * field must be set to the virtual address of the page to be moved * and the node number must contain a valid target node. |
5e9a0f023
|
1170 |
* The pm array ends with node = MAX_NUMNODES. |
742755a1d
|
1171 |
*/ |
5e9a0f023
|
1172 1173 1174 |
static int do_move_page_to_node_array(struct mm_struct *mm, struct page_to_node *pm, int migrate_all) |
742755a1d
|
1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 |
{ int err; struct page_to_node *pp; LIST_HEAD(pagelist); down_read(&mm->mmap_sem); /* * Build a list of pages to migrate */ |
742755a1d
|
1185 1186 1187 |
for (pp = pm; pp->node != MAX_NUMNODES; pp++) { struct vm_area_struct *vma; struct page *page; |
742755a1d
|
1188 1189 |
err = -EFAULT; vma = find_vma(mm, pp->addr); |
70384dc6d
|
1190 |
if (!vma || pp->addr < vma->vm_start || !vma_migratable(vma)) |
742755a1d
|
1191 |
goto set_status; |
500d65d47
|
1192 |
page = follow_page(vma, pp->addr, FOLL_GET|FOLL_SPLIT); |
89f5b7da2
|
1193 1194 1195 1196 |
err = PTR_ERR(page); if (IS_ERR(page)) goto set_status; |
742755a1d
|
1197 1198 1199 |
err = -ENOENT; if (!page) goto set_status; |
62b61f611
|
1200 |
/* Use PageReserved to check for zero page */ |
b79bc0a0c
|
1201 |
if (PageReserved(page)) |
742755a1d
|
1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 |
goto put_and_set; pp->page = page; err = page_to_nid(page); if (err == pp->node) /* * Node already in the right place */ goto put_and_set; err = -EACCES; if (page_mapcount(page) > 1 && !migrate_all) goto put_and_set; |
e632a938d
|
1217 1218 1219 1220 |
if (PageHuge(page)) { isolate_huge_page(page, &pagelist); goto put_and_set; } |
62695a84e
|
1221 |
err = isolate_lru_page(page); |
6d9c285a6
|
1222 |
if (!err) { |
62695a84e
|
1223 |
list_add_tail(&page->lru, &pagelist); |
6d9c285a6
|
1224 1225 1226 |
inc_zone_page_state(page, NR_ISOLATED_ANON + page_is_file_cache(page)); } |
742755a1d
|
1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 |
put_and_set: /* * Either remove the duplicate refcount from * isolate_lru_page() or drop the page ref if it was * not isolated. */ put_page(page); set_status: pp->status = err; } |
e78bbfa82
|
1237 |
err = 0; |
cf608ac19
|
1238 |
if (!list_empty(&pagelist)) { |
68711a746
|
1239 |
err = migrate_pages(&pagelist, new_page_node, NULL, |
9c620e2bc
|
1240 |
(unsigned long)pm, MIGRATE_SYNC, MR_SYSCALL); |
cf608ac19
|
1241 |
if (err) |
e632a938d
|
1242 |
putback_movable_pages(&pagelist); |
cf608ac19
|
1243 |
} |
742755a1d
|
1244 1245 1246 1247 1248 1249 |
up_read(&mm->mmap_sem); return err; } /* |
5e9a0f023
|
1250 1251 1252 |
* Migrate an array of page address onto an array of nodes and fill * the corresponding array of status. */ |
3268c63ed
|
1253 |
static int do_pages_move(struct mm_struct *mm, nodemask_t task_nodes, |
5e9a0f023
|
1254 1255 1256 1257 1258 |
unsigned long nr_pages, const void __user * __user *pages, const int __user *nodes, int __user *status, int flags) { |
3140a2273
|
1259 |
struct page_to_node *pm; |
3140a2273
|
1260 1261 1262 |
unsigned long chunk_nr_pages; unsigned long chunk_start; int err; |
5e9a0f023
|
1263 |
|
3140a2273
|
1264 1265 1266 |
err = -ENOMEM; pm = (struct page_to_node *)__get_free_page(GFP_KERNEL); if (!pm) |
5e9a0f023
|
1267 |
goto out; |
35282a2de
|
1268 1269 |
migrate_prep(); |
5e9a0f023
|
1270 |
/* |
3140a2273
|
1271 1272 |
* Store a chunk of page_to_node array in a page, * but keep the last one as a marker |
5e9a0f023
|
1273 |
*/ |
3140a2273
|
1274 |
chunk_nr_pages = (PAGE_SIZE / sizeof(struct page_to_node)) - 1; |
5e9a0f023
|
1275 |
|
3140a2273
|
1276 1277 1278 1279 |
for (chunk_start = 0; chunk_start < nr_pages; chunk_start += chunk_nr_pages) { int j; |
5e9a0f023
|
1280 |
|
3140a2273
|
1281 1282 1283 1284 1285 1286 |
if (chunk_start + chunk_nr_pages > nr_pages) chunk_nr_pages = nr_pages - chunk_start; /* fill the chunk pm with addrs and nodes from user-space */ for (j = 0; j < chunk_nr_pages; j++) { const void __user *p; |
5e9a0f023
|
1287 |
int node; |
3140a2273
|
1288 1289 1290 1291 1292 1293 |
err = -EFAULT; if (get_user(p, pages + j + chunk_start)) goto out_pm; pm[j].addr = (unsigned long) p; if (get_user(node, nodes + j + chunk_start)) |
5e9a0f023
|
1294 1295 1296 |
goto out_pm; err = -ENODEV; |
6f5a55f1a
|
1297 1298 |
if (node < 0 || node >= MAX_NUMNODES) goto out_pm; |
389162c22
|
1299 |
if (!node_state(node, N_MEMORY)) |
5e9a0f023
|
1300 1301 1302 1303 1304 |
goto out_pm; err = -EACCES; if (!node_isset(node, task_nodes)) goto out_pm; |
3140a2273
|
1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 |
pm[j].node = node; } /* End marker for this chunk */ pm[chunk_nr_pages].node = MAX_NUMNODES; /* Migrate this chunk */ err = do_move_page_to_node_array(mm, pm, flags & MPOL_MF_MOVE_ALL); if (err < 0) goto out_pm; |
5e9a0f023
|
1316 |
|
5e9a0f023
|
1317 |
/* Return status information */ |
3140a2273
|
1318 1319 |
for (j = 0; j < chunk_nr_pages; j++) if (put_user(pm[j].status, status + j + chunk_start)) { |
5e9a0f023
|
1320 |
err = -EFAULT; |
3140a2273
|
1321 1322 1323 1324 |
goto out_pm; } } err = 0; |
5e9a0f023
|
1325 1326 |
out_pm: |
3140a2273
|
1327 |
free_page((unsigned long)pm); |
5e9a0f023
|
1328 1329 1330 1331 1332 |
out: return err; } /* |
2f007e74b
|
1333 |
* Determine the nodes of an array of pages and store it in an array of status. |
742755a1d
|
1334 |
*/ |
80bba1290
|
1335 1336 |
static void do_pages_stat_array(struct mm_struct *mm, unsigned long nr_pages, const void __user **pages, int *status) |
742755a1d
|
1337 |
{ |
2f007e74b
|
1338 |
unsigned long i; |
2f007e74b
|
1339 |
|
742755a1d
|
1340 |
down_read(&mm->mmap_sem); |
2f007e74b
|
1341 |
for (i = 0; i < nr_pages; i++) { |
80bba1290
|
1342 |
unsigned long addr = (unsigned long)(*pages); |
742755a1d
|
1343 1344 |
struct vm_area_struct *vma; struct page *page; |
c095adbc2
|
1345 |
int err = -EFAULT; |
2f007e74b
|
1346 1347 |
vma = find_vma(mm, addr); |
70384dc6d
|
1348 |
if (!vma || addr < vma->vm_start) |
742755a1d
|
1349 |
goto set_status; |
2f007e74b
|
1350 |
page = follow_page(vma, addr, 0); |
89f5b7da2
|
1351 1352 1353 1354 |
err = PTR_ERR(page); if (IS_ERR(page)) goto set_status; |
742755a1d
|
1355 1356 |
err = -ENOENT; /* Use PageReserved to check for zero page */ |
b79bc0a0c
|
1357 |
if (!page || PageReserved(page)) |
742755a1d
|
1358 1359 1360 1361 |
goto set_status; err = page_to_nid(page); set_status: |
80bba1290
|
1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 |
*status = err; pages++; status++; } up_read(&mm->mmap_sem); } /* * Determine the nodes of a user array of pages and store it in * a user array of status. */ static int do_pages_stat(struct mm_struct *mm, unsigned long nr_pages, const void __user * __user *pages, int __user *status) { #define DO_PAGES_STAT_CHUNK_NR 16 const void __user *chunk_pages[DO_PAGES_STAT_CHUNK_NR]; int chunk_status[DO_PAGES_STAT_CHUNK_NR]; |
80bba1290
|
1382 |
|
87b8d1ade
|
1383 1384 |
while (nr_pages) { unsigned long chunk_nr; |
80bba1290
|
1385 |
|
87b8d1ade
|
1386 1387 1388 1389 1390 1391 |
chunk_nr = nr_pages; if (chunk_nr > DO_PAGES_STAT_CHUNK_NR) chunk_nr = DO_PAGES_STAT_CHUNK_NR; if (copy_from_user(chunk_pages, pages, chunk_nr * sizeof(*chunk_pages))) break; |
80bba1290
|
1392 1393 |
do_pages_stat_array(mm, chunk_nr, chunk_pages, chunk_status); |
87b8d1ade
|
1394 1395 |
if (copy_to_user(status, chunk_status, chunk_nr * sizeof(*status))) break; |
742755a1d
|
1396 |
|
87b8d1ade
|
1397 1398 1399 1400 1401 |
pages += chunk_nr; status += chunk_nr; nr_pages -= chunk_nr; } return nr_pages ? -EFAULT : 0; |
742755a1d
|
1402 1403 1404 1405 1406 1407 |
} /* * Move a list of pages in the address space of the currently executing * process. */ |
938bb9f5e
|
1408 1409 1410 1411 |
SYSCALL_DEFINE6(move_pages, pid_t, pid, unsigned long, nr_pages, const void __user * __user *, pages, const int __user *, nodes, int __user *, status, int, flags) |
742755a1d
|
1412 |
{ |
c69e8d9c0
|
1413 |
const struct cred *cred = current_cred(), *tcred; |
742755a1d
|
1414 |
struct task_struct *task; |
742755a1d
|
1415 |
struct mm_struct *mm; |
5e9a0f023
|
1416 |
int err; |
3268c63ed
|
1417 |
nodemask_t task_nodes; |
742755a1d
|
1418 1419 1420 1421 1422 1423 1424 1425 1426 |
/* Check flags */ if (flags & ~(MPOL_MF_MOVE|MPOL_MF_MOVE_ALL)) return -EINVAL; if ((flags & MPOL_MF_MOVE_ALL) && !capable(CAP_SYS_NICE)) return -EPERM; /* Find the mm_struct */ |
a879bf582
|
1427 |
rcu_read_lock(); |
228ebcbe6
|
1428 |
task = pid ? find_task_by_vpid(pid) : current; |
742755a1d
|
1429 |
if (!task) { |
a879bf582
|
1430 |
rcu_read_unlock(); |
742755a1d
|
1431 1432 |
return -ESRCH; } |
3268c63ed
|
1433 |
get_task_struct(task); |
742755a1d
|
1434 1435 1436 1437 1438 1439 1440 |
/* * Check if this process has the right to modify the specified * process. The right exists if the process has administrative * capabilities, superuser privileges or the same * userid as the target process. */ |
c69e8d9c0
|
1441 |
tcred = __task_cred(task); |
b38a86eb1
|
1442 1443 |
if (!uid_eq(cred->euid, tcred->suid) && !uid_eq(cred->euid, tcred->uid) && !uid_eq(cred->uid, tcred->suid) && !uid_eq(cred->uid, tcred->uid) && |
742755a1d
|
1444 |
!capable(CAP_SYS_NICE)) { |
c69e8d9c0
|
1445 |
rcu_read_unlock(); |
742755a1d
|
1446 |
err = -EPERM; |
5e9a0f023
|
1447 |
goto out; |
742755a1d
|
1448 |
} |
c69e8d9c0
|
1449 |
rcu_read_unlock(); |
742755a1d
|
1450 |
|
86c3a7645
|
1451 1452 |
err = security_task_movememory(task); if (err) |
5e9a0f023
|
1453 |
goto out; |
86c3a7645
|
1454 |
|
3268c63ed
|
1455 1456 1457 |
task_nodes = cpuset_mems_allowed(task); mm = get_task_mm(task); put_task_struct(task); |
6e8b09eaf
|
1458 1459 1460 1461 1462 1463 1464 1465 |
if (!mm) return -EINVAL; if (nodes) err = do_pages_move(mm, task_nodes, nr_pages, pages, nodes, status, flags); else err = do_pages_stat(mm, nr_pages, pages, status); |
742755a1d
|
1466 |
|
742755a1d
|
1467 1468 |
mmput(mm); return err; |
3268c63ed
|
1469 1470 1471 1472 |
out: put_task_struct(task); return err; |
742755a1d
|
1473 |
} |
742755a1d
|
1474 |
|
7039e1dbe
|
1475 1476 1477 1478 1479 1480 |
#ifdef CONFIG_NUMA_BALANCING /* * Returns true if this is a safe migration target node for misplaced NUMA * pages. Currently it only checks the watermarks which crude */ static bool migrate_balanced_pgdat(struct pglist_data *pgdat, |
3abef4e6c
|
1481 |
unsigned long nr_migrate_pages) |
7039e1dbe
|
1482 1483 1484 1485 1486 1487 1488 |
{ int z; for (z = pgdat->nr_zones - 1; z >= 0; z--) { struct zone *zone = pgdat->node_zones + z; if (!populated_zone(zone)) continue; |
6e543d578
|
1489 |
if (!zone_reclaimable(zone)) |
7039e1dbe
|
1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 |
continue; /* Avoid waking kswapd by allocating pages_to_migrate pages. */ if (!zone_watermark_ok(zone, 0, high_wmark_pages(zone) + nr_migrate_pages, 0, 0)) continue; return true; } return false; } static struct page *alloc_misplaced_dst_page(struct page *page, unsigned long data, int **result) { int nid = (int) data; struct page *newpage; newpage = alloc_pages_exact_node(nid, |
e97ca8e5b
|
1511 1512 1513 |
(GFP_HIGHUSER_MOVABLE | __GFP_THISNODE | __GFP_NOMEMALLOC | __GFP_NORETRY | __GFP_NOWARN) & |
7039e1dbe
|
1514 |
~GFP_IOFS, 0); |
bac0382c6
|
1515 |
|
7039e1dbe
|
1516 1517 1518 1519 |
return newpage; } /* |
a8f607721
|
1520 1521 1522 |
* page migration rate limiting control. * Do not migrate more than @pages_to_migrate in a @migrate_interval_millisecs * window of time. Default here says do not migrate more than 1280M per second. |
e14808b49
|
1523 1524 1525 1526 |
* If a node is rate-limited then PTE NUMA updates are also rate-limited. However * as it is faults that reset the window, pte updates will happen unconditionally * if there has not been a fault since @pteupdate_interval_millisecs after the * throttle window closed. |
a8f607721
|
1527 1528 |
*/ static unsigned int migrate_interval_millisecs __read_mostly = 100; |
e14808b49
|
1529 |
static unsigned int pteupdate_interval_millisecs __read_mostly = 1000; |
a8f607721
|
1530 |
static unsigned int ratelimit_pages __read_mostly = 128 << (20 - PAGE_SHIFT); |
e14808b49
|
1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 |
/* Returns true if NUMA migration is currently rate limited */ bool migrate_ratelimited(int node) { pg_data_t *pgdat = NODE_DATA(node); if (time_after(jiffies, pgdat->numabalancing_migrate_next_window + msecs_to_jiffies(pteupdate_interval_millisecs))) return false; if (pgdat->numabalancing_migrate_nr_pages < ratelimit_pages) return false; return true; } |
b32967ff1
|
1545 |
/* Returns true if the node is migrate rate-limited after the update */ |
1c30e0177
|
1546 1547 |
static bool numamigrate_update_ratelimit(pg_data_t *pgdat, unsigned long nr_pages) |
7039e1dbe
|
1548 |
{ |
a8f607721
|
1549 1550 1551 1552 1553 |
/* * Rate-limit the amount of data that is being migrated to a node. * Optimal placement is no good if the memory bus is saturated and * all the time is being spent migrating! */ |
a8f607721
|
1554 |
if (time_after(jiffies, pgdat->numabalancing_migrate_next_window)) { |
1c5e9c27c
|
1555 |
spin_lock(&pgdat->numabalancing_migrate_lock); |
a8f607721
|
1556 1557 1558 |
pgdat->numabalancing_migrate_nr_pages = 0; pgdat->numabalancing_migrate_next_window = jiffies + msecs_to_jiffies(migrate_interval_millisecs); |
1c5e9c27c
|
1559 |
spin_unlock(&pgdat->numabalancing_migrate_lock); |
a8f607721
|
1560 |
} |
af1839d72
|
1561 1562 1563 |
if (pgdat->numabalancing_migrate_nr_pages > ratelimit_pages) { trace_mm_numa_migrate_ratelimit(current, pgdat->node_id, nr_pages); |
1c5e9c27c
|
1564 |
return true; |
af1839d72
|
1565 |
} |
1c5e9c27c
|
1566 1567 1568 1569 1570 1571 1572 1573 1574 |
/* * This is an unlocked non-atomic update so errors are possible. * The consequences are failing to migrate when we potentiall should * have which is not severe enough to warrant locking. If it is ever * a problem, it can be converted to a per-cpu counter. */ pgdat->numabalancing_migrate_nr_pages += nr_pages; return false; |
b32967ff1
|
1575 |
} |
1c30e0177
|
1576 |
static int numamigrate_isolate_page(pg_data_t *pgdat, struct page *page) |
b32967ff1
|
1577 |
{ |
340ef3902
|
1578 |
int page_lru; |
a8f607721
|
1579 |
|
309381fea
|
1580 |
VM_BUG_ON_PAGE(compound_order(page) && !PageTransHuge(page), page); |
3abef4e6c
|
1581 |
|
7039e1dbe
|
1582 |
/* Avoid migrating to a node that is nearly full */ |
340ef3902
|
1583 1584 |
if (!migrate_balanced_pgdat(pgdat, 1UL << compound_order(page))) return 0; |
7039e1dbe
|
1585 |
|
340ef3902
|
1586 1587 |
if (isolate_lru_page(page)) return 0; |
7039e1dbe
|
1588 |
|
340ef3902
|
1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 |
/* * migrate_misplaced_transhuge_page() skips page migration's usual * check on page_count(), so we must do it here, now that the page * has been isolated: a GUP pin, or any other pin, prevents migration. * The expected page count is 3: 1 for page's mapcount and 1 for the * caller's pin and 1 for the reference taken by isolate_lru_page(). */ if (PageTransHuge(page) && page_count(page) != 3) { putback_lru_page(page); return 0; |
7039e1dbe
|
1599 |
} |
340ef3902
|
1600 1601 1602 |
page_lru = page_is_file_cache(page); mod_zone_page_state(page_zone(page), NR_ISOLATED_ANON + page_lru, hpage_nr_pages(page)); |
149c33e1c
|
1603 |
/* |
340ef3902
|
1604 1605 1606 |
* Isolating the page has taken another reference, so the * caller's reference can be safely dropped without the page * disappearing underneath us during migration. |
149c33e1c
|
1607 1608 |
*/ put_page(page); |
340ef3902
|
1609 |
return 1; |
b32967ff1
|
1610 |
} |
de466bd62
|
1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 |
bool pmd_trans_migrating(pmd_t pmd) { struct page *page = pmd_page(pmd); return PageLocked(page); } void wait_migrate_huge_page(struct anon_vma *anon_vma, pmd_t *pmd) { struct page *page = pmd_page(*pmd); wait_on_page_locked(page); } |
b32967ff1
|
1622 1623 1624 1625 1626 |
/* * Attempt to migrate a misplaced page to the specified destination * node. Caller is expected to have an elevated reference count on * the page that will be dropped by this function before returning. */ |
1bc115d87
|
1627 1628 |
int migrate_misplaced_page(struct page *page, struct vm_area_struct *vma, int node) |
b32967ff1
|
1629 1630 |
{ pg_data_t *pgdat = NODE_DATA(node); |
340ef3902
|
1631 |
int isolated; |
b32967ff1
|
1632 1633 1634 1635 |
int nr_remaining; LIST_HEAD(migratepages); /* |
1bc115d87
|
1636 1637 |
* Don't migrate file pages that are mapped in multiple processes * with execute permissions as they are probably shared libraries. |
b32967ff1
|
1638 |
*/ |
1bc115d87
|
1639 1640 |
if (page_mapcount(page) != 1 && page_is_file_cache(page) && (vma->vm_flags & VM_EXEC)) |
b32967ff1
|
1641 |
goto out; |
b32967ff1
|
1642 1643 1644 1645 1646 1647 |
/* * Rate-limit the amount of data that is being migrated to a node. * Optimal placement is no good if the memory bus is saturated and * all the time is being spent migrating! */ |
340ef3902
|
1648 |
if (numamigrate_update_ratelimit(pgdat, 1)) |
b32967ff1
|
1649 |
goto out; |
b32967ff1
|
1650 1651 1652 1653 1654 1655 |
isolated = numamigrate_isolate_page(pgdat, page); if (!isolated) goto out; list_add(&page->lru, &migratepages); |
9c620e2bc
|
1656 |
nr_remaining = migrate_pages(&migratepages, alloc_misplaced_dst_page, |
68711a746
|
1657 1658 |
NULL, node, MIGRATE_ASYNC, MR_NUMA_MISPLACED); |
b32967ff1
|
1659 |
if (nr_remaining) { |
59c82b70d
|
1660 1661 1662 1663 1664 1665 |
if (!list_empty(&migratepages)) { list_del(&page->lru); dec_zone_page_state(page, NR_ISOLATED_ANON + page_is_file_cache(page)); putback_lru_page(page); } |
b32967ff1
|
1666 1667 1668 |
isolated = 0; } else count_vm_numa_event(NUMA_PAGE_MIGRATE); |
7039e1dbe
|
1669 |
BUG_ON(!list_empty(&migratepages)); |
7039e1dbe
|
1670 |
return isolated; |
340ef3902
|
1671 1672 1673 1674 |
out: put_page(page); return 0; |
7039e1dbe
|
1675 |
} |
220018d38
|
1676 |
#endif /* CONFIG_NUMA_BALANCING */ |
b32967ff1
|
1677 |
|
220018d38
|
1678 |
#if defined(CONFIG_NUMA_BALANCING) && defined(CONFIG_TRANSPARENT_HUGEPAGE) |
340ef3902
|
1679 1680 1681 1682 |
/* * Migrates a THP to a given target node. page must be locked and is unlocked * before returning. */ |
b32967ff1
|
1683 1684 1685 1686 1687 1688 |
int migrate_misplaced_transhuge_page(struct mm_struct *mm, struct vm_area_struct *vma, pmd_t *pmd, pmd_t entry, unsigned long address, struct page *page, int node) { |
c4088ebdc
|
1689 |
spinlock_t *ptl; |
b32967ff1
|
1690 1691 1692 |
pg_data_t *pgdat = NODE_DATA(node); int isolated = 0; struct page *new_page = NULL; |
b32967ff1
|
1693 |
int page_lru = page_is_file_cache(page); |
f714f4f20
|
1694 1695 |
unsigned long mmun_start = address & HPAGE_PMD_MASK; unsigned long mmun_end = mmun_start + HPAGE_PMD_SIZE; |
2b4847e73
|
1696 |
pmd_t orig_entry; |
b32967ff1
|
1697 1698 |
/* |
b32967ff1
|
1699 1700 1701 1702 |
* Rate-limit the amount of data that is being migrated to a node. * Optimal placement is no good if the memory bus is saturated and * all the time is being spent migrating! */ |
d28d43351
|
1703 |
if (numamigrate_update_ratelimit(pgdat, HPAGE_PMD_NR)) |
b32967ff1
|
1704 1705 1706 |
goto out_dropref; new_page = alloc_pages_node(node, |
e97ca8e5b
|
1707 1708 |
(GFP_TRANSHUGE | __GFP_THISNODE) & ~__GFP_WAIT, HPAGE_PMD_ORDER); |
340ef3902
|
1709 1710 |
if (!new_page) goto out_fail; |
b32967ff1
|
1711 |
isolated = numamigrate_isolate_page(pgdat, page); |
340ef3902
|
1712 |
if (!isolated) { |
b32967ff1
|
1713 |
put_page(new_page); |
340ef3902
|
1714 |
goto out_fail; |
b32967ff1
|
1715 |
} |
b0943d61b
|
1716 1717 |
if (mm_tlb_flush_pending(mm)) flush_tlb_range(vma, mmun_start, mmun_end); |
b32967ff1
|
1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 |
/* Prepare a page as a migration target */ __set_page_locked(new_page); SetPageSwapBacked(new_page); /* anon mapping, we can simply copy page->mapping to the new page: */ new_page->mapping = page->mapping; new_page->index = page->index; migrate_page_copy(new_page, page); WARN_ON(PageLRU(new_page)); /* Recheck the target PMD */ |
f714f4f20
|
1729 |
mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); |
c4088ebdc
|
1730 |
ptl = pmd_lock(mm, pmd); |
2b4847e73
|
1731 1732 |
if (unlikely(!pmd_same(*pmd, entry) || page_count(page) != 2)) { fail_putback: |
c4088ebdc
|
1733 |
spin_unlock(ptl); |
f714f4f20
|
1734 |
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
b32967ff1
|
1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 |
/* Reverse changes made by migrate_page_copy() */ if (TestClearPageActive(new_page)) SetPageActive(page); if (TestClearPageUnevictable(new_page)) SetPageUnevictable(page); mlock_migrate_page(page, new_page); unlock_page(new_page); put_page(new_page); /* Free it */ |
a54a407fb
|
1745 1746 |
/* Retake the callers reference and putback on LRU */ get_page(page); |
b32967ff1
|
1747 |
putback_lru_page(page); |
a54a407fb
|
1748 1749 |
mod_zone_page_state(page_zone(page), NR_ISOLATED_ANON + page_lru, -HPAGE_PMD_NR); |
eb4489f69
|
1750 1751 |
goto out_unlock; |
b32967ff1
|
1752 |
} |
2b4847e73
|
1753 |
orig_entry = *pmd; |
b32967ff1
|
1754 |
entry = mk_pmd(new_page, vma->vm_page_prot); |
b32967ff1
|
1755 |
entry = pmd_mkhuge(entry); |
2b4847e73
|
1756 |
entry = maybe_pmd_mkwrite(pmd_mkdirty(entry), vma); |
b32967ff1
|
1757 |
|
2b4847e73
|
1758 1759 1760 1761 1762 1763 1764 |
/* * Clear the old entry under pagetable lock and establish the new PTE. * Any parallel GUP will either observe the old page blocking on the * page lock, block on the page table lock or observe the new page. * The SetPageUptodate on the new page and page_add_new_anon_rmap * guarantee the copy is visible before the pagetable update. */ |
f714f4f20
|
1765 |
flush_cache_range(vma, mmun_start, mmun_end); |
11de9927f
|
1766 |
page_add_anon_rmap(new_page, vma, mmun_start); |
34ee645e8
|
1767 |
pmdp_clear_flush_notify(vma, mmun_start, pmd); |
f714f4f20
|
1768 1769 |
set_pmd_at(mm, mmun_start, pmd, entry); flush_tlb_range(vma, mmun_start, mmun_end); |
ce4a9cc57
|
1770 |
update_mmu_cache_pmd(vma, address, &entry); |
2b4847e73
|
1771 1772 |
if (page_count(page) != 2) { |
f714f4f20
|
1773 1774 |
set_pmd_at(mm, mmun_start, pmd, orig_entry); flush_tlb_range(vma, mmun_start, mmun_end); |
34ee645e8
|
1775 |
mmu_notifier_invalidate_range(mm, mmun_start, mmun_end); |
2b4847e73
|
1776 1777 1778 1779 |
update_mmu_cache_pmd(vma, address, &entry); page_remove_rmap(new_page); goto fail_putback; } |
0a31bc97c
|
1780 |
mem_cgroup_migrate(page, new_page, false); |
b32967ff1
|
1781 |
page_remove_rmap(page); |
2b4847e73
|
1782 |
|
c4088ebdc
|
1783 |
spin_unlock(ptl); |
f714f4f20
|
1784 |
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
b32967ff1
|
1785 |
|
11de9927f
|
1786 1787 1788 |
/* Take an "isolate" reference and put new page on the LRU. */ get_page(new_page); putback_lru_page(new_page); |
b32967ff1
|
1789 1790 1791 1792 1793 1794 1795 |
unlock_page(new_page); unlock_page(page); put_page(page); /* Drop the rmap reference */ put_page(page); /* Drop the LRU isolation reference */ count_vm_events(PGMIGRATE_SUCCESS, HPAGE_PMD_NR); count_vm_numa_events(NUMA_PAGE_MIGRATE, HPAGE_PMD_NR); |
b32967ff1
|
1796 1797 1798 1799 |
mod_zone_page_state(page_zone(page), NR_ISOLATED_ANON + page_lru, -HPAGE_PMD_NR); return isolated; |
340ef3902
|
1800 1801 |
out_fail: count_vm_events(PGMIGRATE_FAIL, HPAGE_PMD_NR); |
b32967ff1
|
1802 |
out_dropref: |
2b4847e73
|
1803 1804 1805 |
ptl = pmd_lock(mm, pmd); if (pmd_same(*pmd, entry)) { entry = pmd_mknonnuma(entry); |
f714f4f20
|
1806 |
set_pmd_at(mm, mmun_start, pmd, entry); |
2b4847e73
|
1807 1808 1809 |
update_mmu_cache_pmd(vma, address, &entry); } spin_unlock(ptl); |
a54a407fb
|
1810 |
|
eb4489f69
|
1811 |
out_unlock: |
340ef3902
|
1812 |
unlock_page(page); |
b32967ff1
|
1813 |
put_page(page); |
b32967ff1
|
1814 1815 |
return 0; } |
7039e1dbe
|
1816 1817 1818 |
#endif /* CONFIG_NUMA_BALANCING */ #endif /* CONFIG_NUMA */ |