Blame view
mm/hugetlb.c
98.2 KB
1da177e4c
|
1 2 |
/* * Generic hugetlb support. |
6d49e352a
|
3 |
* (C) Nadia Yvette Chambers, April 2004 |
1da177e4c
|
4 |
*/ |
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> |
3b32123d7
|
16 |
#include <linux/compiler.h> |
aea47ff36
|
17 |
#include <linux/cpuset.h> |
3935baa9b
|
18 |
#include <linux/mutex.h> |
aa888a749
|
19 |
#include <linux/bootmem.h> |
a34378701
|
20 |
#include <linux/sysfs.h> |
5a0e3ad6a
|
21 |
#include <linux/slab.h> |
0fe6e20b9
|
22 |
#include <linux/rmap.h> |
fd6a03edd
|
23 24 |
#include <linux/swap.h> #include <linux/swapops.h> |
c8721bbbd
|
25 |
#include <linux/page-isolation.h> |
8382d914e
|
26 |
#include <linux/jhash.h> |
d6606683a
|
27 |
|
63551ae0f
|
28 29 |
#include <asm/page.h> #include <asm/pgtable.h> |
24669e584
|
30 |
#include <asm/tlb.h> |
63551ae0f
|
31 |
|
24669e584
|
32 |
#include <linux/io.h> |
63551ae0f
|
33 |
#include <linux/hugetlb.h> |
9dd540e23
|
34 |
#include <linux/hugetlb_cgroup.h> |
9a3052306
|
35 |
#include <linux/node.h> |
7835e98b2
|
36 |
#include "internal.h" |
1da177e4c
|
37 |
|
396faf030
|
38 |
unsigned long hugepages_treat_as_movable; |
a55164389
|
39 |
|
c3f38a387
|
40 |
int hugetlb_max_hstate __read_mostly; |
e5ff21594
|
41 42 |
unsigned int default_hstate_idx; struct hstate hstates[HUGE_MAX_HSTATE]; |
53ba51d21
|
43 |
__initdata LIST_HEAD(huge_boot_pages); |
e5ff21594
|
44 45 46 |
/* for command line parsing */ static struct hstate * __initdata parsed_hstate; static unsigned long __initdata default_hstate_max_huge_pages; |
e11bfbfcb
|
47 |
static unsigned long __initdata default_hstate_size; |
e5ff21594
|
48 |
|
3935baa9b
|
49 |
/* |
31caf665e
|
50 51 |
* Protects updates to hugepage_freelists, hugepage_activelist, nr_huge_pages, * free_huge_pages, and surplus_huge_pages. |
3935baa9b
|
52 |
*/ |
c3f38a387
|
53 |
DEFINE_SPINLOCK(hugetlb_lock); |
0bd0f9fb1
|
54 |
|
8382d914e
|
55 56 57 58 59 60 |
/* * Serializes faults on the same logical page. This is used to * prevent spurious OOMs when the hugepage pool is fully utilized. */ static int num_fault_mutexes; static struct mutex *htlb_fault_mutex_table ____cacheline_aligned_in_smp; |
90481622d
|
61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 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 |
static inline void unlock_or_release_subpool(struct hugepage_subpool *spool) { bool free = (spool->count == 0) && (spool->used_hpages == 0); spin_unlock(&spool->lock); /* If no pages are used, and no other handles to the subpool * remain, free the subpool the subpool remain */ if (free) kfree(spool); } struct hugepage_subpool *hugepage_new_subpool(long nr_blocks) { struct hugepage_subpool *spool; spool = kmalloc(sizeof(*spool), GFP_KERNEL); if (!spool) return NULL; spin_lock_init(&spool->lock); spool->count = 1; spool->max_hpages = nr_blocks; spool->used_hpages = 0; return spool; } void hugepage_put_subpool(struct hugepage_subpool *spool) { spin_lock(&spool->lock); BUG_ON(!spool->count); spool->count--; unlock_or_release_subpool(spool); } static int hugepage_subpool_get_pages(struct hugepage_subpool *spool, long delta) { int ret = 0; if (!spool) return 0; spin_lock(&spool->lock); if ((spool->used_hpages + delta) <= spool->max_hpages) { spool->used_hpages += delta; } else { ret = -ENOMEM; } spin_unlock(&spool->lock); return ret; } static void hugepage_subpool_put_pages(struct hugepage_subpool *spool, long delta) { if (!spool) return; spin_lock(&spool->lock); spool->used_hpages -= delta; /* If hugetlbfs_put_super couldn't free spool due to * an outstanding quota reference, free it now. */ unlock_or_release_subpool(spool); } static inline struct hugepage_subpool *subpool_inode(struct inode *inode) { return HUGETLBFS_SB(inode->i_sb)->spool; } static inline struct hugepage_subpool *subpool_vma(struct vm_area_struct *vma) { |
496ad9aa8
|
136 |
return subpool_inode(file_inode(vma->vm_file)); |
90481622d
|
137 |
} |
e7c4b0bfd
|
138 |
/* |
968229048
|
139 140 |
* Region tracking -- allows tracking of reservations and instantiated pages * across the pages in a mapping. |
84afd99b8
|
141 |
* |
7b24d8616
|
142 143 |
* The region data structures are embedded into a resv_map and * protected by a resv_map's lock |
968229048
|
144 145 146 147 148 149 |
*/ struct file_region { struct list_head link; long from; long to; }; |
1406ec9ba
|
150 |
static long region_add(struct resv_map *resv, long f, long t) |
968229048
|
151 |
{ |
1406ec9ba
|
152 |
struct list_head *head = &resv->regions; |
968229048
|
153 |
struct file_region *rg, *nrg, *trg; |
7b24d8616
|
154 |
spin_lock(&resv->lock); |
968229048
|
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 |
/* 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; |
7b24d8616
|
184 |
spin_unlock(&resv->lock); |
968229048
|
185 186 |
return 0; } |
1406ec9ba
|
187 |
static long region_chg(struct resv_map *resv, long f, long t) |
968229048
|
188 |
{ |
1406ec9ba
|
189 |
struct list_head *head = &resv->regions; |
7b24d8616
|
190 |
struct file_region *rg, *nrg = NULL; |
968229048
|
191 |
long chg = 0; |
7b24d8616
|
192 193 |
retry: spin_lock(&resv->lock); |
968229048
|
194 195 196 197 198 199 200 201 202 |
/* 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) { |
7b24d8616
|
203 204 205 206 207 208 209 210 211 212 213 |
if (!nrg) { spin_unlock(&resv->lock); nrg = kmalloc(sizeof(*nrg), GFP_KERNEL); if (!nrg) return -ENOMEM; nrg->from = f; nrg->to = f; INIT_LIST_HEAD(&nrg->link); goto retry; } |
968229048
|
214 |
|
7b24d8616
|
215 216 217 |
list_add(&nrg->link, rg->link.prev); chg = t - f; goto out_nrg; |
968229048
|
218 219 220 221 222 223 224 225 226 227 228 229 |
} /* 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) |
7b24d8616
|
230 |
goto out; |
968229048
|
231 |
|
25985edce
|
232 |
/* We overlap with this area, if it extends further than |
968229048
|
233 234 235 236 237 238 239 240 |
* 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; } |
7b24d8616
|
241 242 243 244 245 246 247 248 |
out: spin_unlock(&resv->lock); /* We already know we raced and no longer need the new region */ kfree(nrg); return chg; out_nrg: spin_unlock(&resv->lock); |
968229048
|
249 250 |
return chg; } |
1406ec9ba
|
251 |
static long region_truncate(struct resv_map *resv, long end) |
968229048
|
252 |
{ |
1406ec9ba
|
253 |
struct list_head *head = &resv->regions; |
968229048
|
254 255 |
struct file_region *rg, *trg; long chg = 0; |
7b24d8616
|
256 |
spin_lock(&resv->lock); |
968229048
|
257 258 259 260 261 |
/* 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) |
7b24d8616
|
262 |
goto out; |
968229048
|
263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 |
/* 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); } |
7b24d8616
|
279 280 281 |
out: spin_unlock(&resv->lock); |
968229048
|
282 283 |
return chg; } |
1406ec9ba
|
284 |
static long region_count(struct resv_map *resv, long f, long t) |
84afd99b8
|
285 |
{ |
1406ec9ba
|
286 |
struct list_head *head = &resv->regions; |
84afd99b8
|
287 288 |
struct file_region *rg; long chg = 0; |
7b24d8616
|
289 |
spin_lock(&resv->lock); |
84afd99b8
|
290 291 |
/* Locate each segment we overlap with, and count that overlap. */ list_for_each_entry(rg, head, link) { |
f2135a4a5
|
292 293 |
long seg_from; long seg_to; |
84afd99b8
|
294 295 296 297 298 299 300 301 302 303 304 |
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; } |
7b24d8616
|
305 |
spin_unlock(&resv->lock); |
84afd99b8
|
306 307 308 |
return chg; } |
968229048
|
309 |
/* |
e7c4b0bfd
|
310 |
* Convert the address within this vma to the page offset within |
e7c4b0bfd
|
311 312 |
* the mapping, in pagecache page units; huge pages here. */ |
a55164389
|
313 314 |
static pgoff_t vma_hugecache_offset(struct hstate *h, struct vm_area_struct *vma, unsigned long address) |
e7c4b0bfd
|
315 |
{ |
a55164389
|
316 317 |
return ((address - vma->vm_start) >> huge_page_shift(h)) + (vma->vm_pgoff >> huge_page_order(h)); |
e7c4b0bfd
|
318 |
} |
0fe6e20b9
|
319 320 321 322 323 |
pgoff_t linear_hugepage_index(struct vm_area_struct *vma, unsigned long address) { return vma_hugecache_offset(hstate_vma(vma), vma, address); } |
84afd99b8
|
324 |
/* |
08fba6998
|
325 326 327 328 329 330 331 332 333 334 335 |
* 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); |
2415cf12e
|
336 |
return 1UL << huge_page_shift(hstate); |
08fba6998
|
337 |
} |
f340ca0f0
|
338 |
EXPORT_SYMBOL_GPL(vma_kernel_pagesize); |
08fba6998
|
339 340 |
/* |
3340289dd
|
341 342 343 344 345 346 347 348 349 350 351 352 353 |
* 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
|
354 355 356 357 358 359 |
* 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
|
360 |
#define HPAGE_RESV_MASK (HPAGE_RESV_OWNER | HPAGE_RESV_UNMAPPED) |
84afd99b8
|
361 |
|
a1e78772d
|
362 363 364 365 366 367 368 369 370 |
/* * 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
|
371 372 373 374 375 376 377 378 379 |
* * 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
|
380 |
*/ |
e7c4b0bfd
|
381 382 383 384 385 386 387 388 389 390 |
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; } |
9119a41e9
|
391 |
struct resv_map *resv_map_alloc(void) |
84afd99b8
|
392 393 394 395 396 397 |
{ struct resv_map *resv_map = kmalloc(sizeof(*resv_map), GFP_KERNEL); if (!resv_map) return NULL; kref_init(&resv_map->refs); |
7b24d8616
|
398 |
spin_lock_init(&resv_map->lock); |
84afd99b8
|
399 400 401 402 |
INIT_LIST_HEAD(&resv_map->regions); return resv_map; } |
9119a41e9
|
403 |
void resv_map_release(struct kref *ref) |
84afd99b8
|
404 405 406 407 |
{ struct resv_map *resv_map = container_of(ref, struct resv_map, refs); /* Clear out any active regions before we release the map. */ |
1406ec9ba
|
408 |
region_truncate(resv_map, 0); |
84afd99b8
|
409 410 |
kfree(resv_map); } |
4e35f4838
|
411 412 413 414 |
static inline struct resv_map *inode_resv_map(struct inode *inode) { return inode->i_mapping->private_data; } |
84afd99b8
|
415 |
static struct resv_map *vma_resv_map(struct vm_area_struct *vma) |
a1e78772d
|
416 |
{ |
81d1b09c6
|
417 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); |
4e35f4838
|
418 419 420 421 422 423 424 |
if (vma->vm_flags & VM_MAYSHARE) { struct address_space *mapping = vma->vm_file->f_mapping; struct inode *inode = mapping->host; return inode_resv_map(inode); } else { |
84afd99b8
|
425 426 |
return (struct resv_map *)(get_vma_private_data(vma) & ~HPAGE_RESV_MASK); |
4e35f4838
|
427 |
} |
a1e78772d
|
428 |
} |
84afd99b8
|
429 |
static void set_vma_resv_map(struct vm_area_struct *vma, struct resv_map *map) |
a1e78772d
|
430 |
{ |
81d1b09c6
|
431 432 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); VM_BUG_ON_VMA(vma->vm_flags & VM_MAYSHARE, vma); |
a1e78772d
|
433 |
|
84afd99b8
|
434 435 |
set_vma_private_data(vma, (get_vma_private_data(vma) & HPAGE_RESV_MASK) | (unsigned long)map); |
04f2cbe35
|
436 437 438 439 |
} static void set_vma_resv_flags(struct vm_area_struct *vma, unsigned long flags) { |
81d1b09c6
|
440 441 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); VM_BUG_ON_VMA(vma->vm_flags & VM_MAYSHARE, vma); |
e7c4b0bfd
|
442 443 |
set_vma_private_data(vma, get_vma_private_data(vma) | flags); |
04f2cbe35
|
444 445 446 447 |
} static int is_vma_resv_set(struct vm_area_struct *vma, unsigned long flag) { |
81d1b09c6
|
448 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); |
e7c4b0bfd
|
449 450 |
return (get_vma_private_data(vma) & flag) != 0; |
a1e78772d
|
451 |
} |
04f2cbe35
|
452 |
/* Reset counters to 0 and clear all HPAGE_RESV_* flags */ |
a1e78772d
|
453 454 |
void reset_vma_resv_huge_pages(struct vm_area_struct *vma) { |
81d1b09c6
|
455 |
VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); |
f83a275db
|
456 |
if (!(vma->vm_flags & VM_MAYSHARE)) |
a1e78772d
|
457 458 459 460 |
vma->vm_private_data = (void *)0; } /* Returns true if the VMA has associated reserve pages */ |
af0ed73e6
|
461 |
static int vma_has_reserves(struct vm_area_struct *vma, long chg) |
a1e78772d
|
462 |
{ |
af0ed73e6
|
463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 |
if (vma->vm_flags & VM_NORESERVE) { /* * This address is already reserved by other process(chg == 0), * so, we should decrement reserved count. Without decrementing, * reserve count remains after releasing inode, because this * allocated page will go into page cache and is regarded as * coming from reserved pool in releasing step. Currently, we * don't have any other solution to deal with this situation * properly, so add work-around here. */ if (vma->vm_flags & VM_MAYSHARE && chg == 0) return 1; else return 0; } |
a63884e92
|
478 479 |
/* Shared mappings always use reserves */ |
f83a275db
|
480 |
if (vma->vm_flags & VM_MAYSHARE) |
7f09ca51e
|
481 |
return 1; |
a63884e92
|
482 483 484 485 486 |
/* * Only the process that called mmap() has reserves for * private mappings. */ |
7f09ca51e
|
487 488 |
if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) return 1; |
a63884e92
|
489 |
|
7f09ca51e
|
490 |
return 0; |
a1e78772d
|
491 |
} |
a55164389
|
492 |
static void enqueue_huge_page(struct hstate *h, struct page *page) |
1da177e4c
|
493 494 |
{ int nid = page_to_nid(page); |
0edaecfab
|
495 |
list_move(&page->lru, &h->hugepage_freelists[nid]); |
a55164389
|
496 497 |
h->free_huge_pages++; h->free_huge_pages_node[nid]++; |
1da177e4c
|
498 |
} |
bf50bab2b
|
499 500 501 |
static struct page *dequeue_huge_page_node(struct hstate *h, int nid) { struct page *page; |
c8721bbbd
|
502 503 504 505 506 507 508 509 |
list_for_each_entry(page, &h->hugepage_freelists[nid], lru) if (!is_migrate_isolate_page(page)) break; /* * if 'non-isolated free hugepage' not found on the list, * the allocation fails. */ if (&h->hugepage_freelists[nid] == &page->lru) |
bf50bab2b
|
510 |
return NULL; |
0edaecfab
|
511 |
list_move(&page->lru, &h->hugepage_activelist); |
a9869b837
|
512 |
set_page_refcounted(page); |
bf50bab2b
|
513 514 515 516 |
h->free_huge_pages--; h->free_huge_pages_node[nid]--; return page; } |
86cdb465c
|
517 518 519 |
/* Movability of hugepages depends on migration support. */ static inline gfp_t htlb_alloc_mask(struct hstate *h) { |
100873d7a
|
520 |
if (hugepages_treat_as_movable || hugepage_migration_supported(h)) |
86cdb465c
|
521 522 523 524 |
return GFP_HIGHUSER_MOVABLE; else return GFP_HIGHUSER; } |
a55164389
|
525 526 |
static struct page *dequeue_huge_page_vma(struct hstate *h, struct vm_area_struct *vma, |
af0ed73e6
|
527 528 |
unsigned long address, int avoid_reserve, long chg) |
1da177e4c
|
529 |
{ |
b1c12cbcd
|
530 |
struct page *page = NULL; |
480eccf9a
|
531 |
struct mempolicy *mpol; |
19770b326
|
532 |
nodemask_t *nodemask; |
c0ff7453b
|
533 |
struct zonelist *zonelist; |
dd1a239f6
|
534 535 |
struct zone *zone; struct zoneref *z; |
cc9a6c877
|
536 |
unsigned int cpuset_mems_cookie; |
1da177e4c
|
537 |
|
a1e78772d
|
538 539 540 541 542 |
/* * 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 */ |
af0ed73e6
|
543 |
if (!vma_has_reserves(vma, chg) && |
a55164389
|
544 |
h->free_huge_pages - h->resv_huge_pages == 0) |
c0ff7453b
|
545 |
goto err; |
a1e78772d
|
546 |
|
04f2cbe35
|
547 |
/* If reserves cannot be used, ensure enough pages are in the pool */ |
a55164389
|
548 |
if (avoid_reserve && h->free_huge_pages - h->resv_huge_pages == 0) |
6eab04a87
|
549 |
goto err; |
04f2cbe35
|
550 |
|
9966c4bbb
|
551 |
retry_cpuset: |
d26914d11
|
552 |
cpuset_mems_cookie = read_mems_allowed_begin(); |
9966c4bbb
|
553 |
zonelist = huge_zonelist(vma, address, |
86cdb465c
|
554 |
htlb_alloc_mask(h), &mpol, &nodemask); |
9966c4bbb
|
555 |
|
19770b326
|
556 557 |
for_each_zone_zonelist_nodemask(zone, z, zonelist, MAX_NR_ZONES - 1, nodemask) { |
344736f29
|
558 |
if (cpuset_zone_allowed(zone, htlb_alloc_mask(h))) { |
bf50bab2b
|
559 560 |
page = dequeue_huge_page_node(h, zone_to_nid(zone)); if (page) { |
af0ed73e6
|
561 562 563 564 |
if (avoid_reserve) break; if (!vma_has_reserves(vma, chg)) break; |
07443a85a
|
565 |
SetPagePrivate(page); |
af0ed73e6
|
566 |
h->resv_huge_pages--; |
bf50bab2b
|
567 568 |
break; } |
3abf7afd4
|
569 |
} |
1da177e4c
|
570 |
} |
cc9a6c877
|
571 |
|
52cd3b074
|
572 |
mpol_cond_put(mpol); |
d26914d11
|
573 |
if (unlikely(!page && read_mems_allowed_retry(cpuset_mems_cookie))) |
cc9a6c877
|
574 |
goto retry_cpuset; |
1da177e4c
|
575 |
return page; |
cc9a6c877
|
576 577 |
err: |
cc9a6c877
|
578 |
return NULL; |
1da177e4c
|
579 |
} |
1cac6f2c0
|
580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 |
/* * 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. */ static int next_node_allowed(int nid, nodemask_t *nodes_allowed) { nid = next_node(nid, *nodes_allowed); if (nid == MAX_NUMNODES) nid = first_node(*nodes_allowed); VM_BUG_ON(nid >= MAX_NUMNODES); return nid; } 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; } /* * 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. */ static int hstate_next_node_to_alloc(struct hstate *h, nodemask_t *nodes_allowed) { 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); return nid; } /* * 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. */ static int hstate_next_node_to_free(struct hstate *h, nodemask_t *nodes_allowed) { 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); return nid; } #define for_each_node_mask_to_alloc(hs, nr_nodes, node, mask) \ for (nr_nodes = nodes_weight(*mask); \ nr_nodes > 0 && \ ((node = hstate_next_node_to_alloc(hs, mask)) || 1); \ nr_nodes--) #define for_each_node_mask_to_free(hs, nr_nodes, node, mask) \ for (nr_nodes = nodes_weight(*mask); \ nr_nodes > 0 && \ ((node = hstate_next_node_to_free(hs, mask)) || 1); \ nr_nodes--) |
944d9fec8
|
652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 |
#if defined(CONFIG_CMA) && defined(CONFIG_X86_64) static void destroy_compound_gigantic_page(struct page *page, unsigned long order) { int i; int nr_pages = 1 << order; struct page *p = page + 1; for (i = 1; i < nr_pages; i++, p = mem_map_next(p, page, i)) { __ClearPageTail(p); set_page_refcounted(p); p->first_page = NULL; } set_compound_order(page, 0); __ClearPageHead(page); } static void free_gigantic_page(struct page *page, unsigned order) { free_contig_range(page_to_pfn(page), 1 << order); } static int __alloc_gigantic_page(unsigned long start_pfn, unsigned long nr_pages) { unsigned long end_pfn = start_pfn + nr_pages; return alloc_contig_range(start_pfn, end_pfn, MIGRATE_MOVABLE); } static bool pfn_range_valid_gigantic(unsigned long start_pfn, unsigned long nr_pages) { unsigned long i, end_pfn = start_pfn + nr_pages; struct page *page; for (i = start_pfn; i < end_pfn; i++) { if (!pfn_valid(i)) return false; page = pfn_to_page(i); if (PageReserved(page)) return false; if (page_count(page) > 0) return false; if (PageHuge(page)) return false; } return true; } static bool zone_spans_last_pfn(const struct zone *zone, unsigned long start_pfn, unsigned long nr_pages) { unsigned long last_pfn = start_pfn + nr_pages - 1; return zone_spans_pfn(zone, last_pfn); } static struct page *alloc_gigantic_page(int nid, unsigned order) { unsigned long nr_pages = 1 << order; unsigned long ret, pfn, flags; struct zone *z; z = NODE_DATA(nid)->node_zones; for (; z - NODE_DATA(nid)->node_zones < MAX_NR_ZONES; z++) { spin_lock_irqsave(&z->lock, flags); pfn = ALIGN(z->zone_start_pfn, nr_pages); while (zone_spans_last_pfn(z, pfn, nr_pages)) { if (pfn_range_valid_gigantic(pfn, nr_pages)) { /* * We release the zone lock here because * alloc_contig_range() will also lock the zone * at some point. If there's an allocation * spinning on this lock, it may win the race * and cause alloc_contig_range() to fail... */ spin_unlock_irqrestore(&z->lock, flags); ret = __alloc_gigantic_page(pfn, nr_pages); if (!ret) return pfn_to_page(pfn); spin_lock_irqsave(&z->lock, flags); } pfn += nr_pages; } spin_unlock_irqrestore(&z->lock, flags); } return NULL; } static void prep_new_huge_page(struct hstate *h, struct page *page, int nid); static void prep_compound_gigantic_page(struct page *page, unsigned long order); static struct page *alloc_fresh_gigantic_page_node(struct hstate *h, int nid) { struct page *page; page = alloc_gigantic_page(nid, huge_page_order(h)); if (page) { prep_compound_gigantic_page(page, huge_page_order(h)); prep_new_huge_page(h, page, nid); } return page; } static int alloc_fresh_gigantic_page(struct hstate *h, nodemask_t *nodes_allowed) { struct page *page = NULL; int nr_nodes, node; for_each_node_mask_to_alloc(h, nr_nodes, node, nodes_allowed) { page = alloc_fresh_gigantic_page_node(h, node); if (page) return 1; } return 0; } static inline bool gigantic_page_supported(void) { return true; } #else static inline bool gigantic_page_supported(void) { return false; } static inline void free_gigantic_page(struct page *page, unsigned order) { } static inline void destroy_compound_gigantic_page(struct page *page, unsigned long order) { } static inline int alloc_fresh_gigantic_page(struct hstate *h, nodemask_t *nodes_allowed) { return 0; } #endif |
a55164389
|
789 |
static void update_and_free_page(struct hstate *h, struct page *page) |
6af2acb66
|
790 791 |
{ int i; |
a55164389
|
792 |
|
944d9fec8
|
793 794 |
if (hstate_is_gigantic(h) && !gigantic_page_supported()) return; |
18229df5b
|
795 |
|
a55164389
|
796 797 798 |
h->nr_huge_pages--; h->nr_huge_pages_node[page_to_nid(page)]--; for (i = 0; i < pages_per_huge_page(h); i++) { |
32f84528f
|
799 800 |
page[i].flags &= ~(1 << PG_locked | 1 << PG_error | 1 << PG_referenced | 1 << PG_dirty | |
a7407a27c
|
801 802 |
1 << PG_active | 1 << PG_private | 1 << PG_writeback); |
6af2acb66
|
803 |
} |
309381fea
|
804 |
VM_BUG_ON_PAGE(hugetlb_cgroup_from_page(page), page); |
6af2acb66
|
805 806 |
set_compound_page_dtor(page, NULL); set_page_refcounted(page); |
944d9fec8
|
807 808 809 810 811 812 813 |
if (hstate_is_gigantic(h)) { destroy_compound_gigantic_page(page, huge_page_order(h)); free_gigantic_page(page, huge_page_order(h)); } else { arch_release_hugepage(page); __free_pages(page, huge_page_order(h)); } |
6af2acb66
|
814 |
} |
e5ff21594
|
815 816 817 818 819 820 821 822 823 824 |
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; } |
8f1d26d0e
|
825 |
void free_huge_page(struct page *page) |
27a85ef1b
|
826 |
{ |
a55164389
|
827 828 829 830 |
/* * Can't pass hstate in here because it is called from the * compound page destructor. */ |
e5ff21594
|
831 |
struct hstate *h = page_hstate(page); |
7893d1d50
|
832 |
int nid = page_to_nid(page); |
90481622d
|
833 834 |
struct hugepage_subpool *spool = (struct hugepage_subpool *)page_private(page); |
07443a85a
|
835 |
bool restore_reserve; |
27a85ef1b
|
836 |
|
e5df70ab1
|
837 |
set_page_private(page, 0); |
23be7468e
|
838 |
page->mapping = NULL; |
7893d1d50
|
839 |
BUG_ON(page_count(page)); |
0fe6e20b9
|
840 |
BUG_ON(page_mapcount(page)); |
07443a85a
|
841 |
restore_reserve = PagePrivate(page); |
16c794b4f
|
842 |
ClearPagePrivate(page); |
27a85ef1b
|
843 844 |
spin_lock(&hugetlb_lock); |
6d76dcf40
|
845 846 |
hugetlb_cgroup_uncharge_page(hstate_index(h), pages_per_huge_page(h), page); |
07443a85a
|
847 848 |
if (restore_reserve) h->resv_huge_pages++; |
944d9fec8
|
849 |
if (h->surplus_huge_pages_node[nid]) { |
0edaecfab
|
850 851 |
/* remove the page from active list */ list_del(&page->lru); |
a55164389
|
852 853 854 |
update_and_free_page(h, page); h->surplus_huge_pages--; h->surplus_huge_pages_node[nid]--; |
7893d1d50
|
855 |
} else { |
5d3a551c2
|
856 |
arch_clear_hugepage_flags(page); |
a55164389
|
857 |
enqueue_huge_page(h, page); |
7893d1d50
|
858 |
} |
27a85ef1b
|
859 |
spin_unlock(&hugetlb_lock); |
90481622d
|
860 |
hugepage_subpool_put_pages(spool, 1); |
27a85ef1b
|
861 |
} |
a55164389
|
862 |
static void prep_new_huge_page(struct hstate *h, struct page *page, int nid) |
b7ba30c67
|
863 |
{ |
0edaecfab
|
864 |
INIT_LIST_HEAD(&page->lru); |
b7ba30c67
|
865 866 |
set_compound_page_dtor(page, free_huge_page); spin_lock(&hugetlb_lock); |
9dd540e23
|
867 |
set_hugetlb_cgroup(page, NULL); |
a55164389
|
868 869 |
h->nr_huge_pages++; h->nr_huge_pages_node[nid]++; |
b7ba30c67
|
870 871 872 |
spin_unlock(&hugetlb_lock); put_page(page); /* free it into the hugepage allocator */ } |
2906dd528
|
873 |
static void prep_compound_gigantic_page(struct page *page, unsigned long order) |
20a0307c0
|
874 875 876 877 878 879 880 881 |
{ 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); |
ef5a22be2
|
882 |
__ClearPageReserved(page); |
20a0307c0
|
883 884 |
for (i = 1; i < nr_pages; i++, p = mem_map_next(p, page, i)) { __SetPageTail(p); |
ef5a22be2
|
885 886 887 888 889 890 891 892 893 894 895 896 897 |
/* * For gigantic hugepages allocated through bootmem at * boot, it's safer to be consistent with the not-gigantic * hugepages and clear the PG_reserved bit from all tail pages * too. Otherwse drivers using get_user_pages() to access tail * pages may get the reference counting wrong if they see * PG_reserved set on a tail page (despite the head page not * having PG_reserved set). Enforcing this consistency between * head and tail pages allows drivers to optimize away a check * on the head page when they need know if put_page() is needed * after get_user_pages(). */ __ClearPageReserved(p); |
58a84aa92
|
898 |
set_page_count(p, 0); |
20a0307c0
|
899 900 901 |
p->first_page = page; } } |
7795912c2
|
902 903 904 905 906 |
/* * PageHuge() only returns true for hugetlbfs pages, but not for normal or * transparent huge pages. See the PageTransHuge() documentation for more * details. */ |
20a0307c0
|
907 908 |
int PageHuge(struct page *page) { |
20a0307c0
|
909 910 911 912 |
if (!PageCompound(page)) return 0; page = compound_head(page); |
758f66a29
|
913 |
return get_compound_page_dtor(page) == free_huge_page; |
20a0307c0
|
914 |
} |
43131e141
|
915 |
EXPORT_SYMBOL_GPL(PageHuge); |
27c73ae75
|
916 917 918 919 920 921 |
/* * PageHeadHuge() only returns true for hugetlbfs head page, but not for * normal or transparent huge pages. */ int PageHeadHuge(struct page *page_head) { |
27c73ae75
|
922 923 |
if (!PageHead(page_head)) return 0; |
758f66a29
|
924 |
return get_compound_page_dtor(page_head) == free_huge_page; |
27c73ae75
|
925 |
} |
27c73ae75
|
926 |
|
13d60f4b6
|
927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 |
pgoff_t __basepage_index(struct page *page) { struct page *page_head = compound_head(page); pgoff_t index = page_index(page_head); unsigned long compound_idx; if (!PageHuge(page_head)) return page_index(page); if (compound_order(page_head) >= MAX_ORDER) compound_idx = page_to_pfn(page) - page_to_pfn(page_head); else compound_idx = page - page_head; return (index << compound_order(page_head)) + compound_idx; } |
a55164389
|
943 |
static struct page *alloc_fresh_huge_page_node(struct hstate *h, int nid) |
1da177e4c
|
944 |
{ |
1da177e4c
|
945 |
struct page *page; |
f96efd585
|
946 |
|
6484eb3e2
|
947 |
page = alloc_pages_exact_node(nid, |
86cdb465c
|
948 |
htlb_alloc_mask(h)|__GFP_COMP|__GFP_THISNODE| |
551883ae8
|
949 |
__GFP_REPEAT|__GFP_NOWARN, |
a55164389
|
950 |
huge_page_order(h)); |
1da177e4c
|
951 |
if (page) { |
7f2e9525b
|
952 |
if (arch_prepare_hugepage(page)) { |
caff3a2c3
|
953 |
__free_pages(page, huge_page_order(h)); |
7b8ee84d8
|
954 |
return NULL; |
7f2e9525b
|
955 |
} |
a55164389
|
956 |
prep_new_huge_page(h, page, nid); |
1da177e4c
|
957 |
} |
63b4613c3
|
958 959 960 |
return page; } |
b22610268
|
961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 |
static int alloc_fresh_huge_page(struct hstate *h, nodemask_t *nodes_allowed) { struct page *page; int nr_nodes, node; int ret = 0; for_each_node_mask_to_alloc(h, nr_nodes, node, nodes_allowed) { page = alloc_fresh_huge_page_node(h, node); if (page) { ret = 1; break; } } if (ret) count_vm_event(HTLB_BUDDY_PGALLOC); else count_vm_event(HTLB_BUDDY_PGALLOC_FAIL); return ret; } |
e8c5c8249
|
982 983 984 985 986 987 |
/* * 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
|
988 989 |
static int free_pool_huge_page(struct hstate *h, nodemask_t *nodes_allowed, bool acct_surplus) |
e8c5c8249
|
990 |
{ |
b22610268
|
991 |
int nr_nodes, node; |
e8c5c8249
|
992 |
int ret = 0; |
b22610268
|
993 |
for_each_node_mask_to_free(h, nr_nodes, node, nodes_allowed) { |
685f34570
|
994 995 996 997 |
/* * If we're returning unused surplus pages, only examine * nodes with surplus pages. */ |
b22610268
|
998 999 |
if ((!acct_surplus || h->surplus_huge_pages_node[node]) && !list_empty(&h->hugepage_freelists[node])) { |
e8c5c8249
|
1000 |
struct page *page = |
b22610268
|
1001 |
list_entry(h->hugepage_freelists[node].next, |
e8c5c8249
|
1002 1003 1004 |
struct page, lru); list_del(&page->lru); h->free_huge_pages--; |
b22610268
|
1005 |
h->free_huge_pages_node[node]--; |
685f34570
|
1006 1007 |
if (acct_surplus) { h->surplus_huge_pages--; |
b22610268
|
1008 |
h->surplus_huge_pages_node[node]--; |
685f34570
|
1009 |
} |
e8c5c8249
|
1010 1011 |
update_and_free_page(h, page); ret = 1; |
9a76db099
|
1012 |
break; |
e8c5c8249
|
1013 |
} |
b22610268
|
1014 |
} |
e8c5c8249
|
1015 1016 1017 |
return ret; } |
c8721bbbd
|
1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 |
/* * Dissolve a given free hugepage into free buddy pages. This function does * nothing for in-use (including surplus) hugepages. */ static void dissolve_free_huge_page(struct page *page) { spin_lock(&hugetlb_lock); if (PageHuge(page) && !page_count(page)) { struct hstate *h = page_hstate(page); int nid = page_to_nid(page); list_del(&page->lru); h->free_huge_pages--; h->free_huge_pages_node[nid]--; update_and_free_page(h, page); } spin_unlock(&hugetlb_lock); } /* * Dissolve free hugepages in a given pfn range. Used by memory hotplug to * make specified memory blocks removable from the system. * Note that start_pfn should aligned with (minimum) hugepage size. */ void dissolve_free_huge_pages(unsigned long start_pfn, unsigned long end_pfn) { unsigned int order = 8 * sizeof(void *); unsigned long pfn; struct hstate *h; |
d01776393
|
1046 1047 |
if (!hugepages_supported()) return; |
c8721bbbd
|
1048 1049 1050 1051 1052 1053 1054 1055 |
/* Set scan step to minimum hugepage size */ for_each_hstate(h) if (order > huge_page_order(h)) order = huge_page_order(h); VM_BUG_ON(!IS_ALIGNED(start_pfn, 1 << order)); for (pfn = start_pfn; pfn < end_pfn; pfn += 1 << order) dissolve_free_huge_page(pfn_to_page(pfn)); } |
bf50bab2b
|
1056 |
static struct page *alloc_buddy_huge_page(struct hstate *h, int nid) |
7893d1d50
|
1057 1058 |
{ struct page *page; |
bf50bab2b
|
1059 |
unsigned int r_nid; |
7893d1d50
|
1060 |
|
bae7f4ae1
|
1061 |
if (hstate_is_gigantic(h)) |
aa888a749
|
1062 |
return NULL; |
d1c3fb1f8
|
1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 |
/* * 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
|
1087 |
if (h->surplus_huge_pages >= h->nr_overcommit_huge_pages) { |
d1c3fb1f8
|
1088 1089 1090 |
spin_unlock(&hugetlb_lock); return NULL; } else { |
a55164389
|
1091 1092 |
h->nr_huge_pages++; h->surplus_huge_pages++; |
d1c3fb1f8
|
1093 1094 |
} spin_unlock(&hugetlb_lock); |
bf50bab2b
|
1095 |
if (nid == NUMA_NO_NODE) |
86cdb465c
|
1096 |
page = alloc_pages(htlb_alloc_mask(h)|__GFP_COMP| |
bf50bab2b
|
1097 1098 1099 1100 |
__GFP_REPEAT|__GFP_NOWARN, huge_page_order(h)); else page = alloc_pages_exact_node(nid, |
86cdb465c
|
1101 |
htlb_alloc_mask(h)|__GFP_COMP|__GFP_THISNODE| |
bf50bab2b
|
1102 |
__GFP_REPEAT|__GFP_NOWARN, huge_page_order(h)); |
d1c3fb1f8
|
1103 |
|
caff3a2c3
|
1104 1105 |
if (page && arch_prepare_hugepage(page)) { __free_pages(page, huge_page_order(h)); |
ea5768c74
|
1106 |
page = NULL; |
caff3a2c3
|
1107 |
} |
d1c3fb1f8
|
1108 |
spin_lock(&hugetlb_lock); |
7893d1d50
|
1109 |
if (page) { |
0edaecfab
|
1110 |
INIT_LIST_HEAD(&page->lru); |
bf50bab2b
|
1111 |
r_nid = page_to_nid(page); |
7893d1d50
|
1112 |
set_compound_page_dtor(page, free_huge_page); |
9dd540e23
|
1113 |
set_hugetlb_cgroup(page, NULL); |
d1c3fb1f8
|
1114 1115 1116 |
/* * We incremented the global counters already */ |
bf50bab2b
|
1117 1118 |
h->nr_huge_pages_node[r_nid]++; h->surplus_huge_pages_node[r_nid]++; |
3b1163006
|
1119 |
__count_vm_event(HTLB_BUDDY_PGALLOC); |
d1c3fb1f8
|
1120 |
} else { |
a55164389
|
1121 1122 |
h->nr_huge_pages--; h->surplus_huge_pages--; |
3b1163006
|
1123 |
__count_vm_event(HTLB_BUDDY_PGALLOC_FAIL); |
7893d1d50
|
1124 |
} |
d1c3fb1f8
|
1125 |
spin_unlock(&hugetlb_lock); |
7893d1d50
|
1126 1127 1128 |
return page; } |
e4e574b76
|
1129 |
/* |
bf50bab2b
|
1130 1131 1132 1133 1134 1135 |
* 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) { |
4ef918480
|
1136 |
struct page *page = NULL; |
bf50bab2b
|
1137 1138 |
spin_lock(&hugetlb_lock); |
4ef918480
|
1139 1140 |
if (h->free_huge_pages - h->resv_huge_pages > 0) page = dequeue_huge_page_node(h, nid); |
bf50bab2b
|
1141 |
spin_unlock(&hugetlb_lock); |
94ae8ba71
|
1142 |
if (!page) |
bf50bab2b
|
1143 1144 1145 1146 1147 1148 |
page = alloc_buddy_huge_page(h, nid); return page; } /* |
25985edce
|
1149 |
* Increase the hugetlb pool such that it can accommodate a reservation |
e4e574b76
|
1150 1151 |
* of size 'delta'. */ |
a55164389
|
1152 |
static int gather_surplus_pages(struct hstate *h, int delta) |
e4e574b76
|
1153 1154 1155 1156 1157 |
{ struct list_head surplus_list; struct page *page, *tmp; int ret, i; int needed, allocated; |
28073b02b
|
1158 |
bool alloc_ok = true; |
e4e574b76
|
1159 |
|
a55164389
|
1160 |
needed = (h->resv_huge_pages + delta) - h->free_huge_pages; |
ac09b3a15
|
1161 |
if (needed <= 0) { |
a55164389
|
1162 |
h->resv_huge_pages += delta; |
e4e574b76
|
1163 |
return 0; |
ac09b3a15
|
1164 |
} |
e4e574b76
|
1165 1166 1167 1168 1169 1170 1171 1172 |
allocated = 0; INIT_LIST_HEAD(&surplus_list); ret = -ENOMEM; retry: spin_unlock(&hugetlb_lock); for (i = 0; i < needed; i++) { |
bf50bab2b
|
1173 |
page = alloc_buddy_huge_page(h, NUMA_NO_NODE); |
28073b02b
|
1174 1175 1176 1177 |
if (!page) { alloc_ok = false; break; } |
e4e574b76
|
1178 1179 |
list_add(&page->lru, &surplus_list); } |
28073b02b
|
1180 |
allocated += i; |
e4e574b76
|
1181 1182 1183 1184 1185 1186 |
/* * 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
|
1187 1188 |
needed = (h->resv_huge_pages + delta) - (h->free_huge_pages + allocated); |
28073b02b
|
1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 |
if (needed > 0) { if (alloc_ok) goto retry; /* * We were not able to allocate enough pages to * satisfy the entire reservation so we free what * we've allocated so far. */ goto free; } |
e4e574b76
|
1199 1200 |
/* * The surplus_list now contains _at_least_ the number of extra pages |
25985edce
|
1201 |
* needed to accommodate the reservation. Add the appropriate number |
e4e574b76
|
1202 |
* of pages to the hugetlb pool and free the extras back to the buddy |
ac09b3a15
|
1203 1204 1205 |
* 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
|
1206 1207 |
*/ needed += allocated; |
a55164389
|
1208 |
h->resv_huge_pages += delta; |
e4e574b76
|
1209 |
ret = 0; |
a9869b837
|
1210 |
|
19fc3f0ac
|
1211 |
/* Free the needed pages to the hugetlb pool */ |
e4e574b76
|
1212 |
list_for_each_entry_safe(page, tmp, &surplus_list, lru) { |
19fc3f0ac
|
1213 1214 |
if ((--needed) < 0) break; |
a9869b837
|
1215 1216 1217 1218 1219 |
/* * This page is now managed by the hugetlb allocator and has * no users -- drop the buddy allocator's reference. */ put_page_testzero(page); |
309381fea
|
1220 |
VM_BUG_ON_PAGE(page_count(page), page); |
a55164389
|
1221 |
enqueue_huge_page(h, page); |
19fc3f0ac
|
1222 |
} |
28073b02b
|
1223 |
free: |
b0365c8d0
|
1224 |
spin_unlock(&hugetlb_lock); |
19fc3f0ac
|
1225 1226 |
/* Free unnecessary surplus pages to the buddy allocator */ |
c0d934ba2
|
1227 1228 |
list_for_each_entry_safe(page, tmp, &surplus_list, lru) put_page(page); |
a9869b837
|
1229 |
spin_lock(&hugetlb_lock); |
e4e574b76
|
1230 1231 1232 1233 1234 1235 1236 1237 |
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
|
1238 |
* Called with hugetlb_lock held. |
e4e574b76
|
1239 |
*/ |
a55164389
|
1240 1241 |
static void return_unused_surplus_pages(struct hstate *h, unsigned long unused_resv_pages) |
e4e574b76
|
1242 |
{ |
e4e574b76
|
1243 |
unsigned long nr_pages; |
ac09b3a15
|
1244 |
/* Uncommit the reservation */ |
a55164389
|
1245 |
h->resv_huge_pages -= unused_resv_pages; |
ac09b3a15
|
1246 |
|
aa888a749
|
1247 |
/* Cannot return gigantic pages currently */ |
bae7f4ae1
|
1248 |
if (hstate_is_gigantic(h)) |
aa888a749
|
1249 |
return; |
a55164389
|
1250 |
nr_pages = min(unused_resv_pages, h->surplus_huge_pages); |
e4e574b76
|
1251 |
|
685f34570
|
1252 1253 |
/* * We want to release as many surplus pages as possible, spread |
9b5e5d0fd
|
1254 1255 1256 1257 1258 |
* 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
|
1259 1260 |
*/ while (nr_pages--) { |
8cebfcd07
|
1261 |
if (!free_pool_huge_page(h, &node_states[N_MEMORY], 1)) |
685f34570
|
1262 |
break; |
7848a4bf5
|
1263 |
cond_resched_lock(&hugetlb_lock); |
e4e574b76
|
1264 1265 |
} } |
c37f9fb11
|
1266 1267 1268 |
/* * Determine if the huge page at addr within the vma has an associated * reservation. Where it does not we will need to logically increase |
90481622d
|
1269 1270 1271 1272 1273 1274 |
* reservation and actually increase subpool usage 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 allocated from the subpool and instantiated the change should * be committed via vma_commit_reservation. No action is required on * failure. |
c37f9fb11
|
1275 |
*/ |
e2f17d945
|
1276 |
static long vma_needs_reservation(struct hstate *h, |
a55164389
|
1277 |
struct vm_area_struct *vma, unsigned long addr) |
c37f9fb11
|
1278 |
{ |
4e35f4838
|
1279 1280 1281 |
struct resv_map *resv; pgoff_t idx; long chg; |
c37f9fb11
|
1282 |
|
4e35f4838
|
1283 1284 |
resv = vma_resv_map(vma); if (!resv) |
84afd99b8
|
1285 |
return 1; |
c37f9fb11
|
1286 |
|
4e35f4838
|
1287 1288 |
idx = vma_hugecache_offset(h, vma, addr); chg = region_chg(resv, idx, idx + 1); |
84afd99b8
|
1289 |
|
4e35f4838
|
1290 1291 1292 1293 |
if (vma->vm_flags & VM_MAYSHARE) return chg; else return chg < 0 ? chg : 0; |
c37f9fb11
|
1294 |
} |
a55164389
|
1295 1296 |
static void vma_commit_reservation(struct hstate *h, struct vm_area_struct *vma, unsigned long addr) |
c37f9fb11
|
1297 |
{ |
4e35f4838
|
1298 1299 |
struct resv_map *resv; pgoff_t idx; |
84afd99b8
|
1300 |
|
4e35f4838
|
1301 1302 1303 |
resv = vma_resv_map(vma); if (!resv) return; |
84afd99b8
|
1304 |
|
4e35f4838
|
1305 1306 |
idx = vma_hugecache_offset(h, vma, addr); region_add(resv, idx, idx + 1); |
c37f9fb11
|
1307 |
} |
a1e78772d
|
1308 |
static struct page *alloc_huge_page(struct vm_area_struct *vma, |
04f2cbe35
|
1309 |
unsigned long addr, int avoid_reserve) |
1da177e4c
|
1310 |
{ |
90481622d
|
1311 |
struct hugepage_subpool *spool = subpool_vma(vma); |
a55164389
|
1312 |
struct hstate *h = hstate_vma(vma); |
348ea204c
|
1313 |
struct page *page; |
e2f17d945
|
1314 |
long chg; |
6d76dcf40
|
1315 1316 |
int ret, idx; struct hugetlb_cgroup *h_cg; |
a1e78772d
|
1317 |
|
6d76dcf40
|
1318 |
idx = hstate_index(h); |
a1e78772d
|
1319 |
/* |
90481622d
|
1320 1321 1322 1323 1324 1325 |
* Processes that did not create the mapping will have no * reserves and will not have accounted against subpool * limit. Check that the subpool limit can be made before * satisfying the allocation MAP_NORESERVE mappings may also * need pages and subpool limit allocated allocated if no reserve * mapping overlaps. |
a1e78772d
|
1326 |
*/ |
a55164389
|
1327 |
chg = vma_needs_reservation(h, vma, addr); |
c37f9fb11
|
1328 |
if (chg < 0) |
76dcee75c
|
1329 |
return ERR_PTR(-ENOMEM); |
8bb3f12e7
|
1330 1331 |
if (chg || avoid_reserve) if (hugepage_subpool_get_pages(spool, 1)) |
76dcee75c
|
1332 |
return ERR_PTR(-ENOSPC); |
1da177e4c
|
1333 |
|
6d76dcf40
|
1334 |
ret = hugetlb_cgroup_charge_cgroup(idx, pages_per_huge_page(h), &h_cg); |
8f34af6f9
|
1335 1336 |
if (ret) goto out_subpool_put; |
1da177e4c
|
1337 |
spin_lock(&hugetlb_lock); |
af0ed73e6
|
1338 |
page = dequeue_huge_page_vma(h, vma, addr, avoid_reserve, chg); |
81a6fcae3
|
1339 |
if (!page) { |
94ae8ba71
|
1340 |
spin_unlock(&hugetlb_lock); |
bf50bab2b
|
1341 |
page = alloc_buddy_huge_page(h, NUMA_NO_NODE); |
8f34af6f9
|
1342 1343 |
if (!page) goto out_uncharge_cgroup; |
79dbb2368
|
1344 1345 |
spin_lock(&hugetlb_lock); list_move(&page->lru, &h->hugepage_activelist); |
81a6fcae3
|
1346 |
/* Fall through */ |
68842c9b9
|
1347 |
} |
81a6fcae3
|
1348 1349 |
hugetlb_cgroup_commit_charge(idx, pages_per_huge_page(h), h_cg, page); spin_unlock(&hugetlb_lock); |
348ea204c
|
1350 |
|
90481622d
|
1351 |
set_page_private(page, (unsigned long)spool); |
90d8b7e61
|
1352 |
|
a55164389
|
1353 |
vma_commit_reservation(h, vma, addr); |
90d8b7e61
|
1354 |
return page; |
8f34af6f9
|
1355 1356 1357 1358 1359 1360 1361 |
out_uncharge_cgroup: hugetlb_cgroup_uncharge_cgroup(idx, pages_per_huge_page(h), h_cg); out_subpool_put: if (chg || avoid_reserve) hugepage_subpool_put_pages(spool, 1); return ERR_PTR(-ENOSPC); |
b45b5bd65
|
1362 |
} |
74060e4d7
|
1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 |
/* * alloc_huge_page()'s wrapper which simply returns the page if allocation * succeeds, otherwise NULL. This function is called from new_vma_page(), * where no ERR_VALUE is expected to be returned. */ struct page *alloc_huge_page_noerr(struct vm_area_struct *vma, unsigned long addr, int avoid_reserve) { struct page *page = alloc_huge_page(vma, addr, avoid_reserve); if (IS_ERR(page)) page = NULL; return page; } |
91f47662d
|
1376 |
int __weak alloc_bootmem_huge_page(struct hstate *h) |
aa888a749
|
1377 1378 |
{ struct huge_bootmem_page *m; |
b22610268
|
1379 |
int nr_nodes, node; |
aa888a749
|
1380 |
|
b22610268
|
1381 |
for_each_node_mask_to_alloc(h, nr_nodes, node, &node_states[N_MEMORY]) { |
aa888a749
|
1382 |
void *addr; |
8b89a1169
|
1383 1384 1385 |
addr = memblock_virt_alloc_try_nid_nopanic( huge_page_size(h), huge_page_size(h), 0, BOOTMEM_ALLOC_ACCESSIBLE, node); |
aa888a749
|
1386 1387 1388 1389 1390 1391 1392 |
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
|
1393 |
goto found; |
aa888a749
|
1394 |
} |
aa888a749
|
1395 1396 1397 1398 |
} return 0; found: |
df994ead5
|
1399 |
BUG_ON(!IS_ALIGNED(virt_to_phys(m), huge_page_size(h))); |
aa888a749
|
1400 1401 1402 1403 1404 |
/* 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; } |
f412c97ab
|
1405 |
static void __init prep_compound_huge_page(struct page *page, int order) |
18229df5b
|
1406 1407 1408 1409 1410 1411 |
{ if (unlikely(order > (MAX_ORDER - 1))) prep_compound_gigantic_page(page, order); else prep_compound_page(page, order); } |
aa888a749
|
1412 1413 1414 1415 1416 1417 |
/* 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) { |
aa888a749
|
1418 |
struct hstate *h = m->hstate; |
ee8f248d2
|
1419 1420 1421 1422 |
struct page *page; #ifdef CONFIG_HIGHMEM page = pfn_to_page(m->phys >> PAGE_SHIFT); |
8b89a1169
|
1423 1424 |
memblock_free_late(__pa(m), sizeof(struct huge_bootmem_page)); |
ee8f248d2
|
1425 1426 1427 |
#else page = virt_to_page(m); #endif |
aa888a749
|
1428 |
WARN_ON(page_count(page) != 1); |
18229df5b
|
1429 |
prep_compound_huge_page(page, h->order); |
ef5a22be2
|
1430 |
WARN_ON(PageReserved(page)); |
aa888a749
|
1431 |
prep_new_huge_page(h, page, page_to_nid(page)); |
b0320c7b7
|
1432 1433 1434 1435 1436 1437 |
/* * 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. */ |
bae7f4ae1
|
1438 |
if (hstate_is_gigantic(h)) |
3dcc0571c
|
1439 |
adjust_managed_page_count(page, 1 << h->order); |
aa888a749
|
1440 1441 |
} } |
8faa8b077
|
1442 |
static void __init hugetlb_hstate_alloc_pages(struct hstate *h) |
1da177e4c
|
1443 1444 |
{ unsigned long i; |
a55164389
|
1445 |
|
e5ff21594
|
1446 |
for (i = 0; i < h->max_huge_pages; ++i) { |
bae7f4ae1
|
1447 |
if (hstate_is_gigantic(h)) { |
aa888a749
|
1448 1449 |
if (!alloc_bootmem_huge_page(h)) break; |
9b5e5d0fd
|
1450 |
} else if (!alloc_fresh_huge_page(h, |
8cebfcd07
|
1451 |
&node_states[N_MEMORY])) |
1da177e4c
|
1452 |
break; |
1da177e4c
|
1453 |
} |
8faa8b077
|
1454 |
h->max_huge_pages = i; |
e5ff21594
|
1455 1456 1457 1458 1459 1460 1461 |
} static void __init hugetlb_init_hstates(void) { struct hstate *h; for_each_hstate(h) { |
8faa8b077
|
1462 |
/* oversize hugepages were init'ed in early boot */ |
bae7f4ae1
|
1463 |
if (!hstate_is_gigantic(h)) |
8faa8b077
|
1464 |
hugetlb_hstate_alloc_pages(h); |
e5ff21594
|
1465 1466 |
} } |
4abd32dba
|
1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 |
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
|
1477 1478 1479 1480 1481 |
static void __init report_hugepages(void) { struct hstate *h; for_each_hstate(h) { |
4abd32dba
|
1482 |
char buf[32]; |
ffb22af5b
|
1483 1484 |
pr_info("HugeTLB registered %s page size, pre-allocated %ld pages ", |
4abd32dba
|
1485 1486 |
memfmt(buf, huge_page_size(h)), h->free_huge_pages); |
e5ff21594
|
1487 1488 |
} } |
1da177e4c
|
1489 |
#ifdef CONFIG_HIGHMEM |
6ae11b278
|
1490 1491 |
static void try_to_free_low(struct hstate *h, unsigned long count, nodemask_t *nodes_allowed) |
1da177e4c
|
1492 |
{ |
4415cc8df
|
1493 |
int i; |
bae7f4ae1
|
1494 |
if (hstate_is_gigantic(h)) |
aa888a749
|
1495 |
return; |
6ae11b278
|
1496 |
for_each_node_mask(i, *nodes_allowed) { |
1da177e4c
|
1497 |
struct page *page, *next; |
a55164389
|
1498 1499 1500 |
struct list_head *freel = &h->hugepage_freelists[i]; list_for_each_entry_safe(page, next, freel, lru) { if (count >= h->nr_huge_pages) |
6b0c880df
|
1501 |
return; |
1da177e4c
|
1502 1503 1504 |
if (PageHighMem(page)) continue; list_del(&page->lru); |
e5ff21594
|
1505 |
update_and_free_page(h, page); |
a55164389
|
1506 1507 |
h->free_huge_pages--; h->free_huge_pages_node[page_to_nid(page)]--; |
1da177e4c
|
1508 1509 1510 1511 |
} } } #else |
6ae11b278
|
1512 1513 |
static inline void try_to_free_low(struct hstate *h, unsigned long count, nodemask_t *nodes_allowed) |
1da177e4c
|
1514 1515 1516 |
{ } #endif |
20a0307c0
|
1517 1518 1519 1520 1521 |
/* * 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
|
1522 1523 |
static int adjust_pool_surplus(struct hstate *h, nodemask_t *nodes_allowed, int delta) |
20a0307c0
|
1524 |
{ |
b22610268
|
1525 |
int nr_nodes, node; |
20a0307c0
|
1526 1527 |
VM_BUG_ON(delta != -1 && delta != 1); |
20a0307c0
|
1528 |
|
b22610268
|
1529 1530 1531 1532 |
if (delta < 0) { for_each_node_mask_to_alloc(h, nr_nodes, node, nodes_allowed) { if (h->surplus_huge_pages_node[node]) goto found; |
e8c5c8249
|
1533 |
} |
b22610268
|
1534 1535 1536 1537 1538 |
} else { for_each_node_mask_to_free(h, nr_nodes, node, nodes_allowed) { if (h->surplus_huge_pages_node[node] < h->nr_huge_pages_node[node]) goto found; |
e8c5c8249
|
1539 |
} |
b22610268
|
1540 1541 |
} return 0; |
20a0307c0
|
1542 |
|
b22610268
|
1543 1544 1545 1546 |
found: h->surplus_huge_pages += delta; h->surplus_huge_pages_node[node] += delta; return 1; |
20a0307c0
|
1547 |
} |
a55164389
|
1548 |
#define persistent_huge_pages(h) (h->nr_huge_pages - h->surplus_huge_pages) |
6ae11b278
|
1549 1550 |
static unsigned long set_max_huge_pages(struct hstate *h, unsigned long count, nodemask_t *nodes_allowed) |
1da177e4c
|
1551 |
{ |
7893d1d50
|
1552 |
unsigned long min_count, ret; |
1da177e4c
|
1553 |
|
944d9fec8
|
1554 |
if (hstate_is_gigantic(h) && !gigantic_page_supported()) |
aa888a749
|
1555 |
return h->max_huge_pages; |
7893d1d50
|
1556 1557 1558 1559 |
/* * Increase the pool size * First take pages out of surplus state. Then make up the * remaining difference by allocating fresh huge pages. |
d1c3fb1f8
|
1560 1561 1562 1563 1564 1565 |
* * 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
|
1566 |
*/ |
1da177e4c
|
1567 |
spin_lock(&hugetlb_lock); |
a55164389
|
1568 |
while (h->surplus_huge_pages && count > persistent_huge_pages(h)) { |
6ae11b278
|
1569 |
if (!adjust_pool_surplus(h, nodes_allowed, -1)) |
7893d1d50
|
1570 1571 |
break; } |
a55164389
|
1572 |
while (count > persistent_huge_pages(h)) { |
7893d1d50
|
1573 1574 1575 1576 1577 1578 |
/* * 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); |
944d9fec8
|
1579 1580 1581 1582 |
if (hstate_is_gigantic(h)) ret = alloc_fresh_gigantic_page(h, nodes_allowed); else ret = alloc_fresh_huge_page(h, nodes_allowed); |
7893d1d50
|
1583 1584 1585 |
spin_lock(&hugetlb_lock); if (!ret) goto out; |
536240f2b
|
1586 1587 1588 |
/* Bail for signals. Probably ctrl-c from user */ if (signal_pending(current)) goto out; |
7893d1d50
|
1589 |
} |
7893d1d50
|
1590 1591 1592 1593 1594 1595 1596 |
/* * 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
|
1597 1598 1599 1600 1601 1602 1603 1604 |
* * 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
|
1605 |
*/ |
a55164389
|
1606 |
min_count = h->resv_huge_pages + h->nr_huge_pages - h->free_huge_pages; |
6b0c880df
|
1607 |
min_count = max(count, min_count); |
6ae11b278
|
1608 |
try_to_free_low(h, min_count, nodes_allowed); |
a55164389
|
1609 |
while (min_count < persistent_huge_pages(h)) { |
6ae11b278
|
1610 |
if (!free_pool_huge_page(h, nodes_allowed, 0)) |
1da177e4c
|
1611 |
break; |
55f67141a
|
1612 |
cond_resched_lock(&hugetlb_lock); |
1da177e4c
|
1613 |
} |
a55164389
|
1614 |
while (count < persistent_huge_pages(h)) { |
6ae11b278
|
1615 |
if (!adjust_pool_surplus(h, nodes_allowed, 1)) |
7893d1d50
|
1616 1617 1618 |
break; } out: |
a55164389
|
1619 |
ret = persistent_huge_pages(h); |
1da177e4c
|
1620 |
spin_unlock(&hugetlb_lock); |
7893d1d50
|
1621 |
return ret; |
1da177e4c
|
1622 |
} |
a34378701
|
1623 1624 1625 1626 1627 1628 1629 1630 1631 |
#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
|
1632 1633 1634 |
static struct hstate *kobj_to_node_hstate(struct kobject *kobj, int *nidp); static struct hstate *kobj_to_hstate(struct kobject *kobj, int *nidp) |
a34378701
|
1635 1636 |
{ int i; |
9a3052306
|
1637 |
|
a34378701
|
1638 |
for (i = 0; i < HUGE_MAX_HSTATE; i++) |
9a3052306
|
1639 1640 1641 |
if (hstate_kobjs[i] == kobj) { if (nidp) *nidp = NUMA_NO_NODE; |
a34378701
|
1642 |
return &hstates[i]; |
9a3052306
|
1643 1644 1645 |
} return kobj_to_node_hstate(kobj, nidp); |
a34378701
|
1646 |
} |
06808b082
|
1647 |
static ssize_t nr_hugepages_show_common(struct kobject *kobj, |
a34378701
|
1648 1649 |
struct kobj_attribute *attr, char *buf) { |
9a3052306
|
1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 |
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
|
1662 |
} |
adbe8726d
|
1663 |
|
238d3c13f
|
1664 1665 1666 |
static ssize_t __nr_hugepages_store_common(bool obey_mempolicy, struct hstate *h, int nid, unsigned long count, size_t len) |
a34378701
|
1667 1668 |
{ int err; |
bad44b5be
|
1669 |
NODEMASK_ALLOC(nodemask_t, nodes_allowed, GFP_KERNEL | __GFP_NORETRY); |
a34378701
|
1670 |
|
944d9fec8
|
1671 |
if (hstate_is_gigantic(h) && !gigantic_page_supported()) { |
adbe8726d
|
1672 1673 1674 |
err = -EINVAL; goto out; } |
9a3052306
|
1675 1676 1677 1678 1679 1680 1681 |
if (nid == NUMA_NO_NODE) { /* * global hstate attribute */ if (!(obey_mempolicy && init_nodemask_of_mempolicy(nodes_allowed))) { NODEMASK_FREE(nodes_allowed); |
8cebfcd07
|
1682 |
nodes_allowed = &node_states[N_MEMORY]; |
9a3052306
|
1683 1684 1685 1686 1687 1688 1689 1690 1691 |
} } 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 |
8cebfcd07
|
1692 |
nodes_allowed = &node_states[N_MEMORY]; |
9a3052306
|
1693 |
|
06808b082
|
1694 |
h->max_huge_pages = set_max_huge_pages(h, count, nodes_allowed); |
a34378701
|
1695 |
|
8cebfcd07
|
1696 |
if (nodes_allowed != &node_states[N_MEMORY]) |
06808b082
|
1697 1698 1699 |
NODEMASK_FREE(nodes_allowed); return len; |
adbe8726d
|
1700 1701 1702 |
out: NODEMASK_FREE(nodes_allowed); return err; |
06808b082
|
1703 |
} |
238d3c13f
|
1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 |
static ssize_t nr_hugepages_store_common(bool obey_mempolicy, struct kobject *kobj, const char *buf, size_t len) { struct hstate *h; unsigned long count; int nid; int err; err = kstrtoul(buf, 10, &count); if (err) return err; h = kobj_to_hstate(kobj, &nid); return __nr_hugepages_store_common(obey_mempolicy, h, nid, count, len); } |
06808b082
|
1720 1721 1722 1723 1724 1725 1726 1727 1728 |
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) { |
238d3c13f
|
1729 |
return nr_hugepages_store_common(false, kobj, buf, len); |
a34378701
|
1730 1731 |
} HSTATE_ATTR(nr_hugepages); |
06808b082
|
1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 |
#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) { |
238d3c13f
|
1747 |
return nr_hugepages_store_common(true, kobj, buf, len); |
06808b082
|
1748 1749 1750 |
} HSTATE_ATTR(nr_hugepages_mempolicy); #endif |
a34378701
|
1751 1752 1753 |
static ssize_t nr_overcommit_hugepages_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { |
9a3052306
|
1754 |
struct hstate *h = kobj_to_hstate(kobj, NULL); |
a34378701
|
1755 1756 1757 |
return sprintf(buf, "%lu ", h->nr_overcommit_huge_pages); } |
adbe8726d
|
1758 |
|
a34378701
|
1759 1760 1761 1762 1763 |
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
|
1764 |
struct hstate *h = kobj_to_hstate(kobj, NULL); |
a34378701
|
1765 |
|
bae7f4ae1
|
1766 |
if (hstate_is_gigantic(h)) |
adbe8726d
|
1767 |
return -EINVAL; |
3dbb95f78
|
1768 |
err = kstrtoul(buf, 10, &input); |
a34378701
|
1769 |
if (err) |
73ae31e59
|
1770 |
return err; |
a34378701
|
1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 |
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
|
1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 |
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
|
1795 1796 1797 1798 1799 1800 |
} HSTATE_ATTR_RO(free_hugepages); static ssize_t resv_hugepages_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { |
9a3052306
|
1801 |
struct hstate *h = kobj_to_hstate(kobj, NULL); |
a34378701
|
1802 1803 1804 1805 1806 1807 1808 1809 |
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
|
1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 |
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
|
1822 1823 1824 1825 1826 1827 1828 1829 1830 |
} 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
|
1831 1832 1833 |
#ifdef CONFIG_NUMA &nr_hugepages_mempolicy_attr.attr, #endif |
a34378701
|
1834 1835 1836 1837 1838 1839 |
NULL, }; static struct attribute_group hstate_attr_group = { .attrs = hstate_attrs, }; |
094e9539b
|
1840 1841 1842 |
static int hugetlb_sysfs_add_hstate(struct hstate *h, struct kobject *parent, struct kobject **hstate_kobjs, struct attribute_group *hstate_attr_group) |
a34378701
|
1843 1844 |
{ int retval; |
972dc4de1
|
1845 |
int hi = hstate_index(h); |
a34378701
|
1846 |
|
9a3052306
|
1847 1848 |
hstate_kobjs[hi] = kobject_create_and_add(h->name, parent); if (!hstate_kobjs[hi]) |
a34378701
|
1849 |
return -ENOMEM; |
9a3052306
|
1850 |
retval = sysfs_create_group(hstate_kobjs[hi], hstate_attr_group); |
a34378701
|
1851 |
if (retval) |
9a3052306
|
1852 |
kobject_put(hstate_kobjs[hi]); |
a34378701
|
1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 |
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
|
1867 1868 |
err = hugetlb_sysfs_add_hstate(h, hugepages_kobj, hstate_kobjs, &hstate_attr_group); |
a34378701
|
1869 |
if (err) |
ffb22af5b
|
1870 |
pr_err("Hugetlb: Unable to add hstate %s", h->name); |
a34378701
|
1871 1872 |
} } |
9a3052306
|
1873 1874 1875 1876 |
#ifdef CONFIG_NUMA /* * node_hstate/s - associate per node hstate attributes, via their kobjects, |
10fbcf4c6
|
1877 1878 1879 |
* with node devices in node_devices[] using a parallel array. The array * index of a node device or _hstate == node id. * This is here to avoid any static dependency of the node device driver, in |
9a3052306
|
1880 1881 1882 1883 1884 1885 1886 1887 1888 |
* 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]; /* |
10fbcf4c6
|
1889 |
* A subset of global hstate attributes for node devices |
9a3052306
|
1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 |
*/ 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, }; /* |
10fbcf4c6
|
1903 |
* kobj_to_node_hstate - lookup global hstate for node device hstate attr kobj. |
9a3052306
|
1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 |
* 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; } /* |
10fbcf4c6
|
1926 |
* Unregister hstate attributes from a single node device. |
9a3052306
|
1927 1928 |
* No-op if no hstate attributes attached. */ |
3cd8b44fa
|
1929 |
static void hugetlb_unregister_node(struct node *node) |
9a3052306
|
1930 1931 |
{ struct hstate *h; |
10fbcf4c6
|
1932 |
struct node_hstate *nhs = &node_hstates[node->dev.id]; |
9a3052306
|
1933 1934 |
if (!nhs->hugepages_kobj) |
9b5e5d0fd
|
1935 |
return; /* no hstate attributes */ |
9a3052306
|
1936 |
|
972dc4de1
|
1937 1938 1939 1940 1941 |
for_each_hstate(h) { int idx = hstate_index(h); if (nhs->hstate_kobjs[idx]) { kobject_put(nhs->hstate_kobjs[idx]); nhs->hstate_kobjs[idx] = NULL; |
9a3052306
|
1942 |
} |
972dc4de1
|
1943 |
} |
9a3052306
|
1944 1945 1946 1947 1948 1949 |
kobject_put(nhs->hugepages_kobj); nhs->hugepages_kobj = NULL; } /* |
10fbcf4c6
|
1950 |
* hugetlb module exit: unregister hstate attributes from node devices |
9a3052306
|
1951 1952 1953 1954 1955 1956 1957 |
* that have them. */ static void hugetlb_unregister_all_nodes(void) { int nid; /* |
10fbcf4c6
|
1958 |
* disable node device registrations. |
9a3052306
|
1959 1960 1961 1962 1963 1964 1965 |
*/ register_hugetlbfs_with_node(NULL, NULL); /* * remove hstate attributes from any nodes that have them. */ for (nid = 0; nid < nr_node_ids; nid++) |
8732794b1
|
1966 |
hugetlb_unregister_node(node_devices[nid]); |
9a3052306
|
1967 1968 1969 |
} /* |
10fbcf4c6
|
1970 |
* Register hstate attributes for a single node device. |
9a3052306
|
1971 1972 |
* No-op if attributes already registered. */ |
3cd8b44fa
|
1973 |
static void hugetlb_register_node(struct node *node) |
9a3052306
|
1974 1975 |
{ struct hstate *h; |
10fbcf4c6
|
1976 |
struct node_hstate *nhs = &node_hstates[node->dev.id]; |
9a3052306
|
1977 1978 1979 1980 1981 1982 |
int err; if (nhs->hugepages_kobj) return; /* already allocated */ nhs->hugepages_kobj = kobject_create_and_add("hugepages", |
10fbcf4c6
|
1983 |
&node->dev.kobj); |
9a3052306
|
1984 1985 1986 1987 1988 1989 1990 1991 |
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) { |
ffb22af5b
|
1992 1993 1994 |
pr_err("Hugetlb: Unable to add hstate %s for node %d ", h->name, node->dev.id); |
9a3052306
|
1995 1996 1997 1998 1999 2000 2001 |
hugetlb_unregister_node(node); break; } } } /* |
9b5e5d0fd
|
2002 |
* hugetlb init time: register hstate attributes for all registered node |
10fbcf4c6
|
2003 2004 |
* devices of nodes that have memory. All on-line nodes should have * registered their associated device by this time. |
9a3052306
|
2005 |
*/ |
7d9ca0004
|
2006 |
static void __init hugetlb_register_all_nodes(void) |
9a3052306
|
2007 2008 |
{ int nid; |
8cebfcd07
|
2009 |
for_each_node_state(nid, N_MEMORY) { |
8732794b1
|
2010 |
struct node *node = node_devices[nid]; |
10fbcf4c6
|
2011 |
if (node->dev.id == nid) |
9a3052306
|
2012 2013 2014 2015 |
hugetlb_register_node(node); } /* |
10fbcf4c6
|
2016 |
* Let the node device driver know we're here so it can |
9a3052306
|
2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 |
* [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
|
2037 2038 2039 |
static void __exit hugetlb_exit(void) { struct hstate *h; |
9a3052306
|
2040 |
hugetlb_unregister_all_nodes(); |
a34378701
|
2041 |
for_each_hstate(h) { |
972dc4de1
|
2042 |
kobject_put(hstate_kobjs[hstate_index(h)]); |
a34378701
|
2043 2044 2045 |
} kobject_put(hugepages_kobj); |
8382d914e
|
2046 |
kfree(htlb_fault_mutex_table); |
a34378701
|
2047 2048 2049 2050 2051 |
} module_exit(hugetlb_exit); static int __init hugetlb_init(void) { |
8382d914e
|
2052 |
int i; |
457c1b27e
|
2053 |
if (!hugepages_supported()) |
0ef89d25d
|
2054 |
return 0; |
a34378701
|
2055 |
|
e11bfbfcb
|
2056 2057 2058 2059 |
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
|
2060 |
} |
972dc4de1
|
2061 |
default_hstate_idx = hstate_index(size_to_hstate(default_hstate_size)); |
e11bfbfcb
|
2062 2063 |
if (default_hstate_max_huge_pages) default_hstate.max_huge_pages = default_hstate_max_huge_pages; |
a34378701
|
2064 2065 |
hugetlb_init_hstates(); |
aa888a749
|
2066 |
gather_bootmem_prealloc(); |
a34378701
|
2067 2068 2069 |
report_hugepages(); hugetlb_sysfs_init(); |
9a3052306
|
2070 |
hugetlb_register_all_nodes(); |
7179e7bf4
|
2071 |
hugetlb_cgroup_file_init(); |
9a3052306
|
2072 |
|
8382d914e
|
2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 |
#ifdef CONFIG_SMP num_fault_mutexes = roundup_pow_of_two(8 * num_possible_cpus()); #else num_fault_mutexes = 1; #endif htlb_fault_mutex_table = kmalloc(sizeof(struct mutex) * num_fault_mutexes, GFP_KERNEL); BUG_ON(!htlb_fault_mutex_table); for (i = 0; i < num_fault_mutexes; i++) mutex_init(&htlb_fault_mutex_table[i]); |
a34378701
|
2084 2085 2086 2087 2088 2089 2090 2091 |
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
|
2092 |
unsigned long i; |
a34378701
|
2093 |
if (size_to_hstate(PAGE_SIZE << order)) { |
ffb22af5b
|
2094 2095 |
pr_warning("hugepagesz= specified twice, ignoring "); |
a34378701
|
2096 2097 |
return; } |
47d38344a
|
2098 |
BUG_ON(hugetlb_max_hstate >= HUGE_MAX_HSTATE); |
a34378701
|
2099 |
BUG_ON(order == 0); |
47d38344a
|
2100 |
h = &hstates[hugetlb_max_hstate++]; |
a34378701
|
2101 2102 |
h->order = order; h->mask = ~((1ULL << (order + PAGE_SHIFT)) - 1); |
8faa8b077
|
2103 2104 2105 2106 |
h->nr_huge_pages = 0; h->free_huge_pages = 0; for (i = 0; i < MAX_NUMNODES; ++i) INIT_LIST_HEAD(&h->hugepage_freelists[i]); |
0edaecfab
|
2107 |
INIT_LIST_HEAD(&h->hugepage_activelist); |
8cebfcd07
|
2108 2109 |
h->next_nid_to_alloc = first_node(node_states[N_MEMORY]); h->next_nid_to_free = first_node(node_states[N_MEMORY]); |
a34378701
|
2110 2111 |
snprintf(h->name, HSTATE_NAME_LEN, "hugepages-%lukB", huge_page_size(h)/1024); |
8faa8b077
|
2112 |
|
a34378701
|
2113 2114 |
parsed_hstate = h; } |
e11bfbfcb
|
2115 |
static int __init hugetlb_nrpages_setup(char *s) |
a34378701
|
2116 2117 |
{ unsigned long *mhp; |
8faa8b077
|
2118 |
static unsigned long *last_mhp; |
a34378701
|
2119 2120 |
/* |
47d38344a
|
2121 |
* !hugetlb_max_hstate means we haven't parsed a hugepagesz= parameter yet, |
a34378701
|
2122 2123 |
* so this hugepages= parameter goes to the "default hstate". */ |
47d38344a
|
2124 |
if (!hugetlb_max_hstate) |
a34378701
|
2125 2126 2127 |
mhp = &default_hstate_max_huge_pages; else mhp = &parsed_hstate->max_huge_pages; |
8faa8b077
|
2128 |
if (mhp == last_mhp) { |
ffb22af5b
|
2129 2130 2131 |
pr_warning("hugepages= specified twice without " "interleaving hugepagesz=, ignoring "); |
8faa8b077
|
2132 2133 |
return 1; } |
a34378701
|
2134 2135 |
if (sscanf(s, "%lu", mhp) <= 0) *mhp = 0; |
8faa8b077
|
2136 2137 2138 2139 2140 |
/* * 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. */ |
47d38344a
|
2141 |
if (hugetlb_max_hstate && parsed_hstate->order >= MAX_ORDER) |
8faa8b077
|
2142 2143 2144 |
hugetlb_hstate_alloc_pages(parsed_hstate); last_mhp = mhp; |
a34378701
|
2145 2146 |
return 1; } |
e11bfbfcb
|
2147 2148 2149 2150 2151 2152 2153 2154 |
__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
|
2155 |
|
8a2134605
|
2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 |
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
|
2168 2169 2170 |
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
|
2171 |
{ |
e5ff21594
|
2172 |
struct hstate *h = &default_hstate; |
238d3c13f
|
2173 |
unsigned long tmp = h->max_huge_pages; |
08d4a2465
|
2174 |
int ret; |
e5ff21594
|
2175 |
|
457c1b27e
|
2176 2177 |
if (!hugepages_supported()) return -ENOTSUPP; |
e5ff21594
|
2178 2179 |
table->data = &tmp; table->maxlen = sizeof(unsigned long); |
08d4a2465
|
2180 2181 2182 |
ret = proc_doulongvec_minmax(table, write, buffer, length, ppos); if (ret) goto out; |
e5ff21594
|
2183 |
|
238d3c13f
|
2184 2185 2186 |
if (write) ret = __nr_hugepages_store_common(obey_mempolicy, h, NUMA_NO_NODE, tmp, *length); |
08d4a2465
|
2187 2188 |
out: return ret; |
1da177e4c
|
2189 |
} |
396faf030
|
2190 |
|
06808b082
|
2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 |
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 */ |
a3d0c6aa1
|
2207 |
int hugetlb_overcommit_handler(struct ctl_table *table, int write, |
8d65af789
|
2208 |
void __user *buffer, |
a3d0c6aa1
|
2209 2210 |
size_t *length, loff_t *ppos) { |
a55164389
|
2211 |
struct hstate *h = &default_hstate; |
e5ff21594
|
2212 |
unsigned long tmp; |
08d4a2465
|
2213 |
int ret; |
e5ff21594
|
2214 |
|
457c1b27e
|
2215 2216 |
if (!hugepages_supported()) return -ENOTSUPP; |
c033a93c0
|
2217 |
tmp = h->nr_overcommit_huge_pages; |
e5ff21594
|
2218 |
|
bae7f4ae1
|
2219 |
if (write && hstate_is_gigantic(h)) |
adbe8726d
|
2220 |
return -EINVAL; |
e5ff21594
|
2221 2222 |
table->data = &tmp; table->maxlen = sizeof(unsigned long); |
08d4a2465
|
2223 2224 2225 |
ret = proc_doulongvec_minmax(table, write, buffer, length, ppos); if (ret) goto out; |
e5ff21594
|
2226 2227 2228 2229 2230 2231 |
if (write) { spin_lock(&hugetlb_lock); h->nr_overcommit_huge_pages = tmp; spin_unlock(&hugetlb_lock); } |
08d4a2465
|
2232 2233 |
out: return ret; |
a3d0c6aa1
|
2234 |
} |
1da177e4c
|
2235 |
#endif /* CONFIG_SYSCTL */ |
e1759c215
|
2236 |
void hugetlb_report_meminfo(struct seq_file *m) |
1da177e4c
|
2237 |
{ |
a55164389
|
2238 |
struct hstate *h = &default_hstate; |
457c1b27e
|
2239 2240 |
if (!hugepages_supported()) return; |
e1759c215
|
2241 |
seq_printf(m, |
4f98a2fee
|
2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 |
"HugePages_Total: %5lu " "HugePages_Free: %5lu " "HugePages_Rsvd: %5lu " "HugePages_Surp: %5lu " "Hugepagesize: %8lu kB ", |
a55164389
|
2252 2253 2254 2255 2256 |
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
|
2257 2258 2259 2260 |
} int hugetlb_report_node_meminfo(int nid, char *buf) { |
a55164389
|
2261 |
struct hstate *h = &default_hstate; |
457c1b27e
|
2262 2263 |
if (!hugepages_supported()) return 0; |
1da177e4c
|
2264 2265 2266 |
return sprintf(buf, "Node %d HugePages_Total: %5u " |
a1de09195
|
2267 2268 2269 2270 |
"Node %d HugePages_Free: %5u " "Node %d HugePages_Surp: %5u ", |
a55164389
|
2271 2272 2273 |
nid, h->nr_huge_pages_node[nid], nid, h->free_huge_pages_node[nid], nid, h->surplus_huge_pages_node[nid]); |
1da177e4c
|
2274 |
} |
949f7ec57
|
2275 2276 2277 2278 |
void hugetlb_show_meminfo(void) { struct hstate *h; int nid; |
457c1b27e
|
2279 2280 |
if (!hugepages_supported()) return; |
949f7ec57
|
2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 |
for_each_node_state(nid, N_MEMORY) for_each_hstate(h) pr_info("Node %d hugepages_total=%u hugepages_free=%u hugepages_surp=%u hugepages_size=%lukB ", nid, h->nr_huge_pages_node[nid], h->free_huge_pages_node[nid], h->surplus_huge_pages_node[nid], 1UL << (huge_page_order(h) + PAGE_SHIFT - 10)); } |
1da177e4c
|
2291 2292 2293 |
/* Return the number pages of memory we physically have, in PAGE_SIZE units. */ unsigned long hugetlb_total_pages(void) { |
d00285884
|
2294 2295 2296 2297 2298 2299 |
struct hstate *h; unsigned long nr_total_pages = 0; for_each_hstate(h) nr_total_pages += h->nr_huge_pages * pages_per_huge_page(h); return nr_total_pages; |
1da177e4c
|
2300 |
} |
1da177e4c
|
2301 |
|
a55164389
|
2302 |
static int hugetlb_acct_memory(struct hstate *h, long delta) |
fc1b8a73d
|
2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 |
{ 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
|
2325 |
if (gather_surplus_pages(h, delta) < 0) |
fc1b8a73d
|
2326 |
goto out; |
a55164389
|
2327 2328 |
if (delta > cpuset_mems_nr(h->free_huge_pages_node)) { return_unused_surplus_pages(h, delta); |
fc1b8a73d
|
2329 2330 2331 2332 2333 2334 |
goto out; } } ret = 0; if (delta < 0) |
a55164389
|
2335 |
return_unused_surplus_pages(h, (unsigned long) -delta); |
fc1b8a73d
|
2336 2337 2338 2339 2340 |
out: spin_unlock(&hugetlb_lock); return ret; } |
84afd99b8
|
2341 2342 |
static void hugetlb_vm_op_open(struct vm_area_struct *vma) { |
f522c3ac0
|
2343 |
struct resv_map *resv = vma_resv_map(vma); |
84afd99b8
|
2344 2345 2346 2347 2348 |
/* * 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
|
2349 |
* has a reference to the reservation map it cannot disappear until |
84afd99b8
|
2350 2351 2352 |
* after this open call completes. It is therefore safe to take a * new reference here without additional locking. */ |
4e35f4838
|
2353 |
if (resv && is_vma_resv_set(vma, HPAGE_RESV_OWNER)) |
f522c3ac0
|
2354 |
kref_get(&resv->refs); |
84afd99b8
|
2355 |
} |
a1e78772d
|
2356 2357 |
static void hugetlb_vm_op_close(struct vm_area_struct *vma) { |
a55164389
|
2358 |
struct hstate *h = hstate_vma(vma); |
f522c3ac0
|
2359 |
struct resv_map *resv = vma_resv_map(vma); |
90481622d
|
2360 |
struct hugepage_subpool *spool = subpool_vma(vma); |
4e35f4838
|
2361 |
unsigned long reserve, start, end; |
84afd99b8
|
2362 |
|
4e35f4838
|
2363 2364 |
if (!resv || !is_vma_resv_set(vma, HPAGE_RESV_OWNER)) return; |
84afd99b8
|
2365 |
|
4e35f4838
|
2366 2367 |
start = vma_hugecache_offset(h, vma, vma->vm_start); end = vma_hugecache_offset(h, vma, vma->vm_end); |
84afd99b8
|
2368 |
|
4e35f4838
|
2369 |
reserve = (end - start) - region_count(resv, start, end); |
84afd99b8
|
2370 |
|
4e35f4838
|
2371 2372 2373 2374 2375 |
kref_put(&resv->refs, resv_map_release); if (reserve) { hugetlb_acct_memory(h, -reserve); hugepage_subpool_put_pages(spool, reserve); |
84afd99b8
|
2376 |
} |
a1e78772d
|
2377 |
} |
1da177e4c
|
2378 2379 2380 2381 2382 2383 |
/* * 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
|
2384 |
static int hugetlb_vm_op_fault(struct vm_area_struct *vma, struct vm_fault *vmf) |
1da177e4c
|
2385 2386 |
{ BUG(); |
d0217ac04
|
2387 |
return 0; |
1da177e4c
|
2388 |
} |
f0f37e2f7
|
2389 |
const struct vm_operations_struct hugetlb_vm_ops = { |
d0217ac04
|
2390 |
.fault = hugetlb_vm_op_fault, |
84afd99b8
|
2391 |
.open = hugetlb_vm_op_open, |
a1e78772d
|
2392 |
.close = hugetlb_vm_op_close, |
1da177e4c
|
2393 |
}; |
1e8f889b1
|
2394 2395 |
static pte_t make_huge_pte(struct vm_area_struct *vma, struct page *page, int writable) |
63551ae0f
|
2396 2397 |
{ pte_t entry; |
1e8f889b1
|
2398 |
if (writable) { |
106c992a5
|
2399 2400 |
entry = huge_pte_mkwrite(huge_pte_mkdirty(mk_huge_pte(page, vma->vm_page_prot))); |
63551ae0f
|
2401 |
} else { |
106c992a5
|
2402 2403 |
entry = huge_pte_wrprotect(mk_huge_pte(page, vma->vm_page_prot)); |
63551ae0f
|
2404 2405 2406 |
} entry = pte_mkyoung(entry); entry = pte_mkhuge(entry); |
d9ed9faac
|
2407 |
entry = arch_make_huge_pte(entry, vma, page, writable); |
63551ae0f
|
2408 2409 2410 |
return entry; } |
1e8f889b1
|
2411 2412 2413 2414 |
static void set_huge_ptep_writable(struct vm_area_struct *vma, unsigned long address, pte_t *ptep) { pte_t entry; |
106c992a5
|
2415 |
entry = huge_pte_mkwrite(huge_pte_mkdirty(huge_ptep_get(ptep))); |
32f84528f
|
2416 |
if (huge_ptep_set_access_flags(vma, address, ptep, entry, 1)) |
4b3073e1c
|
2417 |
update_mmu_cache(vma, address, ptep); |
1e8f889b1
|
2418 |
} |
4a705fef9
|
2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 |
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; } 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; } |
1e8f889b1
|
2444 |
|
63551ae0f
|
2445 2446 2447 2448 2449 |
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
|
2450 |
unsigned long addr; |
1e8f889b1
|
2451 |
int cow; |
a55164389
|
2452 2453 |
struct hstate *h = hstate_vma(vma); unsigned long sz = huge_page_size(h); |
e8569dd29
|
2454 2455 2456 |
unsigned long mmun_start; /* For mmu_notifiers */ unsigned long mmun_end; /* For mmu_notifiers */ int ret = 0; |
1e8f889b1
|
2457 2458 |
cow = (vma->vm_flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE; |
63551ae0f
|
2459 |
|
e8569dd29
|
2460 2461 2462 2463 |
mmun_start = vma->vm_start; mmun_end = vma->vm_end; if (cow) mmu_notifier_invalidate_range_start(src, mmun_start, mmun_end); |
a55164389
|
2464 |
for (addr = vma->vm_start; addr < vma->vm_end; addr += sz) { |
cb900f412
|
2465 |
spinlock_t *src_ptl, *dst_ptl; |
c74df32c7
|
2466 2467 2468 |
src_pte = huge_pte_offset(src, addr); if (!src_pte) continue; |
a55164389
|
2469 |
dst_pte = huge_pte_alloc(dst, addr, sz); |
e8569dd29
|
2470 2471 2472 2473 |
if (!dst_pte) { ret = -ENOMEM; break; } |
c5c99429f
|
2474 2475 2476 2477 |
/* If the pagetables are shared don't copy or take references */ if (dst_pte == src_pte) continue; |
cb900f412
|
2478 2479 2480 |
dst_ptl = huge_pte_lock(h, dst, dst_pte); src_ptl = huge_pte_lockptr(h, src, src_pte); spin_lock_nested(src_ptl, SINGLE_DEPTH_NESTING); |
4a705fef9
|
2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 |
entry = huge_ptep_get(src_pte); if (huge_pte_none(entry)) { /* skip none entry */ ; } else if (unlikely(is_hugetlb_entry_migration(entry) || is_hugetlb_entry_hwpoisoned(entry))) { swp_entry_t swp_entry = pte_to_swp_entry(entry); if (is_write_migration_entry(swp_entry) && cow) { /* * COW mappings require pages in both * parent and child to be set to read. */ make_migration_entry_read(&swp_entry); entry = swp_entry_to_pte(swp_entry); set_huge_pte_at(src, addr, src_pte, entry); } set_huge_pte_at(dst, addr, dst_pte, entry); } else { |
34ee645e8
|
2499 |
if (cow) { |
7f2e9525b
|
2500 |
huge_ptep_set_wrprotect(src, addr, src_pte); |
34ee645e8
|
2501 2502 2503 |
mmu_notifier_invalidate_range(src, mmun_start, mmun_end); } |
0253d634e
|
2504 |
entry = huge_ptep_get(src_pte); |
1c59827d1
|
2505 2506 |
ptepage = pte_page(entry); get_page(ptepage); |
0fe6e20b9
|
2507 |
page_dup_rmap(ptepage); |
1c59827d1
|
2508 2509 |
set_huge_pte_at(dst, addr, dst_pte, entry); } |
cb900f412
|
2510 2511 |
spin_unlock(src_ptl); spin_unlock(dst_ptl); |
63551ae0f
|
2512 |
} |
63551ae0f
|
2513 |
|
e8569dd29
|
2514 2515 2516 2517 |
if (cow) mmu_notifier_invalidate_range_end(src, mmun_start, mmun_end); return ret; |
63551ae0f
|
2518 |
} |
24669e584
|
2519 2520 2521 |
void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma, unsigned long start, unsigned long end, struct page *ref_page) |
63551ae0f
|
2522 |
{ |
24669e584
|
2523 |
int force_flush = 0; |
63551ae0f
|
2524 2525 |
struct mm_struct *mm = vma->vm_mm; unsigned long address; |
c7546f8f0
|
2526 |
pte_t *ptep; |
63551ae0f
|
2527 |
pte_t pte; |
cb900f412
|
2528 |
spinlock_t *ptl; |
63551ae0f
|
2529 |
struct page *page; |
a55164389
|
2530 2531 |
struct hstate *h = hstate_vma(vma); unsigned long sz = huge_page_size(h); |
2ec74c3ef
|
2532 2533 |
const unsigned long mmun_start = start; /* For mmu_notifiers */ const unsigned long mmun_end = end; /* For mmu_notifiers */ |
a55164389
|
2534 |
|
63551ae0f
|
2535 |
WARN_ON(!is_vm_hugetlb_page(vma)); |
a55164389
|
2536 2537 |
BUG_ON(start & ~huge_page_mask(h)); BUG_ON(end & ~huge_page_mask(h)); |
63551ae0f
|
2538 |
|
24669e584
|
2539 |
tlb_start_vma(tlb, vma); |
2ec74c3ef
|
2540 |
mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); |
569f48b85
|
2541 |
address = start; |
24669e584
|
2542 |
again: |
569f48b85
|
2543 |
for (; address < end; address += sz) { |
c7546f8f0
|
2544 |
ptep = huge_pte_offset(mm, address); |
4c8872659
|
2545 |
if (!ptep) |
c7546f8f0
|
2546 |
continue; |
cb900f412
|
2547 |
ptl = huge_pte_lock(h, mm, ptep); |
39dde65c9
|
2548 |
if (huge_pmd_unshare(mm, &address, ptep)) |
cb900f412
|
2549 |
goto unlock; |
39dde65c9
|
2550 |
|
6629326b8
|
2551 2552 |
pte = huge_ptep_get(ptep); if (huge_pte_none(pte)) |
cb900f412
|
2553 |
goto unlock; |
6629326b8
|
2554 2555 2556 2557 |
/* * HWPoisoned hugepage is already unmapped and dropped reference */ |
8c4894c6b
|
2558 |
if (unlikely(is_hugetlb_entry_hwpoisoned(pte))) { |
106c992a5
|
2559 |
huge_pte_clear(mm, address, ptep); |
cb900f412
|
2560 |
goto unlock; |
8c4894c6b
|
2561 |
} |
6629326b8
|
2562 2563 |
page = pte_page(pte); |
04f2cbe35
|
2564 2565 2566 2567 2568 2569 |
/* * 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) { |
04f2cbe35
|
2570 |
if (page != ref_page) |
cb900f412
|
2571 |
goto unlock; |
04f2cbe35
|
2572 2573 2574 2575 2576 2577 2578 2579 |
/* * 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
|
2580 |
pte = huge_ptep_get_and_clear(mm, address, ptep); |
24669e584
|
2581 |
tlb_remove_tlb_entry(tlb, ptep, address); |
106c992a5
|
2582 |
if (huge_pte_dirty(pte)) |
6649a3863
|
2583 |
set_page_dirty(page); |
9e81130b7
|
2584 |
|
24669e584
|
2585 2586 |
page_remove_rmap(page); force_flush = !__tlb_remove_page(tlb, page); |
cb900f412
|
2587 |
if (force_flush) { |
569f48b85
|
2588 |
address += sz; |
cb900f412
|
2589 |
spin_unlock(ptl); |
24669e584
|
2590 |
break; |
cb900f412
|
2591 |
} |
9e81130b7
|
2592 |
/* Bail out after unmapping reference page if supplied */ |
cb900f412
|
2593 2594 |
if (ref_page) { spin_unlock(ptl); |
9e81130b7
|
2595 |
break; |
cb900f412
|
2596 2597 2598 |
} unlock: spin_unlock(ptl); |
63551ae0f
|
2599 |
} |
24669e584
|
2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 |
/* * mmu_gather ran out of room to batch pages, we break out of * the PTE lock to avoid doing the potential expensive TLB invalidate * and page-free while holding it. */ if (force_flush) { force_flush = 0; tlb_flush_mmu(tlb); if (address < end && !ref_page) goto again; |
fe1668ae5
|
2610 |
} |
2ec74c3ef
|
2611 |
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
24669e584
|
2612 |
tlb_end_vma(tlb, vma); |
1da177e4c
|
2613 |
} |
63551ae0f
|
2614 |
|
d833352a4
|
2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 |
void __unmap_hugepage_range_final(struct mmu_gather *tlb, struct vm_area_struct *vma, unsigned long start, unsigned long end, struct page *ref_page) { __unmap_hugepage_range(tlb, vma, start, end, ref_page); /* * Clear this flag so that x86's huge_pmd_share page_table_shareable * test will fail on a vma being torn down, and not grab a page table * on its way out. We're lucky that the flag has such an appropriate * name, and can in fact be safely cleared here. We could clear it * before the __unmap_hugepage_range above, but all that's necessary |
c8c06efa8
|
2627 |
* is to clear it before releasing the i_mmap_rwsem. This works |
d833352a4
|
2628 |
* because in the context this is called, the VMA is about to be |
c8c06efa8
|
2629 |
* destroyed and the i_mmap_rwsem is held. |
d833352a4
|
2630 2631 2632 |
*/ vma->vm_flags &= ~VM_MAYSHARE; } |
502717f4e
|
2633 |
void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start, |
04f2cbe35
|
2634 |
unsigned long end, struct page *ref_page) |
502717f4e
|
2635 |
{ |
24669e584
|
2636 2637 2638 2639 |
struct mm_struct *mm; struct mmu_gather tlb; mm = vma->vm_mm; |
2b047252d
|
2640 |
tlb_gather_mmu(&tlb, mm, start, end); |
24669e584
|
2641 2642 |
__unmap_hugepage_range(&tlb, vma, start, end, ref_page); tlb_finish_mmu(&tlb, start, end); |
502717f4e
|
2643 |
} |
04f2cbe35
|
2644 2645 2646 2647 2648 2649 |
/* * 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. */ |
2f4612af4
|
2650 2651 |
static void unmap_ref_private(struct mm_struct *mm, struct vm_area_struct *vma, struct page *page, unsigned long address) |
04f2cbe35
|
2652 |
{ |
7526674de
|
2653 |
struct hstate *h = hstate_vma(vma); |
04f2cbe35
|
2654 2655 |
struct vm_area_struct *iter_vma; struct address_space *mapping; |
04f2cbe35
|
2656 2657 2658 2659 2660 2661 |
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
|
2662 |
address = address & huge_page_mask(h); |
36e4f20af
|
2663 2664 |
pgoff = ((address - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff; |
496ad9aa8
|
2665 |
mapping = file_inode(vma->vm_file)->i_mapping; |
04f2cbe35
|
2666 |
|
4eb2b1dcd
|
2667 2668 2669 2670 2671 |
/* * 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 */ |
83cde9e8b
|
2672 |
i_mmap_lock_write(mapping); |
6b2dbba8b
|
2673 |
vma_interval_tree_foreach(iter_vma, &mapping->i_mmap, pgoff, pgoff) { |
04f2cbe35
|
2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 |
/* 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)) |
24669e584
|
2686 2687 |
unmap_hugepage_range(iter_vma, address, address + huge_page_size(h), page); |
04f2cbe35
|
2688 |
} |
83cde9e8b
|
2689 |
i_mmap_unlock_write(mapping); |
04f2cbe35
|
2690 |
} |
0fe6e20b9
|
2691 2692 |
/* * Hugetlb_cow() should be called with page lock of the original hugepage held. |
ef009b25f
|
2693 2694 2695 |
* Called with hugetlb_instantiation_mutex held and pte_page locked so we * cannot race with other handlers or page migration. * Keep the pte_same checks anyway to make transition from the mutex easier. |
0fe6e20b9
|
2696 |
*/ |
1e8f889b1
|
2697 |
static int hugetlb_cow(struct mm_struct *mm, struct vm_area_struct *vma, |
04f2cbe35
|
2698 |
unsigned long address, pte_t *ptep, pte_t pte, |
cb900f412
|
2699 |
struct page *pagecache_page, spinlock_t *ptl) |
1e8f889b1
|
2700 |
{ |
a55164389
|
2701 |
struct hstate *h = hstate_vma(vma); |
1e8f889b1
|
2702 |
struct page *old_page, *new_page; |
ad4404a22
|
2703 |
int ret = 0, outside_reserve = 0; |
2ec74c3ef
|
2704 2705 |
unsigned long mmun_start; /* For mmu_notifiers */ unsigned long mmun_end; /* For mmu_notifiers */ |
1e8f889b1
|
2706 2707 |
old_page = pte_page(pte); |
04f2cbe35
|
2708 |
retry_avoidcopy: |
1e8f889b1
|
2709 2710 |
/* If no-one else is actually using this page, avoid the copy * and just make the page writable */ |
37a2140dc
|
2711 2712 |
if (page_mapcount(old_page) == 1 && PageAnon(old_page)) { page_move_anon_rmap(old_page, vma, address); |
1e8f889b1
|
2713 |
set_huge_ptep_writable(vma, address, ptep); |
83c54070e
|
2714 |
return 0; |
1e8f889b1
|
2715 |
} |
04f2cbe35
|
2716 2717 2718 2719 2720 2721 2722 2723 2724 |
/* * 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. */ |
5944d0116
|
2725 |
if (is_vma_resv_set(vma, HPAGE_RESV_OWNER) && |
04f2cbe35
|
2726 2727 |
old_page != pagecache_page) outside_reserve = 1; |
1e8f889b1
|
2728 |
page_cache_get(old_page); |
b76c8cfbf
|
2729 |
|
ad4404a22
|
2730 2731 2732 2733 |
/* * Drop page table lock as buddy allocator may be called. It will * be acquired again before returning to the caller, as expected. */ |
cb900f412
|
2734 |
spin_unlock(ptl); |
04f2cbe35
|
2735 |
new_page = alloc_huge_page(vma, address, outside_reserve); |
1e8f889b1
|
2736 |
|
2fc39cec6
|
2737 |
if (IS_ERR(new_page)) { |
04f2cbe35
|
2738 2739 2740 2741 2742 2743 2744 2745 |
/* * 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) { |
ad4404a22
|
2746 |
page_cache_release(old_page); |
04f2cbe35
|
2747 |
BUG_ON(huge_pte_none(pte)); |
2f4612af4
|
2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 |
unmap_ref_private(mm, vma, old_page, address); BUG_ON(huge_pte_none(pte)); spin_lock(ptl); ptep = huge_pte_offset(mm, address & huge_page_mask(h)); if (likely(ptep && pte_same(huge_ptep_get(ptep), pte))) goto retry_avoidcopy; /* * race occurs while re-acquiring page table * lock, and our job is done. */ return 0; |
04f2cbe35
|
2760 |
} |
ad4404a22
|
2761 2762 2763 |
ret = (PTR_ERR(new_page) == -ENOMEM) ? VM_FAULT_OOM : VM_FAULT_SIGBUS; goto out_release_old; |
1e8f889b1
|
2764 |
} |
0fe6e20b9
|
2765 2766 2767 2768 |
/* * When the original hugepage is shared one, it does not have * anon_vma prepared. */ |
44e2aa937
|
2769 |
if (unlikely(anon_vma_prepare(vma))) { |
ad4404a22
|
2770 2771 |
ret = VM_FAULT_OOM; goto out_release_all; |
44e2aa937
|
2772 |
} |
0fe6e20b9
|
2773 |
|
47ad8475c
|
2774 2775 |
copy_user_huge_page(new_page, old_page, address, vma, pages_per_huge_page(h)); |
0ed361dec
|
2776 |
__SetPageUptodate(new_page); |
1e8f889b1
|
2777 |
|
2ec74c3ef
|
2778 2779 2780 |
mmun_start = address & huge_page_mask(h); mmun_end = mmun_start + huge_page_size(h); mmu_notifier_invalidate_range_start(mm, mmun_start, mmun_end); |
ad4404a22
|
2781 |
|
b76c8cfbf
|
2782 |
/* |
cb900f412
|
2783 |
* Retake the page table lock to check for racing updates |
b76c8cfbf
|
2784 2785 |
* before the page tables are altered */ |
cb900f412
|
2786 |
spin_lock(ptl); |
a55164389
|
2787 |
ptep = huge_pte_offset(mm, address & huge_page_mask(h)); |
a9af0c5df
|
2788 |
if (likely(ptep && pte_same(huge_ptep_get(ptep), pte))) { |
07443a85a
|
2789 |
ClearPagePrivate(new_page); |
1e8f889b1
|
2790 |
/* Break COW */ |
8fe627ec5
|
2791 |
huge_ptep_clear_flush(vma, address, ptep); |
34ee645e8
|
2792 |
mmu_notifier_invalidate_range(mm, mmun_start, mmun_end); |
1e8f889b1
|
2793 2794 |
set_huge_pte_at(mm, address, ptep, make_huge_pte(vma, new_page, 1)); |
0fe6e20b9
|
2795 |
page_remove_rmap(old_page); |
cd67f0d2a
|
2796 |
hugepage_add_new_anon_rmap(new_page, vma, address); |
1e8f889b1
|
2797 2798 2799 |
/* Make the old page be freed below */ new_page = old_page; } |
cb900f412
|
2800 |
spin_unlock(ptl); |
2ec74c3ef
|
2801 |
mmu_notifier_invalidate_range_end(mm, mmun_start, mmun_end); |
ad4404a22
|
2802 |
out_release_all: |
1e8f889b1
|
2803 |
page_cache_release(new_page); |
ad4404a22
|
2804 |
out_release_old: |
1e8f889b1
|
2805 |
page_cache_release(old_page); |
8312034f3
|
2806 |
|
ad4404a22
|
2807 2808 |
spin_lock(ptl); /* Caller expects lock to be held */ return ret; |
1e8f889b1
|
2809 |
} |
04f2cbe35
|
2810 |
/* Return the pagecache page at a given address within a VMA */ |
a55164389
|
2811 2812 |
static struct page *hugetlbfs_pagecache_page(struct hstate *h, struct vm_area_struct *vma, unsigned long address) |
04f2cbe35
|
2813 2814 |
{ struct address_space *mapping; |
e7c4b0bfd
|
2815 |
pgoff_t idx; |
04f2cbe35
|
2816 2817 |
mapping = vma->vm_file->f_mapping; |
a55164389
|
2818 |
idx = vma_hugecache_offset(h, vma, address); |
04f2cbe35
|
2819 2820 2821 |
return find_lock_page(mapping, idx); } |
3ae77f43b
|
2822 2823 2824 2825 2826 |
/* * 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
|
2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 |
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
|
2841 |
static int hugetlb_no_page(struct mm_struct *mm, struct vm_area_struct *vma, |
8382d914e
|
2842 2843 |
struct address_space *mapping, pgoff_t idx, unsigned long address, pte_t *ptep, unsigned int flags) |
ac9b9c667
|
2844 |
{ |
a55164389
|
2845 |
struct hstate *h = hstate_vma(vma); |
ac9b9c667
|
2846 |
int ret = VM_FAULT_SIGBUS; |
409eb8c26
|
2847 |
int anon_rmap = 0; |
4c8872659
|
2848 |
unsigned long size; |
4c8872659
|
2849 |
struct page *page; |
1e8f889b1
|
2850 |
pte_t new_pte; |
cb900f412
|
2851 |
spinlock_t *ptl; |
4c8872659
|
2852 |
|
04f2cbe35
|
2853 2854 2855 |
/* * 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
|
2856 |
* COW. Warn that such a situation has occurred as it may not be obvious |
04f2cbe35
|
2857 2858 |
*/ if (is_vma_resv_set(vma, HPAGE_RESV_UNMAPPED)) { |
ffb22af5b
|
2859 2860 2861 |
pr_warning("PID %d killed due to inadequate hugepage pool ", current->pid); |
04f2cbe35
|
2862 2863 |
return ret; } |
4c8872659
|
2864 2865 2866 2867 |
/* * Use page lock to guard against racing truncation * before we get page_table_lock. */ |
6bda666a0
|
2868 2869 2870 |
retry: page = find_lock_page(mapping, idx); if (!page) { |
a55164389
|
2871 |
size = i_size_read(mapping->host) >> huge_page_shift(h); |
ebed4bfc8
|
2872 2873 |
if (idx >= size) goto out; |
04f2cbe35
|
2874 |
page = alloc_huge_page(vma, address, 0); |
2fc39cec6
|
2875 |
if (IS_ERR(page)) { |
76dcee75c
|
2876 2877 2878 2879 2880 |
ret = PTR_ERR(page); if (ret == -ENOMEM) ret = VM_FAULT_OOM; else ret = VM_FAULT_SIGBUS; |
6bda666a0
|
2881 2882 |
goto out; } |
47ad8475c
|
2883 |
clear_huge_page(page, address, pages_per_huge_page(h)); |
0ed361dec
|
2884 |
__SetPageUptodate(page); |
ac9b9c667
|
2885 |
|
f83a275db
|
2886 |
if (vma->vm_flags & VM_MAYSHARE) { |
6bda666a0
|
2887 |
int err; |
45c682a68
|
2888 |
struct inode *inode = mapping->host; |
6bda666a0
|
2889 2890 2891 2892 |
err = add_to_page_cache(page, mapping, idx, GFP_KERNEL); if (err) { put_page(page); |
6bda666a0
|
2893 2894 2895 2896 |
if (err == -EEXIST) goto retry; goto out; } |
07443a85a
|
2897 |
ClearPagePrivate(page); |
45c682a68
|
2898 2899 |
spin_lock(&inode->i_lock); |
a55164389
|
2900 |
inode->i_blocks += blocks_per_huge_page(h); |
45c682a68
|
2901 |
spin_unlock(&inode->i_lock); |
23be7468e
|
2902 |
} else { |
6bda666a0
|
2903 |
lock_page(page); |
0fe6e20b9
|
2904 2905 2906 2907 |
if (unlikely(anon_vma_prepare(vma))) { ret = VM_FAULT_OOM; goto backout_unlocked; } |
409eb8c26
|
2908 |
anon_rmap = 1; |
23be7468e
|
2909 |
} |
0fe6e20b9
|
2910 |
} else { |
998b4382c
|
2911 2912 2913 2914 2915 2916 |
/* * 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))) { |
32f84528f
|
2917 |
ret = VM_FAULT_HWPOISON | |
972dc4de1
|
2918 |
VM_FAULT_SET_HINDEX(hstate_index(h)); |
998b4382c
|
2919 2920 |
goto backout_unlocked; } |
6bda666a0
|
2921 |
} |
1e8f889b1
|
2922 |
|
57303d801
|
2923 2924 2925 2926 2927 2928 |
/* * 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
|
2929 |
if ((flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED)) |
2b26736c8
|
2930 2931 2932 2933 |
if (vma_needs_reservation(h, vma, address) < 0) { ret = VM_FAULT_OOM; goto backout_unlocked; } |
57303d801
|
2934 |
|
cb900f412
|
2935 2936 |
ptl = huge_pte_lockptr(h, mm, ptep); spin_lock(ptl); |
a55164389
|
2937 |
size = i_size_read(mapping->host) >> huge_page_shift(h); |
4c8872659
|
2938 2939 |
if (idx >= size) goto backout; |
83c54070e
|
2940 |
ret = 0; |
7f2e9525b
|
2941 |
if (!huge_pte_none(huge_ptep_get(ptep))) |
4c8872659
|
2942 |
goto backout; |
07443a85a
|
2943 2944 |
if (anon_rmap) { ClearPagePrivate(page); |
409eb8c26
|
2945 |
hugepage_add_new_anon_rmap(page, vma, address); |
ac7149045
|
2946 |
} else |
409eb8c26
|
2947 |
page_dup_rmap(page); |
1e8f889b1
|
2948 2949 2950 |
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
|
2951 |
if ((flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED)) { |
1e8f889b1
|
2952 |
/* Optimization, do the COW without a second fault */ |
cb900f412
|
2953 |
ret = hugetlb_cow(mm, vma, address, ptep, new_pte, page, ptl); |
1e8f889b1
|
2954 |
} |
cb900f412
|
2955 |
spin_unlock(ptl); |
4c8872659
|
2956 2957 |
unlock_page(page); out: |
ac9b9c667
|
2958 |
return ret; |
4c8872659
|
2959 2960 |
backout: |
cb900f412
|
2961 |
spin_unlock(ptl); |
2b26736c8
|
2962 |
backout_unlocked: |
4c8872659
|
2963 2964 2965 |
unlock_page(page); put_page(page); goto out; |
ac9b9c667
|
2966 |
} |
8382d914e
|
2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 |
#ifdef CONFIG_SMP static u32 fault_mutex_hash(struct hstate *h, struct mm_struct *mm, struct vm_area_struct *vma, struct address_space *mapping, pgoff_t idx, unsigned long address) { unsigned long key[2]; u32 hash; if (vma->vm_flags & VM_SHARED) { key[0] = (unsigned long) mapping; key[1] = idx; } else { key[0] = (unsigned long) mm; key[1] = address >> huge_page_shift(h); } hash = jhash2((u32 *)&key, sizeof(key)/sizeof(u32), 0); return hash & (num_fault_mutexes - 1); } #else /* * For uniprocesor systems we always use a single mutex, so just * return 0 and avoid the hashing overhead. */ static u32 fault_mutex_hash(struct hstate *h, struct mm_struct *mm, struct vm_area_struct *vma, struct address_space *mapping, pgoff_t idx, unsigned long address) { return 0; } #endif |
86e5216f8
|
3001 |
int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma, |
788c7df45
|
3002 |
unsigned long address, unsigned int flags) |
86e5216f8
|
3003 |
{ |
8382d914e
|
3004 |
pte_t *ptep, entry; |
cb900f412
|
3005 |
spinlock_t *ptl; |
1e8f889b1
|
3006 |
int ret; |
8382d914e
|
3007 3008 |
u32 hash; pgoff_t idx; |
0fe6e20b9
|
3009 |
struct page *page = NULL; |
57303d801
|
3010 |
struct page *pagecache_page = NULL; |
a55164389
|
3011 |
struct hstate *h = hstate_vma(vma); |
8382d914e
|
3012 |
struct address_space *mapping; |
86e5216f8
|
3013 |
|
1e16a539a
|
3014 |
address &= huge_page_mask(h); |
fd6a03edd
|
3015 3016 3017 |
ptep = huge_pte_offset(mm, address); if (ptep) { entry = huge_ptep_get(ptep); |
290408d4a
|
3018 |
if (unlikely(is_hugetlb_entry_migration(entry))) { |
cb900f412
|
3019 |
migration_entry_wait_huge(vma, mm, ptep); |
290408d4a
|
3020 3021 |
return 0; } else if (unlikely(is_hugetlb_entry_hwpoisoned(entry))) |
32f84528f
|
3022 |
return VM_FAULT_HWPOISON_LARGE | |
972dc4de1
|
3023 |
VM_FAULT_SET_HINDEX(hstate_index(h)); |
fd6a03edd
|
3024 |
} |
a55164389
|
3025 |
ptep = huge_pte_alloc(mm, address, huge_page_size(h)); |
86e5216f8
|
3026 3027 |
if (!ptep) return VM_FAULT_OOM; |
8382d914e
|
3028 3029 |
mapping = vma->vm_file->f_mapping; idx = vma_hugecache_offset(h, vma, address); |
3935baa9b
|
3030 3031 3032 3033 3034 |
/* * 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. */ |
8382d914e
|
3035 3036 |
hash = fault_mutex_hash(h, mm, vma, mapping, idx, address); mutex_lock(&htlb_fault_mutex_table[hash]); |
7f2e9525b
|
3037 3038 |
entry = huge_ptep_get(ptep); if (huge_pte_none(entry)) { |
8382d914e
|
3039 |
ret = hugetlb_no_page(mm, vma, mapping, idx, address, ptep, flags); |
b4d1d99fd
|
3040 |
goto out_mutex; |
3935baa9b
|
3041 |
} |
86e5216f8
|
3042 |
|
83c54070e
|
3043 |
ret = 0; |
1e8f889b1
|
3044 |
|
57303d801
|
3045 3046 3047 3048 3049 3050 3051 3052 |
/* * 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. */ |
106c992a5
|
3053 |
if ((flags & FAULT_FLAG_WRITE) && !huge_pte_write(entry)) { |
2b26736c8
|
3054 3055 |
if (vma_needs_reservation(h, vma, address) < 0) { ret = VM_FAULT_OOM; |
b4d1d99fd
|
3056 |
goto out_mutex; |
2b26736c8
|
3057 |
} |
57303d801
|
3058 |
|
f83a275db
|
3059 |
if (!(vma->vm_flags & VM_MAYSHARE)) |
57303d801
|
3060 3061 3062 |
pagecache_page = hugetlbfs_pagecache_page(h, vma, address); } |
56c9cfb13
|
3063 3064 3065 3066 3067 3068 3069 3070 |
/* * 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); |
66aebce74
|
3071 |
get_page(page); |
56c9cfb13
|
3072 |
if (page != pagecache_page) |
0fe6e20b9
|
3073 |
lock_page(page); |
0fe6e20b9
|
3074 |
|
cb900f412
|
3075 3076 |
ptl = huge_pte_lockptr(h, mm, ptep); spin_lock(ptl); |
1e8f889b1
|
3077 |
/* Check for a racing update before calling hugetlb_cow */ |
b4d1d99fd
|
3078 |
if (unlikely(!pte_same(entry, huge_ptep_get(ptep)))) |
cb900f412
|
3079 |
goto out_ptl; |
b4d1d99fd
|
3080 |
|
788c7df45
|
3081 |
if (flags & FAULT_FLAG_WRITE) { |
106c992a5
|
3082 |
if (!huge_pte_write(entry)) { |
57303d801
|
3083 |
ret = hugetlb_cow(mm, vma, address, ptep, entry, |
cb900f412
|
3084 3085 |
pagecache_page, ptl); goto out_ptl; |
b4d1d99fd
|
3086 |
} |
106c992a5
|
3087 |
entry = huge_pte_mkdirty(entry); |
b4d1d99fd
|
3088 3089 |
} entry = pte_mkyoung(entry); |
788c7df45
|
3090 3091 |
if (huge_ptep_set_access_flags(vma, address, ptep, entry, flags & FAULT_FLAG_WRITE)) |
4b3073e1c
|
3092 |
update_mmu_cache(vma, address, ptep); |
b4d1d99fd
|
3093 |
|
cb900f412
|
3094 3095 |
out_ptl: spin_unlock(ptl); |
57303d801
|
3096 3097 3098 3099 3100 |
if (pagecache_page) { unlock_page(pagecache_page); put_page(pagecache_page); } |
1f64d69c7
|
3101 3102 |
if (page != pagecache_page) unlock_page(page); |
66aebce74
|
3103 |
put_page(page); |
57303d801
|
3104 |
|
b4d1d99fd
|
3105 |
out_mutex: |
8382d914e
|
3106 |
mutex_unlock(&htlb_fault_mutex_table[hash]); |
1e8f889b1
|
3107 |
return ret; |
86e5216f8
|
3108 |
} |
28a35716d
|
3109 3110 3111 3112 |
long follow_hugetlb_page(struct mm_struct *mm, struct vm_area_struct *vma, struct page **pages, struct vm_area_struct **vmas, unsigned long *position, unsigned long *nr_pages, long i, unsigned int flags) |
63551ae0f
|
3113 |
{ |
d5d4b0aa4
|
3114 3115 |
unsigned long pfn_offset; unsigned long vaddr = *position; |
28a35716d
|
3116 |
unsigned long remainder = *nr_pages; |
a55164389
|
3117 |
struct hstate *h = hstate_vma(vma); |
63551ae0f
|
3118 |
|
63551ae0f
|
3119 |
while (vaddr < vma->vm_end && remainder) { |
4c8872659
|
3120 |
pte_t *pte; |
cb900f412
|
3121 |
spinlock_t *ptl = NULL; |
2a15efc95
|
3122 |
int absent; |
4c8872659
|
3123 |
struct page *page; |
63551ae0f
|
3124 |
|
4c8872659
|
3125 3126 |
/* * Some archs (sparc64, sh*) have multiple pte_ts to |
2a15efc95
|
3127 |
* each hugepage. We have to make sure we get the |
4c8872659
|
3128 |
* first, for the page indexing below to work. |
cb900f412
|
3129 3130 |
* * Note that page table lock is not held when pte is null. |
4c8872659
|
3131 |
*/ |
a55164389
|
3132 |
pte = huge_pte_offset(mm, vaddr & huge_page_mask(h)); |
cb900f412
|
3133 3134 |
if (pte) ptl = huge_pte_lock(h, mm, pte); |
2a15efc95
|
3135 3136 3137 3138 |
absent = !pte || huge_pte_none(huge_ptep_get(pte)); /* * When coredumping, it suits get_dump_page if we just return |
3ae77f43b
|
3139 3140 3141 3142 |
* 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
|
3143 |
*/ |
3ae77f43b
|
3144 3145 |
if (absent && (flags & FOLL_DUMP) && !hugetlbfs_pagecache_present(h, vma, vaddr)) { |
cb900f412
|
3146 3147 |
if (pte) spin_unlock(ptl); |
2a15efc95
|
3148 3149 3150 |
remainder = 0; break; } |
63551ae0f
|
3151 |
|
9cc3a5bd4
|
3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 |
/* * We need call hugetlb_fault for both hugepages under migration * (in which case hugetlb_fault waits for the migration,) and * hwpoisoned hugepages (in which case we need to prevent the * caller from accessing to them.) In order to do this, we use * here is_swap_pte instead of is_hugetlb_entry_migration and * is_hugetlb_entry_hwpoisoned. This is because it simply covers * both cases, and because we can't follow correct pages * directly from any kind of swap entries. */ if (absent || is_swap_pte(huge_ptep_get(pte)) || |
106c992a5
|
3163 3164 |
((flags & FOLL_WRITE) && !huge_pte_write(huge_ptep_get(pte)))) { |
4c8872659
|
3165 |
int ret; |
63551ae0f
|
3166 |
|
cb900f412
|
3167 3168 |
if (pte) spin_unlock(ptl); |
2a15efc95
|
3169 3170 |
ret = hugetlb_fault(mm, vma, vaddr, (flags & FOLL_WRITE) ? FAULT_FLAG_WRITE : 0); |
a89182c76
|
3171 |
if (!(ret & VM_FAULT_ERROR)) |
4c8872659
|
3172 |
continue; |
63551ae0f
|
3173 |
|
4c8872659
|
3174 |
remainder = 0; |
4c8872659
|
3175 3176 |
break; } |
a55164389
|
3177 |
pfn_offset = (vaddr & ~huge_page_mask(h)) >> PAGE_SHIFT; |
7f2e9525b
|
3178 |
page = pte_page(huge_ptep_get(pte)); |
d5d4b0aa4
|
3179 |
same_page: |
d6692183a
|
3180 |
if (pages) { |
2a15efc95
|
3181 |
pages[i] = mem_map_offset(page, pfn_offset); |
a0368d4e4
|
3182 |
get_page_foll(pages[i]); |
d6692183a
|
3183 |
} |
63551ae0f
|
3184 3185 3186 3187 3188 |
if (vmas) vmas[i] = vma; vaddr += PAGE_SIZE; |
d5d4b0aa4
|
3189 |
++pfn_offset; |
63551ae0f
|
3190 3191 |
--remainder; ++i; |
d5d4b0aa4
|
3192 |
if (vaddr < vma->vm_end && remainder && |
a55164389
|
3193 |
pfn_offset < pages_per_huge_page(h)) { |
d5d4b0aa4
|
3194 3195 3196 3197 3198 3199 |
/* * We use pfn_offset to avoid touching the pageframes * of this compound page. */ goto same_page; } |
cb900f412
|
3200 |
spin_unlock(ptl); |
63551ae0f
|
3201 |
} |
28a35716d
|
3202 |
*nr_pages = remainder; |
63551ae0f
|
3203 |
*position = vaddr; |
2a15efc95
|
3204 |
return i ? i : -EFAULT; |
63551ae0f
|
3205 |
} |
8f860591f
|
3206 |
|
7da4d641c
|
3207 |
unsigned long hugetlb_change_protection(struct vm_area_struct *vma, |
8f860591f
|
3208 3209 3210 3211 3212 3213 |
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
|
3214 |
struct hstate *h = hstate_vma(vma); |
7da4d641c
|
3215 |
unsigned long pages = 0; |
8f860591f
|
3216 3217 3218 |
BUG_ON(address >= end); flush_cache_range(vma, address, end); |
a5338093b
|
3219 |
mmu_notifier_invalidate_range_start(mm, start, end); |
83cde9e8b
|
3220 |
i_mmap_lock_write(vma->vm_file->f_mapping); |
a55164389
|
3221 |
for (; address < end; address += huge_page_size(h)) { |
cb900f412
|
3222 |
spinlock_t *ptl; |
8f860591f
|
3223 3224 3225 |
ptep = huge_pte_offset(mm, address); if (!ptep) continue; |
cb900f412
|
3226 |
ptl = huge_pte_lock(h, mm, ptep); |
7da4d641c
|
3227 3228 |
if (huge_pmd_unshare(mm, &address, ptep)) { pages++; |
cb900f412
|
3229 |
spin_unlock(ptl); |
39dde65c9
|
3230 |
continue; |
7da4d641c
|
3231 |
} |
7f2e9525b
|
3232 |
if (!huge_pte_none(huge_ptep_get(ptep))) { |
8f860591f
|
3233 |
pte = huge_ptep_get_and_clear(mm, address, ptep); |
106c992a5
|
3234 |
pte = pte_mkhuge(huge_pte_modify(pte, newprot)); |
be7517d6a
|
3235 |
pte = arch_make_huge_pte(pte, vma, NULL, 0); |
8f860591f
|
3236 |
set_huge_pte_at(mm, address, ptep, pte); |
7da4d641c
|
3237 |
pages++; |
8f860591f
|
3238 |
} |
cb900f412
|
3239 |
spin_unlock(ptl); |
8f860591f
|
3240 |
} |
d833352a4
|
3241 |
/* |
c8c06efa8
|
3242 |
* Must flush TLB before releasing i_mmap_rwsem: x86's huge_pmd_unshare |
d833352a4
|
3243 |
* may have cleared our pud entry and done put_page on the page table: |
c8c06efa8
|
3244 |
* once we release i_mmap_rwsem, another task can do the final put_page |
d833352a4
|
3245 3246 |
* and that page table be reused and filled with junk. */ |
8f860591f
|
3247 |
flush_tlb_range(vma, start, end); |
34ee645e8
|
3248 |
mmu_notifier_invalidate_range(mm, start, end); |
83cde9e8b
|
3249 |
i_mmap_unlock_write(vma->vm_file->f_mapping); |
a5338093b
|
3250 |
mmu_notifier_invalidate_range_end(mm, start, end); |
7da4d641c
|
3251 3252 |
return pages << h->order; |
8f860591f
|
3253 |
} |
a1e78772d
|
3254 3255 |
int hugetlb_reserve_pages(struct inode *inode, long from, long to, |
5a6fe1259
|
3256 |
struct vm_area_struct *vma, |
ca16d140a
|
3257 |
vm_flags_t vm_flags) |
e4e574b76
|
3258 |
{ |
17c9d12e1
|
3259 |
long ret, chg; |
a55164389
|
3260 |
struct hstate *h = hstate_inode(inode); |
90481622d
|
3261 |
struct hugepage_subpool *spool = subpool_inode(inode); |
9119a41e9
|
3262 |
struct resv_map *resv_map; |
e4e574b76
|
3263 |
|
a1e78772d
|
3264 |
/* |
17c9d12e1
|
3265 3266 |
* Only apply hugepage reservation if asked. At fault time, an * attempt will be made for VM_NORESERVE to allocate a page |
90481622d
|
3267 |
* without using reserves |
17c9d12e1
|
3268 |
*/ |
ca16d140a
|
3269 |
if (vm_flags & VM_NORESERVE) |
17c9d12e1
|
3270 3271 3272 |
return 0; /* |
a1e78772d
|
3273 3274 3275 3276 3277 |
* 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 */ |
9119a41e9
|
3278 |
if (!vma || vma->vm_flags & VM_MAYSHARE) { |
4e35f4838
|
3279 |
resv_map = inode_resv_map(inode); |
9119a41e9
|
3280 |
|
1406ec9ba
|
3281 |
chg = region_chg(resv_map, from, to); |
9119a41e9
|
3282 3283 3284 |
} else { resv_map = resv_map_alloc(); |
17c9d12e1
|
3285 3286 |
if (!resv_map) return -ENOMEM; |
a1e78772d
|
3287 |
chg = to - from; |
84afd99b8
|
3288 |
|
17c9d12e1
|
3289 3290 3291 |
set_vma_resv_map(vma, resv_map); set_vma_resv_flags(vma, HPAGE_RESV_OWNER); } |
c50ac0508
|
3292 3293 3294 3295 |
if (chg < 0) { ret = chg; goto out_err; } |
8a6301127
|
3296 |
|
90481622d
|
3297 |
/* There must be enough pages in the subpool for the mapping */ |
c50ac0508
|
3298 3299 3300 3301 |
if (hugepage_subpool_get_pages(spool, chg)) { ret = -ENOSPC; goto out_err; } |
5a6fe1259
|
3302 3303 |
/* |
17c9d12e1
|
3304 |
* Check enough hugepages are available for the reservation. |
90481622d
|
3305 |
* Hand the pages back to the subpool if there are not |
5a6fe1259
|
3306 |
*/ |
a55164389
|
3307 |
ret = hugetlb_acct_memory(h, chg); |
68842c9b9
|
3308 |
if (ret < 0) { |
90481622d
|
3309 |
hugepage_subpool_put_pages(spool, chg); |
c50ac0508
|
3310 |
goto out_err; |
68842c9b9
|
3311 |
} |
17c9d12e1
|
3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 |
/* * 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
|
3324 |
if (!vma || vma->vm_flags & VM_MAYSHARE) |
1406ec9ba
|
3325 |
region_add(resv_map, from, to); |
a43a8c39b
|
3326 |
return 0; |
c50ac0508
|
3327 |
out_err: |
f031dd274
|
3328 3329 |
if (vma && is_vma_resv_set(vma, HPAGE_RESV_OWNER)) kref_put(&resv_map->refs, resv_map_release); |
c50ac0508
|
3330 |
return ret; |
a43a8c39b
|
3331 3332 3333 3334 |
} void hugetlb_unreserve_pages(struct inode *inode, long offset, long freed) { |
a55164389
|
3335 |
struct hstate *h = hstate_inode(inode); |
4e35f4838
|
3336 |
struct resv_map *resv_map = inode_resv_map(inode); |
9119a41e9
|
3337 |
long chg = 0; |
90481622d
|
3338 |
struct hugepage_subpool *spool = subpool_inode(inode); |
45c682a68
|
3339 |
|
9119a41e9
|
3340 |
if (resv_map) |
1406ec9ba
|
3341 |
chg = region_truncate(resv_map, offset); |
45c682a68
|
3342 |
spin_lock(&inode->i_lock); |
e4c6f8bed
|
3343 |
inode->i_blocks -= (blocks_per_huge_page(h) * freed); |
45c682a68
|
3344 |
spin_unlock(&inode->i_lock); |
90481622d
|
3345 |
hugepage_subpool_put_pages(spool, (chg - freed)); |
a55164389
|
3346 |
hugetlb_acct_memory(h, -(chg - freed)); |
a43a8c39b
|
3347 |
} |
93f70f900
|
3348 |
|
3212b535f
|
3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 |
#ifdef CONFIG_ARCH_WANT_HUGE_PMD_SHARE static unsigned long page_table_shareable(struct vm_area_struct *svma, struct vm_area_struct *vma, unsigned long addr, pgoff_t idx) { unsigned long saddr = ((idx - svma->vm_pgoff) << PAGE_SHIFT) + svma->vm_start; unsigned long sbase = saddr & PUD_MASK; unsigned long s_end = sbase + PUD_SIZE; /* Allow segments to share if only one is marked locked */ unsigned long vm_flags = vma->vm_flags & ~VM_LOCKED; unsigned long svm_flags = svma->vm_flags & ~VM_LOCKED; /* * match the virtual addresses, permission and the alignment of the * page table page. */ if (pmd_index(addr) != pmd_index(saddr) || vm_flags != svm_flags || sbase < svma->vm_start || svma->vm_end < s_end) return 0; return saddr; } static int vma_shareable(struct vm_area_struct *vma, unsigned long addr) { unsigned long base = addr & PUD_MASK; unsigned long end = base + PUD_SIZE; /* * check on proper vm_flags and page table alignment */ if (vma->vm_flags & VM_MAYSHARE && vma->vm_start <= base && end <= vma->vm_end) return 1; return 0; } /* * Search for a shareable pmd page for hugetlb. In any case calls pmd_alloc() * and returns the corresponding pte. While this is not necessary for the * !shared pmd case because we can allocate the pmd later as well, it makes the * code much cleaner. pmd allocation is essential for the shared case because |
c8c06efa8
|
3394 |
* pud has to be populated inside the same i_mmap_rwsem section - otherwise |
3212b535f
|
3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 |
* racing tasks could either miss the sharing (see huge_pte_offset) or select a * bad pmd for sharing. */ pte_t *huge_pmd_share(struct mm_struct *mm, unsigned long addr, pud_t *pud) { struct vm_area_struct *vma = find_vma(mm, addr); struct address_space *mapping = vma->vm_file->f_mapping; pgoff_t idx = ((addr - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff; struct vm_area_struct *svma; unsigned long saddr; pte_t *spte = NULL; pte_t *pte; |
cb900f412
|
3408 |
spinlock_t *ptl; |
3212b535f
|
3409 3410 3411 |
if (!vma_shareable(vma, addr)) return (pte_t *)pmd_alloc(mm, pud, addr); |
83cde9e8b
|
3412 |
i_mmap_lock_write(mapping); |
3212b535f
|
3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 |
vma_interval_tree_foreach(svma, &mapping->i_mmap, idx, idx) { if (svma == vma) continue; saddr = page_table_shareable(svma, vma, addr, idx); if (saddr) { spte = huge_pte_offset(svma->vm_mm, saddr); if (spte) { get_page(virt_to_page(spte)); break; } } } if (!spte) goto out; |
cb900f412
|
3429 3430 |
ptl = huge_pte_lockptr(hstate_vma(vma), mm, spte); spin_lock(ptl); |
3212b535f
|
3431 3432 3433 3434 3435 |
if (pud_none(*pud)) pud_populate(mm, pud, (pmd_t *)((unsigned long)spte & PAGE_MASK)); else put_page(virt_to_page(spte)); |
cb900f412
|
3436 |
spin_unlock(ptl); |
3212b535f
|
3437 3438 |
out: pte = (pte_t *)pmd_alloc(mm, pud, addr); |
83cde9e8b
|
3439 |
i_mmap_unlock_write(mapping); |
3212b535f
|
3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 |
return pte; } /* * unmap huge page backed by shared pte. * * Hugetlb pte page is ref counted at the time of mapping. If pte is shared * indicated by page_count > 1, unmap is achieved by clearing pud and * decrementing the ref count. If count == 1, the pte page is not shared. * |
cb900f412
|
3450 |
* called with page table lock held. |
3212b535f
|
3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 |
* * returns: 1 successfully unmapped a shared pte page * 0 the underlying pte page is not shared, or it is the last user */ int huge_pmd_unshare(struct mm_struct *mm, unsigned long *addr, pte_t *ptep) { pgd_t *pgd = pgd_offset(mm, *addr); pud_t *pud = pud_offset(pgd, *addr); BUG_ON(page_count(virt_to_page(ptep)) == 0); if (page_count(virt_to_page(ptep)) == 1) return 0; pud_clear(pud); put_page(virt_to_page(ptep)); *addr = ALIGN(*addr, HPAGE_SIZE * PTRS_PER_PTE) - HPAGE_SIZE; return 1; } |
9e5fc74c3
|
3469 3470 3471 3472 3473 3474 3475 |
#define want_pmd_share() (1) #else /* !CONFIG_ARCH_WANT_HUGE_PMD_SHARE */ pte_t *huge_pmd_share(struct mm_struct *mm, unsigned long addr, pud_t *pud) { return NULL; } #define want_pmd_share() (0) |
3212b535f
|
3476 |
#endif /* CONFIG_ARCH_WANT_HUGE_PMD_SHARE */ |
9e5fc74c3
|
3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 |
#ifdef CONFIG_ARCH_WANT_GENERAL_HUGETLB pte_t *huge_pte_alloc(struct mm_struct *mm, unsigned long addr, unsigned long sz) { pgd_t *pgd; pud_t *pud; pte_t *pte = NULL; pgd = pgd_offset(mm, addr); pud = pud_alloc(mm, pgd, addr); if (pud) { if (sz == PUD_SIZE) { pte = (pte_t *)pud; } else { BUG_ON(sz != PMD_SIZE); if (want_pmd_share() && pud_none(*pud)) pte = huge_pmd_share(mm, addr, pud); else pte = (pte_t *)pmd_alloc(mm, pud, addr); } } BUG_ON(pte && !pte_none(*pte) && !pte_huge(*pte)); return pte; } pte_t *huge_pte_offset(struct mm_struct *mm, unsigned long addr) { pgd_t *pgd; pud_t *pud; pmd_t *pmd = NULL; pgd = pgd_offset(mm, addr); if (pgd_present(*pgd)) { pud = pud_offset(pgd, addr); if (pud_present(*pud)) { if (pud_huge(*pud)) return (pte_t *)pud; pmd = pmd_offset(pud, addr); } } return (pte_t *) pmd; } struct page * follow_huge_pmd(struct mm_struct *mm, unsigned long address, pmd_t *pmd, int write) { struct page *page; page = pte_page(*(pte_t *)pmd); if (page) page += ((address & ~PMD_MASK) >> PAGE_SHIFT); return page; } struct page * follow_huge_pud(struct mm_struct *mm, unsigned long address, pud_t *pud, int write) { struct page *page; page = pte_page(*(pte_t *)pud); if (page) page += ((address & ~PUD_MASK) >> PAGE_SHIFT); return page; } #else /* !CONFIG_ARCH_WANT_GENERAL_HUGETLB */ /* Can be overriden by architectures */ |
3b32123d7
|
3548 |
struct page * __weak |
9e5fc74c3
|
3549 3550 3551 3552 3553 3554 3555 3556 |
follow_huge_pud(struct mm_struct *mm, unsigned long address, pud_t *pud, int write) { BUG(); return NULL; } #endif /* CONFIG_ARCH_WANT_GENERAL_HUGETLB */ |
d5bd91069
|
3557 |
#ifdef CONFIG_MEMORY_FAILURE |
6de2b1aab
|
3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 |
/* 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
|
3571 3572 3573 3574 |
/* * This function is called from memory failure code. * Assume the caller holds page lock of the head page. */ |
6de2b1aab
|
3575 |
int dequeue_hwpoisoned_huge_page(struct page *hpage) |
93f70f900
|
3576 3577 3578 |
{ struct hstate *h = page_hstate(hpage); int nid = page_to_nid(hpage); |
6de2b1aab
|
3579 |
int ret = -EBUSY; |
93f70f900
|
3580 3581 |
spin_lock(&hugetlb_lock); |
6de2b1aab
|
3582 |
if (is_hugepage_on_freelist(hpage)) { |
56f2fb147
|
3583 3584 3585 3586 3587 3588 3589 |
/* * Hwpoisoned hugepage isn't linked to activelist or freelist, * but dangling hpage->lru can trigger list-debug warnings * (this happens when we call unpoison_memory() on it), * so let it point to itself with list_del_init(). */ list_del_init(&hpage->lru); |
8c6c2ecb4
|
3590 |
set_page_refcounted(hpage); |
6de2b1aab
|
3591 3592 3593 3594 |
h->free_huge_pages--; h->free_huge_pages_node[nid]--; ret = 0; } |
93f70f900
|
3595 |
spin_unlock(&hugetlb_lock); |
6de2b1aab
|
3596 |
return ret; |
93f70f900
|
3597 |
} |
6de2b1aab
|
3598 |
#endif |
31caf665e
|
3599 3600 3601 |
bool isolate_huge_page(struct page *page, struct list_head *list) { |
309381fea
|
3602 |
VM_BUG_ON_PAGE(!PageHead(page), page); |
31caf665e
|
3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 |
if (!get_page_unless_zero(page)) return false; spin_lock(&hugetlb_lock); list_move_tail(&page->lru, list); spin_unlock(&hugetlb_lock); return true; } void putback_active_hugepage(struct page *page) { |
309381fea
|
3613 |
VM_BUG_ON_PAGE(!PageHead(page), page); |
31caf665e
|
3614 3615 3616 3617 3618 |
spin_lock(&hugetlb_lock); list_move_tail(&page->lru, &(page_hstate(page))->hugepage_activelist); spin_unlock(&hugetlb_lock); put_page(page); } |
c8721bbbd
|
3619 3620 3621 |
bool is_hugepage_active(struct page *page) { |
309381fea
|
3622 |
VM_BUG_ON_PAGE(!PageHuge(page), page); |
c8721bbbd
|
3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 |
/* * This function can be called for a tail page because the caller, * scan_movable_pages, scans through a given pfn-range which typically * covers one memory block. In systems using gigantic hugepage (1GB * for x86_64,) a hugepage is larger than a memory block, and we don't * support migrating such large hugepages for now, so return false * when called for tail pages. */ if (PageTail(page)) return false; /* * Refcount of a hwpoisoned hugepages is 1, but they are not active, * so we should return false for them. */ if (unlikely(PageHWPoison(page))) return false; return page_count(page) > 0; } |