Blame view
mm/hugetlb.c
76.2 KB
1da177e4c
|
1 2 3 4 |
/* * Generic hugetlb support. * (C) William Irwin, April 2004 */ |
1da177e4c
|
5 6 7 8 |
#include <linux/list.h> #include <linux/init.h> #include <linux/module.h> #include <linux/mm.h> |
e1759c215
|
9 |
#include <linux/seq_file.h> |
1da177e4c
|
10 11 |
#include <linux/sysctl.h> #include <linux/highmem.h> |
cddb8a5c1
|
12 |
#include <linux/mmu_notifier.h> |
1da177e4c
|
13 |
#include <linux/nodemask.h> |
63551ae0f
|
14 |
#include <linux/pagemap.h> |
5da7ca860
|
15 |
#include <linux/mempolicy.h> |
aea47ff36
|
16 |
#include <linux/cpuset.h> |
3935baa9b
|
17 |
#include <linux/mutex.h> |
aa888a749
|
18 |
#include <linux/bootmem.h> |
a34378701
|
19 |
#include <linux/sysfs.h> |
5a0e3ad6a
|
20 |
#include <linux/slab.h> |
0fe6e20b9
|
21 |
#include <linux/rmap.h> |
fd6a03edd
|
22 23 |
#include <linux/swap.h> #include <linux/swapops.h> |
d6606683a
|
24 |
|
63551ae0f
|
25 26 |
#include <asm/page.h> #include <asm/pgtable.h> |
78a34ae29
|
27 |
#include <asm/io.h> |
63551ae0f
|
28 29 |
#include <linux/hugetlb.h> |
9a3052306
|
30 |
#include <linux/node.h> |
7835e98b2
|
31 |
#include "internal.h" |
1da177e4c
|
32 33 |
const unsigned long hugetlb_zero = 0, hugetlb_infinity = ~0UL; |
396faf030
|
34 35 |
static gfp_t htlb_alloc_mask = GFP_HIGHUSER; unsigned long hugepages_treat_as_movable; |
a55164389
|
36 |
|
e5ff21594
|
37 38 39 |
static int max_hstate; unsigned int default_hstate_idx; struct hstate hstates[HUGE_MAX_HSTATE]; |
53ba51d21
|
40 |
__initdata LIST_HEAD(huge_boot_pages); |
e5ff21594
|
41 42 43 |
/* for command line parsing */ static struct hstate * __initdata parsed_hstate; static unsigned long __initdata default_hstate_max_huge_pages; |
e11bfbfcb
|
44 |
static unsigned long __initdata default_hstate_size; |
e5ff21594
|
45 46 47 |
#define for_each_hstate(h) \ for ((h) = hstates; (h) < &hstates[max_hstate]; (h)++) |
396faf030
|
48 |
|
3935baa9b
|
49 50 51 52 |
/* * Protects updates to hugepage_freelists, nr_huge_pages, and free_huge_pages */ static DEFINE_SPINLOCK(hugetlb_lock); |
0bd0f9fb1
|
53 |
|
e7c4b0bfd
|
54 |
/* |
968229048
|
55 56 |
* Region tracking -- allows tracking of reservations and instantiated pages * across the pages in a mapping. |
84afd99b8
|
57 58 59 60 61 62 63 64 65 66 |
* * The region data structures are protected by a combination of the mmap_sem * and the hugetlb_instantion_mutex. To access or modify a region the caller * must either hold the mmap_sem for write, or the mmap_sem for read and * the hugetlb_instantiation mutex: * * down_write(&mm->mmap_sem); * or * down_read(&mm->mmap_sem); * mutex_lock(&hugetlb_instantiation_mutex); |
968229048
|
67 68 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 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 |
*/ struct file_region { struct list_head link; long from; long to; }; static long region_add(struct list_head *head, long f, long t) { struct file_region *rg, *nrg, *trg; /* Locate the region we are either in or before. */ list_for_each_entry(rg, head, link) if (f <= rg->to) break; /* Round our left edge to the current segment if it encloses us. */ if (f > rg->from) f = rg->from; /* Check for and consume any regions we now overlap with. */ nrg = rg; list_for_each_entry_safe(rg, trg, rg->link.prev, link) { if (&rg->link == head) break; if (rg->from > t) break; /* If this area reaches higher then extend our area to * include it completely. If this is not the first area * which we intend to reuse, free it. */ if (rg->to > t) t = rg->to; if (rg != nrg) { list_del(&rg->link); kfree(rg); } } nrg->from = f; nrg->to = t; return 0; } static long region_chg(struct list_head *head, long f, long t) { struct file_region *rg, *nrg; long chg = 0; /* Locate the region we are before or in. */ list_for_each_entry(rg, head, link) if (f <= rg->to) break; /* If we are below the current region then a new region is required. * Subtle, allocate a new region at the position but make it zero * size such that we can guarantee to record the reservation. */ if (&rg->link == head || t < rg->from) { nrg = kmalloc(sizeof(*nrg), GFP_KERNEL); if (!nrg) return -ENOMEM; nrg->from = f; nrg->to = f; INIT_LIST_HEAD(&nrg->link); list_add(&nrg->link, rg->link.prev); return t - f; } /* Round our left edge to the current segment if it encloses us. */ if (f > rg->from) f = rg->from; chg = t - f; /* Check for and consume any regions we now overlap with. */ list_for_each_entry(rg, rg->link.prev, link) { if (&rg->link == head) break; if (rg->from > t) return chg; |
25985edce
|
146 |
/* We overlap with this area, if it extends further than |
968229048
|
147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 |
* us then we must extend ourselves. Account for its * existing reservation. */ if (rg->to > t) { chg += rg->to - t; t = rg->to; } chg -= rg->to - rg->from; } return chg; } static long region_truncate(struct list_head *head, long end) { struct file_region *rg, *trg; long chg = 0; /* Locate the region we are either in or before. */ list_for_each_entry(rg, head, link) if (end <= rg->to) break; if (&rg->link == head) return 0; /* If we are in the middle of a region then adjust it. */ if (end > rg->from) { chg = rg->to - end; rg->to = end; rg = list_entry(rg->link.next, typeof(*rg), link); } /* Drop any remaining regions. */ list_for_each_entry_safe(rg, trg, rg->link.prev, link) { if (&rg->link == head) break; chg += rg->to - rg->from; list_del(&rg->link); kfree(rg); } return chg; } |
84afd99b8
|
187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 |
static long region_count(struct list_head *head, long f, long t) { struct file_region *rg; long chg = 0; /* Locate each segment we overlap with, and count that overlap. */ list_for_each_entry(rg, head, link) { int seg_from; int seg_to; if (rg->to <= f) continue; if (rg->from >= t) break; seg_from = max(rg->from, f); seg_to = min(rg->to, t); chg += seg_to - seg_from; } return chg; } |
968229048
|
210 |
/* |
e7c4b0bfd
|
211 |
* Convert the address within this vma to the page offset within |
e7c4b0bfd
|
212 213 |
* the mapping, in pagecache page units; huge pages here. */ |
a55164389
|
214 215 |
static pgoff_t vma_hugecache_offset(struct hstate *h, struct vm_area_struct *vma, unsigned long address) |
e7c4b0bfd
|
216 |
{ |
a55164389
|
217 218 |
return ((address - vma->vm_start) >> huge_page_shift(h)) + (vma->vm_pgoff >> huge_page_order(h)); |
e7c4b0bfd
|
219 |
} |
0fe6e20b9
|
220 221 222 223 224 |
pgoff_t linear_hugepage_index(struct vm_area_struct *vma, unsigned long address) { return vma_hugecache_offset(hstate_vma(vma), vma, address); } |
84afd99b8
|
225 |
/* |
08fba6998
|
226 227 228 229 230 231 232 233 234 235 236 237 238 239 |
* Return the size of the pages allocated when backing a VMA. In the majority * cases this will be same size as used by the page table entries. */ unsigned long vma_kernel_pagesize(struct vm_area_struct *vma) { struct hstate *hstate; if (!is_vm_hugetlb_page(vma)) return PAGE_SIZE; hstate = hstate_vma(vma); return 1UL << (hstate->order + PAGE_SHIFT); } |
f340ca0f0
|
240 |
EXPORT_SYMBOL_GPL(vma_kernel_pagesize); |
08fba6998
|
241 242 |
/* |
3340289dd
|
243 244 245 246 247 248 249 250 251 252 253 254 255 |
* Return the page size being used by the MMU to back a VMA. In the majority * of cases, the page size used by the kernel matches the MMU size. On * architectures where it differs, an architecture-specific version of this * function is required. */ #ifndef vma_mmu_pagesize unsigned long vma_mmu_pagesize(struct vm_area_struct *vma) { return vma_kernel_pagesize(vma); } #endif /* |
84afd99b8
|
256 257 258 259 260 261 |
* Flags for MAP_PRIVATE reservations. These are stored in the bottom * bits of the reservation map pointer, which are always clear due to * alignment. */ #define HPAGE_RESV_OWNER (1UL << 0) #define HPAGE_RESV_UNMAPPED (1UL << 1) |
04f2cbe35
|
262 |
#define HPAGE_RESV_MASK (HPAGE_RESV_OWNER | HPAGE_RESV_UNMAPPED) |
84afd99b8
|
263 |
|
a1e78772d
|
264 265 266 267 268 269 270 271 272 |
/* * These helpers are used to track how many pages are reserved for * faults in a MAP_PRIVATE mapping. Only the process that called mmap() * is guaranteed to have their future faults succeed. * * With the exception of reset_vma_resv_huge_pages() which is called at fork(), * the reserve counters are updated with the hugetlb_lock held. It is safe * to reset the VMA at fork() time as it is not in use yet and there is no * chance of the global counters getting corrupted as a result of the values. |
84afd99b8
|
273 274 275 276 277 278 279 280 281 |
* * The private mapping reservation is represented in a subtly different * manner to a shared mapping. A shared mapping has a region map associated * with the underlying file, this region map represents the backing file * pages which have ever had a reservation assigned which this persists even * after the page is instantiated. A private mapping has a region map * associated with the original mmap which is attached to all VMAs which * reference it, this region map represents those offsets which have consumed * reservation ie. where pages have been instantiated. |
a1e78772d
|
282 |
*/ |
e7c4b0bfd
|
283 284 285 286 287 288 289 290 291 292 |
static unsigned long get_vma_private_data(struct vm_area_struct *vma) { return (unsigned long)vma->vm_private_data; } static void set_vma_private_data(struct vm_area_struct *vma, unsigned long value) { vma->vm_private_data = (void *)value; } |
84afd99b8
|
293 294 295 296 |
struct resv_map { struct kref refs; struct list_head regions; }; |
2a4b3ded5
|
297 |
static struct resv_map *resv_map_alloc(void) |
84afd99b8
|
298 299 300 301 302 303 304 305 306 307 |
{ struct resv_map *resv_map = kmalloc(sizeof(*resv_map), GFP_KERNEL); if (!resv_map) return NULL; kref_init(&resv_map->refs); INIT_LIST_HEAD(&resv_map->regions); return resv_map; } |
2a4b3ded5
|
308 |
static void resv_map_release(struct kref *ref) |
84afd99b8
|
309 310 311 312 313 314 315 316 317 |
{ struct resv_map *resv_map = container_of(ref, struct resv_map, refs); /* Clear out any active regions before we release the map. */ region_truncate(&resv_map->regions, 0); kfree(resv_map); } static struct resv_map *vma_resv_map(struct vm_area_struct *vma) |
a1e78772d
|
318 319 |
{ VM_BUG_ON(!is_vm_hugetlb_page(vma)); |
f83a275db
|
320 |
if (!(vma->vm_flags & VM_MAYSHARE)) |
84afd99b8
|
321 322 |
return (struct resv_map *)(get_vma_private_data(vma) & ~HPAGE_RESV_MASK); |
2a4b3ded5
|
323 |
return NULL; |
a1e78772d
|
324 |
} |
84afd99b8
|
325 |
static void set_vma_resv_map(struct vm_area_struct *vma, struct resv_map *map) |
a1e78772d
|
326 327 |
{ VM_BUG_ON(!is_vm_hugetlb_page(vma)); |
f83a275db
|
328 |
VM_BUG_ON(vma->vm_flags & VM_MAYSHARE); |
a1e78772d
|
329 |
|
84afd99b8
|
330 331 |
set_vma_private_data(vma, (get_vma_private_data(vma) & HPAGE_RESV_MASK) | (unsigned long)map); |
04f2cbe35
|
332 333 334 335 |
} static void set_vma_resv_flags(struct vm_area_struct *vma, unsigned long flags) { |
04f2cbe35
|
336 |
VM_BUG_ON(!is_vm_hugetlb_page(vma)); |
f83a275db
|
337 |
VM_BUG_ON(vma->vm_flags & VM_MAYSHARE); |
e7c4b0bfd
|
338 339 |
set_vma_private_data(vma, get_vma_private_data(vma) | flags); |
04f2cbe35
|
340 341 342 343 344 |
} static int is_vma_resv_set(struct vm_area_struct *vma, unsigned long flag) { VM_BUG_ON(!is_vm_hugetlb_page(vma)); |
e7c4b0bfd
|
345 346 |
return (get_vma_private_data(vma) & flag) != 0; |
a1e78772d
|
347 348 349 |
} /* Decrement the reserved pages in the hugepage pool by one */ |
a55164389
|
350 351 |
static void decrement_hugepage_resv_vma(struct hstate *h, struct vm_area_struct *vma) |
a1e78772d
|
352 |
{ |
c37f9fb11
|
353 354 |
if (vma->vm_flags & VM_NORESERVE) return; |
f83a275db
|
355 |
if (vma->vm_flags & VM_MAYSHARE) { |
a1e78772d
|
356 |
/* Shared mappings always use reserves */ |
a55164389
|
357 |
h->resv_huge_pages--; |
84afd99b8
|
358 |
} else if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) { |
a1e78772d
|
359 360 361 362 |
/* * Only the process that called mmap() has reserves for * private mappings. */ |
a55164389
|
363 |
h->resv_huge_pages--; |
a1e78772d
|
364 365 |
} } |
04f2cbe35
|
366 |
/* Reset counters to 0 and clear all HPAGE_RESV_* flags */ |
a1e78772d
|
367 368 369 |
void reset_vma_resv_huge_pages(struct vm_area_struct *vma) { VM_BUG_ON(!is_vm_hugetlb_page(vma)); |
f83a275db
|
370 |
if (!(vma->vm_flags & VM_MAYSHARE)) |
a1e78772d
|
371 372 373 374 |
vma->vm_private_data = (void *)0; } /* Returns true if the VMA has associated reserve pages */ |
7f09ca51e
|
375 |
static int vma_has_reserves(struct vm_area_struct *vma) |
a1e78772d
|
376 |
{ |
f83a275db
|
377 |
if (vma->vm_flags & VM_MAYSHARE) |
7f09ca51e
|
378 379 380 381 |
return 1; if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) return 1; return 0; |
a1e78772d
|
382 |
} |
0ebabb416
|
383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 |
static void copy_gigantic_page(struct page *dst, struct page *src) { int i; struct hstate *h = page_hstate(src); struct page *dst_base = dst; struct page *src_base = src; for (i = 0; i < pages_per_huge_page(h); ) { cond_resched(); copy_highpage(dst, src); i++; dst = mem_map_next(dst, dst_base, i); src = mem_map_next(src, src_base, i); } } void copy_huge_page(struct page *dst, struct page *src) { int i; struct hstate *h = page_hstate(src); if (unlikely(pages_per_huge_page(h) > MAX_ORDER_NR_PAGES)) { copy_gigantic_page(dst, src); return; } might_sleep(); for (i = 0; i < pages_per_huge_page(h); i++) { cond_resched(); copy_highpage(dst + i, src + i); } } |
a55164389
|
416 |
static void enqueue_huge_page(struct hstate *h, struct page *page) |
1da177e4c
|
417 418 |
{ int nid = page_to_nid(page); |
a55164389
|
419 420 421 |
list_add(&page->lru, &h->hugepage_freelists[nid]); h->free_huge_pages++; h->free_huge_pages_node[nid]++; |
1da177e4c
|
422 |
} |
bf50bab2b
|
423 424 425 426 427 428 429 430 |
static struct page *dequeue_huge_page_node(struct hstate *h, int nid) { struct page *page; if (list_empty(&h->hugepage_freelists[nid])) return NULL; page = list_entry(h->hugepage_freelists[nid].next, struct page, lru); list_del(&page->lru); |
a9869b837
|
431 |
set_page_refcounted(page); |
bf50bab2b
|
432 433 434 435 |
h->free_huge_pages--; h->free_huge_pages_node[nid]--; return page; } |
a55164389
|
436 437 |
static struct page *dequeue_huge_page_vma(struct hstate *h, struct vm_area_struct *vma, |
04f2cbe35
|
438 |
unsigned long address, int avoid_reserve) |
1da177e4c
|
439 |
{ |
1da177e4c
|
440 |
struct page *page = NULL; |
480eccf9a
|
441 |
struct mempolicy *mpol; |
19770b326
|
442 |
nodemask_t *nodemask; |
c0ff7453b
|
443 |
struct zonelist *zonelist; |
dd1a239f6
|
444 445 |
struct zone *zone; struct zoneref *z; |
1da177e4c
|
446 |
|
c0ff7453b
|
447 448 449 |
get_mems_allowed(); zonelist = huge_zonelist(vma, address, htlb_alloc_mask, &mpol, &nodemask); |
a1e78772d
|
450 451 452 453 454 |
/* * A child process with MAP_PRIVATE mappings created by their parent * have no page reserves. This check ensures that reservations are * not "stolen". The child may still get SIGKILLed */ |
7f09ca51e
|
455 |
if (!vma_has_reserves(vma) && |
a55164389
|
456 |
h->free_huge_pages - h->resv_huge_pages == 0) |
c0ff7453b
|
457 |
goto err; |
a1e78772d
|
458 |
|
04f2cbe35
|
459 |
/* If reserves cannot be used, ensure enough pages are in the pool */ |
a55164389
|
460 |
if (avoid_reserve && h->free_huge_pages - h->resv_huge_pages == 0) |
6eab04a87
|
461 |
goto err; |
04f2cbe35
|
462 |
|
19770b326
|
463 464 |
for_each_zone_zonelist_nodemask(zone, z, zonelist, MAX_NR_ZONES - 1, nodemask) { |
bf50bab2b
|
465 466 467 468 469 470 471 |
if (cpuset_zone_allowed_softwall(zone, htlb_alloc_mask)) { page = dequeue_huge_page_node(h, zone_to_nid(zone)); if (page) { if (!avoid_reserve) decrement_hugepage_resv_vma(h, vma); break; } |
3abf7afd4
|
472 |
} |
1da177e4c
|
473 |
} |
c0ff7453b
|
474 |
err: |
52cd3b074
|
475 |
mpol_cond_put(mpol); |
c0ff7453b
|
476 |
put_mems_allowed(); |
1da177e4c
|
477 478 |
return page; } |
a55164389
|
479 |
static void update_and_free_page(struct hstate *h, struct page *page) |
6af2acb66
|
480 481 |
{ int i; |
a55164389
|
482 |
|
18229df5b
|
483 |
VM_BUG_ON(h->order >= MAX_ORDER); |
a55164389
|
484 485 486 |
h->nr_huge_pages--; h->nr_huge_pages_node[page_to_nid(page)]--; for (i = 0; i < pages_per_huge_page(h); i++) { |
6af2acb66
|
487 488 489 490 491 492 |
page[i].flags &= ~(1 << PG_locked | 1 << PG_error | 1 << PG_referenced | 1 << PG_dirty | 1 << PG_active | 1 << PG_reserved | 1 << PG_private | 1<< PG_writeback); } set_compound_page_dtor(page, NULL); set_page_refcounted(page); |
7f2e9525b
|
493 |
arch_release_hugepage(page); |
a55164389
|
494 |
__free_pages(page, huge_page_order(h)); |
6af2acb66
|
495 |
} |
e5ff21594
|
496 497 498 499 500 501 502 503 504 505 |
struct hstate *size_to_hstate(unsigned long size) { struct hstate *h; for_each_hstate(h) { if (huge_page_size(h) == size) return h; } return NULL; } |
27a85ef1b
|
506 507 |
static void free_huge_page(struct page *page) { |
a55164389
|
508 509 510 511 |
/* * Can't pass hstate in here because it is called from the * compound page destructor. */ |
e5ff21594
|
512 |
struct hstate *h = page_hstate(page); |
7893d1d50
|
513 |
int nid = page_to_nid(page); |
c79fb75e5
|
514 |
struct address_space *mapping; |
27a85ef1b
|
515 |
|
c79fb75e5
|
516 |
mapping = (struct address_space *) page_private(page); |
e5df70ab1
|
517 |
set_page_private(page, 0); |
23be7468e
|
518 |
page->mapping = NULL; |
7893d1d50
|
519 |
BUG_ON(page_count(page)); |
0fe6e20b9
|
520 |
BUG_ON(page_mapcount(page)); |
27a85ef1b
|
521 522 523 |
INIT_LIST_HEAD(&page->lru); spin_lock(&hugetlb_lock); |
aa888a749
|
524 |
if (h->surplus_huge_pages_node[nid] && huge_page_order(h) < MAX_ORDER) { |
a55164389
|
525 526 527 |
update_and_free_page(h, page); h->surplus_huge_pages--; h->surplus_huge_pages_node[nid]--; |
7893d1d50
|
528 |
} else { |
a55164389
|
529 |
enqueue_huge_page(h, page); |
7893d1d50
|
530 |
} |
27a85ef1b
|
531 |
spin_unlock(&hugetlb_lock); |
c79fb75e5
|
532 |
if (mapping) |
9a119c056
|
533 |
hugetlb_put_quota(mapping, 1); |
27a85ef1b
|
534 |
} |
a55164389
|
535 |
static void prep_new_huge_page(struct hstate *h, struct page *page, int nid) |
b7ba30c67
|
536 537 538 |
{ set_compound_page_dtor(page, free_huge_page); spin_lock(&hugetlb_lock); |
a55164389
|
539 540 |
h->nr_huge_pages++; h->nr_huge_pages_node[nid]++; |
b7ba30c67
|
541 542 543 |
spin_unlock(&hugetlb_lock); put_page(page); /* free it into the hugepage allocator */ } |
20a0307c0
|
544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 |
static void prep_compound_gigantic_page(struct page *page, unsigned long order) { int i; int nr_pages = 1 << order; struct page *p = page + 1; /* we rely on prep_new_huge_page to set the destructor */ set_compound_order(page, order); __SetPageHead(page); for (i = 1; i < nr_pages; i++, p = mem_map_next(p, page, i)) { __SetPageTail(p); p->first_page = page; } } int PageHuge(struct page *page) { compound_page_dtor *dtor; if (!PageCompound(page)) return 0; page = compound_head(page); dtor = get_compound_page_dtor(page); return dtor == free_huge_page; } |
43131e141
|
571 |
EXPORT_SYMBOL_GPL(PageHuge); |
a55164389
|
572 |
static struct page *alloc_fresh_huge_page_node(struct hstate *h, int nid) |
1da177e4c
|
573 |
{ |
1da177e4c
|
574 |
struct page *page; |
f96efd585
|
575 |
|
aa888a749
|
576 577 |
if (h->order >= MAX_ORDER) return NULL; |
6484eb3e2
|
578 |
page = alloc_pages_exact_node(nid, |
551883ae8
|
579 580 |
htlb_alloc_mask|__GFP_COMP|__GFP_THISNODE| __GFP_REPEAT|__GFP_NOWARN, |
a55164389
|
581 |
huge_page_order(h)); |
1da177e4c
|
582 |
if (page) { |
7f2e9525b
|
583 |
if (arch_prepare_hugepage(page)) { |
caff3a2c3
|
584 |
__free_pages(page, huge_page_order(h)); |
7b8ee84d8
|
585 |
return NULL; |
7f2e9525b
|
586 |
} |
a55164389
|
587 |
prep_new_huge_page(h, page, nid); |
1da177e4c
|
588 |
} |
63b4613c3
|
589 590 591 |
return page; } |
5ced66c90
|
592 |
/* |
6ae11b278
|
593 594 595 596 597 |
* common helper functions for hstate_next_node_to_{alloc|free}. * We may have allocated or freed a huge page based on a different * nodes_allowed previously, so h->next_node_to_{alloc|free} might * be outside of *nodes_allowed. Ensure that we use an allowed * node for alloc or free. |
9a76db099
|
598 |
*/ |
6ae11b278
|
599 |
static int next_node_allowed(int nid, nodemask_t *nodes_allowed) |
9a76db099
|
600 |
{ |
6ae11b278
|
601 |
nid = next_node(nid, *nodes_allowed); |
9a76db099
|
602 |
if (nid == MAX_NUMNODES) |
6ae11b278
|
603 |
nid = first_node(*nodes_allowed); |
9a76db099
|
604 605 606 607 |
VM_BUG_ON(nid >= MAX_NUMNODES); return nid; } |
6ae11b278
|
608 609 610 611 612 613 |
static int get_valid_node_allowed(int nid, nodemask_t *nodes_allowed) { if (!node_isset(nid, *nodes_allowed)) nid = next_node_allowed(nid, nodes_allowed); return nid; } |
9a76db099
|
614 |
/* |
6ae11b278
|
615 616 617 618 |
* returns the previously saved node ["this node"] from which to * allocate a persistent huge page for the pool and advance the * next node from which to allocate, handling wrap at end of node * mask. |
5ced66c90
|
619 |
*/ |
6ae11b278
|
620 621 |
static int hstate_next_node_to_alloc(struct hstate *h, nodemask_t *nodes_allowed) |
5ced66c90
|
622 |
{ |
6ae11b278
|
623 624 625 626 627 628 |
int nid; VM_BUG_ON(!nodes_allowed); nid = get_valid_node_allowed(h->next_nid_to_alloc, nodes_allowed); h->next_nid_to_alloc = next_node_allowed(nid, nodes_allowed); |
9a76db099
|
629 |
|
9a76db099
|
630 |
return nid; |
5ced66c90
|
631 |
} |
6ae11b278
|
632 |
static int alloc_fresh_huge_page(struct hstate *h, nodemask_t *nodes_allowed) |
63b4613c3
|
633 634 635 636 637 |
{ struct page *page; int start_nid; int next_nid; int ret = 0; |
6ae11b278
|
638 |
start_nid = hstate_next_node_to_alloc(h, nodes_allowed); |
e8c5c8249
|
639 |
next_nid = start_nid; |
63b4613c3
|
640 641 |
do { |
e8c5c8249
|
642 |
page = alloc_fresh_huge_page_node(h, next_nid); |
9a76db099
|
643 |
if (page) { |
63b4613c3
|
644 |
ret = 1; |
9a76db099
|
645 646 |
break; } |
6ae11b278
|
647 |
next_nid = hstate_next_node_to_alloc(h, nodes_allowed); |
9a76db099
|
648 |
} while (next_nid != start_nid); |
63b4613c3
|
649 |
|
3b1163006
|
650 651 652 653 |
if (ret) count_vm_event(HTLB_BUDDY_PGALLOC); else count_vm_event(HTLB_BUDDY_PGALLOC_FAIL); |
63b4613c3
|
654 |
return ret; |
1da177e4c
|
655 |
} |
e8c5c8249
|
656 |
/* |
6ae11b278
|
657 658 659 660 |
* helper for free_pool_huge_page() - return the previously saved * node ["this node"] from which to free a huge page. Advance the * next node id whether or not we find a free huge page to free so * that the next attempt to free addresses the next node. |
e8c5c8249
|
661 |
*/ |
6ae11b278
|
662 |
static int hstate_next_node_to_free(struct hstate *h, nodemask_t *nodes_allowed) |
e8c5c8249
|
663 |
{ |
6ae11b278
|
664 665 666 667 668 669 |
int nid; VM_BUG_ON(!nodes_allowed); nid = get_valid_node_allowed(h->next_nid_to_free, nodes_allowed); h->next_nid_to_free = next_node_allowed(nid, nodes_allowed); |
9a76db099
|
670 |
|
9a76db099
|
671 |
return nid; |
e8c5c8249
|
672 673 674 675 676 677 678 679 |
} /* * Free huge page from pool from next node to free. * Attempt to keep persistent huge pages more or less * balanced over allowed nodes. * Called with hugetlb_lock locked. */ |
6ae11b278
|
680 681 |
static int free_pool_huge_page(struct hstate *h, nodemask_t *nodes_allowed, bool acct_surplus) |
e8c5c8249
|
682 683 684 685 |
{ int start_nid; int next_nid; int ret = 0; |
6ae11b278
|
686 |
start_nid = hstate_next_node_to_free(h, nodes_allowed); |
e8c5c8249
|
687 688 689 |
next_nid = start_nid; do { |
685f34570
|
690 691 692 693 694 695 |
/* * If we're returning unused surplus pages, only examine * nodes with surplus pages. */ if ((!acct_surplus || h->surplus_huge_pages_node[next_nid]) && !list_empty(&h->hugepage_freelists[next_nid])) { |
e8c5c8249
|
696 697 698 699 700 701 |
struct page *page = list_entry(h->hugepage_freelists[next_nid].next, struct page, lru); list_del(&page->lru); h->free_huge_pages--; h->free_huge_pages_node[next_nid]--; |
685f34570
|
702 703 704 705 |
if (acct_surplus) { h->surplus_huge_pages--; h->surplus_huge_pages_node[next_nid]--; } |
e8c5c8249
|
706 707 |
update_and_free_page(h, page); ret = 1; |
9a76db099
|
708 |
break; |
e8c5c8249
|
709 |
} |
6ae11b278
|
710 |
next_nid = hstate_next_node_to_free(h, nodes_allowed); |
9a76db099
|
711 |
} while (next_nid != start_nid); |
e8c5c8249
|
712 713 714 |
return ret; } |
bf50bab2b
|
715 |
static struct page *alloc_buddy_huge_page(struct hstate *h, int nid) |
7893d1d50
|
716 717 |
{ struct page *page; |
bf50bab2b
|
718 |
unsigned int r_nid; |
7893d1d50
|
719 |
|
aa888a749
|
720 721 |
if (h->order >= MAX_ORDER) return NULL; |
d1c3fb1f8
|
722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 |
/* * Assume we will successfully allocate the surplus page to * prevent racing processes from causing the surplus to exceed * overcommit * * This however introduces a different race, where a process B * tries to grow the static hugepage pool while alloc_pages() is * called by process A. B will only examine the per-node * counters in determining if surplus huge pages can be * converted to normal huge pages in adjust_pool_surplus(). A * won't be able to increment the per-node counter, until the * lock is dropped by B, but B doesn't drop hugetlb_lock until * no more huge pages can be converted from surplus to normal * state (and doesn't try to convert again). Thus, we have a * case where a surplus huge page exists, the pool is grown, and * the surplus huge page still exists after, even though it * should just have been converted to a normal huge page. This * does not leak memory, though, as the hugepage will be freed * once it is out of use. It also does not allow the counters to * go out of whack in adjust_pool_surplus() as we don't modify * the node values until we've gotten the hugepage and only the * per-node value is checked there. */ spin_lock(&hugetlb_lock); |
a55164389
|
746 |
if (h->surplus_huge_pages >= h->nr_overcommit_huge_pages) { |
d1c3fb1f8
|
747 748 749 |
spin_unlock(&hugetlb_lock); return NULL; } else { |
a55164389
|
750 751 |
h->nr_huge_pages++; h->surplus_huge_pages++; |
d1c3fb1f8
|
752 753 |
} spin_unlock(&hugetlb_lock); |
bf50bab2b
|
754 755 756 757 758 759 760 761 |
if (nid == NUMA_NO_NODE) page = alloc_pages(htlb_alloc_mask|__GFP_COMP| __GFP_REPEAT|__GFP_NOWARN, huge_page_order(h)); else page = alloc_pages_exact_node(nid, htlb_alloc_mask|__GFP_COMP|__GFP_THISNODE| __GFP_REPEAT|__GFP_NOWARN, huge_page_order(h)); |
d1c3fb1f8
|
762 |
|
caff3a2c3
|
763 764 765 766 |
if (page && arch_prepare_hugepage(page)) { __free_pages(page, huge_page_order(h)); return NULL; } |
d1c3fb1f8
|
767 |
spin_lock(&hugetlb_lock); |
7893d1d50
|
768 |
if (page) { |
bf50bab2b
|
769 |
r_nid = page_to_nid(page); |
7893d1d50
|
770 |
set_compound_page_dtor(page, free_huge_page); |
d1c3fb1f8
|
771 772 773 |
/* * We incremented the global counters already */ |
bf50bab2b
|
774 775 |
h->nr_huge_pages_node[r_nid]++; h->surplus_huge_pages_node[r_nid]++; |
3b1163006
|
776 |
__count_vm_event(HTLB_BUDDY_PGALLOC); |
d1c3fb1f8
|
777 |
} else { |
a55164389
|
778 779 |
h->nr_huge_pages--; h->surplus_huge_pages--; |
3b1163006
|
780 |
__count_vm_event(HTLB_BUDDY_PGALLOC_FAIL); |
7893d1d50
|
781 |
} |
d1c3fb1f8
|
782 |
spin_unlock(&hugetlb_lock); |
7893d1d50
|
783 784 785 |
return page; } |
e4e574b76
|
786 |
/* |
bf50bab2b
|
787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 |
* This allocation function is useful in the context where vma is irrelevant. * E.g. soft-offlining uses this function because it only cares physical * address of error page. */ struct page *alloc_huge_page_node(struct hstate *h, int nid) { struct page *page; spin_lock(&hugetlb_lock); page = dequeue_huge_page_node(h, nid); spin_unlock(&hugetlb_lock); if (!page) page = alloc_buddy_huge_page(h, nid); return page; } /* |
25985edce
|
806 |
* Increase the hugetlb pool such that it can accommodate a reservation |
e4e574b76
|
807 808 |
* of size 'delta'. */ |
a55164389
|
809 |
static int gather_surplus_pages(struct hstate *h, int delta) |
e4e574b76
|
810 811 812 813 814 |
{ struct list_head surplus_list; struct page *page, *tmp; int ret, i; int needed, allocated; |
a55164389
|
815 |
needed = (h->resv_huge_pages + delta) - h->free_huge_pages; |
ac09b3a15
|
816 |
if (needed <= 0) { |
a55164389
|
817 |
h->resv_huge_pages += delta; |
e4e574b76
|
818 |
return 0; |
ac09b3a15
|
819 |
} |
e4e574b76
|
820 821 822 823 824 825 826 827 |
allocated = 0; INIT_LIST_HEAD(&surplus_list); ret = -ENOMEM; retry: spin_unlock(&hugetlb_lock); for (i = 0; i < needed; i++) { |
bf50bab2b
|
828 |
page = alloc_buddy_huge_page(h, NUMA_NO_NODE); |
a9869b837
|
829 |
if (!page) |
e4e574b76
|
830 831 832 833 834 |
/* * We were not able to allocate enough pages to * satisfy the entire reservation so we free what * we've allocated so far. */ |
e4e574b76
|
835 |
goto free; |
e4e574b76
|
836 837 838 839 840 841 842 843 844 845 |
list_add(&page->lru, &surplus_list); } allocated += needed; /* * After retaking hugetlb_lock, we need to recalculate 'needed' * because either resv_huge_pages or free_huge_pages may have changed. */ spin_lock(&hugetlb_lock); |
a55164389
|
846 847 |
needed = (h->resv_huge_pages + delta) - (h->free_huge_pages + allocated); |
e4e574b76
|
848 849 850 851 852 |
if (needed > 0) goto retry; /* * The surplus_list now contains _at_least_ the number of extra pages |
25985edce
|
853 |
* needed to accommodate the reservation. Add the appropriate number |
e4e574b76
|
854 |
* of pages to the hugetlb pool and free the extras back to the buddy |
ac09b3a15
|
855 856 857 |
* allocator. Commit the entire reservation here to prevent another * process from stealing the pages as they are added to the pool but * before they are reserved. |
e4e574b76
|
858 859 |
*/ needed += allocated; |
a55164389
|
860 |
h->resv_huge_pages += delta; |
e4e574b76
|
861 |
ret = 0; |
a9869b837
|
862 863 |
spin_unlock(&hugetlb_lock); |
19fc3f0ac
|
864 |
/* Free the needed pages to the hugetlb pool */ |
e4e574b76
|
865 |
list_for_each_entry_safe(page, tmp, &surplus_list, lru) { |
19fc3f0ac
|
866 867 |
if ((--needed) < 0) break; |
e4e574b76
|
868 |
list_del(&page->lru); |
a9869b837
|
869 870 871 872 873 874 |
/* * This page is now managed by the hugetlb allocator and has * no users -- drop the buddy allocator's reference. */ put_page_testzero(page); VM_BUG_ON(page_count(page)); |
a55164389
|
875 |
enqueue_huge_page(h, page); |
19fc3f0ac
|
876 877 878 |
} /* Free unnecessary surplus pages to the buddy allocator */ |
a9869b837
|
879 |
free: |
19fc3f0ac
|
880 |
if (!list_empty(&surplus_list)) { |
19fc3f0ac
|
881 882 |
list_for_each_entry_safe(page, tmp, &surplus_list, lru) { list_del(&page->lru); |
a9869b837
|
883 |
put_page(page); |
af767cbdd
|
884 |
} |
e4e574b76
|
885 |
} |
a9869b837
|
886 |
spin_lock(&hugetlb_lock); |
e4e574b76
|
887 888 889 890 891 892 893 894 |
return ret; } /* * When releasing a hugetlb pool reservation, any surplus pages that were * allocated to satisfy the reservation must be explicitly freed if they were * never used. |
685f34570
|
895 |
* Called with hugetlb_lock held. |
e4e574b76
|
896 |
*/ |
a55164389
|
897 898 |
static void return_unused_surplus_pages(struct hstate *h, unsigned long unused_resv_pages) |
e4e574b76
|
899 |
{ |
e4e574b76
|
900 |
unsigned long nr_pages; |
ac09b3a15
|
901 |
/* Uncommit the reservation */ |
a55164389
|
902 |
h->resv_huge_pages -= unused_resv_pages; |
ac09b3a15
|
903 |
|
aa888a749
|
904 905 906 |
/* Cannot return gigantic pages currently */ if (h->order >= MAX_ORDER) return; |
a55164389
|
907 |
nr_pages = min(unused_resv_pages, h->surplus_huge_pages); |
e4e574b76
|
908 |
|
685f34570
|
909 910 |
/* * We want to release as many surplus pages as possible, spread |
9b5e5d0fd
|
911 912 913 914 915 |
* evenly across all nodes with memory. Iterate across these nodes * until we can no longer free unreserved surplus pages. This occurs * when the nodes with surplus pages have no free pages. * free_pool_huge_page() will balance the the freed pages across the * on-line nodes with memory and will handle the hstate accounting. |
685f34570
|
916 917 |
*/ while (nr_pages--) { |
9b5e5d0fd
|
918 |
if (!free_pool_huge_page(h, &node_states[N_HIGH_MEMORY], 1)) |
685f34570
|
919 |
break; |
e4e574b76
|
920 921 |
} } |
c37f9fb11
|
922 923 924 925 926 927 928 929 930 |
/* * Determine if the huge page at addr within the vma has an associated * reservation. Where it does not we will need to logically increase * reservation and actually increase quota before an allocation can occur. * Where any new reservation would be required the reservation change is * prepared, but not committed. Once the page has been quota'd allocated * an instantiated the change should be committed via vma_commit_reservation. * No action is required on failure. */ |
e2f17d945
|
931 |
static long vma_needs_reservation(struct hstate *h, |
a55164389
|
932 |
struct vm_area_struct *vma, unsigned long addr) |
c37f9fb11
|
933 934 935 |
{ struct address_space *mapping = vma->vm_file->f_mapping; struct inode *inode = mapping->host; |
f83a275db
|
936 |
if (vma->vm_flags & VM_MAYSHARE) { |
a55164389
|
937 |
pgoff_t idx = vma_hugecache_offset(h, vma, addr); |
c37f9fb11
|
938 939 |
return region_chg(&inode->i_mapping->private_list, idx, idx + 1); |
84afd99b8
|
940 941 |
} else if (!is_vma_resv_set(vma, HPAGE_RESV_OWNER)) { return 1; |
c37f9fb11
|
942 |
|
84afd99b8
|
943 |
} else { |
e2f17d945
|
944 |
long err; |
a55164389
|
945 |
pgoff_t idx = vma_hugecache_offset(h, vma, addr); |
84afd99b8
|
946 947 948 949 950 951 952 |
struct resv_map *reservations = vma_resv_map(vma); err = region_chg(&reservations->regions, idx, idx + 1); if (err < 0) return err; return 0; } |
c37f9fb11
|
953 |
} |
a55164389
|
954 955 |
static void vma_commit_reservation(struct hstate *h, struct vm_area_struct *vma, unsigned long addr) |
c37f9fb11
|
956 957 958 |
{ struct address_space *mapping = vma->vm_file->f_mapping; struct inode *inode = mapping->host; |
f83a275db
|
959 |
if (vma->vm_flags & VM_MAYSHARE) { |
a55164389
|
960 |
pgoff_t idx = vma_hugecache_offset(h, vma, addr); |
c37f9fb11
|
961 |
region_add(&inode->i_mapping->private_list, idx, idx + 1); |
84afd99b8
|
962 963 |
} else if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) { |
a55164389
|
964 |
pgoff_t idx = vma_hugecache_offset(h, vma, addr); |
84afd99b8
|
965 966 967 968 |
struct resv_map *reservations = vma_resv_map(vma); /* Mark this page used in the map. */ region_add(&reservations->regions, idx, idx + 1); |
c37f9fb11
|
969 970 |
} } |
a1e78772d
|
971 |
static struct page *alloc_huge_page(struct vm_area_struct *vma, |
04f2cbe35
|
972 |
unsigned long addr, int avoid_reserve) |
1da177e4c
|
973 |
{ |
a55164389
|
974 |
struct hstate *h = hstate_vma(vma); |
348ea204c
|
975 |
struct page *page; |
a1e78772d
|
976 977 |
struct address_space *mapping = vma->vm_file->f_mapping; struct inode *inode = mapping->host; |
e2f17d945
|
978 |
long chg; |
a1e78772d
|
979 980 981 982 983 |
/* * Processes that did not create the mapping will have no reserves and * will not have accounted against quota. Check that the quota can be * made before satisfying the allocation |
c37f9fb11
|
984 985 |
* MAP_NORESERVE mappings may also need pages and quota allocated * if no reserve mapping overlaps. |
a1e78772d
|
986 |
*/ |
a55164389
|
987 |
chg = vma_needs_reservation(h, vma, addr); |
c37f9fb11
|
988 |
if (chg < 0) |
e0dcd8a05
|
989 |
return ERR_PTR(-VM_FAULT_OOM); |
c37f9fb11
|
990 |
if (chg) |
a1e78772d
|
991 |
if (hugetlb_get_quota(inode->i_mapping, chg)) |
e0dcd8a05
|
992 |
return ERR_PTR(-VM_FAULT_SIGBUS); |
1da177e4c
|
993 994 |
spin_lock(&hugetlb_lock); |
a55164389
|
995 |
page = dequeue_huge_page_vma(h, vma, addr, avoid_reserve); |
1da177e4c
|
996 |
spin_unlock(&hugetlb_lock); |
b45b5bd65
|
997 |
|
68842c9b9
|
998 |
if (!page) { |
bf50bab2b
|
999 |
page = alloc_buddy_huge_page(h, NUMA_NO_NODE); |
68842c9b9
|
1000 |
if (!page) { |
a1e78772d
|
1001 |
hugetlb_put_quota(inode->i_mapping, chg); |
4a6018f7f
|
1002 |
return ERR_PTR(-VM_FAULT_SIGBUS); |
68842c9b9
|
1003 1004 |
} } |
348ea204c
|
1005 |
|
a1e78772d
|
1006 |
set_page_private(page, (unsigned long) mapping); |
90d8b7e61
|
1007 |
|
a55164389
|
1008 |
vma_commit_reservation(h, vma, addr); |
c37f9fb11
|
1009 |
|
90d8b7e61
|
1010 |
return page; |
b45b5bd65
|
1011 |
} |
91f47662d
|
1012 |
int __weak alloc_bootmem_huge_page(struct hstate *h) |
aa888a749
|
1013 1014 |
{ struct huge_bootmem_page *m; |
9b5e5d0fd
|
1015 |
int nr_nodes = nodes_weight(node_states[N_HIGH_MEMORY]); |
aa888a749
|
1016 1017 1018 1019 1020 |
while (nr_nodes) { void *addr; addr = __alloc_bootmem_node_nopanic( |
6ae11b278
|
1021 |
NODE_DATA(hstate_next_node_to_alloc(h, |
9b5e5d0fd
|
1022 |
&node_states[N_HIGH_MEMORY])), |
aa888a749
|
1023 1024 1025 1026 1027 1028 1029 1030 1031 |
huge_page_size(h), huge_page_size(h), 0); if (addr) { /* * Use the beginning of the huge page to store the * huge_bootmem_page struct (until gather_bootmem * puts them into the mem_map). */ m = addr; |
91f47662d
|
1032 |
goto found; |
aa888a749
|
1033 |
} |
aa888a749
|
1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 |
nr_nodes--; } return 0; found: BUG_ON((unsigned long)virt_to_phys(m) & (huge_page_size(h) - 1)); /* Put them into a private list first because mem_map is not up yet */ list_add(&m->list, &huge_boot_pages); m->hstate = h; return 1; } |
18229df5b
|
1045 1046 1047 1048 1049 1050 1051 |
static void prep_compound_huge_page(struct page *page, int order) { if (unlikely(order > (MAX_ORDER - 1))) prep_compound_gigantic_page(page, order); else prep_compound_page(page, order); } |
aa888a749
|
1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 |
/* Put bootmem huge pages into the standard lists after mem_map is up */ static void __init gather_bootmem_prealloc(void) { struct huge_bootmem_page *m; list_for_each_entry(m, &huge_boot_pages, list) { struct page *page = virt_to_page(m); struct hstate *h = m->hstate; __ClearPageReserved(page); WARN_ON(page_count(page) != 1); |
18229df5b
|
1062 |
prep_compound_huge_page(page, h->order); |
aa888a749
|
1063 |
prep_new_huge_page(h, page, page_to_nid(page)); |
b0320c7b7
|
1064 1065 1066 1067 1068 1069 1070 1071 |
/* * If we had gigantic hugepages allocated at boot time, we need * to restore the 'stolen' pages to totalram_pages in order to * fix confusing memory reports from free(1) and another * side-effects, like CommitLimit going negative. */ if (h->order > (MAX_ORDER - 1)) totalram_pages += 1 << h->order; |
aa888a749
|
1072 1073 |
} } |
8faa8b077
|
1074 |
static void __init hugetlb_hstate_alloc_pages(struct hstate *h) |
1da177e4c
|
1075 1076 |
{ unsigned long i; |
a55164389
|
1077 |
|
e5ff21594
|
1078 |
for (i = 0; i < h->max_huge_pages; ++i) { |
aa888a749
|
1079 1080 1081 |
if (h->order >= MAX_ORDER) { if (!alloc_bootmem_huge_page(h)) break; |
9b5e5d0fd
|
1082 1083 |
} else if (!alloc_fresh_huge_page(h, &node_states[N_HIGH_MEMORY])) |
1da177e4c
|
1084 |
break; |
1da177e4c
|
1085 |
} |
8faa8b077
|
1086 |
h->max_huge_pages = i; |
e5ff21594
|
1087 1088 1089 1090 1091 1092 1093 |
} static void __init hugetlb_init_hstates(void) { struct hstate *h; for_each_hstate(h) { |
8faa8b077
|
1094 1095 1096 |
/* oversize hugepages were init'ed in early boot */ if (h->order < MAX_ORDER) hugetlb_hstate_alloc_pages(h); |
e5ff21594
|
1097 1098 |
} } |
4abd32dba
|
1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 |
static char * __init memfmt(char *buf, unsigned long n) { if (n >= (1UL << 30)) sprintf(buf, "%lu GB", n >> 30); else if (n >= (1UL << 20)) sprintf(buf, "%lu MB", n >> 20); else sprintf(buf, "%lu KB", n >> 10); return buf; } |
e5ff21594
|
1109 1110 1111 1112 1113 |
static void __init report_hugepages(void) { struct hstate *h; for_each_hstate(h) { |
4abd32dba
|
1114 1115 1116 1117 1118 1119 |
char buf[32]; printk(KERN_INFO "HugeTLB registered %s page size, " "pre-allocated %ld pages ", memfmt(buf, huge_page_size(h)), h->free_huge_pages); |
e5ff21594
|
1120 1121 |
} } |
1da177e4c
|
1122 |
#ifdef CONFIG_HIGHMEM |
6ae11b278
|
1123 1124 |
static void try_to_free_low(struct hstate *h, unsigned long count, nodemask_t *nodes_allowed) |
1da177e4c
|
1125 |
{ |
4415cc8df
|
1126 |
int i; |
aa888a749
|
1127 1128 |
if (h->order >= MAX_ORDER) return; |
6ae11b278
|
1129 |
for_each_node_mask(i, *nodes_allowed) { |
1da177e4c
|
1130 |
struct page *page, *next; |
a55164389
|
1131 1132 1133 |
struct list_head *freel = &h->hugepage_freelists[i]; list_for_each_entry_safe(page, next, freel, lru) { if (count >= h->nr_huge_pages) |
6b0c880df
|
1134 |
return; |
1da177e4c
|
1135 1136 1137 |
if (PageHighMem(page)) continue; list_del(&page->lru); |
e5ff21594
|
1138 |
update_and_free_page(h, page); |
a55164389
|
1139 1140 |
h->free_huge_pages--; h->free_huge_pages_node[page_to_nid(page)]--; |
1da177e4c
|
1141 1142 1143 1144 |
} } } #else |
6ae11b278
|
1145 1146 |
static inline void try_to_free_low(struct hstate *h, unsigned long count, nodemask_t *nodes_allowed) |
1da177e4c
|
1147 1148 1149 |
{ } #endif |
20a0307c0
|
1150 1151 1152 1153 1154 |
/* * Increment or decrement surplus_huge_pages. Keep node-specific counters * balanced by operating on them in a round-robin fashion. * Returns 1 if an adjustment was made. */ |
6ae11b278
|
1155 1156 |
static int adjust_pool_surplus(struct hstate *h, nodemask_t *nodes_allowed, int delta) |
20a0307c0
|
1157 |
{ |
e8c5c8249
|
1158 |
int start_nid, next_nid; |
20a0307c0
|
1159 1160 1161 |
int ret = 0; VM_BUG_ON(delta != -1 && delta != 1); |
20a0307c0
|
1162 |
|
e8c5c8249
|
1163 |
if (delta < 0) |
6ae11b278
|
1164 |
start_nid = hstate_next_node_to_alloc(h, nodes_allowed); |
e8c5c8249
|
1165 |
else |
6ae11b278
|
1166 |
start_nid = hstate_next_node_to_free(h, nodes_allowed); |
e8c5c8249
|
1167 1168 1169 1170 1171 |
next_nid = start_nid; do { int nid = next_nid; if (delta < 0) { |
e8c5c8249
|
1172 1173 1174 |
/* * To shrink on this node, there must be a surplus page */ |
9a76db099
|
1175 |
if (!h->surplus_huge_pages_node[nid]) { |
6ae11b278
|
1176 1177 |
next_nid = hstate_next_node_to_alloc(h, nodes_allowed); |
e8c5c8249
|
1178 |
continue; |
9a76db099
|
1179 |
} |
e8c5c8249
|
1180 1181 |
} if (delta > 0) { |
e8c5c8249
|
1182 1183 1184 1185 |
/* * Surplus cannot exceed the total number of pages */ if (h->surplus_huge_pages_node[nid] >= |
9a76db099
|
1186 |
h->nr_huge_pages_node[nid]) { |
6ae11b278
|
1187 1188 |
next_nid = hstate_next_node_to_free(h, nodes_allowed); |
e8c5c8249
|
1189 |
continue; |
9a76db099
|
1190 |
} |
e8c5c8249
|
1191 |
} |
20a0307c0
|
1192 1193 1194 1195 1196 |
h->surplus_huge_pages += delta; h->surplus_huge_pages_node[nid] += delta; ret = 1; break; |
e8c5c8249
|
1197 |
} while (next_nid != start_nid); |
20a0307c0
|
1198 |
|
20a0307c0
|
1199 1200 |
return ret; } |
a55164389
|
1201 |
#define persistent_huge_pages(h) (h->nr_huge_pages - h->surplus_huge_pages) |
6ae11b278
|
1202 1203 |
static unsigned long set_max_huge_pages(struct hstate *h, unsigned long count, nodemask_t *nodes_allowed) |
1da177e4c
|
1204 |
{ |
7893d1d50
|
1205 |
unsigned long min_count, ret; |
1da177e4c
|
1206 |
|
aa888a749
|
1207 1208 |
if (h->order >= MAX_ORDER) return h->max_huge_pages; |
7893d1d50
|
1209 1210 1211 1212 |
/* * Increase the pool size * First take pages out of surplus state. Then make up the * remaining difference by allocating fresh huge pages. |
d1c3fb1f8
|
1213 1214 1215 1216 1217 1218 |
* * We might race with alloc_buddy_huge_page() here and be unable * to convert a surplus huge page to a normal huge page. That is * not critical, though, it just means the overall size of the * pool might be one hugepage larger than it needs to be, but * within all the constraints specified by the sysctls. |
7893d1d50
|
1219 |
*/ |
1da177e4c
|
1220 |
spin_lock(&hugetlb_lock); |
a55164389
|
1221 |
while (h->surplus_huge_pages && count > persistent_huge_pages(h)) { |
6ae11b278
|
1222 |
if (!adjust_pool_surplus(h, nodes_allowed, -1)) |
7893d1d50
|
1223 1224 |
break; } |
a55164389
|
1225 |
while (count > persistent_huge_pages(h)) { |
7893d1d50
|
1226 1227 1228 1229 1230 1231 |
/* * If this allocation races such that we no longer need the * page, free_huge_page will handle it by freeing the page * and reducing the surplus. */ spin_unlock(&hugetlb_lock); |
6ae11b278
|
1232 |
ret = alloc_fresh_huge_page(h, nodes_allowed); |
7893d1d50
|
1233 1234 1235 |
spin_lock(&hugetlb_lock); if (!ret) goto out; |
536240f2b
|
1236 1237 1238 |
/* Bail for signals. Probably ctrl-c from user */ if (signal_pending(current)) goto out; |
7893d1d50
|
1239 |
} |
7893d1d50
|
1240 1241 1242 1243 1244 1245 1246 |
/* * Decrease the pool size * First return free pages to the buddy allocator (being careful * to keep enough around to satisfy reservations). Then place * pages into surplus state as needed so the pool will shrink * to the desired size as pages become free. |
d1c3fb1f8
|
1247 1248 1249 1250 1251 1252 1253 1254 |
* * By placing pages into the surplus state independent of the * overcommit value, we are allowing the surplus pool size to * exceed overcommit. There are few sane options here. Since * alloc_buddy_huge_page() is checking the global counter, * though, we'll note that we're not allowed to exceed surplus * and won't grow the pool anywhere else. Not until one of the * sysctls are changed, or the surplus pages go out of use. |
7893d1d50
|
1255 |
*/ |
a55164389
|
1256 |
min_count = h->resv_huge_pages + h->nr_huge_pages - h->free_huge_pages; |
6b0c880df
|
1257 |
min_count = max(count, min_count); |
6ae11b278
|
1258 |
try_to_free_low(h, min_count, nodes_allowed); |
a55164389
|
1259 |
while (min_count < persistent_huge_pages(h)) { |
6ae11b278
|
1260 |
if (!free_pool_huge_page(h, nodes_allowed, 0)) |
1da177e4c
|
1261 |
break; |
1da177e4c
|
1262 |
} |
a55164389
|
1263 |
while (count < persistent_huge_pages(h)) { |
6ae11b278
|
1264 |
if (!adjust_pool_surplus(h, nodes_allowed, 1)) |
7893d1d50
|
1265 1266 1267 |
break; } out: |
a55164389
|
1268 |
ret = persistent_huge_pages(h); |
1da177e4c
|
1269 |
spin_unlock(&hugetlb_lock); |
7893d1d50
|
1270 |
return ret; |
1da177e4c
|
1271 |
} |
a34378701
|
1272 1273 1274 1275 1276 1277 1278 1279 1280 |
#define HSTATE_ATTR_RO(_name) \ static struct kobj_attribute _name##_attr = __ATTR_RO(_name) #define HSTATE_ATTR(_name) \ static struct kobj_attribute _name##_attr = \ __ATTR(_name, 0644, _name##_show, _name##_store) static struct kobject *hugepages_kobj; static struct kobject *hstate_kobjs[HUGE_MAX_HSTATE]; |
9a3052306
|
1281 1282 1283 |
static struct hstate *kobj_to_node_hstate(struct kobject *kobj, int *nidp); static struct hstate *kobj_to_hstate(struct kobject *kobj, int *nidp) |
a34378701
|
1284 1285 |
{ int i; |
9a3052306
|
1286 |
|
a34378701
|
1287 |
for (i = 0; i < HUGE_MAX_HSTATE; i++) |
9a3052306
|
1288 1289 1290 |
if (hstate_kobjs[i] == kobj) { if (nidp) *nidp = NUMA_NO_NODE; |
a34378701
|
1291 |
return &hstates[i]; |
9a3052306
|
1292 1293 1294 |
} return kobj_to_node_hstate(kobj, nidp); |
a34378701
|
1295 |
} |
06808b082
|
1296 |
static ssize_t nr_hugepages_show_common(struct kobject *kobj, |
a34378701
|
1297 1298 |
struct kobj_attribute *attr, char *buf) { |
9a3052306
|
1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 |
struct hstate *h; unsigned long nr_huge_pages; int nid; h = kobj_to_hstate(kobj, &nid); if (nid == NUMA_NO_NODE) nr_huge_pages = h->nr_huge_pages; else nr_huge_pages = h->nr_huge_pages_node[nid]; return sprintf(buf, "%lu ", nr_huge_pages); |
a34378701
|
1311 |
} |
adbe8726d
|
1312 |
|
06808b082
|
1313 1314 1315 |
static ssize_t nr_hugepages_store_common(bool obey_mempolicy, struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t len) |
a34378701
|
1316 1317 |
{ int err; |
9a3052306
|
1318 |
int nid; |
06808b082
|
1319 |
unsigned long count; |
9a3052306
|
1320 |
struct hstate *h; |
bad44b5be
|
1321 |
NODEMASK_ALLOC(nodemask_t, nodes_allowed, GFP_KERNEL | __GFP_NORETRY); |
a34378701
|
1322 |
|
06808b082
|
1323 |
err = strict_strtoul(buf, 10, &count); |
73ae31e59
|
1324 |
if (err) |
adbe8726d
|
1325 |
goto out; |
a34378701
|
1326 |
|
9a3052306
|
1327 |
h = kobj_to_hstate(kobj, &nid); |
adbe8726d
|
1328 1329 1330 1331 |
if (h->order >= MAX_ORDER) { err = -EINVAL; goto out; } |
9a3052306
|
1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 |
if (nid == NUMA_NO_NODE) { /* * global hstate attribute */ if (!(obey_mempolicy && init_nodemask_of_mempolicy(nodes_allowed))) { NODEMASK_FREE(nodes_allowed); nodes_allowed = &node_states[N_HIGH_MEMORY]; } } else if (nodes_allowed) { /* * per node hstate attribute: adjust count to global, * but restrict alloc/free to the specified node. */ count += h->nr_huge_pages - h->nr_huge_pages_node[nid]; init_nodemask_of_node(nodes_allowed, nid); } else nodes_allowed = &node_states[N_HIGH_MEMORY]; |
06808b082
|
1350 |
h->max_huge_pages = set_max_huge_pages(h, count, nodes_allowed); |
a34378701
|
1351 |
|
9b5e5d0fd
|
1352 |
if (nodes_allowed != &node_states[N_HIGH_MEMORY]) |
06808b082
|
1353 1354 1355 |
NODEMASK_FREE(nodes_allowed); return len; |
adbe8726d
|
1356 1357 1358 |
out: NODEMASK_FREE(nodes_allowed); return err; |
06808b082
|
1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 |
} static ssize_t nr_hugepages_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return nr_hugepages_show_common(kobj, attr, buf); } static ssize_t nr_hugepages_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t len) { return nr_hugepages_store_common(false, kobj, attr, buf, len); |
a34378701
|
1371 1372 |
} HSTATE_ATTR(nr_hugepages); |
06808b082
|
1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 |
#ifdef CONFIG_NUMA /* * hstate attribute for optionally mempolicy-based constraint on persistent * huge page alloc/free. */ static ssize_t nr_hugepages_mempolicy_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { return nr_hugepages_show_common(kobj, attr, buf); } static ssize_t nr_hugepages_mempolicy_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t len) { return nr_hugepages_store_common(true, kobj, attr, buf, len); } HSTATE_ATTR(nr_hugepages_mempolicy); #endif |
a34378701
|
1392 1393 1394 |
static ssize_t nr_overcommit_hugepages_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { |
9a3052306
|
1395 |
struct hstate *h = kobj_to_hstate(kobj, NULL); |
a34378701
|
1396 1397 1398 |
return sprintf(buf, "%lu ", h->nr_overcommit_huge_pages); } |
adbe8726d
|
1399 |
|
a34378701
|
1400 1401 1402 1403 1404 |
static ssize_t nr_overcommit_hugepages_store(struct kobject *kobj, struct kobj_attribute *attr, const char *buf, size_t count) { int err; unsigned long input; |
9a3052306
|
1405 |
struct hstate *h = kobj_to_hstate(kobj, NULL); |
a34378701
|
1406 |
|
adbe8726d
|
1407 1408 |
if (h->order >= MAX_ORDER) return -EINVAL; |
a34378701
|
1409 1410 |
err = strict_strtoul(buf, 10, &input); if (err) |
73ae31e59
|
1411 |
return err; |
a34378701
|
1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 |
spin_lock(&hugetlb_lock); h->nr_overcommit_huge_pages = input; spin_unlock(&hugetlb_lock); return count; } HSTATE_ATTR(nr_overcommit_hugepages); static ssize_t free_hugepages_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { |
9a3052306
|
1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 |
struct hstate *h; unsigned long free_huge_pages; int nid; h = kobj_to_hstate(kobj, &nid); if (nid == NUMA_NO_NODE) free_huge_pages = h->free_huge_pages; else free_huge_pages = h->free_huge_pages_node[nid]; return sprintf(buf, "%lu ", free_huge_pages); |
a34378701
|
1436 1437 1438 1439 1440 1441 |
} HSTATE_ATTR_RO(free_hugepages); static ssize_t resv_hugepages_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { |
9a3052306
|
1442 |
struct hstate *h = kobj_to_hstate(kobj, NULL); |
a34378701
|
1443 1444 1445 1446 1447 1448 1449 1450 |
return sprintf(buf, "%lu ", h->resv_huge_pages); } HSTATE_ATTR_RO(resv_hugepages); static ssize_t surplus_hugepages_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { |
9a3052306
|
1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 |
struct hstate *h; unsigned long surplus_huge_pages; int nid; h = kobj_to_hstate(kobj, &nid); if (nid == NUMA_NO_NODE) surplus_huge_pages = h->surplus_huge_pages; else surplus_huge_pages = h->surplus_huge_pages_node[nid]; return sprintf(buf, "%lu ", surplus_huge_pages); |
a34378701
|
1463 1464 1465 1466 1467 1468 1469 1470 1471 |
} HSTATE_ATTR_RO(surplus_hugepages); static struct attribute *hstate_attrs[] = { &nr_hugepages_attr.attr, &nr_overcommit_hugepages_attr.attr, &free_hugepages_attr.attr, &resv_hugepages_attr.attr, &surplus_hugepages_attr.attr, |
06808b082
|
1472 1473 1474 |
#ifdef CONFIG_NUMA &nr_hugepages_mempolicy_attr.attr, #endif |
a34378701
|
1475 1476 1477 1478 1479 1480 |
NULL, }; static struct attribute_group hstate_attr_group = { .attrs = hstate_attrs, }; |
094e9539b
|
1481 1482 1483 |
static int hugetlb_sysfs_add_hstate(struct hstate *h, struct kobject *parent, struct kobject **hstate_kobjs, struct attribute_group *hstate_attr_group) |
a34378701
|
1484 1485 |
{ int retval; |
9a3052306
|
1486 |
int hi = h - hstates; |
a34378701
|
1487 |
|
9a3052306
|
1488 1489 |
hstate_kobjs[hi] = kobject_create_and_add(h->name, parent); if (!hstate_kobjs[hi]) |
a34378701
|
1490 |
return -ENOMEM; |
9a3052306
|
1491 |
retval = sysfs_create_group(hstate_kobjs[hi], hstate_attr_group); |
a34378701
|
1492 |
if (retval) |
9a3052306
|
1493 |
kobject_put(hstate_kobjs[hi]); |
a34378701
|
1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 |
return retval; } static void __init hugetlb_sysfs_init(void) { struct hstate *h; int err; hugepages_kobj = kobject_create_and_add("hugepages", mm_kobj); if (!hugepages_kobj) return; for_each_hstate(h) { |
9a3052306
|
1508 1509 |
err = hugetlb_sysfs_add_hstate(h, hugepages_kobj, hstate_kobjs, &hstate_attr_group); |
a34378701
|
1510 1511 1512 1513 1514 |
if (err) printk(KERN_ERR "Hugetlb: Unable to add hstate %s", h->name); } } |
9a3052306
|
1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 |
#ifdef CONFIG_NUMA /* * node_hstate/s - associate per node hstate attributes, via their kobjects, * with node sysdevs in node_devices[] using a parallel array. The array * index of a node sysdev or _hstate == node id. * This is here to avoid any static dependency of the node sysdev driver, in * the base kernel, on the hugetlb module. */ struct node_hstate { struct kobject *hugepages_kobj; struct kobject *hstate_kobjs[HUGE_MAX_HSTATE]; }; struct node_hstate node_hstates[MAX_NUMNODES]; /* * A subset of global hstate attributes for node sysdevs */ static struct attribute *per_node_hstate_attrs[] = { &nr_hugepages_attr.attr, &free_hugepages_attr.attr, &surplus_hugepages_attr.attr, NULL, }; static struct attribute_group per_node_hstate_attr_group = { .attrs = per_node_hstate_attrs, }; /* * kobj_to_node_hstate - lookup global hstate for node sysdev hstate attr kobj. * Returns node id via non-NULL nidp. */ static struct hstate *kobj_to_node_hstate(struct kobject *kobj, int *nidp) { int nid; for (nid = 0; nid < nr_node_ids; nid++) { struct node_hstate *nhs = &node_hstates[nid]; int i; for (i = 0; i < HUGE_MAX_HSTATE; i++) if (nhs->hstate_kobjs[i] == kobj) { if (nidp) *nidp = nid; return &hstates[i]; } } BUG(); return NULL; } /* * Unregister hstate attributes from a single node sysdev. * No-op if no hstate attributes attached. */ void hugetlb_unregister_node(struct node *node) { struct hstate *h; struct node_hstate *nhs = &node_hstates[node->sysdev.id]; if (!nhs->hugepages_kobj) |
9b5e5d0fd
|
1577 |
return; /* no hstate attributes */ |
9a3052306
|
1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 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 |
for_each_hstate(h) if (nhs->hstate_kobjs[h - hstates]) { kobject_put(nhs->hstate_kobjs[h - hstates]); nhs->hstate_kobjs[h - hstates] = NULL; } kobject_put(nhs->hugepages_kobj); nhs->hugepages_kobj = NULL; } /* * hugetlb module exit: unregister hstate attributes from node sysdevs * that have them. */ static void hugetlb_unregister_all_nodes(void) { int nid; /* * disable node sysdev registrations. */ register_hugetlbfs_with_node(NULL, NULL); /* * remove hstate attributes from any nodes that have them. */ for (nid = 0; nid < nr_node_ids; nid++) hugetlb_unregister_node(&node_devices[nid]); } /* * Register hstate attributes for a single node sysdev. * No-op if attributes already registered. */ void hugetlb_register_node(struct node *node) { struct hstate *h; struct node_hstate *nhs = &node_hstates[node->sysdev.id]; int err; if (nhs->hugepages_kobj) return; /* already allocated */ nhs->hugepages_kobj = kobject_create_and_add("hugepages", &node->sysdev.kobj); if (!nhs->hugepages_kobj) return; for_each_hstate(h) { err = hugetlb_sysfs_add_hstate(h, nhs->hugepages_kobj, nhs->hstate_kobjs, &per_node_hstate_attr_group); if (err) { printk(KERN_ERR "Hugetlb: Unable to add hstate %s" " for node %d ", h->name, node->sysdev.id); hugetlb_unregister_node(node); break; } } } /* |
9b5e5d0fd
|
1643 1644 1645 |
* hugetlb init time: register hstate attributes for all registered node * sysdevs of nodes that have memory. All on-line nodes should have * registered their associated sysdev by this time. |
9a3052306
|
1646 1647 1648 1649 |
*/ static void hugetlb_register_all_nodes(void) { int nid; |
9b5e5d0fd
|
1650 |
for_each_node_state(nid, N_HIGH_MEMORY) { |
9a3052306
|
1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 |
struct node *node = &node_devices[nid]; if (node->sysdev.id == nid) hugetlb_register_node(node); } /* * Let the node sysdev driver know we're here so it can * [un]register hstate attributes on node hotplug. */ register_hugetlbfs_with_node(hugetlb_register_node, hugetlb_unregister_node); } #else /* !CONFIG_NUMA */ static struct hstate *kobj_to_node_hstate(struct kobject *kobj, int *nidp) { BUG(); if (nidp) *nidp = -1; return NULL; } static void hugetlb_unregister_all_nodes(void) { } static void hugetlb_register_all_nodes(void) { } #endif |
a34378701
|
1678 1679 1680 |
static void __exit hugetlb_exit(void) { struct hstate *h; |
9a3052306
|
1681 |
hugetlb_unregister_all_nodes(); |
a34378701
|
1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 |
for_each_hstate(h) { kobject_put(hstate_kobjs[h - hstates]); } kobject_put(hugepages_kobj); } module_exit(hugetlb_exit); static int __init hugetlb_init(void) { |
0ef89d25d
|
1692 1693 1694 1695 1696 1697 |
/* Some platform decide whether they support huge pages at boot * time. On these, such as powerpc, HPAGE_SHIFT is set to 0 when * there is no such support */ if (HPAGE_SHIFT == 0) return 0; |
a34378701
|
1698 |
|
e11bfbfcb
|
1699 1700 1701 1702 |
if (!size_to_hstate(default_hstate_size)) { default_hstate_size = HPAGE_SIZE; if (!size_to_hstate(default_hstate_size)) hugetlb_add_hstate(HUGETLB_PAGE_ORDER); |
a34378701
|
1703 |
} |
e11bfbfcb
|
1704 1705 1706 |
default_hstate_idx = size_to_hstate(default_hstate_size) - hstates; if (default_hstate_max_huge_pages) default_hstate.max_huge_pages = default_hstate_max_huge_pages; |
a34378701
|
1707 1708 |
hugetlb_init_hstates(); |
aa888a749
|
1709 |
gather_bootmem_prealloc(); |
a34378701
|
1710 1711 1712 |
report_hugepages(); hugetlb_sysfs_init(); |
9a3052306
|
1713 |
hugetlb_register_all_nodes(); |
a34378701
|
1714 1715 1716 1717 1718 1719 1720 1721 |
return 0; } module_init(hugetlb_init); /* Should be called on processing a hugepagesz=... option */ void __init hugetlb_add_hstate(unsigned order) { struct hstate *h; |
8faa8b077
|
1722 |
unsigned long i; |
a34378701
|
1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 |
if (size_to_hstate(PAGE_SIZE << order)) { printk(KERN_WARNING "hugepagesz= specified twice, ignoring "); return; } BUG_ON(max_hstate >= HUGE_MAX_HSTATE); BUG_ON(order == 0); h = &hstates[max_hstate++]; h->order = order; h->mask = ~((1ULL << (order + PAGE_SHIFT)) - 1); |
8faa8b077
|
1733 1734 1735 1736 |
h->nr_huge_pages = 0; h->free_huge_pages = 0; for (i = 0; i < MAX_NUMNODES; ++i) INIT_LIST_HEAD(&h->hugepage_freelists[i]); |
9b5e5d0fd
|
1737 1738 |
h->next_nid_to_alloc = first_node(node_states[N_HIGH_MEMORY]); h->next_nid_to_free = first_node(node_states[N_HIGH_MEMORY]); |
a34378701
|
1739 1740 |
snprintf(h->name, HSTATE_NAME_LEN, "hugepages-%lukB", huge_page_size(h)/1024); |
8faa8b077
|
1741 |
|
a34378701
|
1742 1743 |
parsed_hstate = h; } |
e11bfbfcb
|
1744 |
static int __init hugetlb_nrpages_setup(char *s) |
a34378701
|
1745 1746 |
{ unsigned long *mhp; |
8faa8b077
|
1747 |
static unsigned long *last_mhp; |
a34378701
|
1748 1749 1750 1751 1752 1753 1754 1755 1756 |
/* * !max_hstate means we haven't parsed a hugepagesz= parameter yet, * so this hugepages= parameter goes to the "default hstate". */ if (!max_hstate) mhp = &default_hstate_max_huge_pages; else mhp = &parsed_hstate->max_huge_pages; |
8faa8b077
|
1757 1758 1759 1760 1761 1762 |
if (mhp == last_mhp) { printk(KERN_WARNING "hugepages= specified twice without " "interleaving hugepagesz=, ignoring "); return 1; } |
a34378701
|
1763 1764 |
if (sscanf(s, "%lu", mhp) <= 0) *mhp = 0; |
8faa8b077
|
1765 1766 1767 1768 1769 1770 1771 1772 1773 |
/* * Global state is always initialized later in hugetlb_init. * But we need to allocate >= MAX_ORDER hstates here early to still * use the bootmem allocator. */ if (max_hstate && parsed_hstate->order >= MAX_ORDER) hugetlb_hstate_alloc_pages(parsed_hstate); last_mhp = mhp; |
a34378701
|
1774 1775 |
return 1; } |
e11bfbfcb
|
1776 1777 1778 1779 1780 1781 1782 1783 |
__setup("hugepages=", hugetlb_nrpages_setup); static int __init hugetlb_default_setup(char *s) { default_hstate_size = memparse(s, &s); return 1; } __setup("default_hugepagesz=", hugetlb_default_setup); |
a34378701
|
1784 |
|
8a2134605
|
1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 |
static unsigned int cpuset_mems_nr(unsigned int *array) { int node; unsigned int nr = 0; for_each_node_mask(node, cpuset_current_mems_allowed) nr += array[node]; return nr; } #ifdef CONFIG_SYSCTL |
06808b082
|
1797 1798 1799 |
static int hugetlb_sysctl_handler_common(bool obey_mempolicy, struct ctl_table *table, int write, void __user *buffer, size_t *length, loff_t *ppos) |
1da177e4c
|
1800 |
{ |
e5ff21594
|
1801 1802 |
struct hstate *h = &default_hstate; unsigned long tmp; |
08d4a2465
|
1803 |
int ret; |
e5ff21594
|
1804 |
|
c033a93c0
|
1805 |
tmp = h->max_huge_pages; |
e5ff21594
|
1806 |
|
adbe8726d
|
1807 1808 |
if (write && h->order >= MAX_ORDER) return -EINVAL; |
e5ff21594
|
1809 1810 |
table->data = &tmp; table->maxlen = sizeof(unsigned long); |
08d4a2465
|
1811 1812 1813 |
ret = proc_doulongvec_minmax(table, write, buffer, length, ppos); if (ret) goto out; |
e5ff21594
|
1814 |
|
06808b082
|
1815 |
if (write) { |
bad44b5be
|
1816 1817 |
NODEMASK_ALLOC(nodemask_t, nodes_allowed, GFP_KERNEL | __GFP_NORETRY); |
06808b082
|
1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 |
if (!(obey_mempolicy && init_nodemask_of_mempolicy(nodes_allowed))) { NODEMASK_FREE(nodes_allowed); nodes_allowed = &node_states[N_HIGH_MEMORY]; } h->max_huge_pages = set_max_huge_pages(h, tmp, nodes_allowed); if (nodes_allowed != &node_states[N_HIGH_MEMORY]) NODEMASK_FREE(nodes_allowed); } |
08d4a2465
|
1828 1829 |
out: return ret; |
1da177e4c
|
1830 |
} |
396faf030
|
1831 |
|
06808b082
|
1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 |
int hugetlb_sysctl_handler(struct ctl_table *table, int write, void __user *buffer, size_t *length, loff_t *ppos) { return hugetlb_sysctl_handler_common(false, table, write, buffer, length, ppos); } #ifdef CONFIG_NUMA int hugetlb_mempolicy_sysctl_handler(struct ctl_table *table, int write, void __user *buffer, size_t *length, loff_t *ppos) { return hugetlb_sysctl_handler_common(true, table, write, buffer, length, ppos); } #endif /* CONFIG_NUMA */ |
396faf030
|
1848 |
int hugetlb_treat_movable_handler(struct ctl_table *table, int write, |
8d65af789
|
1849 |
void __user *buffer, |
396faf030
|
1850 1851 |
size_t *length, loff_t *ppos) { |
8d65af789
|
1852 |
proc_dointvec(table, write, buffer, length, ppos); |
396faf030
|
1853 1854 1855 1856 1857 1858 |
if (hugepages_treat_as_movable) htlb_alloc_mask = GFP_HIGHUSER_MOVABLE; else htlb_alloc_mask = GFP_HIGHUSER; return 0; } |
a3d0c6aa1
|
1859 |
int hugetlb_overcommit_handler(struct ctl_table *table, int write, |
8d65af789
|
1860 |
void __user *buffer, |
a3d0c6aa1
|
1861 1862 |
size_t *length, loff_t *ppos) { |
a55164389
|
1863 |
struct hstate *h = &default_hstate; |
e5ff21594
|
1864 |
unsigned long tmp; |
08d4a2465
|
1865 |
int ret; |
e5ff21594
|
1866 |
|
c033a93c0
|
1867 |
tmp = h->nr_overcommit_huge_pages; |
e5ff21594
|
1868 |
|
adbe8726d
|
1869 1870 |
if (write && h->order >= MAX_ORDER) return -EINVAL; |
e5ff21594
|
1871 1872 |
table->data = &tmp; table->maxlen = sizeof(unsigned long); |
08d4a2465
|
1873 1874 1875 |
ret = proc_doulongvec_minmax(table, write, buffer, length, ppos); if (ret) goto out; |
e5ff21594
|
1876 1877 1878 1879 1880 1881 |
if (write) { spin_lock(&hugetlb_lock); h->nr_overcommit_huge_pages = tmp; spin_unlock(&hugetlb_lock); } |
08d4a2465
|
1882 1883 |
out: return ret; |
a3d0c6aa1
|
1884 |
} |
1da177e4c
|
1885 |
#endif /* CONFIG_SYSCTL */ |
e1759c215
|
1886 |
void hugetlb_report_meminfo(struct seq_file *m) |
1da177e4c
|
1887 |
{ |
a55164389
|
1888 |
struct hstate *h = &default_hstate; |
e1759c215
|
1889 |
seq_printf(m, |
4f98a2fee
|
1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 |
"HugePages_Total: %5lu " "HugePages_Free: %5lu " "HugePages_Rsvd: %5lu " "HugePages_Surp: %5lu " "Hugepagesize: %8lu kB ", |
a55164389
|
1900 1901 1902 1903 1904 |
h->nr_huge_pages, h->free_huge_pages, h->resv_huge_pages, h->surplus_huge_pages, 1UL << (huge_page_order(h) + PAGE_SHIFT - 10)); |
1da177e4c
|
1905 1906 1907 1908 |
} int hugetlb_report_node_meminfo(int nid, char *buf) { |
a55164389
|
1909 |
struct hstate *h = &default_hstate; |
1da177e4c
|
1910 1911 1912 |
return sprintf(buf, "Node %d HugePages_Total: %5u " |
a1de09195
|
1913 1914 1915 1916 |
"Node %d HugePages_Free: %5u " "Node %d HugePages_Surp: %5u ", |
a55164389
|
1917 1918 1919 |
nid, h->nr_huge_pages_node[nid], nid, h->free_huge_pages_node[nid], nid, h->surplus_huge_pages_node[nid]); |
1da177e4c
|
1920 |
} |
1da177e4c
|
1921 1922 1923 |
/* Return the number pages of memory we physically have, in PAGE_SIZE units. */ unsigned long hugetlb_total_pages(void) { |
a55164389
|
1924 1925 |
struct hstate *h = &default_hstate; return h->nr_huge_pages * pages_per_huge_page(h); |
1da177e4c
|
1926 |
} |
1da177e4c
|
1927 |
|
a55164389
|
1928 |
static int hugetlb_acct_memory(struct hstate *h, long delta) |
fc1b8a73d
|
1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 |
{ int ret = -ENOMEM; spin_lock(&hugetlb_lock); /* * When cpuset is configured, it breaks the strict hugetlb page * reservation as the accounting is done on a global variable. Such * reservation is completely rubbish in the presence of cpuset because * the reservation is not checked against page availability for the * current cpuset. Application can still potentially OOM'ed by kernel * with lack of free htlb page in cpuset that the task is in. * Attempt to enforce strict accounting with cpuset is almost * impossible (or too ugly) because cpuset is too fluid that * task or memory node can be dynamically moved between cpusets. * * The change of semantics for shared hugetlb mapping with cpuset is * undesirable. However, in order to preserve some of the semantics, * we fall back to check against current free page availability as * a best attempt and hopefully to minimize the impact of changing * semantics that cpuset has. */ if (delta > 0) { |
a55164389
|
1951 |
if (gather_surplus_pages(h, delta) < 0) |
fc1b8a73d
|
1952 |
goto out; |
a55164389
|
1953 1954 |
if (delta > cpuset_mems_nr(h->free_huge_pages_node)) { return_unused_surplus_pages(h, delta); |
fc1b8a73d
|
1955 1956 1957 1958 1959 1960 |
goto out; } } ret = 0; if (delta < 0) |
a55164389
|
1961 |
return_unused_surplus_pages(h, (unsigned long) -delta); |
fc1b8a73d
|
1962 1963 1964 1965 1966 |
out: spin_unlock(&hugetlb_lock); return ret; } |
84afd99b8
|
1967 1968 1969 1970 1971 1972 1973 1974 |
static void hugetlb_vm_op_open(struct vm_area_struct *vma) { struct resv_map *reservations = vma_resv_map(vma); /* * This new VMA should share its siblings reservation map if present. * The VMA will only ever have a valid reservation map pointer where * it is being copied for another still existing VMA. As that VMA |
25985edce
|
1975 |
* has a reference to the reservation map it cannot disappear until |
84afd99b8
|
1976 1977 1978 1979 1980 1981 |
* after this open call completes. It is therefore safe to take a * new reference here without additional locking. */ if (reservations) kref_get(&reservations->refs); } |
a1e78772d
|
1982 1983 |
static void hugetlb_vm_op_close(struct vm_area_struct *vma) { |
a55164389
|
1984 |
struct hstate *h = hstate_vma(vma); |
84afd99b8
|
1985 1986 1987 1988 1989 1990 |
struct resv_map *reservations = vma_resv_map(vma); unsigned long reserve; unsigned long start; unsigned long end; if (reservations) { |
a55164389
|
1991 1992 |
start = vma_hugecache_offset(h, vma, vma->vm_start); end = vma_hugecache_offset(h, vma, vma->vm_end); |
84afd99b8
|
1993 1994 1995 1996 1997 |
reserve = (end - start) - region_count(&reservations->regions, start, end); kref_put(&reservations->refs, resv_map_release); |
7251ff78b
|
1998 |
if (reserve) { |
a55164389
|
1999 |
hugetlb_acct_memory(h, -reserve); |
7251ff78b
|
2000 2001 |
hugetlb_put_quota(vma->vm_file->f_mapping, reserve); } |
84afd99b8
|
2002 |
} |
a1e78772d
|
2003 |
} |
1da177e4c
|
2004 2005 2006 2007 2008 2009 |
/* * We cannot handle pagefaults against hugetlb pages at all. They cause * handle_mm_fault() to try to instantiate regular-sized pages in the * hugegpage VMA. do_page_fault() is supposed to trap this, so BUG is we get * this far. */ |
d0217ac04
|
2010 |
static int hugetlb_vm_op_fault(struct vm_area_struct *vma, struct vm_fault *vmf) |
1da177e4c
|
2011 2012 |
{ BUG(); |
d0217ac04
|
2013 |
return 0; |
1da177e4c
|
2014 |
} |
f0f37e2f7
|
2015 |
const struct vm_operations_struct hugetlb_vm_ops = { |
d0217ac04
|
2016 |
.fault = hugetlb_vm_op_fault, |
84afd99b8
|
2017 |
.open = hugetlb_vm_op_open, |
a1e78772d
|
2018 |
.close = hugetlb_vm_op_close, |
1da177e4c
|
2019 |
}; |
1e8f889b1
|
2020 2021 |
static pte_t make_huge_pte(struct vm_area_struct *vma, struct page *page, int writable) |
63551ae0f
|
2022 2023 |
{ pte_t entry; |
1e8f889b1
|
2024 |
if (writable) { |
63551ae0f
|
2025 2026 2027 |
entry = pte_mkwrite(pte_mkdirty(mk_pte(page, vma->vm_page_prot))); } else { |
7f2e9525b
|
2028 |
entry = huge_pte_wrprotect(mk_pte(page, vma->vm_page_prot)); |
63551ae0f
|
2029 2030 2031 2032 2033 2034 |
} entry = pte_mkyoung(entry); entry = pte_mkhuge(entry); return entry; } |
1e8f889b1
|
2035 2036 2037 2038 |
static void set_huge_ptep_writable(struct vm_area_struct *vma, unsigned long address, pte_t *ptep) { pte_t entry; |
7f2e9525b
|
2039 2040 |
entry = pte_mkwrite(pte_mkdirty(huge_ptep_get(ptep))); if (huge_ptep_set_access_flags(vma, address, ptep, entry, 1)) { |
4b3073e1c
|
2041 |
update_mmu_cache(vma, address, ptep); |
8dab5241d
|
2042 |
} |
1e8f889b1
|
2043 |
} |
63551ae0f
|
2044 2045 2046 2047 2048 |
int copy_hugetlb_page_range(struct mm_struct *dst, struct mm_struct *src, struct vm_area_struct *vma) { pte_t *src_pte, *dst_pte, entry; struct page *ptepage; |
1c59827d1
|
2049 |
unsigned long addr; |
1e8f889b1
|
2050 |
int cow; |
a55164389
|
2051 2052 |
struct hstate *h = hstate_vma(vma); unsigned long sz = huge_page_size(h); |
1e8f889b1
|
2053 2054 |
cow = (vma->vm_flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE; |
63551ae0f
|
2055 |
|
a55164389
|
2056 |
for (addr = vma->vm_start; addr < vma->vm_end; addr += sz) { |
c74df32c7
|
2057 2058 2059 |
src_pte = huge_pte_offset(src, addr); if (!src_pte) continue; |
a55164389
|
2060 |
dst_pte = huge_pte_alloc(dst, addr, sz); |
63551ae0f
|
2061 2062 |
if (!dst_pte) goto nomem; |
c5c99429f
|
2063 2064 2065 2066 |
/* If the pagetables are shared don't copy or take references */ if (dst_pte == src_pte) continue; |
c74df32c7
|
2067 |
spin_lock(&dst->page_table_lock); |
464787581
|
2068 |
spin_lock_nested(&src->page_table_lock, SINGLE_DEPTH_NESTING); |
7f2e9525b
|
2069 |
if (!huge_pte_none(huge_ptep_get(src_pte))) { |
1e8f889b1
|
2070 |
if (cow) |
7f2e9525b
|
2071 2072 |
huge_ptep_set_wrprotect(src, addr, src_pte); entry = huge_ptep_get(src_pte); |
1c59827d1
|
2073 2074 |
ptepage = pte_page(entry); get_page(ptepage); |
0fe6e20b9
|
2075 |
page_dup_rmap(ptepage); |
1c59827d1
|
2076 2077 2078 |
set_huge_pte_at(dst, addr, dst_pte, entry); } spin_unlock(&src->page_table_lock); |
c74df32c7
|
2079 |
spin_unlock(&dst->page_table_lock); |
63551ae0f
|
2080 2081 2082 2083 2084 2085 |
} return 0; nomem: return -ENOMEM; } |
290408d4a
|
2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 |
static int is_hugetlb_entry_migration(pte_t pte) { swp_entry_t swp; if (huge_pte_none(pte) || pte_present(pte)) return 0; swp = pte_to_swp_entry(pte); if (non_swap_entry(swp) && is_migration_entry(swp)) { return 1; } else return 0; } |
fd6a03edd
|
2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 |
static int is_hugetlb_entry_hwpoisoned(pte_t pte) { swp_entry_t swp; if (huge_pte_none(pte) || pte_present(pte)) return 0; swp = pte_to_swp_entry(pte); if (non_swap_entry(swp) && is_hwpoison_entry(swp)) { return 1; } else return 0; } |
502717f4e
|
2110 |
void __unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start, |
04f2cbe35
|
2111 |
unsigned long end, struct page *ref_page) |
63551ae0f
|
2112 2113 2114 |
{ struct mm_struct *mm = vma->vm_mm; unsigned long address; |
c7546f8f0
|
2115 |
pte_t *ptep; |
63551ae0f
|
2116 2117 |
pte_t pte; struct page *page; |
fe1668ae5
|
2118 |
struct page *tmp; |
a55164389
|
2119 2120 |
struct hstate *h = hstate_vma(vma); unsigned long sz = huge_page_size(h); |
c0a499c2c
|
2121 |
/* |
3d48ae45e
|
2122 |
* A page gathering list, protected by per file i_mmap_mutex. The |
c0a499c2c
|
2123 2124 2125 |
* lock is used to avoid list corruption from multiple unmapping * of the same page since we are using page->lru. */ |
fe1668ae5
|
2126 |
LIST_HEAD(page_list); |
63551ae0f
|
2127 2128 |
WARN_ON(!is_vm_hugetlb_page(vma)); |
a55164389
|
2129 2130 |
BUG_ON(start & ~huge_page_mask(h)); BUG_ON(end & ~huge_page_mask(h)); |
63551ae0f
|
2131 |
|
cddb8a5c1
|
2132 |
mmu_notifier_invalidate_range_start(mm, start, end); |
508034a32
|
2133 |
spin_lock(&mm->page_table_lock); |
a55164389
|
2134 |
for (address = start; address < end; address += sz) { |
c7546f8f0
|
2135 |
ptep = huge_pte_offset(mm, address); |
4c8872659
|
2136 |
if (!ptep) |
c7546f8f0
|
2137 |
continue; |
39dde65c9
|
2138 2139 |
if (huge_pmd_unshare(mm, &address, ptep)) continue; |
04f2cbe35
|
2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 |
/* * If a reference page is supplied, it is because a specific * page is being unmapped, not a range. Ensure the page we * are about to unmap is the actual page of interest. */ if (ref_page) { pte = huge_ptep_get(ptep); if (huge_pte_none(pte)) continue; page = pte_page(pte); if (page != ref_page) continue; /* * Mark the VMA as having unmapped its page so that * future faults in this VMA will fail rather than * looking like data was lost */ set_vma_resv_flags(vma, HPAGE_RESV_UNMAPPED); } |
c7546f8f0
|
2160 |
pte = huge_ptep_get_and_clear(mm, address, ptep); |
7f2e9525b
|
2161 |
if (huge_pte_none(pte)) |
63551ae0f
|
2162 |
continue; |
c7546f8f0
|
2163 |
|
fd6a03edd
|
2164 2165 2166 2167 2168 |
/* * HWPoisoned hugepage is already unmapped and dropped reference */ if (unlikely(is_hugetlb_entry_hwpoisoned(pte))) continue; |
63551ae0f
|
2169 |
page = pte_page(pte); |
6649a3863
|
2170 2171 |
if (pte_dirty(pte)) set_page_dirty(page); |
fe1668ae5
|
2172 |
list_add(&page->lru, &page_list); |
63551ae0f
|
2173 |
} |
1da177e4c
|
2174 |
spin_unlock(&mm->page_table_lock); |
508034a32
|
2175 |
flush_tlb_range(vma, start, end); |
cddb8a5c1
|
2176 |
mmu_notifier_invalidate_range_end(mm, start, end); |
fe1668ae5
|
2177 |
list_for_each_entry_safe(page, tmp, &page_list, lru) { |
0fe6e20b9
|
2178 |
page_remove_rmap(page); |
fe1668ae5
|
2179 2180 2181 |
list_del(&page->lru); put_page(page); } |
1da177e4c
|
2182 |
} |
63551ae0f
|
2183 |
|
502717f4e
|
2184 |
void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start, |
04f2cbe35
|
2185 |
unsigned long end, struct page *ref_page) |
502717f4e
|
2186 |
{ |
3d48ae45e
|
2187 |
mutex_lock(&vma->vm_file->f_mapping->i_mmap_mutex); |
a137e1cc6
|
2188 |
__unmap_hugepage_range(vma, start, end, ref_page); |
3d48ae45e
|
2189 |
mutex_unlock(&vma->vm_file->f_mapping->i_mmap_mutex); |
502717f4e
|
2190 |
} |
04f2cbe35
|
2191 2192 2193 2194 2195 2196 |
/* * This is called when the original mapper is failing to COW a MAP_PRIVATE * mappping it owns the reserve page for. The intention is to unmap the page * from other VMAs and let the children be SIGKILLed if they are faulting the * same region. */ |
2a4b3ded5
|
2197 2198 |
static int unmap_ref_private(struct mm_struct *mm, struct vm_area_struct *vma, struct page *page, unsigned long address) |
04f2cbe35
|
2199 |
{ |
7526674de
|
2200 |
struct hstate *h = hstate_vma(vma); |
04f2cbe35
|
2201 2202 2203 2204 2205 2206 2207 2208 2209 |
struct vm_area_struct *iter_vma; struct address_space *mapping; struct prio_tree_iter iter; pgoff_t pgoff; /* * vm_pgoff is in PAGE_SIZE units, hence the different calculation * from page cache lookup which is in HPAGE_SIZE units. */ |
7526674de
|
2210 |
address = address & huge_page_mask(h); |
04f2cbe35
|
2211 2212 2213 |
pgoff = ((address - vma->vm_start) >> PAGE_SHIFT) + (vma->vm_pgoff >> PAGE_SHIFT); mapping = (struct address_space *)page_private(page); |
4eb2b1dcd
|
2214 2215 2216 2217 2218 |
/* * Take the mapping lock for the duration of the table walk. As * this mapping should be shared between all the VMAs, * __unmap_hugepage_range() is called as the lock is already held */ |
3d48ae45e
|
2219 |
mutex_lock(&mapping->i_mmap_mutex); |
04f2cbe35
|
2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 |
vma_prio_tree_foreach(iter_vma, &iter, &mapping->i_mmap, pgoff, pgoff) { /* Do not unmap the current VMA */ if (iter_vma == vma) continue; /* * Unmap the page from other VMAs without their own reserves. * They get marked to be SIGKILLed if they fault in these * areas. This is because a future no-page fault on this VMA * could insert a zeroed page instead of the data existing * from the time of fork. This would look like data corruption */ if (!is_vma_resv_set(iter_vma, HPAGE_RESV_OWNER)) |
4eb2b1dcd
|
2233 |
__unmap_hugepage_range(iter_vma, |
7526674de
|
2234 |
address, address + huge_page_size(h), |
04f2cbe35
|
2235 2236 |
page); } |
3d48ae45e
|
2237 |
mutex_unlock(&mapping->i_mmap_mutex); |
04f2cbe35
|
2238 2239 2240 |
return 1; } |
0fe6e20b9
|
2241 2242 2243 |
/* * Hugetlb_cow() should be called with page lock of the original hugepage held. */ |
1e8f889b1
|
2244 |
static int hugetlb_cow(struct mm_struct *mm, struct vm_area_struct *vma, |
04f2cbe35
|
2245 2246 |
unsigned long address, pte_t *ptep, pte_t pte, struct page *pagecache_page) |
1e8f889b1
|
2247 |
{ |
a55164389
|
2248 |
struct hstate *h = hstate_vma(vma); |
1e8f889b1
|
2249 |
struct page *old_page, *new_page; |
79ac6ba40
|
2250 |
int avoidcopy; |
04f2cbe35
|
2251 |
int outside_reserve = 0; |
1e8f889b1
|
2252 2253 |
old_page = pte_page(pte); |
04f2cbe35
|
2254 |
retry_avoidcopy: |
1e8f889b1
|
2255 2256 |
/* If no-one else is actually using this page, avoid the copy * and just make the page writable */ |
0fe6e20b9
|
2257 |
avoidcopy = (page_mapcount(old_page) == 1); |
1e8f889b1
|
2258 |
if (avoidcopy) { |
56c9cfb13
|
2259 2260 |
if (PageAnon(old_page)) page_move_anon_rmap(old_page, vma, address); |
1e8f889b1
|
2261 |
set_huge_ptep_writable(vma, address, ptep); |
83c54070e
|
2262 |
return 0; |
1e8f889b1
|
2263 |
} |
04f2cbe35
|
2264 2265 2266 2267 2268 2269 2270 2271 2272 |
/* * If the process that created a MAP_PRIVATE mapping is about to * perform a COW due to a shared page count, attempt to satisfy * the allocation without using the existing reserves. The pagecache * page is used to determine if the reserve at this address was * consumed or not. If reserves were used, a partial faulted mapping * at the time of fork() could consume its reserves on COW instead * of the full address range. */ |
f83a275db
|
2273 |
if (!(vma->vm_flags & VM_MAYSHARE) && |
04f2cbe35
|
2274 2275 2276 |
is_vma_resv_set(vma, HPAGE_RESV_OWNER) && old_page != pagecache_page) outside_reserve = 1; |
1e8f889b1
|
2277 |
page_cache_get(old_page); |
b76c8cfbf
|
2278 2279 2280 |
/* Drop page_table_lock as buddy allocator may be called */ spin_unlock(&mm->page_table_lock); |
04f2cbe35
|
2281 |
new_page = alloc_huge_page(vma, address, outside_reserve); |
1e8f889b1
|
2282 |
|
2fc39cec6
|
2283 |
if (IS_ERR(new_page)) { |
1e8f889b1
|
2284 |
page_cache_release(old_page); |
04f2cbe35
|
2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 |
/* * If a process owning a MAP_PRIVATE mapping fails to COW, * it is due to references held by a child and an insufficient * huge page pool. To guarantee the original mappers * reliability, unmap the page from child processes. The child * may get SIGKILLed if it later faults. */ if (outside_reserve) { BUG_ON(huge_pte_none(pte)); if (unmap_ref_private(mm, vma, old_page, address)) { BUG_ON(page_count(old_page) != 1); BUG_ON(huge_pte_none(pte)); |
b76c8cfbf
|
2298 |
spin_lock(&mm->page_table_lock); |
04f2cbe35
|
2299 2300 2301 2302 |
goto retry_avoidcopy; } WARN_ON_ONCE(1); } |
b76c8cfbf
|
2303 2304 |
/* Caller expects lock to be held */ spin_lock(&mm->page_table_lock); |
2fc39cec6
|
2305 |
return -PTR_ERR(new_page); |
1e8f889b1
|
2306 |
} |
0fe6e20b9
|
2307 2308 2309 2310 |
/* * When the original hugepage is shared one, it does not have * anon_vma prepared. */ |
44e2aa937
|
2311 2312 2313 |
if (unlikely(anon_vma_prepare(vma))) { /* Caller expects lock to be held */ spin_lock(&mm->page_table_lock); |
0fe6e20b9
|
2314 |
return VM_FAULT_OOM; |
44e2aa937
|
2315 |
} |
0fe6e20b9
|
2316 |
|
47ad8475c
|
2317 2318 |
copy_user_huge_page(new_page, old_page, address, vma, pages_per_huge_page(h)); |
0ed361dec
|
2319 |
__SetPageUptodate(new_page); |
1e8f889b1
|
2320 |
|
b76c8cfbf
|
2321 2322 2323 2324 2325 |
/* * Retake the page_table_lock to check for racing updates * before the page tables are altered */ spin_lock(&mm->page_table_lock); |
a55164389
|
2326 |
ptep = huge_pte_offset(mm, address & huge_page_mask(h)); |
7f2e9525b
|
2327 |
if (likely(pte_same(huge_ptep_get(ptep), pte))) { |
1e8f889b1
|
2328 |
/* Break COW */ |
3edd4fc95
|
2329 2330 2331 |
mmu_notifier_invalidate_range_start(mm, address & huge_page_mask(h), (address & huge_page_mask(h)) + huge_page_size(h)); |
8fe627ec5
|
2332 |
huge_ptep_clear_flush(vma, address, ptep); |
1e8f889b1
|
2333 2334 |
set_huge_pte_at(mm, address, ptep, make_huge_pte(vma, new_page, 1)); |
0fe6e20b9
|
2335 |
page_remove_rmap(old_page); |
cd67f0d2a
|
2336 |
hugepage_add_new_anon_rmap(new_page, vma, address); |
1e8f889b1
|
2337 2338 |
/* Make the old page be freed below */ new_page = old_page; |
3edd4fc95
|
2339 2340 2341 |
mmu_notifier_invalidate_range_end(mm, address & huge_page_mask(h), (address & huge_page_mask(h)) + huge_page_size(h)); |
1e8f889b1
|
2342 2343 2344 |
} page_cache_release(new_page); page_cache_release(old_page); |
83c54070e
|
2345 |
return 0; |
1e8f889b1
|
2346 |
} |
04f2cbe35
|
2347 |
/* Return the pagecache page at a given address within a VMA */ |
a55164389
|
2348 2349 |
static struct page *hugetlbfs_pagecache_page(struct hstate *h, struct vm_area_struct *vma, unsigned long address) |
04f2cbe35
|
2350 2351 |
{ struct address_space *mapping; |
e7c4b0bfd
|
2352 |
pgoff_t idx; |
04f2cbe35
|
2353 2354 |
mapping = vma->vm_file->f_mapping; |
a55164389
|
2355 |
idx = vma_hugecache_offset(h, vma, address); |
04f2cbe35
|
2356 2357 2358 |
return find_lock_page(mapping, idx); } |
3ae77f43b
|
2359 2360 2361 2362 2363 |
/* * Return whether there is a pagecache page to back given address within VMA. * Caller follow_hugetlb_page() holds page_table_lock so we cannot lock_page. */ static bool hugetlbfs_pagecache_present(struct hstate *h, |
2a15efc95
|
2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 |
struct vm_area_struct *vma, unsigned long address) { struct address_space *mapping; pgoff_t idx; struct page *page; mapping = vma->vm_file->f_mapping; idx = vma_hugecache_offset(h, vma, address); page = find_get_page(mapping, idx); if (page) put_page(page); return page != NULL; } |
a1ed3dda0
|
2378 |
static int hugetlb_no_page(struct mm_struct *mm, struct vm_area_struct *vma, |
788c7df45
|
2379 |
unsigned long address, pte_t *ptep, unsigned int flags) |
ac9b9c667
|
2380 |
{ |
a55164389
|
2381 |
struct hstate *h = hstate_vma(vma); |
ac9b9c667
|
2382 |
int ret = VM_FAULT_SIGBUS; |
e7c4b0bfd
|
2383 |
pgoff_t idx; |
4c8872659
|
2384 |
unsigned long size; |
4c8872659
|
2385 2386 |
struct page *page; struct address_space *mapping; |
1e8f889b1
|
2387 |
pte_t new_pte; |
4c8872659
|
2388 |
|
04f2cbe35
|
2389 2390 2391 |
/* * Currently, we are forced to kill the process in the event the * original mapper has unmapped pages from the child due to a failed |
25985edce
|
2392 |
* COW. Warn that such a situation has occurred as it may not be obvious |
04f2cbe35
|
2393 2394 2395 2396 2397 2398 2399 2400 |
*/ if (is_vma_resv_set(vma, HPAGE_RESV_UNMAPPED)) { printk(KERN_WARNING "PID %d killed due to inadequate hugepage pool ", current->pid); return ret; } |
4c8872659
|
2401 |
mapping = vma->vm_file->f_mapping; |
a55164389
|
2402 |
idx = vma_hugecache_offset(h, vma, address); |
4c8872659
|
2403 2404 2405 2406 2407 |
/* * Use page lock to guard against racing truncation * before we get page_table_lock. */ |
6bda666a0
|
2408 2409 2410 |
retry: page = find_lock_page(mapping, idx); if (!page) { |
a55164389
|
2411 |
size = i_size_read(mapping->host) >> huge_page_shift(h); |
ebed4bfc8
|
2412 2413 |
if (idx >= size) goto out; |
04f2cbe35
|
2414 |
page = alloc_huge_page(vma, address, 0); |
2fc39cec6
|
2415 2416 |
if (IS_ERR(page)) { ret = -PTR_ERR(page); |
6bda666a0
|
2417 2418 |
goto out; } |
47ad8475c
|
2419 |
clear_huge_page(page, address, pages_per_huge_page(h)); |
0ed361dec
|
2420 |
__SetPageUptodate(page); |
ac9b9c667
|
2421 |
|
f83a275db
|
2422 |
if (vma->vm_flags & VM_MAYSHARE) { |
6bda666a0
|
2423 |
int err; |
45c682a68
|
2424 |
struct inode *inode = mapping->host; |
6bda666a0
|
2425 2426 2427 2428 |
err = add_to_page_cache(page, mapping, idx, GFP_KERNEL); if (err) { put_page(page); |
6bda666a0
|
2429 2430 2431 2432 |
if (err == -EEXIST) goto retry; goto out; } |
45c682a68
|
2433 2434 |
spin_lock(&inode->i_lock); |
a55164389
|
2435 |
inode->i_blocks += blocks_per_huge_page(h); |
45c682a68
|
2436 |
spin_unlock(&inode->i_lock); |
0fe6e20b9
|
2437 |
page_dup_rmap(page); |
23be7468e
|
2438 |
} else { |
6bda666a0
|
2439 |
lock_page(page); |
0fe6e20b9
|
2440 2441 2442 2443 2444 |
if (unlikely(anon_vma_prepare(vma))) { ret = VM_FAULT_OOM; goto backout_unlocked; } hugepage_add_new_anon_rmap(page, vma, address); |
23be7468e
|
2445 |
} |
0fe6e20b9
|
2446 |
} else { |
998b4382c
|
2447 2448 2449 2450 2451 2452 |
/* * If memory error occurs between mmap() and fault, some process * don't have hwpoisoned swap entry for errored virtual address. * So we need to block hugepage fault by PG_hwpoison bit check. */ if (unlikely(PageHWPoison(page))) { |
aa50d3a7a
|
2453 2454 |
ret = VM_FAULT_HWPOISON | VM_FAULT_SET_HINDEX(h - hstates); |
998b4382c
|
2455 2456 |
goto backout_unlocked; } |
0fe6e20b9
|
2457 |
page_dup_rmap(page); |
6bda666a0
|
2458 |
} |
1e8f889b1
|
2459 |
|
57303d801
|
2460 2461 2462 2463 2464 2465 |
/* * If we are going to COW a private mapping later, we examine the * pending reservations for this page now. This will ensure that * any allocations necessary to record that reservation occur outside * the spinlock. */ |
788c7df45
|
2466 |
if ((flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED)) |
2b26736c8
|
2467 2468 2469 2470 |
if (vma_needs_reservation(h, vma, address) < 0) { ret = VM_FAULT_OOM; goto backout_unlocked; } |
57303d801
|
2471 |
|
ac9b9c667
|
2472 |
spin_lock(&mm->page_table_lock); |
a55164389
|
2473 |
size = i_size_read(mapping->host) >> huge_page_shift(h); |
4c8872659
|
2474 2475 |
if (idx >= size) goto backout; |
83c54070e
|
2476 |
ret = 0; |
7f2e9525b
|
2477 |
if (!huge_pte_none(huge_ptep_get(ptep))) |
4c8872659
|
2478 |
goto backout; |
1e8f889b1
|
2479 2480 2481 |
new_pte = make_huge_pte(vma, page, ((vma->vm_flags & VM_WRITE) && (vma->vm_flags & VM_SHARED))); set_huge_pte_at(mm, address, ptep, new_pte); |
788c7df45
|
2482 |
if ((flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED)) { |
1e8f889b1
|
2483 |
/* Optimization, do the COW without a second fault */ |
04f2cbe35
|
2484 |
ret = hugetlb_cow(mm, vma, address, ptep, new_pte, page); |
1e8f889b1
|
2485 |
} |
ac9b9c667
|
2486 |
spin_unlock(&mm->page_table_lock); |
4c8872659
|
2487 2488 |
unlock_page(page); out: |
ac9b9c667
|
2489 |
return ret; |
4c8872659
|
2490 2491 2492 |
backout: spin_unlock(&mm->page_table_lock); |
2b26736c8
|
2493 |
backout_unlocked: |
4c8872659
|
2494 2495 2496 |
unlock_page(page); put_page(page); goto out; |
ac9b9c667
|
2497 |
} |
86e5216f8
|
2498 |
int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma, |
788c7df45
|
2499 |
unsigned long address, unsigned int flags) |
86e5216f8
|
2500 2501 2502 |
{ pte_t *ptep; pte_t entry; |
1e8f889b1
|
2503 |
int ret; |
0fe6e20b9
|
2504 |
struct page *page = NULL; |
57303d801
|
2505 |
struct page *pagecache_page = NULL; |
3935baa9b
|
2506 |
static DEFINE_MUTEX(hugetlb_instantiation_mutex); |
a55164389
|
2507 |
struct hstate *h = hstate_vma(vma); |
86e5216f8
|
2508 |
|
fd6a03edd
|
2509 2510 2511 |
ptep = huge_pte_offset(mm, address); if (ptep) { entry = huge_ptep_get(ptep); |
290408d4a
|
2512 2513 2514 2515 |
if (unlikely(is_hugetlb_entry_migration(entry))) { migration_entry_wait(mm, (pmd_t *)ptep, address); return 0; } else if (unlikely(is_hugetlb_entry_hwpoisoned(entry))) |
aa50d3a7a
|
2516 2517 |
return VM_FAULT_HWPOISON_LARGE | VM_FAULT_SET_HINDEX(h - hstates); |
fd6a03edd
|
2518 |
} |
a55164389
|
2519 |
ptep = huge_pte_alloc(mm, address, huge_page_size(h)); |
86e5216f8
|
2520 2521 |
if (!ptep) return VM_FAULT_OOM; |
3935baa9b
|
2522 2523 2524 2525 2526 2527 |
/* * Serialize hugepage allocation and instantiation, so that we don't * get spurious allocation failures if two CPUs race to instantiate * the same page in the page cache. */ mutex_lock(&hugetlb_instantiation_mutex); |
7f2e9525b
|
2528 2529 |
entry = huge_ptep_get(ptep); if (huge_pte_none(entry)) { |
788c7df45
|
2530 |
ret = hugetlb_no_page(mm, vma, address, ptep, flags); |
b4d1d99fd
|
2531 |
goto out_mutex; |
3935baa9b
|
2532 |
} |
86e5216f8
|
2533 |
|
83c54070e
|
2534 |
ret = 0; |
1e8f889b1
|
2535 |
|
57303d801
|
2536 2537 2538 2539 2540 2541 2542 2543 |
/* * If we are going to COW the mapping later, we examine the pending * reservations for this page now. This will ensure that any * allocations necessary to record that reservation occur outside the * spinlock. For private mappings, we also lookup the pagecache * page now as it is used to determine if a reservation has been * consumed. */ |
788c7df45
|
2544 |
if ((flags & FAULT_FLAG_WRITE) && !pte_write(entry)) { |
2b26736c8
|
2545 2546 |
if (vma_needs_reservation(h, vma, address) < 0) { ret = VM_FAULT_OOM; |
b4d1d99fd
|
2547 |
goto out_mutex; |
2b26736c8
|
2548 |
} |
57303d801
|
2549 |
|
f83a275db
|
2550 |
if (!(vma->vm_flags & VM_MAYSHARE)) |
57303d801
|
2551 2552 2553 |
pagecache_page = hugetlbfs_pagecache_page(h, vma, address); } |
56c9cfb13
|
2554 2555 2556 2557 2558 2559 2560 2561 2562 |
/* * hugetlb_cow() requires page locks of pte_page(entry) and * pagecache_page, so here we need take the former one * when page != pagecache_page or !pagecache_page. * Note that locking order is always pagecache_page -> page, * so no worry about deadlock. */ page = pte_page(entry); if (page != pagecache_page) |
0fe6e20b9
|
2563 |
lock_page(page); |
0fe6e20b9
|
2564 |
|
1e8f889b1
|
2565 2566 |
spin_lock(&mm->page_table_lock); /* Check for a racing update before calling hugetlb_cow */ |
b4d1d99fd
|
2567 2568 |
if (unlikely(!pte_same(entry, huge_ptep_get(ptep)))) goto out_page_table_lock; |
788c7df45
|
2569 |
if (flags & FAULT_FLAG_WRITE) { |
b4d1d99fd
|
2570 |
if (!pte_write(entry)) { |
57303d801
|
2571 2572 |
ret = hugetlb_cow(mm, vma, address, ptep, entry, pagecache_page); |
b4d1d99fd
|
2573 2574 2575 2576 2577 |
goto out_page_table_lock; } entry = pte_mkdirty(entry); } entry = pte_mkyoung(entry); |
788c7df45
|
2578 2579 |
if (huge_ptep_set_access_flags(vma, address, ptep, entry, flags & FAULT_FLAG_WRITE)) |
4b3073e1c
|
2580 |
update_mmu_cache(vma, address, ptep); |
b4d1d99fd
|
2581 2582 |
out_page_table_lock: |
1e8f889b1
|
2583 |
spin_unlock(&mm->page_table_lock); |
57303d801
|
2584 2585 2586 2587 2588 |
if (pagecache_page) { unlock_page(pagecache_page); put_page(pagecache_page); } |
1f64d69c7
|
2589 2590 |
if (page != pagecache_page) unlock_page(page); |
57303d801
|
2591 |
|
b4d1d99fd
|
2592 |
out_mutex: |
3935baa9b
|
2593 |
mutex_unlock(&hugetlb_instantiation_mutex); |
1e8f889b1
|
2594 2595 |
return ret; |
86e5216f8
|
2596 |
} |
ceb868796
|
2597 2598 2599 2600 2601 2602 2603 2604 |
/* Can be overriden by architectures */ __attribute__((weak)) struct page * follow_huge_pud(struct mm_struct *mm, unsigned long address, pud_t *pud, int write) { BUG(); return NULL; } |
63551ae0f
|
2605 2606 |
int follow_hugetlb_page(struct mm_struct *mm, struct vm_area_struct *vma, struct page **pages, struct vm_area_struct **vmas, |
5b23dbe81
|
2607 |
unsigned long *position, int *length, int i, |
2a15efc95
|
2608 |
unsigned int flags) |
63551ae0f
|
2609 |
{ |
d5d4b0aa4
|
2610 2611 |
unsigned long pfn_offset; unsigned long vaddr = *position; |
63551ae0f
|
2612 |
int remainder = *length; |
a55164389
|
2613 |
struct hstate *h = hstate_vma(vma); |
63551ae0f
|
2614 |
|
1c59827d1
|
2615 |
spin_lock(&mm->page_table_lock); |
63551ae0f
|
2616 |
while (vaddr < vma->vm_end && remainder) { |
4c8872659
|
2617 |
pte_t *pte; |
2a15efc95
|
2618 |
int absent; |
4c8872659
|
2619 |
struct page *page; |
63551ae0f
|
2620 |
|
4c8872659
|
2621 2622 |
/* * Some archs (sparc64, sh*) have multiple pte_ts to |
2a15efc95
|
2623 |
* each hugepage. We have to make sure we get the |
4c8872659
|
2624 2625 |
* first, for the page indexing below to work. */ |
a55164389
|
2626 |
pte = huge_pte_offset(mm, vaddr & huge_page_mask(h)); |
2a15efc95
|
2627 2628 2629 2630 |
absent = !pte || huge_pte_none(huge_ptep_get(pte)); /* * When coredumping, it suits get_dump_page if we just return |
3ae77f43b
|
2631 2632 2633 2634 |
* an error where there's an empty slot with no huge pagecache * to back it. This way, we avoid allocating a hugepage, and * the sparse dumpfile avoids allocating disk blocks, but its * huge holes still show up with zeroes where they need to be. |
2a15efc95
|
2635 |
*/ |
3ae77f43b
|
2636 2637 |
if (absent && (flags & FOLL_DUMP) && !hugetlbfs_pagecache_present(h, vma, vaddr)) { |
2a15efc95
|
2638 2639 2640 |
remainder = 0; break; } |
63551ae0f
|
2641 |
|
2a15efc95
|
2642 2643 |
if (absent || ((flags & FOLL_WRITE) && !pte_write(huge_ptep_get(pte)))) { |
4c8872659
|
2644 |
int ret; |
63551ae0f
|
2645 |
|
4c8872659
|
2646 |
spin_unlock(&mm->page_table_lock); |
2a15efc95
|
2647 2648 |
ret = hugetlb_fault(mm, vma, vaddr, (flags & FOLL_WRITE) ? FAULT_FLAG_WRITE : 0); |
4c8872659
|
2649 |
spin_lock(&mm->page_table_lock); |
a89182c76
|
2650 |
if (!(ret & VM_FAULT_ERROR)) |
4c8872659
|
2651 |
continue; |
63551ae0f
|
2652 |
|
4c8872659
|
2653 |
remainder = 0; |
4c8872659
|
2654 2655 |
break; } |
a55164389
|
2656 |
pfn_offset = (vaddr & ~huge_page_mask(h)) >> PAGE_SHIFT; |
7f2e9525b
|
2657 |
page = pte_page(huge_ptep_get(pte)); |
d5d4b0aa4
|
2658 |
same_page: |
d6692183a
|
2659 |
if (pages) { |
2a15efc95
|
2660 |
pages[i] = mem_map_offset(page, pfn_offset); |
4b2e38ad7
|
2661 |
get_page(pages[i]); |
d6692183a
|
2662 |
} |
63551ae0f
|
2663 2664 2665 2666 2667 |
if (vmas) vmas[i] = vma; vaddr += PAGE_SIZE; |
d5d4b0aa4
|
2668 |
++pfn_offset; |
63551ae0f
|
2669 2670 |
--remainder; ++i; |
d5d4b0aa4
|
2671 |
if (vaddr < vma->vm_end && remainder && |
a55164389
|
2672 |
pfn_offset < pages_per_huge_page(h)) { |
d5d4b0aa4
|
2673 2674 2675 2676 2677 2678 |
/* * We use pfn_offset to avoid touching the pageframes * of this compound page. */ goto same_page; } |
63551ae0f
|
2679 |
} |
1c59827d1
|
2680 |
spin_unlock(&mm->page_table_lock); |
63551ae0f
|
2681 2682 |
*length = remainder; *position = vaddr; |
2a15efc95
|
2683 |
return i ? i : -EFAULT; |
63551ae0f
|
2684 |
} |
8f860591f
|
2685 2686 2687 2688 2689 2690 2691 2692 |
void hugetlb_change_protection(struct vm_area_struct *vma, unsigned long address, unsigned long end, pgprot_t newprot) { struct mm_struct *mm = vma->vm_mm; unsigned long start = address; pte_t *ptep; pte_t pte; |
a55164389
|
2693 |
struct hstate *h = hstate_vma(vma); |
8f860591f
|
2694 2695 2696 |
BUG_ON(address >= end); flush_cache_range(vma, address, end); |
3d48ae45e
|
2697 |
mutex_lock(&vma->vm_file->f_mapping->i_mmap_mutex); |
8f860591f
|
2698 |
spin_lock(&mm->page_table_lock); |
a55164389
|
2699 |
for (; address < end; address += huge_page_size(h)) { |
8f860591f
|
2700 2701 2702 |
ptep = huge_pte_offset(mm, address); if (!ptep) continue; |
39dde65c9
|
2703 2704 |
if (huge_pmd_unshare(mm, &address, ptep)) continue; |
7f2e9525b
|
2705 |
if (!huge_pte_none(huge_ptep_get(ptep))) { |
8f860591f
|
2706 2707 2708 |
pte = huge_ptep_get_and_clear(mm, address, ptep); pte = pte_mkhuge(pte_modify(pte, newprot)); set_huge_pte_at(mm, address, ptep, pte); |
8f860591f
|
2709 2710 2711 |
} } spin_unlock(&mm->page_table_lock); |
3d48ae45e
|
2712 |
mutex_unlock(&vma->vm_file->f_mapping->i_mmap_mutex); |
8f860591f
|
2713 2714 2715 |
flush_tlb_range(vma, start, end); } |
a1e78772d
|
2716 2717 |
int hugetlb_reserve_pages(struct inode *inode, long from, long to, |
5a6fe1259
|
2718 |
struct vm_area_struct *vma, |
ca16d140a
|
2719 |
vm_flags_t vm_flags) |
e4e574b76
|
2720 |
{ |
17c9d12e1
|
2721 |
long ret, chg; |
a55164389
|
2722 |
struct hstate *h = hstate_inode(inode); |
e4e574b76
|
2723 |
|
a1e78772d
|
2724 |
/* |
17c9d12e1
|
2725 2726 2727 2728 |
* Only apply hugepage reservation if asked. At fault time, an * attempt will be made for VM_NORESERVE to allocate a page * and filesystem quota without using reserves */ |
ca16d140a
|
2729 |
if (vm_flags & VM_NORESERVE) |
17c9d12e1
|
2730 2731 2732 |
return 0; /* |
a1e78772d
|
2733 2734 2735 2736 2737 |
* Shared mappings base their reservation on the number of pages that * are already allocated on behalf of the file. Private mappings need * to reserve the full area even if read-only as mprotect() may be * called to make the mapping read-write. Assume !vma is a shm mapping */ |
f83a275db
|
2738 |
if (!vma || vma->vm_flags & VM_MAYSHARE) |
a1e78772d
|
2739 |
chg = region_chg(&inode->i_mapping->private_list, from, to); |
17c9d12e1
|
2740 2741 2742 2743 |
else { struct resv_map *resv_map = resv_map_alloc(); if (!resv_map) return -ENOMEM; |
a1e78772d
|
2744 |
chg = to - from; |
84afd99b8
|
2745 |
|
17c9d12e1
|
2746 2747 2748 |
set_vma_resv_map(vma, resv_map); set_vma_resv_flags(vma, HPAGE_RESV_OWNER); } |
e4e574b76
|
2749 2750 |
if (chg < 0) return chg; |
8a6301127
|
2751 |
|
17c9d12e1
|
2752 |
/* There must be enough filesystem quota for the mapping */ |
90d8b7e61
|
2753 2754 |
if (hugetlb_get_quota(inode->i_mapping, chg)) return -ENOSPC; |
5a6fe1259
|
2755 2756 |
/* |
17c9d12e1
|
2757 2758 |
* Check enough hugepages are available for the reservation. * Hand back the quota if there are not |
5a6fe1259
|
2759 |
*/ |
a55164389
|
2760 |
ret = hugetlb_acct_memory(h, chg); |
68842c9b9
|
2761 2762 |
if (ret < 0) { hugetlb_put_quota(inode->i_mapping, chg); |
a43a8c39b
|
2763 |
return ret; |
68842c9b9
|
2764 |
} |
17c9d12e1
|
2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 |
/* * Account for the reservations made. Shared mappings record regions * that have reservations as they are shared by multiple VMAs. * When the last VMA disappears, the region map says how much * the reservation was and the page cache tells how much of * the reservation was consumed. Private mappings are per-VMA and * only the consumed reservations are tracked. When the VMA * disappears, the original reservation is the VMA size and the * consumed reservations are stored in the map. Hence, nothing * else has to be done for private mappings here */ |
f83a275db
|
2777 |
if (!vma || vma->vm_flags & VM_MAYSHARE) |
a1e78772d
|
2778 |
region_add(&inode->i_mapping->private_list, from, to); |
a43a8c39b
|
2779 2780 2781 2782 2783 |
return 0; } void hugetlb_unreserve_pages(struct inode *inode, long offset, long freed) { |
a55164389
|
2784 |
struct hstate *h = hstate_inode(inode); |
a43a8c39b
|
2785 |
long chg = region_truncate(&inode->i_mapping->private_list, offset); |
45c682a68
|
2786 2787 |
spin_lock(&inode->i_lock); |
e4c6f8bed
|
2788 |
inode->i_blocks -= (blocks_per_huge_page(h) * freed); |
45c682a68
|
2789 |
spin_unlock(&inode->i_lock); |
90d8b7e61
|
2790 |
hugetlb_put_quota(inode->i_mapping, (chg - freed)); |
a55164389
|
2791 |
hugetlb_acct_memory(h, -(chg - freed)); |
a43a8c39b
|
2792 |
} |
93f70f900
|
2793 |
|
d5bd91069
|
2794 |
#ifdef CONFIG_MEMORY_FAILURE |
6de2b1aab
|
2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 |
/* Should be called in hugetlb_lock */ static int is_hugepage_on_freelist(struct page *hpage) { struct page *page; struct page *tmp; struct hstate *h = page_hstate(hpage); int nid = page_to_nid(hpage); list_for_each_entry_safe(page, tmp, &h->hugepage_freelists[nid], lru) if (page == hpage) return 1; return 0; } |
93f70f900
|
2808 2809 2810 2811 |
/* * This function is called from memory failure code. * Assume the caller holds page lock of the head page. */ |
6de2b1aab
|
2812 |
int dequeue_hwpoisoned_huge_page(struct page *hpage) |
93f70f900
|
2813 2814 2815 |
{ struct hstate *h = page_hstate(hpage); int nid = page_to_nid(hpage); |
6de2b1aab
|
2816 |
int ret = -EBUSY; |
93f70f900
|
2817 2818 |
spin_lock(&hugetlb_lock); |
6de2b1aab
|
2819 2820 |
if (is_hugepage_on_freelist(hpage)) { list_del(&hpage->lru); |
8c6c2ecb4
|
2821 |
set_page_refcounted(hpage); |
6de2b1aab
|
2822 2823 2824 2825 |
h->free_huge_pages--; h->free_huge_pages_node[nid]--; ret = 0; } |
93f70f900
|
2826 |
spin_unlock(&hugetlb_lock); |
6de2b1aab
|
2827 |
return ret; |
93f70f900
|
2828 |
} |
6de2b1aab
|
2829 |
#endif |